Skip to main content

Full text of "Philological essays"

See other formats





HARVARD 

COLLEGE 

LIBRARY 




PHILOLOGICAL 

ESSAYS. 



PHILOLOGICAL 

ESSAYS 


BY 


T. HEWITT KEY, M.A., F.R.S., 

FOUMr.IU.y PROPFSI=IOR or UATIIRMATICS in the UNIVEUSITY OF VmOINIA, D.S. J 
THEN OF LATIN IN nNIVERSITH COLLEOK, LONDON; 

NOW OF COMrARATlVR GRAMMAR IN TUB NAME. 

AND READ UAHTFR OF TflF RCTIOOL. 


^LONDON: 

BELL & DALDY, YORK STREET, CO VENT GARDEN. 

1868. 


(./// riflht!: ri‘xrrrdl.\ 


Digitized by Google 



. B 

b ^ i '6 


/ S y Of k. 


!£t I >1(1^ . 

(f 


LONDON : 

K. «*LAY, tMN, AND TAVIjOH, |■BtNr|fD8, 
IIKRAD BTUK^rr IlIf.U 


■■“■ HAKVATwONiitHSTY 
UBRARY 

AUG 2. 0 1^91 


Digitized by Google 



PEEFACE. 


The following Papers are in part now published for 
the first time, in part reprinted from the ‘Transac- 
tions of the Philological Society,’ but with many 
ehanges or additions. In the selection I have passed 
over several papers, because I wish to. reserve them 
for a systematic treatise on language, the greater part 
of which is already ripe for publication. 

I fear that some traces of haste will be visible in 
what I am now putting out; but I have two pleas 
in excuse : first, that the head-master of a school of 
nearly four hundred boys has little leisure for other 
work ; and secondly, that such leisure cannot now be 
expected to be very productive for one who has entered 
his seventieth year. 

I take the present opportunity of enumerating the 
various philological papers which have proceeded from 
my pen during the forty years or so in which I hs.ve 
been connected with University College, first as Pro- 
fessor of Latin (thirteen years) and then as Pro- 
fessor of Comparative Grammar (nearly twenty-seven 
years). ' 


Digitized by Google 



VI 


PREFACE. 


A. Quarterly Journal of Education, published by 
the Society for the Diffusion of Useful Knowledge : — 

Vol. i. p. 89, Review of Zumpt’s Grammar, in 
which I first put forth several of the new ideas which 
appear in my Latin Grammar. 

Vol. ii. p. 143, Review of Sallust’s Catiline and 
Jugurtha, as edited (1) by the Rev. W. Trollope, M.A. ; 
(2) by Professor Charles Anthon, of New York 

Vol. ii. p. 344, School Editions of Terence, where 
(pp. 349-364, &c.) I first give my theory of Terentian 
metres. 

Vol. iii. p. 312, Review of Crombie’s ‘Gymnasium.’ 

Vol. iv. p. 134, Review of Allen’s Treatise ‘ On 
Latin Pai-ticles.’ 

Vol. iv. p. 336, Review of Carey’s ‘ Latin Prosody 
made Easy.’ 

B. Various articles in the Penny Cyclopaedia, 
chiefly bearing on language. Among these papers 
was one on Terentian metres, in agreement with the 
article already mentioned as published in the Journal 
of Education. 'Fhese papers were for the most part 
collected in a little volume entitled : — 

C. ‘ The Alphabet, Terentian Metres, &c.;’ with a new 
paper on ‘ Good, Better, Best, Well,’ &c. 1844. To 
this was prefixed ‘A Prefatory Letter’ (to Mr. Long), 
with ‘Remarks on the Viuronianus of the Rev. J. W. 
Donaldson, 1844,’ where I brought against him several 
distinct charges of plagiarism. The first published 
copy of this book was sent to him August 23d. 

In 1849 there was a ‘second issue’ of this book, 
with a paper ‘ On the Pronouns of the Third Person.’ 

J). The publication of the above-mentioned ‘ Pro- 


Digitized by Google 



PREFACE. vii 

fatory Letter’ led to a controversy with Mr. Donaldson, 
which took the shape of four additional pamphlets : 
first firom him what he was pleased to call, ‘A Keply 
to the Calumnies and Misrepresentations of Professor 
T. H. Key.’ A copy of this I received on the 20th 
of November 1844. I at once wrote, and on the 
30th of the same month sent him a printed copy of 

A Rejoinder to the Reply of the Rev. J. W. Donald- 
son, B.D., in a second ‘ Letter to G. Long, Esq. M.A. 
&c. ;’ to which I attached a paper on 

The Formation of the Latin Perfects amavi, &c. 

On the 13th of December I received his second 
pamphlet, entitled ‘A Brief Examination of Professor 
Key’s Rejoinder;’ and on the 8th of the following 
January I sent him again in print — 

Comments on Mr. Donaldson’s Brief Examination 
of a Rejoinder, &c. 

All the five pamphlets in the year 1845 I had 
reprinted, and distributed in private circulation nearly 
five hundred copies, which produced from him a threat 
that he would apply to the Court of Chancery for an in- 
junction, — a threat however that ended, as I expected, 
in nothing; for although of course I had no legal 
right to print what he had written, yet I felt justified 
in so doing, because he had implied that I looked 
forward to a verdict in my favour in the sole hope 
that my readers might not see his replies. 

E. In 1846 I published — 

A Latin Grammar on the System of Crude Forms ; 
and in 1862 what I may call a third edition of the 
same, ‘ corrected and somewhat enlarged.’ 

F, Proceedings of the Philological Society, vol. ii. 


Digitized by Google 



viii PEEFACE, 

p. 50 : On the English Verb do and the Latin dare ; 
and On the Formation of the English Weak Perfecta 

Ibid. p. 143: On the Relations which exist between 
the Preterite went and the Verb go ; and also between 
va and the Verbs alter and andare. 

Ibid. p. 180 : The Lapp and Finn Tongues not un- 
connected with the Indo-European Family. 

Ibid. p. 249 : On the Origin of certain Latin Words. 

Ibid. vol. iii. p. 45 : On the Misuse of the Terms 
Epenthesis and Euphony. 

Ibid. p. 57 : On the Origin of the Demonstrative 
Pronoun, the Definite Article, the Pronouns of the 
Third Person, the Relative and the Interrogative. 

Ibid. p. 115 : On the Names of the Parts of the 
Human Body, as common to the several Families of 
the Indo-European Language. 

Ibid. p. 130 : On apparent Exceptions from the Tri- 
literal Form of Monosyllabic Roots. 

Ibid. p. 136 : On the Chronology of the Catilinarian 
Orations. 

Ibid. p. 205 : On the Origin of certain Latin Words. 

Ibid. vol. iv. p. 25 : On the Pronouns of the First 
and Second Persons. 

Ibid. p. 87 : An Attempt to prove the Identity of 
the Roots is, was, and be. 

Ibid. vol. V. p. 51 : On the Nature of the Verb, par- 
ticularly on the Formation of the Middle or Passive 
Voice. 

Ibid. p. 89 : On the Derivation and Meaning of 
certain Latin Words. 

Ibid. p. 103 ; On the Etymology of certain Latin 
Words. 


Digitized by Google 



PEEFACE. 


IX 


Ibid. p. 191 : On Vowel-assimilation, especially in 
relation to Professor Willis’s Experiment on Vowel 
Sound& 

Ibid. vol. vi. p. 63 : On the Imperfect Infinitive, 
Imperfect Participle, and those Substantives which 
fall under the definition ‘ Nomen actionis.’ 

Ibid. p. 93 : Miscellaneous Remarks on some Latin 
Words. 

Ibid. p. 117: On some alleged Distinctions in Lan- 
guages believed to be without foundation. 

Ibid. p. 127 : On the Etymology of ottXoos, ZmXoos, 
&c. 

Ibid. p. 138 : On the Etymology of aroa. 

Ibid. p. 139 : Some Remarks on the Speech ‘ Pro 
Plancio.’ 

Ibid. p. 152 : On the Etymology of circumforaneus, 
circulator, cento. 

Ibid. p. 155 : A Translation (from the German) of 
Ahren’s Paper, ‘ On Feminines in and ms ; and on the 
word ywr/.’ 

Ibid. p. 188 : A Translation (from the Italian) 
of Dr. G. Henzen’s Paper, ‘ On the Inscription 
of Sora.’ 

G. Transactions of the Philological Society — vol. 
for 1854, p. 26 : On the Derivation of maritimus, 
aeditimus, Jinitivius, legitimus, miles, and diues. 

Ibid. p. 29 : A Search in some European Languages 
after the Representatives of the Greek Preposition ava 
as prefixed to Verbs. 

An unfavourable review of this paper appeared in 
Kuhn’s ‘ Zeitschrift fur vergleichende Sprachforschung,’ 
vol. iv. pp. 217 — 219; but the editor, with somewhat 


Digitized by Google 


X 


PKEFACE. 


unusual courtesy, gave admission to a reply from me 
(zur Erwiederung) in vol. v. pp. 72 — 80. 

Ihid, p. 72 : On the Meaning of adaequare. 

Ibid. p. 85 : On the Prepositions evi, in, and related 
Words. 

Ibid. p. 131 ; On the Et3rmology of the Verb obso- 
lescere (but see pp. 200, 220). 

Ibid. p. 199 : On the Use of the Reflective form 
deuerti. 

Ibid. p. 206 : On Metathesis. 

Ibid. vol. for 1855, p. 1 : On the Latin Verb mittere, 
its Origin and Affinities ; and generally on Verbs 
signifying ‘to <70’ in the Indo-European Family. 

Ibid. p. 96 : On the Derivation and Meaning of the 
Latin Verb usurjxire. 

Ibid. p. 119 ; On Greek Accentuation. 

Ibid. vol. for 1856, p. 219 : On Diminutives. — I. 
English. 

Ibid. p. 195 : (On Diminutives. — II. Latin; or) On 
the Representatives of the Keltic suffix agh or ach 
‘ little,’ in the Latin Vocabulary. 

Ihid. vol. for 1857, p. 115 : On the Word Inkling. 

Ibid. vol. for 1859, p. 136 : On the Derivations of 
duntaxat, tranquillus, and si in si dis placet. 

Ibid. p. 140 : On the Derivation of the Gothic hanfs 
‘ one-handed.’ 

Ibid. p. 143: On the Derivation of the Word 

Ibid. p. 145 : On the Convertibility of n and d. 

Ibid. p. 273 : A Supplemental Paper on the Keltic 
Suffix agh, &c. tis occurring in Latin, Greek, and other 
Languages. 


Digitized by Google 



PREFACE. 


XI 


7Z»ic?. vol. for 1860-1, p. 172: Miscellaneous Remarks 
suggested by Ritschl’s Plautus, especially on the For- 
mation of the Latin Perfect. 

Ihid. vol. for 1862-3, p. 1 : Miscellaneous Papers; 
(A) On altero- and its Analogues ; (B) On Words which 
denote ‘Waterfowl’ and ‘Swimming;’ (C) On eir of 
eiireiu {inquit) and fV of iiro/iai {scquw) ; (D) On 
alaceri- and some related Greek Words ; (E) On uiu- 
ere, &c. 

Ibid. p. 113 : The Sanskrit Language as the Basis 
of Linguistic Science, and the Labours of the German 
School in that Field — are they not over-valued ? 

Ibid. p. 213 ; On titillare and riKTeip. 

Ibid. p. 216 : The Anglo-Saxon Language called 
in aid to support the Doctrine which attributes a Suffix 
agh OT ag to Latin Verbs. 

Ibid. vol. for 1865: On the so-called ‘A priva- 
tivum.’ 

Ibid, vol for 1866, p. 1 ; Daiighter and Fille, are 
they connected ? 

Ibid. p. 25 : On the Latin Words temere and teme- 
rare. 

Ibid. p. 30 : On the Latin Prepositions re and ; 
their Origin and Primitive Meaning. 

Ibid. p. 49 : The Latin et, que, atque (ac), and the 
Greek «ai, re, all of one Origin. 

Ibid. vol. for 1867, p. 1 : On the Formation of Greek 
Futures and First Aorists. 

H. Bell’s English Journal of Education. TTiutcen 
papers on Latin Etymology, signed ‘ Claudius,’ viz. : 
I. July 1850, p. 254 ; II. August, p. 292; III. Sep- 
touibcr, p. 310 ; IV. October, p. 354 ; V. November, 


Digitized by Google 



xii PREFACE. 

p. 402 ; VI. January 1851, p. 1 ; VII. March, p. 69 ; 
VIII. April, p. 109 ; IX. May, p. 149 ; X. June, 
p. 196; XL July, p. 240; XII. August, p. 281; 
XIII. September, p. 313. 

I. A Review of Smith’s Latin Dictionary in the 
Westminster, July 1855. 

As my arguments touched upon moral questions 
as well as matters of scholarship, I was the more 
unwilling to take shelter under the anonymous, and 
accordingly gave him formal notice that I was the 
writer. 

J. Knight’s English Cyclopaedia, article ‘ Language.’ 


One object in entering into these particulars has 
been to correct some errors which have appeared in 
certain classical publications, and are believed to pre- 
vail somewhat widely. In an edition of the Adelphi 
by the Rev. Wharton B. Marriott (formerly Fellow 
of Exeter College, Oxford, and late Assistant Master 
of Eton), 1863, the second part of the Introduction 
deals with the metres of Latin comedy, and to the 
views of Bentley are opposed those of ‘ more recent 
scholars’ (p. 13) ; and he then proceeds to discuss ‘the 
main ground these “ recent scholars ” take up,’ adding 
a note, — 

‘ See particularly the article on Terentian Metres in 
the Penny Cyclopaedia, evidently by Professor Donald- 
son ; and the chapter of the same author’s Varronianus 
(xiv.), on the Constitution and Pathology of the Latin 
Language.' 

Soon after, in § 5, he speaks of ‘ the three kinds of 


Digitized by Google 



PREFACE. 


XUl 


evidence they adduce in support of (their) view and 
then says, ‘The two first of these arguments have 
already been stated by Donaldson (Varron. pp. 433, 
437, sqq.) in a way that leaves nothing for others 
to add.’ 

Again, his notes (w. 688 and 899) on the pronuncia- 
tion first of item quidem modo, and then of student 
and facere, refer to Donaldson’s ‘ interesting chapter ’ 
as the source of his information. 

On my first seeing Mr. Marriott’s book, I wrote a 
note to him in which I pointed out that ho had fallen 
into a very natural mistake in ascribing to the author 
of the Varronianus the paternity of the article 
Terentian Metres in the Penny Cyclopaedia, for this 
article and the corresponding chapter in the Varroni- 
anus evidently proceeded from the same pen ; but that 
the simple fact was that the article in the Penny 
Cyclopjedia was written by me, and dishonestly 
appropriated by Donaldson. He at once favoured me 
with a courteous reply, in which he admitted his error, 
and at the same time assured me that the mistake 
he had made was one of general currency. 

And in fact I find that the editor of Terence in the 
‘ Bibliotheca Classica’ shares the error, for he also com- 
mences his Introduction on the Metres of Terence 
(p. xxviii.) thus : — 

‘ This subject has been noticed by the author of the 
Varronianus (chap, x.), who refers to the Journal of 
Education (voL ii. p. 344, &c.), where it is treated gene- 
rally in a manner which leaves nothing for others.’ 

When he wrote these words, I have little doubt that, 
in his own mmd, Mr. Parry ascribed to Dr. Donaldson 


Digitized by Google 


XIV 


PREFACE. 


the authorship of tlie article in the Journal of Education, 
for in his preface (p. ix.), after saying that ‘ the ques- 
tion of Terentian language and metre is a subject which 
has till lately lain fallow in England since the time of 
Bentley and Hare,’ he adds in a note, ‘ When I wrote 
this sentence, I had not seen Mr. Key’s Essay on the 
Metres of Terence and Plautus. My only acquaint- 
ance with his researches was through the Varronianus.’ 
Now the whole of the chapter of the Varronianus in 
question is, with two petty exceptions, an unmitigated 
plagiarism by Donaldson from my two articles on 
Terentian Metres, that in the Journal of Education 
and that in the Cyclopmdia. One exception is the 
doctrine that homines (in the Phormio, v. 1, 37) is 
a monosyllable (!). The honour of this is all his own. 
The other is that puellam in Haut. v. 5, 16 is to be 
pronounced ptdlam. It must have been somewhat 
grating to Dr. Donaldson’s conscience to find that 
Mr. Parry, amid his general approval of his chapter 
on the comic metres, selects this one remark for dis- 
approval (p. XXXV. note). 

I confess that it is not satisfactory to me that 
my views should come to the knowledge of scholars 
through the medium of Donaldsonian writings ; and 
I have deemed it a duty to print these pages, although 
the offender has now passed away. I never thought 
it necessary to notice the abusive or contemptuous 
tenns in which I have been habitually alluded to in 
the later editions of the Cratylus and Varronianus, 
for such abuse I heeded as little as I did the flattery 
he bestowed upon me in the first edition of the former 
work. 


Digitized by Google 



CONTENTS. 


rAOR 

Pbbfacb y 

I. On the Representativea of gya in Allied Languages . 1 

II. On the Piepogitions tyi, »«, and related Words ■ ■ 57 

III. On the Latin Prepositions re and pro 74 

IV. On tlifl German Prefix mrr itnd AllinH Forma ^ IQl 

V. and q/i!«r the same Word 117 

VI. On the Bo^called ‘ g privative* 127 

VII. The Latin e(, que, atgut (ac), and the Greek tcu, rt, all 

of one Origin '. 149 

VIII. On the Latin Particles out, an, nS 170 

IX. On Plnral Forma in Latin with a Singular Meaning, 

and eapecially on VuKil’a tibo of menta . ■ . . 185 

X. Eicreacent ConsonantB ^ . . . . . . . . . 2M 

XL On Palae Diviaion of Suflixea . ^ . . . 225 

XII. QvseritTr ; The Sanskrit LanRuage as the Basis of 
Lingnistic Science ; and the Laboura of the German 
School in that Field — are they not over-valued 1 — 

First Part . . . . . . . . . . . . 218 

XIII. Qyaeritvr — Second Part 270 

XIV, Poatecript to Qraerityr 309 

Addenda, CorriRenda, &c 317 

Ikdex . . . . . , , . . 323 


Digitized by Google 


PHILOLOGICAL ESSAYS. 


1 . 

ON THE REPRESENTATIVES OF ai>a 
IN ALLIED LANGUAGF.S. 

The little syllables which are prefixed or affixed to 
roots in the process of word-building were probably 
at the outset possessed of an importance equal to 
that claimed for the roots to which they are attached ; 
in other words, they also are roots ; but, supporting 
for the time an inferior part, are of course subject to 
be treated with some indignity beside the greater per- 
sonage on whom they wait. Thus it will be often 
found that both prefixes and suffixes are curtailed of 
their fair proportions. But among such secondary 
syllables, none perhaps suffer more abridgement or 
alteration than the prepositions used in the compo- 
sition of verbs, especially those which fall under the 
class called by German grammarians inseparable. 
Grimm has particularly noticed this liability (Deutsche 
Grammatik, ii. 865). 

‘The doctrine,’ says he, ‘which holds true gene- 
rally of particles, that they become obscure in signi- 
fication and disguised in form, is specially applicable 

B' 


Digitized by Google 



2 


ON THE EEPRESENTATIVES OP 


to the inseparable particles. The notion which they 
express wavers between increased intensity and a 
privative character, or occupies an intermediate posi- 
tion. nie form again passes commonly through all the 
vowels, and at last fades away into an unaccented e, 
while the consonants either drop off" or are modified by 
the influence of the initial consonant of the word which 
is brought into contact with them. One particle indeed 
(our ge-), in the vulgar dialect, has sunk down into an 
almost imperceptible breathing. The more this corrup- 
tion of a particle develops itself, the less capable does 
it become of maintaining the independent and separate 
character which it first possessed.’ 

Greek scholars in this country will probably give a 
ready assent to the power which Kiihner assigns to 
the preposition ava in the etymological portion of his 
grammar (§ 365, 2) : ‘ dva, auf (hinauf).’ At any rate 
up is the notion which distinctly presents itself in a 
large number of the verbs compounded with ava.' 
But German authors have allowed themselves to be 
biassed by the tempting similarity between the Greek 
ava and the Gothic and bid German preposition ana, 
which in modern German takes the shorter form of 
an, the equivalent of our on ; and hence in his 
syntax, § 602, Kuhner writes ; ‘ 'Ava [old Germ, ana, 
and as still written an with the dative and accusa- 


^ It ia a Bomewhat strange fact that Matthiae, in his large 
grammar of more than a thousand pages (at least in the English 
translation), gives not a word which can lead his readers to the 
true sense of am. His examples are limited to such as he trans- 
lates hy on, in, throughout, agaitut, with, or by phrases of dittri- 
bulion. On the sense which ava brings to verbs in composition he 
is utterly silent. 


Digitized by Google 



THE GREEK PREPOSITION ava. 


3 


tive]. The fundamental signification of the preposition 
ova is on, up {an, auf).’ In justification of the sense 
on, Kiihner gives no examples but awa av 

u>fup, ava rapyiip(p aKptp in Honier, and tihtt, S’ ava 
. ffKaiTTfp Alos alerds in Pindar. Now in all these 
examples elevation is a prevailing idea ; and the 
English translation ‘ upon,' or rather ‘ up on the 
sceptre,’ duly represents the first of the above phrases, 
where ava contributes no more to the sentence than 
the English up>, for the second preposition on represents 
what the Greek expresses by the dative case-ending. 

In the course of this paper the real representative 
of the Greek ava on German ground will be pointed 
out, together with the arguments necessary to estab- 
lish its claim. For the present I must deal with ava 
alone. Now the chief meanings which belong to this 
preposition are the following ; 1, up, as ava tov irora- 
fiov, ava poov ir\eiv ; ava vara Otovaa ', 2, it is often 
convenient to fancy an acclivity, where none may 
actuafiy exist, and thus on the most level ground we 
may speak of going up tliis line and down that. 
Hence we get the meaning of along, through, as a»« 
V7)as, aarv, ireStov. 3, from through in place we pass 
readily to through in time, during, as ava wktu ‘all 
night long,’ ava rov iroXefiov ‘throughout the war.’ 
4, that the idea of distribution, which is so common 
in this preposition, is in immediate relation with that 
of along or through, is often seen physically; for 
example, when a postman distributing his cargo of 
letters passes along the streets as he leaves them at 
the successive houses. So an epidemic passes through 
a camp, attacking one soldier after another. I pur- 
posely pass over the statement that ava with numerals 

B 2 


Digitized by Google 



4 


OK THE REPRESENTATIVES OF 


signifies up to, full, as is stated in a lexicon of repute, 
or auf (circa) as Kuliner would translate it, because 
in the passages (Horn. Od. ix. 209, Herod, iv. 101) 
quoted or referred to, the distributival sense seems to 
prevail;* but of course, when more decisive instances 
are produced, I shall readily welcome a usage which is 
perfectly consistent with the sense of the preposition, 
as our own construction, ‘ up to three hundred,’ serves 
to show. 

I next pass to what more concerns me, the use of 
apa in composition with verbs ; and here the important 
bearing of the subject upon the future arguments must 
be my apology for entering into fuller detail. 

1. The sense of up is, as I have already said, too 
evidently exhibited in the compound verbs to render 
a collection of instances necessary to establish it. Still, 
with a view to matter which will subsequently come 
under consideration, I would draw attention to certain 
classes of verba in which this sense of tip is prominent ; 
as, a. verbs with the idea of flame, heat, &c. ascending: 
av-ai6-, -aiOvaa-, -dtrz-, -av-, Ppaaa-, -8o^-, -fe-, -Ovfua-, 
-Kai-, -Acap^Xaf-, -\apir-, -irpijO-, -Trvpo-, -(f>aiv-, -^Xey-, 
-^\v~ ; b. verbs of searching or investigating, in which 
ava seems to signify up to the very sources: au-eip- 
(r.), -epevva-, -eptora-, -eraf-; apa-^r]re-, -xpip-, -pavBav-, 
-p,acrrev-, -prfKo-, -iTVvOav- (r.), -iTKOire- ; — C. loud noise, 
Avhere the loudness is attributed to ava, just as we 
ourselves say ‘ speak up, raise your voice, you speak 
too low to be heard.’ Under this head Liddell 
and Scott’s Lexicon furnishes some thirty or forty 


Of the phrases ora aropa, oro Bvpof tx***” j ond ora tovs 
irpo»Toi>5 tiroi, mention is made below. 


Digitized by Google 



TUE GREEK PREPOSITION ava. 


5 


examples, some of which however perhaps belong 
to §6. 

2. As downward motion, by the law of gravity, is 
the natural course of most visible bodies, the idea of 
up is connected with reversed action. Hence the sense 
of back is found in more than thirty compounds in 
the same lexicon. 

3. But to go back is to go over the same ground 
again. This idea, again, occurs as ficquently as the 
last. I will only quote the examples ava-<yi~/v(oaK- and 
ava yvapi^- ‘ know again, recognise’ ; and ava pu/AvriaK-, 
‘ remind.’ 

4. But to retrace one’s steps is another phrase for 
the reversal of some preceding action, where the 
English prefix is commonly un-. Hence av-apa- (r.) 
‘ recall a curse,’ ava-BiSaaK- ‘ untcach,’ -eXto-o-- ‘ unroll,’ 
-ei^- (r.) ‘ recall a prayer,’ -KaXvTrr- ‘ unwrap,’ -*Xa>^- 
‘ untwist (what has been spun),’ -KoXvfifia- ‘ come to 
the surface again after diving,’ -kvttt- ‘ raise (the head) 
again after stooping,’ -p,aintv- (r.) ‘ make an oracle in- 
valid,’ -wTvo’a- ‘ unfold,, -<r/f€uof- * ‘ dismantle,’ -<r^oXX- 
‘ rise up after a fall,’ -tT<f>payi^~ ‘ unseal,’ -rvXtira- ‘ unroll.’ 

5. Sometimes the simple verb already in itself ex- 
presses the idea of loosening, stripping, opening ; and 
then the prefix appeals only to strengthen the idea of 
relaxation : and yet there will often be found some- 


1 Amo-K€vaC- we are told means ‘ strictly to pack up the baggage 
(ri vKtiri), Lat. vasa colligere, and so to carry away, Xen. An. vi. 
2, 8 : usu. iu Med. to break up, march away.’ Why not ‘ dis- 
mantle ’ here, as in the other uses of the word ) This would be in 
agreement with the phrase just quoted from the lexicon, ‘ break 
up ; ’ and indeed it is usual for a series of acts to take their collective 
name from the first in iho series. 


Digitized by Google 



G ON THE REPRESENTATIVES OP 

thing more than this, viz. a reference to a previous act 
of binding, &c. This in English is the case with ‘ to 
unloose,’ not so with ‘ to loosen and similarly ti?*- 
‘ melt ’ is applicable to things which in their natural 
condition are solid, whereas ava-rriK- implies a return 
to a former condition, and can only be used of thawing 
congealed fluids. Examples of such words are ava- 
•yufivo-, -Sep-, -oiy-, -irav-, -rqK-, -^aXa-, av-iTifu, and 
ava-verav-w/ii, 

6. The idea of opening or discovery is also seen in 

other compounds with ava, where the simple verb de- 
notes some means by which the opening is effected. 
Here again not unfrequently our own language also 
consistently expresses the idea by up : av-evptoK- ‘ find 
out,’ -evpvv- ‘ widen,’ ava-xea^- ‘ split up,’ -xXiP- ‘ bend 
back (a door) and so open,’ -fati/- ‘ tear up or open a 
wound,’ -^yvvfii ‘ break up or open,’ ‘ split up,’ 

-Te/ii>- ‘ cut up,’ -xatv- ‘ gape open.’ 

7. From the idea of opening we readily pass to that 
of commencing, where again up is at times used in 
English. Thus we say : ‘ open a ball, open fire, strike 
up a tune.’ To this head perhaps belong the following 
words, where the translation is borrowed from the 
lexicon already named : -ava-KOKKv- ‘ begin to crow,’ 
-fcpe/c- (r.) ‘begin to play (a tune),’ -fcpoo- (r.) ‘strike 
up (a tune) or begin a speech,’ -/leXv- ‘begin to 
sing,’ -fiaXX- (r.) ‘begin (anything), -oSvp- (r.) ‘break 
into waibng,’ -<f>vaar ‘ begin to blow,’ -pa^wSe- ‘ begin 
singing ;’ and perhaps we should not be wrong in 
translating ava-yeXa- ‘burst out laughing, set up a 
laugh.’ 

8. The idea of back is in close connexion with those 
of escaping, removal, aivay : ava-Kopu^- (r.)‘ get safe 


Digitized by Google 



THE GEEEK PREPOSITION ava. 


7 


away, escape,’ -<f>evy- ‘ escape,’ ‘ frighten away,’ 

-pv~ (r.) ‘ rescue.’ 

9. Indeed the idea of removal also connects itself 
directly with the idea of up, inasmuch as motion 
upward is in many cases a convenient or even essen- 
tial preliminary. Thus in Latin ferre, toUere, sustuli 
have for their first sense ‘to raise,’ and only in a 
secondary way signify ‘ carry off.’ Examples are 
av-aipe- ‘ take up and so carry away,’ av-apva^- ‘ snatch 
up and carry off,’ ava-Kadcup- ‘ clean up or clear up,’ 
-iTTroyyi^- ‘ sponge up,’ -i^a- ‘ wipe up,’ -weT- (r.) ‘ fly 
away.’ 

10. As the idea of through is often expressed by ava 
in company with nouns, so we have ava-mip- * pierce 
through, spit,’ -nrpa- ‘bore through,’ -Trijyi/w/it' transfix.’ 

11. Hence we may perhaps deduee thorough distri- 
bution, an act p>^'>'vading all j^arts, as seen more or less 
in: ava-SiBw/ju, ‘distribute,’ -fu/io- ‘leaven thoroughly,’ 
-Kepavpvfu, -KipvapMi, -fiiyvv/ju and -fiiay-, -<f>vp-, ‘mix 
thoroughly, mix up.’ But very possibly a better inter- 
pretation, so far as regards the verbs of mixing, may 
be obtained directly from the idea of upward move- 
ment, seeing that the process of mixing is a constant 
battle with the heavier ingredients which persist in 
sinking. The truth of this will be felt by any one 
who has mixed a bowl of salad or a powder containing 
calomel. 

12. The idea of completeness or thoroughly might 

well be expected in compounds with ava, and accord- 
ingly we find this meaning attributed to av-ap/xo^-, 
-annp,o-, ; ava-fii^puax-, -^o)ypa<f)e-, -rrpi-. Even 

of these some may be doubted, and at best the list is 
very short. The explanation of the paucity may perhaps 


Digitized by Google 



8 


ON THE IIEPRESENTATIVES OF 


be this. It was noticed alx)ve tliat ava obtained its 
sense of ‘ through ’ from the notion of a fictitious 
acclivity, wliere a person goes up this line and down 
that. Hence Kara ‘ down’ would be entitled to share 
the privilege, and accordingly this preposition is equally 
used in distributival phrases, as Kara <f>v\a ‘ by tribes,’ 
Kar’ avBpa ‘ man by man,’ &c. On the same principle 
it is well calculated to express ‘ thoroughness ’ with 
verbs. This office it performs in the Greek vocabulary 
to a great extent, being in much higher favour for the 
puipose than ava, whereas with us the word up is more 
in vogue. Hence Kar-eaOi- ‘ eat up,’ Karu-mv- ‘ drink 
up,’ Kara-xpa- (r.) ‘use up.’^ 

In a few instances the idea of on or is said to be 
the signification of ava, as in av-etp- ‘ fasten on,’ ava- 
KoWa,- ‘ glue on,’ avappavr-, ‘ sew on.’ But here we 
seem to have a totally different preposition, the ana- 
logue of the Gothic ana, German a?i, signifying ‘on.’ 
I find that 1 have spent many more words uixm 
this preliminary matter than I had intended. My 
apology must be, th.at I knew of no grammar or dic- 
tionaiy in which the subject was handled in sufficient 
detail. Nor indeed is there any part of language more 
commonly neglected in grammars, for to them the 
question pi*operly belongs, than the power of prepo- 
sitions as prefixed to verbs. At the same time, what I 
have said seems necessary for the just appreciation of 
tlie evidence I shall have to adduce ; and I have now 


I It should be stated that, in drawing up these lists of compound 
verbs, I have relied almost exclusively on the excellent lexicon of 
Liddell and Scott, an acknowledgment I am the more bound to 
make, as I have ventured at times to criticise some of their state- 

inciiU. 


Digitized by Google 



THE GKEEK PREPOSITION ava. 


9 


the satisfaction of knowing that Pott in his new 
edition of the Etymologische Forschungen (p. 305), 
has adopted my distribution of the meanings of apa. 

After this preface, the first problem is, whether the 
Latin language has any representative of ava. My 
answer is, that it has at least one, and, as I believe, no 
less than three, or even four representatives. That 
apa should appear in Latin without a final vowel is 
what is to be expected when we compare the cases of 
OTTO and ab, evi and oh, irapa and per in perjurus, €Pi 
and in. Further, in Greece itself apa was reduced to 
ap in some forms and dialects, just as /eara, trapa, ept 
were to kut, vap, and ep, and this especially in those 
dialects which have the closest affinity to Latin, the 
Doric, and J5olic. Now in three words, ancisus, 
(Lucr. iii. 660), anquir-, and anhela-, the form an has 
been preserved ; but for all of these a word of remark 
seems necessary. Lucretius is speaking of a snake 
suddenly divided into many parts, and yet in these 
several parts still exhibiting signs of vitality for a 
while : 

“ Onmia iam sorsum cernes ancisa recent! 

Volnere tortari.” 

Here therefore ancisa is no compound of am, as 
Forcelluii would make it ; but clearly means ‘ cut off, 
or cut through.’ Anquir- seems to have for its mean- 
ing ‘ search up to the sources,’ and indeed apapuiaTev- is 
explained by Liddell and Scott as = anquir-. AuheUi- 
is used of those violent up-blowings which follow 
volcanic action, as in Cic. anhelitus terrae, and Ovid, 
Fast. iv. 49 1 ; also of the flame driven out by a fumacc- 
blast, or from the nostrils of Colchian bulls. Comp. 


Digitized by Google 



10 


ON THE REPHESENTATIVE3 OF 


ava-(f>v<ra-, whence ai/a-<f>vat}-<Tt- ‘ the blow-hole of a 
crater,’ and ava-^vaia- ‘ blow as a dolphin.' At any 
rate such an explanation of anhela- seems more satis- 
factory than what we find in Andrews’s Lexicon, who 
gives as the ‘ literal’ sense ; ‘ to draw the breath from 
around the whole body.’ This translation no doubt 
proceeded on the assumption that the word contains 
the prefix am ‘ round,’, although in this case it should 
have been am-hela-, if we may judge from am-ici-. 
But besides this, an-hela- clearly means an expiration 
rather than an inspiration. 

Of course before a labial tin would pass into am, 
and accordingly we have am-puta- ‘ cut oflF,’ am-mone- 
‘ remind ’ avafii/jiv7j<rK-, am-bur- ‘ begin to bum, singe,’ 
am-bed- ‘ begin to eat,’ ‘ nibble at.’ Cf. our own burn, 
bite. The notion of am ‘round,’ is inconsistent with the 
meaning of both am-puta- and am^ione- , and the form 
ammone is that which for Ovid’s Fasti has by far the 
best authority, if we may take for our guides those 
MSS. which Merkel himself collated. 

Assimilation also accounts for the forms allevor- 
‘lift up,’ alliga- ‘tie up,’ and the impersonal allub- 
escit ' of a commencing love, for in all these the 
notion of ad ‘ to,’ seems out of place. The first has 
for synonyms in Forcellini, ‘ sublevo, in altum tollo, 
sursum levo.’ As for alliga-, it is enough to quote 
the phrase alligare minus, and to note that Pliny, 
when he has occasion for the idea ‘ tie to,’ or rather, 
‘tie up to,’ uses adalliga-; but a verb twice com- 
pounded with the same preposition would be some- 
thing strange. Alloqui too is very insufficiently 
translated by the verb ‘address.’ It means to ‘console, 
cheer up,’ and so is clearly a com]jound of an. 


Digitized by Google 



THE GllEEK PREPOSITION ava. 11 

Again, before s an n would of course be silent ; and 
so we have an explanation of such forms as assicca- 
‘ diy up ’ = ava^paiv-, ossudesc- ‘ burst out into a 
sweat,’ and assurg- ‘rise again.’ In such phrases as 
majoribus natu assurgere, both the notion of ad and 
that of ava are intelligible, but in all other uses of 
this verb, that of ava alone is admissible, especially 
in sentences where the notion of ‘ get up again ’ after 
a fall is implied, as in ‘ Galli neque sustinere se pro- 
lapsi neque adsurgere {assurgere 1) ex voraginihus 
poterant’ (Liv. xxii. 2) ; and again : ‘ Tetra ibi luctatio 
erat . ... in prono citius pede se fallente ut sen 
manibus in adsurgendo seu genu se adjuvissent, ipsis 
adminiculis prolapsis {ox prolapsi) itermn corruerent' 
(xxi. 36). 

I next take cases where in place of an I find 
but a simple a to represent the prefix. Here again 
we have what is parallel to the usage of Greece. In 
the Doric and ADolic dialects (see Ahrens, De Dialeetis), 
if the simple verb began with a <r, followed imme- 
diately by another consonant, the fuller form av, or 
its equivalent, ov, dropped its nasal. Accordingly we 
find in Latin a-scend- opposed to descend-, a-spira- 
‘ exhale ’ (‘ pulmones se contrahunt aspirantes,’ Cic.), 
a-stru- ‘ build up,’ opposed to de-stru- ; a-sta-, as used 
in Plautus without any meaning of ad, e.g. ‘ Haul 
ineusceme ( for so the MSS. = avevaxVf^o^h) astite- 
runt,’ ‘no inelegant that’ (Trin. iii. 1, 24). On 
the same principle we have a-gnosc- ‘ recognise,' = 
ava-ytyvma-K-. 

Thus already we have a respectable stoek of words 
in which an assumed an, = ava, has all in its favour 
alike as to form and meaning. But I also venture on 


Digitized by Google 



ON THE REPRESENTATIVES OP 


V2 


the assertion, that a visible ad in Latin compound 
verbs not unfrequently stands as a substitute for our 
an; so that the language had in fact for the com- 
position of verbs two prepositions of this shape, 
Avhich it is important to distinguish. The interchange 
of an n and d is what most philologers will readily 
admit, and indeed the relation between these letters 
is precisely the same as that between m and b, and as 
that between the nasal ng and g (of go). Only when 
the nasal passage is in communication with the 
wind-pipe, have we m, n, and ng : but the moment 
this passage is closed by the velum palati, these 
respectively pass into h, d, and g. (See Mr. Wey- 
mouth’s paper on this subject, in the ‘ Transactions 
of the Philological Society’ for 1856,' page 21, and 
the work of Blindeisen, to which he refers.) 

But I cannot now stop to discuss this point at any 
length. Assuming that a preposition an may well 
take the form ad, I request attention to the following : 
accUvi- ‘ up-hill,’ opposed to de-clivi-, ‘ down hill ; ’ 
accresc- ‘ grow up,’ by the side of de-cresc- ‘ grow 
down;’ as in ‘Valitiido mi decrdscit, accresdt labor’ 
(Plant Cure. ii. 1,4); acced- ‘ rise as the tide,’ and 
deced- ‘ ebb ;’ ad-olesc- ‘ grow up,’ but ab-olesc- ‘ cause 
to grow down ;’ ad-aestua- ‘ boil up,’ apprehend- ‘ take 
up,’ accumula- ‘heap up,’ agger- ‘ heap up,’ ad-iniple- 
‘ fill up, ad-aequa- ‘ raise' to a level (with),’ ad-operi- 
‘ cover up,’ atting- ‘begin to touch, lay a finger upon,’ 
ad-juva- ‘ lift up,’ and accumbo ‘ I lie with the body 
raised,' as on a dinner couch, = ava-Kti/xat. 


* In Livy i. 29 we ought to read, onmia tecta tolo aeqnavil (not 
adaequat). 


Digitized by Google 



THK GUKEK I'UErOSITION aia. 


13 


Secondly, the notion of ‘ again ’ has already been 
seen in a-gnosc- and ammone-, which by some arc 
written adgnosc- and admone-. 

Thirdly, the reversal of an act was common with 
ava, but for ad I can only produce acquiesc- ‘ rest 
after labour,’ identical in sense and perhaps in form 
with ava-naveffOai, seeing that the Latin loves to have 
q as the analogue of a Greek tt. 

Fourthly, ad-aperi- bears a close analogy to av-oiy- ; 
and as in discussing the powers of ava I deduced from 
the idea of opening that of commencement, so in Latin 
I find ad-ama- ‘ fall in love,’ ad-mira- (r.) ‘ be sud- 
denly seized with wonder,’ ajffle- ‘burst into tears,’ ad- 
dormisc- ‘ fall asleep,’ ad-hinni- ‘ set up a loud neigh,’ 
accend- and ad-ole-^ ‘ set on fire,’ ad-gem- ‘ all at once 
sigh,’ ad-vesperasc- ‘ begin to be dusk,’ ad-esuri- ‘ be 
seized with a fit of hunger.’ 

The physical notion of through clearly resides in 
ad-ig- ' drive through, pierce, transfix.’ 

Again, the sense of removal growing out of the 
sense of upward movement, as seen in compounds of 
ava, § 9, has its counterpart in ad-im- ‘ take up and so 
take away’ (comp, av-atpe-), ad-aresc- ‘dry up’ (in- 
trans.), ad-bib- ‘ drink up.’ 

With the class of ava-fuay-, ava-<f)vp-, I unite ad- 
misce-, as also assicca- already quoted. 

To the fists already given 1 am not sure but that I 


' This adole- Lb virtually one with adoUiC; the root-syllable ol 
being only a variety of al of aUre. In both the notion of ‘ upward’ 
prevails, only in adole- we have that special sense which occurs in 
the familiar af-cre y/omTuam. A rd-ere and ard-uue, of the same 
stock, also unite the two meanings. 


Digitized by Google 



14 


ON THE REPRESENTATIVES OF 


ought to add many others. Thus ad-i-, aggredi- (r.), 
ad-eqiiita-, accurr-, acced-, cvcci- invite me as it were 
to the translations, ‘ go up, march up, run up, ride up, 
step up to any one, call up at any rate, these phrases 
are quite in agreement with the idiom of our own 
language. Again, admin-iculiim ‘a prop,’ seems to 
imply a verb ad-min- ‘ prop up and ad-juva- in its 
preposition claims affinity with ava, partly because 
verbs of assistance are very apt to appear with a 
preposition signifying up, as sub-leva-, sub-veni-, suc- 
curr-, subsid-iiim, and partly because the simple verb 
juva- seems in itself to have had for its first sense ‘to 
lift or elevate,’ which will at once explain its double 
power ‘ to delight’ and ‘ to assist’ I think, nay, >I 
suspect the root to be identical with that of the verb 
lev-a- and adj. levi-, and our own lift, for an initial ^ in 
Latin raises the suspicion of a lost I ; thus jecur and 
ryirap are brought into connexion with our liver, jocus 
with our laugh. The close connexion between I and 
the y sound (of the Latin j) is well seen in the ‘Z’ 
mouille of the French. 

I am fully aware that some of the compounds with 
ad to which I have laid claim might admit of an 
explanation from the power of the ordinary preposi- 
tion ad. Thus the first element in acclivis might 
have been justified by the prefix of the Greek irpoa- 
avTijs. Yet in many instances this preposition ad fails 
utterly, while the senses of ava are all-sufficient, so 
that I still adhere to what I have said, the doubtful 
instances receiving a borrowed light from the non- 
doubtful. 

With all this I in no way deny that ad ‘ to or near’ 
has contributed its compounds to the Latin language. 


Digitized by Google 



THE GKEEK PREPOSITION ava. 


15 


SO that it may often be difficult to adjudicate between 
the confficting claims of the prepositions ; and at 
times a just judgment will perhaps make a division 
between the two rivals, assigning some uses of the 
same word to the one, some to the other, as in the 
case of acced-. Or possibly the ad — ava may have 
been at first the only prefix admitted to composition 
with verbs, and subsequently compelled to submit to 
invasion of its domain, when the Roman, no longer 
alive to the sense of up, may have allowed himself to 
be unduly biassed by the meanings of the familiar 
preposition ad ‘ to.’ Be this as it may, there will be 
seen in the sequel not a few instances of independent 
prefixes sinking into an identity of form. 

I proceed to yet another variety. It is a peculiarity 
of Latin notation that it often prefers a weak vowel 
to the stronger vowels of other languages. Thus, to 
the Greek o/i^po- and SuktvXo- stand opposed the Latin 
imbei'i- and digitulo-, to the Sanskrit agni- the Latin 
igni- ; and again the Latin sine and lingua are repre- 
sented in French by sans and langue. But the most 
valuable instance for the present question is that of 
the so-called privative particle av of Greek = in of 
Latin. It will presently be seen too, that the prepo- 
sition ava takes in German a form in which the first a 
gives place to i or e. Am I not justified then in 
expressing a suspicion that the Romans in such distri- 
butival phrases as in-dies, in-horas, kc., employed a 
preposition in = aval But the present dealings are 
rather with compound verbs, and here I first throw 
together — intumesc- ‘ swell up,’ ingrandesc- ‘ grow up,’ 
incresc- ‘ grow up,’ inhorre- ‘ bristle up,’ institu- ‘ set 
up,’ insurg- ‘ rise up,’ innutri- ‘ bring up by nursing,’ 


Digitized by Google 



]() 0,\ THE REPRESENTATIVES OF 

injla- ‘ puff up,’ incila- ‘ rouse up,’ incandesc- ‘ blaze 
up,’ inardesc- ‘ blaze up,’ incend-, injlavima- ‘ set on 
fire,’ indaga-, ini'estiga- ‘ trace up to the sources ; ’ 
to which we should probably add the adjective or 
rather participle insolenti- ‘ swelling up with pride,’ 
from a lost verb soh- ‘ swell,’ so that it con-esponds 
to the German participle anschwellend. 

Then with the notion of back ; in-hibe- ‘hold up or 
back = av-ex', and in-fiect- ‘ bend back.’ 

For ‘ again’ I find two clear cases : in-staura- = re- 
staura- and in-gemina- ‘ redouble.’ 

But the most striking use of am is in the sense of 
‘ reversing.’ Now the Latin inconcilia- in some current 
dictionaries is said to have the two somewhat oppo- 
site meanings of ‘ to win over to one’s side, to conciliate,’ 
and ‘ to make an enemy of.’ Concilia- also is for the 
most part mis-explained. But Forcellini had already 
given the right view as to both these verbs. Thus of 
concilia- he says : ‘ Verbum est fullomim,’ quoting 
Varro, ‘ Vestiinentum apud fullonem cum cogitiir con- 
ciliari dicitur ; and supporting this view by Scaliger’s 
derivation of the word, ‘ a ciliis, h. e. pilis! The 
word in fact means to felt cloth, as we still do in 
making drugget or wide-awakes. The metaphorical 
use of the word in the sense 'of promoting the union 
of friendship is sufficiently intelligible ; and of course 
inconcilio is correctly explained in the same work : 

‘ Contrariam signijicationem habet concilio ; ’ in 
other words it means (to invent a new verb) ‘ to un- 
felt (cloth),’ or separate again the woolly fibres which 
had been previously united in the process of felting. 
Thus we have a most expressive metaphor, somewhat 
like our own ‘unravelling,’ and available generally for ^ j 


Digitized by Google 



THE GREEK PREPOSITION ava. 


17 


the idea of breaking up, dissolving, what had been 
closely united. The word occurs in at least four 
passages of Plautus, and in all this idea is most appro- 
priate, due allowance being made for this comic poet’s 
love of bold metaphors. The process of feltiug is no 
longer carried on under our eyes, as it was under the 
eyes of Romans in the age of Plautus ; we shall there- 
fore have a more intelligible, yet at the same time 
equivalent metaphor, if we use in its place the phrase 
‘to make oakum’ of him or it, ‘to tear to rags.’ In 
the Trinummus L 2, 99, and the Mostellaria iii. 1, 85, 
the accompanying accusatives are persons, and the 
idea is breaking them up as regards their property. 
In the Baccides iii. 6, 22 — inconciliare copias omnis 
meas — the idea is substantially the same ; and in the 
Persa v. 2, 53, non inconciliat quom te emo may bo 
rendered by ‘ ho does not tear up,’ that is ‘ annul my 
purchase of you,’ quom in the older writers often 
Wving the power of quod. C. 0. MUller indeed, in 
his edition of Festus, v. inconciliasti, finds an objec- 
tion to the doctrine that this verb is the opposite of 
‘ concilia-’ in that the prefix in, which denotes nega- 
tion {abnuitionem), is never attached to verbs, except 
in the participial form. This difficulty vanishes so 
soon as we make the in, not the negative prefix, but a 
variety of aua. 

A second example is i-gnose- (for in-gnosc) ‘im- 
know,’ to invent another word, that is ‘ forget,’ from 
which the idea of ‘ forgive’ readily fiows. 

Insimulor is on all sides maltreated. But all will 
be smooth if we start from simula-re~ ‘ to make one- 
self like (what one is not),’ or ‘ put on a mask ;’ for 

c 


Digitized by Google 



18 ON THE REPRESENTATIVES OP 

then insimula-re is ‘to unmask (a rogue)/ ‘to expose 
him/ that is ‘ to accuse.’ 

Injitias ire is in the lexicons translated ‘ to deny but 
the incorrectness of this is at once seen in the Flautian 
phrase, ‘ neque nego neque injitias eol But if we derive 
this substantive from a theoretic infari, ‘ to unsay/ i.e. 

‘ to eat one’s words/ ‘ to retract what one has already 
admitted/ the sentence in Plautus has a meaning. 

For the idea of opening I offer in-ara- ‘ plough up,’ 
in-jind- ‘ cleave open,’ ‘ plough up / and for ‘ begin- 
ning/ not only in-cipi- ‘ take up/ ‘ begin’ (and perhaps 
in-coha-), but also in-calesc- ‘begin to get hot,’ and 
in-tepesc- ‘begin to get warm.’ And here I come 
across some verbs the meaning of which deserves con- 
sideration. In Forcellini there is a mixture of what 
is sound with what is unsound, yet even in this latter 
case his articles supply the data for safe conclusions. 
In informa-re he is wholly right ; and yet recent dic- 
tionaries wholly wrong. Thus his words run : ‘Primam 
et rudem alicui rei foi'mam induco and under in- 
formatus his first quotation is : ‘ His informatum mani- 
bus iam parte polita fulmen erat' (Aen. viii. 426), 
while Furlanetto in his edition of the great lexicon 
adds : ‘Varr.’ (Verr. is a misprint) ‘ap. GelL iii. 10: 
Quarta hebdomade caput (of a male foetus) et spina 
quae est in dorso, informatur’ 

As to imbuo, what appears to me to be the correct 
starting-point is what we read near the end of For- 
cellini’s article : ‘ Re intacta adhuc uti incipere.’ But 
the word is probably only a Latin variety of the 
Greek avaBev-w, or in another dialect avSev-u. At any 
rate ‘imbue’ is the translation given by Liddell and Scott 
for the one passage which they quote. The change of 


Digitized by Google 



THU GUEEK I’UEI'OSITION ava. 


19 


consonants is parallel to what is seen between the 
German loide ‘ loins,’ and Latin lumbi ; and again 
between the Italian anda-re and the Latin diminutival 
verb ambula-re. I would therefore translate imbu- 
‘ wet for the first time and it may have got its 
metaphorical meanings from the military idea of wet- 
ting a hitherto maiden sword in the enemy’s blood. 
Those who translate imhutus ‘ steeped in,’ ‘ thoroughly 
imbued with,’ wholly mistake the power of the 
word as understood by Cicero and Catullus. 

The verbs imminu and impell- are usually treated 
as though the preposition were superfluou.s. But here 
we should give a preference to the translation ‘ begin 
to impair,’ i.e. ‘ impair what was previously entire,’ 
and ‘ begin to drive,’ or, in other words, ‘ give to that 
which has hitherto been quite firm its first movement,' 
' start’ it 

The notion of ‘ removal,’ ‘ away,’ is to be seen in 
the verb incid- in the sense of ‘cut ofiT,’ which will 
thus be a difierent word from incid- ' cut into,’ and in 
infring- ‘ break ofiF,’ intabesc- ‘ melt away,’ 

In dealing with the Celtic languages, I shall be 
very brief. I'he Welsh has a representative of ava 
in its inseparable prefix ad- signifying ‘ back,’ 
‘again,’ ‘reversal of an act’ Thus from nojio ‘to 
swim,’ brynu ‘ to buy,’ nabod ' to know,’ gw?ia ‘ to do,’ 
and gwisg sb. ‘ dress,’ there are compounds ad-nofio 
‘ to swim back,’ ad-brynu ‘ to redeem,’ ad-nabod ‘ to 
recognise,’ ad-gwneud ‘ to undo,’ ad-wisg sb. ‘ undress, 
disarray.’ Of verbs alone compounded with this ad 
there exist above one hundred and seventy. 

The Gaelic form corresponding to the Welsh ad- 
is commonly atli-, sometimes as-. Thus- sndmh ‘ to 

c 2 


Digitized by Google 



20 


ON THE BEPRE8ENTAT1VE3 OP 


Bwim,’ ath-shndmh ‘ to swim back ; ’ huail ‘ to strike/ 
ath-bhuail ‘ to strike back or again ; ' loisg ' to bum/ 
ath-loisg ‘ to bum again, bum deeply ; ' ohair ‘ work,’ 
ath-obair ‘ work done over again ; ’ casta or caiste 
‘ twisted,’ ath-chasta ' strongly twisted ; ’ beum ‘ a 
wound,’ ath-bheum ‘ a second wound ; ’ ainm ‘ a 
name,’ ath-ainm ‘ a surname or nickname ; ' eirigh 
‘ rising/ ais-eirigh ‘ resurrection.’ 

In the Breton the particle takes the shape of ad- 
or as-, as ober ‘ faire ’ (I quote from Legonidec), ad- 
ober ‘ refaire ; ’ kmiiza ‘ choir, tomber,’ as-kauSza 
‘ retomber.’ But even ana- in its fullest form has 
left its trace in this language. The verb ana-out has 
also the dialectic varieties ana-^ut and ana-vezout, 
and is in fact a compound of the simple verb gouzout. 
These verbs are of great irregularity, gouzout in 
particular changing the radical syllable gou,z to gwez 
or gwi when the following syllable has one of the 
weak vowels {i or e). But the relation of the two 
verl)s to each other becomes indisputable, when we 
place some of the tenses, as for example the futures, 
alongside of. each other ; — gwez-inn ‘ je saurai,’ gwez-i 
‘tu sauras,’ &c. ana-vez^-inn ‘je connaltrai,’ anorvez-i 
‘ tu connattras/ &c. I take this from the grammar of 
Legonidee ; in his dietionary the verb is also translated, 
and perhaps more correctly, reconnattre. This example 
is the more interesting, because not merely is the prefix 
identical with the Greek ava, but the root of the 
verb also is but a variety in form of the Greek root 
Ft<r or FtS- as seen in laritit. otScf, the Latin vid- of 

' This loss of a ^ is but on instance of a general law in the Celtic 
languages. See another example in ad-wixj from gtoisg, a few lines 
above. 


Digitized by Google 



THK GREEK PREPOSITION ava. 


21 


vide-, and our own wis- or wot-, whence wisdom, 
wise, wit, and the obsolete verb vnt or wot. It is 
also worth whUe to note, that in the ordinary form 
anaout, all trace of the root syllable has vanished, 
just as, to quote an example of Bopp’s, is the case 
with the German im for in dem, where we have 
a remnant of the preposition, and a remnant of the 
case-sufiix, but not a particle of the pronoun signi- 
fying ‘ the.’ We also find in the Breton, both kouna 
‘ to remember ’ and an-kouna ‘ to forget,’ where, 
besides the original form of the prefix, we have in 
its signification what reminds us of one of the most 
important uses of ava, the reversal of an act. 

In Irish there are some three or four prefixes which 
have a claim more or le.ss certain to represent ava. 
1. atJtr, as cruinnighim ‘ I collect,’ ath-chruinnighim 
‘ I collect again rioghaim ‘ I rule,’ aith-rioghaim ‘ I 
dethrone.’ — 2. adh-, as molaim ‘ I praise,’ adh-mholaim 
* I praise warmly.’ — 3. an-^, which unites the two very 
different powers of intensity and reversal: sgairtim 
‘ I cry out,’ an-sgairtim ‘ I cry out loudly ;’ glearaim 
‘I follow’ (sequor), ain-ghleamim *I pursue’ (in- 
sequor) ; aithnim ‘ I know’ (ich kenne), an-aithnim 
(‘ ich kenne nicht,’ says Leo, perhaps rather ‘ I forget’) ; 
icim ' I help,’ ain-icim ‘ I help zealously and a verb 
of the same form icim ‘ I count or reckon,’ ain-icim 
‘ I pass over in counting, I save,’ — 4. amk-, ‘ which 
negatives (or rather reverses) like the Germ.an «n-,’ as 
garaim ‘ I gladden,’ amh-garaim ‘ I torture ;’ riidhim 
‘ I arrange,’ amh-r6idhim ‘ I disarrange.’ 

1 Leo sees in this the Latin prefix in; but bis criticixm is 
damaged by his treating the preposition in and the privative in as 
one word. — Ferionschriften, 1853. 


Digitized by Google 



22 


ON THE UEPRESENTATIVES OF 


I cannot leave this part of the subject without draw- 
ing attention to the light which some of these Celtic 
examples throw on the Latin ignosc-. It is a common 
practice, I believe, to consider the prefix in this verb 
as being the privative in. But to this there is the 
all but insuperable bar that this negative prefix is 
attached solely to adjectives and participles. The 
apparent exception ignora-re is none, as it is imme- 
diately formed from the adjective ignaro-. But if 
in- of ignosc- represents the Greek ava with the sense 
of reversal, we have what we desire, ‘forget,' which 
readily takes the sense of ‘ forgive ; ’ and now the 
Celtic languages confirm this view by the Breton an- 
kouna ‘ to forget,’ and the Irish an-aithnim. Nor is it 
a grave objection that I am here assigning an identity 
of origin to words so different in meaning as ignosc- 
nnd agnosc-. For example, ava-iTKeva^- commonly 
means ‘to dismantle,’ but in Strabo and Dioscorides 
‘ to build again or to repair ;’ av-etXe- ‘ roll up’ in Time, 
and Arist., but ‘unroll’ in Plut.; av-oiKi^- ‘rebuild’ 
Paus., ‘restore to (his) home’ Strab., but ‘cause to 
leave a home ’ Aristoph. The only difference between 
these cases and that of agnosc- and ignosc- is, that the 
Eomans very wisely availed themselves of the variety 
in form to mark the variety in meaning. 

In the Teutonic family I shall first note that the 
form an- was preserved for a while in one verb of old 
English, viz. anhang, as in Chaucer’s ‘ Doctoures Tale ’ 
(v. 12,193 of Tyrwhitt’s ed.) — 

‘ He hail to take him and auhang him fast.’ 

See also Coleridge’s Glossarial Index. 

So again in modern German we must distuiguish 


Digitized by Google 



THE GREEK PREPOSITION ava. 23 

from the compounds with an- ‘ to/ those in which the 
notion of commencement resides. Of these, to place 
the meaning beyond doubt, I quote thirty, and might 
quote perhaps twice as many : — an-hahnen ‘ to break 
a path atirbeissen ‘ to bite the first piece an-blasen 
‘ to blow the first note aiv-bohren ‘ to broach an- 
brennen ' to begin to burn / an-hiilen ' to begin to 
hatch;’ an-faulen ‘to begin to rot;’ an-feilen ‘to 
begin to file ;’ an-geben ‘ to begin to give ;’ an-hacken 
‘ to begin to hack ;’ an-liauen ‘ to Itegin to cut ;’ an- 
hetzen ‘ to begin hunting ;’ an-jagen ‘ to begin to 
chase ;’ an-klingen ‘ to begin to sound ;’ an-laufen ‘ to 
begin to run ;’ an-pfliigen ‘ to begin ploughing ;’ an- 
platzen ‘ to begin to crack ;’ an~raspeln ‘ to begin to 
rasp ;’ anrreissen ‘ to begin to tear ;’ an-reiten ‘ to ride 
for the first time ;’ an-rennen ‘ to start ;’ an-sden ‘ to 
begin to sow ;’ an-sdgen ‘ to begin to saw ;’ an-saugen 
‘to begin to suck;’ atirschaben ‘to begin to scrape;’ 
an-schdlen ‘ to begin to peel ;’ aiv-scharren ‘ to begin 
to rake ;’ an-scheren ‘ to begin to shave ;’ an-schiessen 
{einejlinte) ‘to try (a gun) ;’ an-schmelzen ‘to begin 
to melt.’ 

Still more complete is the similarity of the Gothic 
ana-kunnan, ‘to read,’ which thus preserves one of 
the meanings of the Greek ava-^iofvmaK - ; and of ana- 
kumbj-an in the sense of lying on a dinner-couch, 
so that the word is one in meaning with the Greek 
txvQrKiifzati and the I^atin accuvibot But not rarely 
this language follows its habit of adding to the n an 
excrescent rf, so as to produce the forms anda, and, 
or und in place of ana and an ; and indeed even 
within the limits of the Greek language we find the 
preposition ava itself taking a S. As from the adverb 


Digitized by Google 



24 


ON THE REPKESENTATIVES OF 


an^a ‘ quick/ cornea an adjective aiy^po- ‘ quick/ so 
from ava ‘up/ the adjective avBt}po~ ‘raised/ whence 
av&ripov, ‘a raised hank, dyke, or levde beside a river 
or canal/ This derivation seems more satisfactory than 
those proposed from the verbs ai>aS6<u or av0ea>. 

In old Gennan a < is preferred to a ci ; and besides 
this we find a substitution of weak vowels for the a, 
as in ind-, more commonly int, sometimes in ; but 
old Saxon ant, middle German ent or en, modem 
German ent, Dutch ont, old Frisian oTid, ont, on, as 
well as and, ant, und ; Danish and Swedish und. 
Lastly, in Anglo-Saxon we find, what might be ex- 
pected in a language to which a great variety of immi- 
grants contributed, not only on but o«, cBt, and ed. 
The evidence about to be given is drawn chiefly from 
Grimm ; but it is right to observe that this scholar con- 
nects these prefixes for the most part with the Greek 
aim (not ava), moved thereto in some measure by the 
appearance of the d ov t in so many of the fonns. 
But this seems to be a very insufficient basis for his 
argument ; and the meanings of ava are far more 
suitable in the cases where both afford a tolerably satis- 
factory explanation, while in many the notion of avn 
utterly fails. Thus we find the Gothic anda~ha/thts, 
‘ransom,’ anda-set-s, German ‘ ent-setzlich,' anda- 
stathjis, ‘ adversary and-hindan, ‘ ent-binden,’ ‘ un- 
bind,’ and-hamon, ‘ ent-kleiden,’ and-huljan = ‘ ent- 
hullen,’ and-hruskan = ‘ unter-suchen,’ and-kvithan 
{kvith = our quoth), ‘ent-sagen/ and-letnan, ‘ent-lasseu 
werden,’ and-standan, ‘ resist,’ and-thaggf^an sik, 
‘ ent-sinnen sich,’ and-vasjan, ‘ ent-kleiden.’ 

Having thus paid the Gothic, what is due to it as 
the oldest record of the Teutonic languages, the com- 


Digiiized by Google 



THE GREEK PREPOSITION ava. 


25 


pliment of a separate consideration, I proceed to the 
allied dialects ; but for brevity will mass the evidence, 
taking for my guidance the series of meanings ; and 
in the quotations I shall not unfrequently attach the 
Latin equivalent as supplied by Grimm. 

1. The idea of ‘ up’ is visible in Old Germ, int-hahin, 

‘ sustinere, suflfulcire,’ int-hefan ‘ sustentare,’ in-rihten 
‘erigere,’ in-bWum ‘inflari (be puffed up), turgere;’ 
Old Sax. ant-hebbian ‘ sustinere Ang.-Sax. on-bldvan 
‘ inflare,’ on-hebban ‘ elevare,’ on-hr^ixm ‘ incitare 
(rouse up),’ on-standan ‘ adstare (stand up),’ on-stellan 
‘ incitare,’ on-vacan ‘ expcrgisci (wake up) ; ’ Mid. 
Germ, ent-haben ‘sustinere,’ ent-springen ‘oriri,’ ent- 
«;^en = ‘aufstreben;’ Mod. Germ, ent-stehen ‘arise, 
originate,’ &c. 

2. As we found among the Greek compounds with ava 
many verbs of ‘ flaming up or taking fire,’ so also here 
we have Old Germ, in-liuhtan ‘ iUuminare,’ in-prehtan 
‘ illucescere,' int-prennan ‘ accendere,’ in-prinnan ‘ ex- 
ardescere,’ in-sctnan ‘ illustrare,’ in-zundan ‘incendere;’ 
Mid. Germ, en-blcezen and en-brennen ‘ accendere,’ en- 
hrinnen ‘ accendi,’ en-p/engen ‘ accendere ;’ Ang.-Sax. 
on-dlan ‘ accendere,’ on-bernan ‘ accendere,’ on-tyndan 
‘ accendere / Mod. Germ. ent-Jlammen, ent-glimmen, 
ent-^ilnden, &c. 

3. But if the two classes, which have just been 
given, repudiate aU connexion with avn, and favour 
the cause of ava, still stronger evidence in support of 
ava is found in the extensive series of words, where 
the prefix carries with it the peculiar power of ‘ re- 
versing’ the action of the simple verb. An enumera- 
tion would be idle. The verbs of this class constitute 
the great bulk of Grimm’s third division, yet he has 


Digitized by Google 



2G ON THE REPRESENTATIVES OF 

given but a small fraction of the whole, for the Modern 
German contains a full hundred examples of such com- 
pounds with ent-, the Dutch lexicons contain at least 
a hundred and fifty such compounds with ont- ; and our 
own language might furnish a rich supply, as untie, 
unbind, unloose, &c. In confirmation of the view that 
this sense of reversing a previous act naturally associates 
itself with the idea of ‘ up,’ I may observe that the 
German and Swedish languages at times avail them- 
selves of the prepositions, which in form as well as in 
sense correspond to our own up, in the formation of 
such verbs, for example, mif-decken and upp-tdcka 
‘ to uncover,’ auf-losen and upp-lbsa ‘ to unloose.’ 

I may here be permitted to draw attention to a pre- 
valent error among our own writers on grammar, who 
assume, it must be confessed very naturally, that un- 
as used before verbs {unbind, &c.) is identical with 
un- as used before adjectives and participles {unwise, 
unseen). Grimm has carefully noticed the distinction 
(p. 816) ; but the error still stands in Thorpe’s trans- 
lation of Kask’s Ang.-Sax. Grammar and elsewhere. 
The evidence to the fact that the prefix un in verbs 
and the prefix un in adjectives and perfect participles 
are wholly unconnected, consists of two parts. In the 
first place the meanings differ. The un before adjec- 
tives is, for the most part,^ a simple unqualified nega- 
tive. Thus unwise, unseen, are no more and no less 
than ‘not wise,’ ‘not seen;’ the Latin indicta caussa 
is ‘ caussa non dicta.’ On the other hand, to unjix is 
a positive act ; the loosening of that which was pre- 
viously fixed. Had the English language possessed 


* Sup, however, the paper on ae ‘not.’ 


Digitized by Google 



THE OIIEEK PREPOSITION ava. 


27 


the verb to unknov), like the Latin ignosc-ere and the 
Breton an-kouna, ‘not to know’ would have been a 
mistranslation ; it should have been ‘ to forget,' a word 
of different import ; for although he who forgets is now 
in the position of one who does not know, yet the 
expression carries with it a distinct reference to a 
knowledge once possessed. Still more clearly does the 
difference in the power of the prefix come out, when 
we regard such verbs as unloosen, amrij/teti/, which 
cannot for a moment be held to be equivalents for 
‘ not to loosen,’ ‘ not to melt.’ At the same time it is 
true that now and then the two prefixes may lead to a 
common result Thus our own to unman is a pretty 
correct translation of avavhpo-eiv, and yet this Greek 
verb is derived from the adjective av-avSp-o- ‘ unmanly.’ 
So much for the distinction of sense. The difference 
of form is best seen in a table : 

Eng. un before adj. = 

Grmk. WtUK Gcik. Old Sax. OldGtrn^ (Term. Anf.-Sax. Dutdi. Dan, Sv9d. 

Of I an I un I un I un I un I uu I on I u | o. 

Eng. un before verbs = 

ara I ad I and | and | int | ent | on | ont | und | und. 

Thus English and Latin stand almost alone in 

confovmding the two prefixes imder an identity of 
form. 

4. In p. 813 Grimm dwells at some length on the 
fact ttat the compounds with our prefix often denote 
an incipient sense, as Germ, ent-schlafen ‘ to fall asleep,’ 
Ang.-Sax. on-drcedan ’ to shudder.’ This sense, pecu- 
liar as it is, was marked in the compounds of ava (§7). 
To this division of course belong the large family of 
German compounds with an-, of which I have already 
given abundant instances (p. 23). 


Digitized by Google 



28 


ON THE REPRESENTATIVES OF 


5. The notion of ‘escaping, driving back or off, 
away,’ is also common to ava and the Teutonic pre- 
fixes. Thus I find in Grimm’s list : Old Germ. 
int-ldzan, Ang.-Sax. on-lcBtan ‘ to let off,’ Old Germ. 
int-cdn ‘ evadere,’ ind-^innan ‘ effugere,’ int-sUfen 
‘ elabi,’ in-slingen ‘ evadere,’ int-sagSn and int-rahhdn 
‘ excusare,’ int-fallan ' elabi,’ in-pharan ‘ dilabi,’ in- 
Jliohan ‘ effugere ; ’ and the list might easily be ex- 
tended from existing German languages. 

6. ‘ Opening ’ is a sense found in : Old Germ. 
in-brestan ‘ rumpi,’ in-kinnan ‘ aperire,’ in-geinen 
‘ findere.’ 

7. For ‘again,’ the evidence of the Ang.-Sax. on- 
cndvan ‘know again, recognise,’ would bo most 
valuable even if it stood alone ; but the already 
quoted Gothic and-thaggigan, translated by Grimm 
‘ cognoscero,’ Old Sax. ant-kennjan, translated by 
him ‘ intelligere,’ should probably go with it ; and 
at any rate the modern German ent-sinnen sich ‘ re- 
mind oneself, remember.’ 

8. And this brings me to a special consideration of 
other verbs which Grimm translates by ‘ intelligere.’ 
The notion of mental perception is very commonly 
expressed in language by words which when analysed 
literally signify ‘ take up.’ Thus we often hear such 
a phrase as : ‘ Did you pick up anything at the lec- 
ture V for those who unite attention to fair ability, 
seize what they hear, and make it their own, while 
the stupid or inattentive let the words fall unnoticed. 
Hence the Latin phrase non me praeterit, ‘ it does 
not escape me.’ So again the Scotch have the ex- 
pression gleg at the uptake for ‘ quick of apprehen- 
sion.’ The word which has just been written shows 


Digitized by Google 



THE GREEK PREPOSITION ava. 


29 


that the Latin apprehendere was with reason included 
in the list where ora simulates the form of ad, the 
more so as this interpretation equally suits that other 
use of apprehendere, viz. ‘ to take up or apprehend 
in the sense of arresting a prisoner and of course 
with apprehendere must go the French verb ap- 
prendre ' to learn.’ To this head belong also the 
Old Germ, in-k^n ‘ cognoscere,’ and its represen- 
tative the Ang.-Sax. onrgetan or on-gitan, ‘intelli- 
gere/ with its subst. and-g'et or and-git ‘ intellectus,’ 
and adj. andgitol ‘ intelligibilis.' As for the Gothic 
verb and-standan ‘ resistere,’ Old Sax. and-standan 
' intelligere,’ Old Germ, in-stantan ‘ inteUigere,’ 
Modem Germ, ent-stehen ‘arise,’ various as their 
powers are, they all admit of satisfactory explanation 
if we start from the notion of ‘standing up.’ To 
stand up in spite of difficulties well calculated to 
weigh down the weak, or in other words ‘not to 
succumb,’ is a notion which the Gothic and-standan 
shares with the Latin sub-sistere. The same meta- 
phor applied to the mind gives us the idea, ‘ to be 
equal to a mental task, parem esse negotiis, to be 
strong enough for one’s place, to understand one’s 
work.’ Lastly, the German ent-stehen, Dutch ont- 
staan ‘ to arise,’ express the action, not the mere 
state of ‘ standing up.’ 

9. The verbs which carry with them the idea of 
‘ beginning or undertaking ’ have frequently an iden- 
tical origin with that which in the last paragraph was 
assigned to verbs of perception. ‘ To take a thing .up,’ 
‘ to take a thing in hand,’ are phrases with ourselves 
for ‘ beginning ; ’ and ‘ to take a thing upon one,’ 
means ‘ to take the responsibility of an undertaking.’ 


Digitized by Google 



30 


ON THE REPRESENTATIVES OF 


The Latin susci]>ere acquires its notion of ‘ under- 
taking ’ in this way, and hence it is well calculated to 
translate so many of the compounds in Grimm’s list : 
Goth, and-niman ‘ suscijjere,’ Old Germ, en-newwm, 
Old Germ, int-fdhan,^ Old Sax. ant-fdhan ‘ suscipere,' 
corresponding to Ang.-Sax. on-fangan or contracted 
on-fon ‘ undertake,’ Germ, an-fangen ‘ to begin,’ Old 
Germ, in-kinnan, Ang.-Sax. on-ginnan^ ‘ incipere.’ 

I may here observe that Grimm seems to have in- 
cluded in his lists not a few verbs which belong tp 
compounds with an ‘ on ’ or ‘ to,’ and its representa- 
tives, especially in the Ang.-Sax. series, as on-clifjan 
‘ adhaerere,’ on-feallan ‘ incidere,’ on-iman ‘ incurrere,’ 
on-settan ‘ imponere.’ On the other hand, by a most 
unsatisfactory compensation, in his list of compounds 
with the Gothic arux = our on, there are some which 
must be claimed as compounds with aiia = ava ‘ up,’ 
viz : ana-fang ‘ initium,’ anorsaga ‘ objectio.’ 

In dealing with the German compounds I have 
passed over three which have an initial emp- before an 
f emp-fehlen, emp-fangen, emp-fnden. In the first 
we have a deceitful form, corrupted, as it seems to me, 
from an-hefehlen ‘ to recommend to.’ The argument 
for this lies in the Dutch and Danish forms of the 
word, viz. aan-hevelen and anrhevale. But in the 
others, emp- is but a modification of ent, caused by the 
following lip-letter. The Old Germ, int-fdhan ‘ susci- 
pere,’ and int-fndan ‘ sentire,’ give bail for emp-fangen 
and emp-Jinden ; and the precise meaning of the latter 
was probably ‘ all at once to become sensible of,’ for 

* Hutpicere is merely a misprint in Grimm. 

* Ginnan, the simple verb, is obsolete. Grimm holds that its 
sense must have been ‘ caperc, complocti’ (p. 811). 


Digitized by Google 



THE GREEK PREPOSITION ava. 


31 


‘ to feci ’ is an older meaning of the verb Jinden than 
‘ to find,’ in our English sense. The Scotch indeed 
still possess the verb with this power. ‘ You don’t 
mind what I say,’ says the angry mother to her boy ; 
and giving him a smart box on the ear, she adds, 

‘ D’ye find that.’^ It is easy to see how from the idea 
of feeling that of discovery or finding would arise. 

As regards the Anglo-Saxon, the quotations in 
which the senses of ‘ again ’ and ‘ away ’ have entered, 
have been few. I might indeed make some addition 
to the list» and still more would it be easy to add to 
those in which the idea of ‘ up ’ appears ; but after all, 
the sense of ‘reversing a previous act’ is the one 
which the prefix on = ava usually carries with it. 
’This onesidedness in the Anglo-Saxon preposition 
seems to admit of the following explanation. When 
a word has established itself in several dialectic 
varieties of form, it is a great convenience to distri- 
bute any varieties of meaning which may belong to 
the parent word between them ; and thus a disso- 
lution of partnership as it were takes place, each 
dialectic variety commencing business on its own 
account with its own separate stock. In this way the 
Greek ava, I have said, is represented in Anglo-Saxon 
by four particles, on- which we have already seen, oe-, 
<Bt-, and ed-. Kask, in his Grammar (§ 33), has 
noticed the peculiarity in this language by which the 
aspirate 8 supplants the nasal sounds nn and nd. It 
is probably on this principle that we must account for 
the appearance of the sulEx ax in the plural of the 
present indicative and imperative, while in the other 


* See ftlso JsimicsoTi's IMotionnry. 


Digitized by Google 



32 


ON THE REPBESENTATIVE3 OF 


tenses we have the suf&x en or on (Rask’s Gr. Trans, 
p. 88). I therefore readily assent to this writer (p. 99) 
and to Dr. Bosworth {sub voce), when they tell us that 
015-' represents the German ent- ‘away, from,’ as 
o“-Jle6n = ent-fliehen ‘ flee away,’ oit-gangan — ent- 
gehen ‘ escape,’ ots-sagan = ent-sagen ‘ renounce,’ ots- 
feallan == ent-fallen ‘ fall away.’ But in lieu of this 
o»- we also find at- as a prefix of the same power, and 
probably but a dialectic variety, for the term Anglo- 
Saxon seems to have been applied somewhat vaguely 
to all the variety of Saxon dialects that were spoken 
in this island in early times, although the immigrants 
were supplied from all the coasts between Norway and 
the Zuyder Zee. In the present case there is the 
awkward fact that the language also possessed a pre- 
position at — ‘ to.’ Dr. Bosworth indeed regards the 
two particles as but one, and would explain the change 
of meaning from the idea of ‘ to ’ to that of ‘ from,’ 
on the principle that ‘ you approach a person or thing, 
when you wish to take something away.’ This seems 
unsatisfactory. Examples of at- signifying ‘away,’ 
are at-jie6gan= ent-jliegen ‘fly away;’ at-hledpan = 
ent-laufen ‘run away;’ at-sacan — ent-sagen ‘re- 


^ Os- seems to be a corruption of some such syllable as unS- or 
und- (compare the Ang-Sax. Ufi ‘ tooth,' and the Gothic tunthus, 
Grimm, it 907, muS ‘mouth,’ and Germ, muiul ) ; and at- perhaps 
represents immediately the German mi-, the long vowel compen- 
sating for the disappearance of the liquid. But still ultimately all 
the four little particles are of one origin. As Grimm would dis- 
tinguish between the Gothic prefixes and- and uncL-, so again, in 
p. 715, ho warns his reader against confounding the Ang.-Sax. <m- 
and oS-; yet he himself identilics the Ang.-Sax. on- vith tlie Germ. 
ent - ; and the examples alwvo given are surely suiTicient to identify 
o3- with ent-. 


Digitized by Google 



THE GREEK PREPOSITION apa, 33 

nounce.’ On the other hand, the form ed- is reserved 
for the sense of ‘ again,’ as ed-niwian ‘ to renew,’ ed- 
ledn ‘ to recompense,’ ed-cenning ‘ regeneration.’ Here 
we have, as Rask has remarked, a representative of the 
Kymric or Welsh ad-. Indeed it may be assumed- 
that the form ad- or ed- in the sense of ‘ again,’ parted 
company from the other representatives of apa at an 
early stage of the Indo-European language, so that it 
appears with little variety of form in the Latin, Welsh, 
and Anglo-Saxon. In the Old German too it has its 
distinct representative, though with more considerable 
change, in ita or it (Grimm, p. 757), the vowel being 
such as the above-quoted int- for a?id- and ent- would 
have suggested, and the tenuis t also, as usual in that 
dialect, superseding a medial. It will subsequently be 
seen that derivatives from this ad- or ed- again hold 
themselves somewhat aloof from the other representa- 
tives of apa. 

But I am strongly impressed with the belief that 
the Anglo-Saxon possesses yet another variety of our 
prefix, viz. a- as a corruption of on-, and this the more 
because we find in our modem language instances 
where our ordinary preposition on has been reduced to 
this vowel, as a-foot, Orboard, for on foot, on hoard. The 
adverb a-long, when compared to the Germ, ent-laiig and 
Ang.-Sax. ant-lang, is even a stronger instance of such 
cormption ; but I would rather rely on a perusal of the 
following verbs, which are but a selection from many in- 
stances of a similar kind in Dr. Bosworth’s Dictionary. 

1, Up : a-hebban ‘ lift up,’ a-hreran ‘ raise up,’ a- 
springan ‘ spring up,’ a-timbrian ‘ erect a building,’ 
orioeallan ‘bubble up,’ a-wacan ‘awake,’ a-lichtan 
‘ enlighten,’ a-teiidan ‘ set bn fire.’ 

D 


Digitized by Google 



34 


ON THE REPRESENTATIVES OP 


2. Back : a-hugan ‘ redeem,’ a-cerran ‘ return,' 
a-cw&san ‘ answer,’ a-gefan ‘ give back,’ a-gildan 
‘ repay.’ 

3. Again: a-cucian ‘revive,’ a-gitan ‘know’(‘re- 
.cognise ?’). 

4. Keversal of what the following word denotes, un- : 
a-Jiran ‘emasculate,’ a-le6sian ‘dismember,’ a-scea- 
Ikm ‘shell’ (i.e. ‘ unshell’), a-mansumian ‘ unmarry.’ 

5. Eeversal of a previous act, un- : a-lysan ‘ let 
loose ’ (‘ unloose ?’), a-slackian ‘ slacken,’ a-barian 
‘ make bare,’ a-f6lian ‘ putrefy.’ 

6. Beginning: a-grfnnaw ‘ begin.’ 

7. Eemoval, away : a-carran ‘ avert,’ a-drifan 
‘ drive away,’ orfaran ‘ depart,’ a-lddean ‘excuse.’ 

Before I leave this branch of our subject, I may 
observe that, as Grimm led me to expect, our particle 
has been found to run through the whole gamut of 
vowels, Goth, and-, Old Germ, int-, Modem Germ. 
e7it~, Dutch ont-, Danish tend-. We have also seen 
it written with a single nasal consonant, Greek ava 
and German an- {an-fang, &c.). Old Germ, in-. Mid. 
Germ, en-, Ang.-Sax. on-, and Eng. un-. Further, we 
have seen it reduced to a mere a- in Anglo-Saxon ; 
and our language has still examples in Or-u'oke and 
acknowledge, to say nothing of Shakspere’s acknow 
in the same sense. Lastly, our verb e-lope = Germ. 
ent-laufen, or Dutch ont-loopen ‘ nm off,’ brings us 
to the extreme case of a toneless e. 

I proceed to call a fresh batch of witnesses. It is 
well known that prepositions are fond of assuming a 
certain suffix which has in great measure the form 
and probably the meaning of the comparatival suffix. 
Thus the Latin sab, prae, prope, have secondary forms 


Digitized by Google 



THE GREEK PREPOSITION ava. 35 

super, praeter, and propter. Our own aft and nigh 
lead to after and near, the Gothic uf and ni\> to ufar 
and nidar, the Old Germ, ur (= Goth, us) and hit 
(— our with) to ikar and widar. It is on this prin- 
ciple that Grimm is disposed to deduce from the 
prefix and- a theoretic aTidar- (p. 716), which, though 
not producible in Gothic, he holds to be represented 
by the Old Norse endr- (for evdir-). While he thus 
connects the prefix endr- with the family of the 
Gothic and-, he seems to regard the prefix undr- 
ip. 914) as one no way related to it. But I feel 
compelled to claim undr- as more nearly akin to and- 
than endr- itself, holding the former to be the full 
equivalent of Grimm’s theoretic andar-, while endr- 
appears to me to be for the Old Norse the comparatival 
form of the simple prefix ed- ‘ again,’ so familiar on 
Ang.-Sax. ground. It is not a very strange matter 
that languages should b? capricious in their use of 
these particles, especiallji as the comparatival form 
differs little, if at all, in practical use from the simple 
particle. Thus the Romans abstain from using ad 
‘again’ as an adverb, employing for this object the 
secondary form iterum (comp, the Old Germ, it or 
ita ‘ again’). A final medial in Latin was probably 
pronoimced as a tenuis (comp, ah, oh, suh, with the 
Greek airo, viro, ewt; and with the derivatives from 
sub itself). Hence ad was probably spoken as at, so 
that iterum is entitled to a Again, the Ang.-Saxon 
has a simple prefix ed- ‘again,’ but seems to have 
avoided the formation of a comparative. On the 
other hand, the Norse endr-, Danish alter, Swedish 
Ater, all signifying ‘ again,’ have at home no positive 
to which they may be referred. But while the words 

D 2 


Digitized by Google 



36 


ON THE REPRESENTATIVES OF 


just enumerated all agree in the limitation of their 
power to the one idea of ‘ again/ undr- gives to the 
verbs connected with it meanings of various kinds, 
but amid that variety only such as will flow from the 
idea of ‘up indeed, one half of them are by Grimm 
himself regarded as equivalents of Latin compounds 
with sub. 

But the prefix undr-, or, as Haldorson writes it, 
xmdir-, seems to be identical with the Ang.-Sax., 
Danish, and Swedish under-, as also with the German 
unter- and Dutch onder-. The forms justify the 
assumption that they are only comparatival exten- 
sions of the prefixes we have been considering in 
the preceding pages. Thus the Danish and Swedish 
und-er- stands accurately in the required relation to 
und- ; and nearly so the German tint-er- to ent-, the 
Dutch ond-er- to ont- ; and even the Ang;-Sax. 
und-er- differs in no intolerable degree from the 
simple prefix on-. But if the forms be favourable, 
not less so are the meanings, which the disyllabic 
prefixes give to verbs in composition. The argu- 
ments, if stated at length, would be for the most 
part a repetition of what has been said in discussing 
the simple prefixes ; and the very variety of powers 
which will be found to belong to unter-, &c. wUl only 
strengthen the position, when it appears that this 
variety is in nearly every element the counterpart of 
what has been seen in the compounds with ent-, &c. 
In the German, unter-halten signifies ‘ to sustain, to 
support, to entertain, to keep up,’ the last in all the 
varieties of its use, ‘ to keep up a friendship, a corre- 
spondence, a building, a fire / comp, ap-ex-. Unter- 
nehmen und unter-ziehen ‘ to undertake,' including 


Digitized by Google 



THE GREEK PREPOSITION ava. 


37 


the very word by which I have translated them, 
possess a meaning which has been already seen and 
considered in the Gothic and-niman and Old Germ. 
en-neman (p. 30). Unter-fangen {sich) ‘ to take upon 
oneself, to presume,’ is substantially explained in the 
same place. Unter-stehen (sich) ‘ to be so bold,’ 
brings to mind what was said of the Gothic and- 
standan, to which it is immediately related in both 
elements ; and similarly our own understand is in 
agreement with the Old Saxon andstandan and Old 
Germ, indstantan ‘ intelligere.’ Further, we have 
unterstiitzen ‘ to prop up,’ unter-wUhlen ‘ to grub or 
rummage up’ (like a hog), unter-heilen ‘ to wedge up,’ 
‘raise by wedges;’ unter-bauen, unter-mauem, ‘to 
support an object by building a wall, &c. up to it.’ 
Unter-suchen ‘ to search up to the sources,’ has in its 
prefix the same power that ara has in ava-Kpiv-, &c. 
Unter-richten and unter-weisen, ‘ to instruct,’ may 
well be classed with the numerous verbs of ‘educa- 
tion,’ which owe their power largely to the notion of 
‘ up,’ as bring up, educate, rear, edify, instruct, train 
up, instituere, innutrire, alumnus (fi:om al-ere ‘to 
raise ’). Another power of the Greek ava and German 
ent- shows itself in unter-lassen ‘to leave off.’ Tlie 
idea thus expressed by the fuller prefix is not far 
remote from what belongs to the German ent-lassen or 
Dutch ont-laten ‘ to let off, to release,’ while it pre- 
cisely agrees with what we see in the Danish und-lade 
‘to leave off’ Untersagen einem etvMS ‘to forbid, 
to interdict,’ and ent-sagen {einer sache) ‘ to renounce 
a thing,’ or its equivalent in form, the Ang.-Sax. on- 
sacan ‘ to refuse,’ all agree in expressing a prohibitory 
injunction, and the prohibitory portion of the idea 


Digitized by Google 



38 


ON THE REPRESENTATIVES OF 


must reside in the prefixes. Again, such verbs as 
unter-arbeiten, -grahen, -hohlen, -minen, -spiilen, 
-waschen, speak of an action directed from below, 
i.e. upwards. 

Further, I cannot but attach some little weight 
to the consideration that the Latin preposition 
sub, which truly represents our up in both form 
and sense, forces itself constantly upon us when we 
translate these German compounds into Latin : nay, it 
seems probable that a desire to pve a literal German 
equivalent led to the formation of some among the fol- 
lowing German verbs from the Latin: unter-drucken— 

‘ supprimere,' «n<er-w’e7^en=‘ subjicere,’ unter-jochen 
= ‘ subjugare,’ «n<er-scArei6en=‘subscribere,’ unter- 
siegeln and wn«er-zeic/men=‘subsignare,’ unter-eitem 
and ««<er-scA?caren=‘ suppurare.' Unter-bleiben ‘to 
remain behind,’ expresses the same notion as the 
Greek vvo-Xeitreffffat and the Latin re-manere, and the 
prefixes of these two verbs are in agreement with 
the power of ava. Unter-mischen and unter-mengen 
I would rather translate by the vernacular, ‘ to mix 
up,’ than by ‘intermix,’ for here also is found the 
idea of upward movement, as in ava-yttury-, apa-if>vp. 

But while I have been thus enumerating a long 
series of German compounds with unter-, I have 
probably exposed the theory to a suspicion of some 
weakness, by appearing to ignore that familiar pre- 
position unter-, or, as we English -wite it, under-, 
with the sense of ‘ lower.’ But in truth I have not 
lost sight of this word, nor was it my intention to 
claim as akin to ava all the instances in which the 
German vocabulary presents a compound with unter-. 
In the first place, I resign all claim to those sub- 


Digitized by Google 



THE GREEK PREPOSITION ava. 


39 


stantives which are directly formed from a simple 
substantive by the addition of this prefix, as unler- 
lehrer ‘under-teacher,’ unter-kleid ‘under-garment.’ 
Of the other substantives, I claim only such as are 
deduced from verbs in which the unter- has already 
been claimed. It is therefore solely in the region 
of the verbs that the battle between the rival prefixes 
must be fought ; but, to use a more pacific metaphor, 
it may be asked, Where is the line of demarcation 
to be drawn? Now I find a strong confirmation of 
my theory in the fact, that the compounds which I 
have been led to claim on the evidence of their 
meaning alone, turn out to belong, every one of 
them, to a natural class, and the principle of dis- 
tinction on which this class is formed had wholly 
escaped my attention when first making a collection 
of examples. It is however a familiar fact with 
German scholars, that the compounds with unter- are 
divisible into those which have a separable prefix, as 
unter-(jehen ‘to go down, sink, perish,’ whence ich 
gehe unter and unter-zu-gehen, and, secondly, those 
with an inseparable prefix, as unter-sagen ‘ to inter- 
dict,’ whence ich unter-sage, never ich sage unter, 
zu unter-sagen, not unter-zu-sagen. Further, there 
is an invariable distinction of accent, those with 
a separable prefix accentuating the prefix itself, 
Unter-gehen ‘to go down,’ the others as uniformly 
giving the accent to the verb, unter-sdgen. Thus 
we have two streams of words, which, though they 
meet in a common bed, do not mix their waters, and 
by this distinction seem to justify me in referring 
them to different sources. Now all the verbs which 
I claim possess the inseparable prefix, with the accent 


Digitized by Google 



40 


ON THE REPRESENTATIVES OF 


on the root syllable of tlie verb ; on the other hand, 
to the separable prefix and its peculiar accent is 
regularly attached the notion of ‘ down or under.’ 
My views as to the origin of this other preposition 
do not belong to the present subject ; and as I have 
enough upon my hands, I purpose to reserve them 
for subsequent consideration. It may be observed, 
however, that the compounds with my own unter- 
scem to be the older occupants of the ground. In 
the Old Norse, Grimm expresses his belief that undr- 
is always inseparable ; and at any rate it is not until 
the period of the Middle German that we meet with 
a first attempt to import the Latin inter (from in). 
This was for the purpose of creating a quasi-hybrid 
formation, which however, in obedience to the law 
that holds in the physical world under like circum- 
stances, soon died out. I allude to the use of unter 
as an equivalent to the Latin inter or French entre in 
the formation of reciprocal verbs, as sich unter-kiissen, 
&c. in evident imitation of the French sentre-haiser 
(see Grimm, ii. 878). 

The Ang. -Saxon will also yield to my wooing. Here 
I find the prefixes on- and under- unmistakeably as- 
serting their relationship to each other by the similarity 
of power which they bring with them to the simple 
verb. On-gitan is translated by Dr. Bosworth ‘ to 
know, perceive, understand,’ under-gitan ‘ to under- 
stand, know, perceive ; ’ 2. on-gynnan ‘ to begin, un- 
dertake,’ under-gynnan ‘ to begin ; ’ 3. on-secan ‘ to 
inquire,’ under-secan ‘to seek under, to inquire, to 
examine;’ 4. on-%oendan ‘to turn upon, &c. over- 
throw,’ under-wendan ‘ to turn under, to subvert ; ’ 
5. on-cerran ‘ to turn, to turn from, to invert,’ xtnder- 


Digilized by 



THE GREEK PREPOSITION ava. 


41 


cei-ran ‘ to turn under, to subvert ; ’ 6. on-fon ‘ to 
receive, take,’ under-fon ‘ to undertake.’ 

Now it is plain from the translations, — ‘ to seek 
under’ in 3, ‘to turn upon’ and ‘turn under’ in 4, 
and ‘turn under’ in 6, — that the lexicographer was 
anxious to give in the first place what he deemed a 
literal translation, and that in his endeavour to effect 
this object he was biassed by the supposition that the 
Ang.-Sax. prefixes on- and under- had the power 
which belongs to the two prepositions so written at 
the present time. In truth the words subvert and 
overthrow, for over is but a comparatival form of up, 
give strong evidence in favour of the power here 
claimed for the two Ang.-Sax. prefixes ; and thus up- 
turn or up-set would have been the simplest transla- 
tion. ‘Under-turn’ or ‘turn under’ are both rejected 
by the idiom of our language. 

In what has been said, it has been more than once 
assumed that the original meaning of the Latin svh is 
‘ up.’ But this win not obtain the ready assent of all 
scholars. Those whose matured intellect has been more 
especially devoted to the Greek language, — and this 
condition applies to the great bulk of classical scholars 
both in England and Germany, — are very apt to have 
what I must consider an enoneous bias as to the 
power of this prefix. Grimm also (iii. p. 253) puts 
forward views in which I cannot agree. His sections 
6 and 8 in that chapter seem to me to require re- 
modelling ; and I would put together as equivalent 
forms, Lat. sub, Greek xmo. Go. uf. Old and Mid. 
Germ, uf. Modem Germ, auf. Old Frisian op or up, 
Dutch op, Norse and Swedish upp, Eng. up. The 
Latin sub, as it stands superior to the rest in having 


Digitized by Google 



42 


ON THE REPRESENTATIVES OF 


preserved the initial consonant, so also exhibits the 
true meaning of the word with more clearness than its 
sister language, the Greek. Its power is well seen 
when it is employed as a prefix to verbs, and also in 
its derivatives. Thus we have suh-veh- ‘carry up' 
(see Caesar, B. G. i. 1 6), sum- (= suh-im-) ‘ take up ’ 
(opposed to dem- ‘take down’), sub-due- ‘draw up’ 
(sc. naves, opposed to deduc-), sub-leg- ‘gather up,’ 
sub-leva^ ‘ lift up,’ sub-sili- ‘ leap up,’ subsist- ‘ stand 
up,’ sub-vert- ‘ up-tum,’ sub-i- ‘ ascend,’ suc-ced- ‘ go 
up,’ suc-cing- ‘ gird up,’ sub-veni-, succurr- ‘ come up 
or run up to a person’s support,’ suc-cuti- ‘ toss up,’ 
suf-fer- ‘ bear up, sustain,’ suf-Jicit the opposite to de- 
Jicit, suf-Jla- ‘ blow up,’ suf-fulci- ‘ prop up,’ sug-ger- 
‘ heap up,’ sup-pie- ‘ fill up,’ surg- (= sur-rig-) ‘ rise 
up,’ with sub-iig- ‘raise up,’ sus-dp- ‘take up,’ sus- 
citor ‘ rouse up,’ sus-pend- ‘ hang up,’ suspic- ‘ look 
up,’ suspira- = ‘ an-hela-’, sus-tine- ‘ hold up,’ sus-toll- 
‘ raise up,’ sursum (= sub-vorsum) ‘ upward-’ 

Surely then, so far as sub is concerned, Grimm is 
not justified in the assertion “ that it is merely by the 
addition of the suffix er (as seen in super) that this 
preposition obtains its full sense of upward motion.” 
But let us look to the derivatives from sub and its 
representatives : as, superi, superior, swm/mtis, all of 
which distinctly denote ‘ elevation.’ So in Greek, to 
say nothing of iirep, we have in vvarot,^ an epithet of 


* Yet the following statement has been made : “ viraroc for 
vTrepraTos, like Lat. summus for supremns.’’ Would the sup- 
porters of such doctrines regard postumm, primut, purarot, rpiirrot, 
as contractions of postremut, priorimtu, ptotpraro^, irportpuraToe 'i 
Again, when vrarq is translated ‘ the lowest chord or note,’ it must 
be remembered that the names employed in the Greek musical 


Digitized by Google 



THE GREEK PREPOSITION ava. 


43 


Jupiter on the one hand, and on the other the ordinary- 
title in Greek writers of the Roman consul. Again, 
are not i^ot ‘ height,’ and ‘ on high,’ evidently- 
connected with our preposition? But if these instances 
be not enough, all the Teutonic languages, with the 
exception of the Gothic, conspire in supporting our 
view ; for the prepositions uf, auf, op, up and upp in 
the different branches of this family have a power too 
distinct and too invariable for any doubt. And even 
in the Gothic, though Grimm would assign ‘under' 
to the preposition as its primary sense, his own short 
list of compounds with uf (ii. 902) includes uf-haban 
‘ sustinere ’ (hold up), uf-hrinnan ‘ exardescere ’ (blaze 
up), uf-grahan ‘suffodere’ (dig up), uf-brikan ‘reji- 
cere,’ uf-hmnan ‘cognoscere’ (say rather ‘re-cogno- 
scere’), uf-vdpjan ‘exclamare,’ uf-svdgjan ‘ ingemiscere,’ 
all of which contain senses such as would be suited to 
compounds of apa, and therefore may well reside in 
compounds with another preposition signifying ‘up.’ 
But if we pass from the Gothic to the Old German, the 
e-vidence is of the clearest character. The following 
eleven verbs make up the entire list of Gri mm 
(p. 897) : ufhaben ‘ supportare,’ uf-hefan ‘ suspendere,’ 
uf-kan ‘ Burgere,’ tif-giem ‘ exiit,’ uf-kangit ‘ adolescit,’ 
uf-purgen ‘ suscitare ;’ uf-burren ‘ attollere ;’ uf-qws- 
man ‘oriri, exoriri ;’ uf-richten ‘ erigere ;’ uf-stantan 
‘ surgere ;’ uf-stikan ‘ ascendere, scandere.’ Again, 
the comparatival forms, Lat. super, Gr. inrep, Goth. 
ufar. Old Germ, upar, ubar. Mod. Germ. Uber -with 
oher as an inseparable prefix. Old Sax. obar. Old Fris. 


terminology are precisely the opposite to ours. Compare viari) 
‘ the highest note,’ though the word in itself means lowest. 


Digitized by Google 



44 


ON THK REPRESENTATIVES OF 


over or contracted ur, Dutch over, Ang.-Sax. ofer, 
Eng. over and tipper, Old Norse yfir and ofr, Swed. 
ofver, Dan. over, are not more regular in formation 
than consistent in sense. Grimm himself admits that 
they all express the idea of elevation ; but if this idea 
did not already exist in the root, how could its intro- 
duction be ' effected by the comparatival suffix ? how 
could the addition of a syllable signifying ‘ more’ or 
‘ of two’ bring about the marvellous metamorphosis of 
‘down’ to ‘up?’ To admit this would be to admit 
that after should signify before and nether above; 
and thus all language would be subverted. 

Still there remains a difficulty not to be passed over, 
in the fact that sub, vvo, and the Gothic uf often re- 
quire the translation ‘ under.’ The explanation I would 
offer is this, that movement upward is the first sense 
of sub, &c. ; but that when that movement reaches its 
limit, the body which had been moving ‘ up’ towards 
a certain object, has attained the position of being 
‘under’ it. Accordingly sub murum ire means ‘to 
go up to the wall,’ but sub muro esse ‘ to be under the 
wall’ We hang ‘ up’. a chandelier ; and the operation 
over, the chandelier is ‘ under’ the ceiling. It is there- 
fore habitual to find sub denoting ‘ under’ when com- 
pounded with verbs of rest, as sulriacere, subesse : and 
if it be also at times found with this sense in verbs of 
motion, it should be recollected that the mere verbs 
of ‘ putting,’ though as verbs of motion they should 
require the accompanying preposition to take an accu- 
sative alone, yet often allow the case of rest (abl. in 
Lat, dat. in Greek) to supplant the case of motion. 
Thus we find collocare in navi, in cubili, in custodia, 
where the strict theory of grammar would rather 


Digitized by Google 



THE GHEEK PllEPOSlTION ava. 


45 


demand an accusative, in navem, &c. In the same 
way the syntactical rule which justly admits a dative 
after verbs compounded with prepositions of rest, as 
camptis interiacet Tiberi ac moenibus Romanis, is 
extended also to verbs of mere putting, as anatum ova 
gaUinis supponimus ; and this with some reason, seeing 
that the act of putting is momentary, and the mind 
prefers to dwell on the permanent state of things which 
follows. Hence we find that submittere, though strictly 
signifying ‘to send up,’ as Terra submittit Jlores, is 
also used of ‘ putting under or down,’ especially in the 
perfect participle, where the action is over. Such a prac- 
tice is well calculated to lead to equivocal results. 
Thus submissus is ‘ upraised’ in Silius Italicus, ‘ lowered 
or low’ in Cicero and Caesar. But for the most part 
the verb which it accompanies by its own nature pre- 
vents ambiguity, as submergere. 

There is yet another point of view from which we 
are, apt to attribute to svh the idea of ‘ under.’ In 
the various processes of undermining, as by digging, 
the action of water, &c. the agent is of course below ; 
but on the other hand the action is directed upward, 
so that is stiU in its proper place. A man in a 
cave may dig downward or upward. It is only in the 
latter case that the operation can with strict propriety 
be expressed by suffodere, undermine, untergrahen. 

In Greek the use of wtto as ‘ under’ in compounded 
verbs was carried to the greater excess, because there 
lay at hand the unambiguous ava to express the notion 
of ‘up.’ But even the Greek has distinct traces of 
the original power of vwo in compounds, as viroSex°l*ai 
■ I take upon myself, undertake,’ xnrwxveoiiai the same, 
wTre^w ‘ I uphold,’ imoXanPavto ‘ I take up, apprehend,’ 


Digitized by Coogle 



46 


ON THE REPRESENTATIVES OF 


(wTToX. imrov ‘ pull up a horee,’) {xf>umifu ‘ I support an 
attack’ - subsisto. 

Lastly, when we find two meanings as here attached 
to a word, one of which implies motion, the either rest, 
it seems generally right to give a preference to the 
former, seeing that verbs of the shorter form, and for 
that reason the older, commonly denote action. Indeed, 
if the mimetic origin of language be admitted, this 
follows as a necessary consequence. 

But to leave this digression. In dealing with the 
German unterhalten there was given for one of its 
translations ‘ to entertain,' a word which in power is 
nearly equivalent to ‘ sustain.’ As sustenance is con- 
nected with the one word, so we have the idea of food 
implied in the phrase ‘ good entertainment for man 
and horse.’ Even to entertain in the sense of ‘ amus- 
ing’ is to keep up the interest and spirits of friends, 
’fake too the following passage from the “Life of Col. 
Hutchinson,” by his widow (Bohn, 1846, p. 319) : — 
“Col. Hutchinson’s cheerful and constant spirit never 
anticipated any evil with fear. His prudence wanted 
not foresight that it might come, yet his faith and 
courage entertained his hope that God would either 
prevent it or help him to bear it.” But the word enter- 
tain belongs to the Norman element of our language, 
being the representative of the French entretenir and 
the Italian intertenere. We are thus brought to the 
Latin domain, and as tenere is the precise equivalent 
in sense of the German ‘halten,’ the question arises 
whether there ean be any connexion in blood, as there 
is undoubtedly much external similarity, between the 
Latin inter (Fr. entre) and the German prefix unter 
‘ up.’ Enter-prise, entrc-prise, entre-prendre compared 


Digitized by Google 



THE GKEEK PREPOSITION ava. 47 

with unter-nehmen suggest the same inquiry, since the 
verb prendre is identical with the Latin prehendere 
or prendere. But we have also the poetical emprise, 
which conducts us in like manner to the Italian noun 
impresa and verb imprendere ‘ to undertake.' This 
verb is the more interesting as it also has the sense 
‘to learn,’ thus giving a double assurance that its 
prefix is connected with the particle ava ‘up.’ But 
besides this, I am led to assume that the Latin lan- 
guage also, some time or other, in some part of Italy, 
possessed two verbs of nearly equal import, impren- 
dere and inler-prendere, where wc have an exact 
counterpart in the prefixes to the German ent-nehmen 
and unter-nehmen. 

Invited in this decided manner to the consideration 
of the Latin compounds with inter, I find among 
them nearly all the varieties of power which ava and 
its representatives possess. At the same time the 
Latin, like the German, has also compounds with a 
second inter of distinct origin. With this admission 
I lay claim to the following : Intel-lig- ‘ to pick or 
gather up,’ and hence ‘ to perceive inter-misce- ‘ to 
mix up,’ and inter-turbo- (Plant., Ter.) ‘to stir up' 
(for the true sense of turbo-re is simply ‘ to stir,’ hence 
turbida oqua ‘muddy water’). Inter-iung- (equos, 
boves) ‘ unyoke,’ is a distinct example of inter in the 
to us uninteresting sense of reversing an act. As the 
literal meaning of iungere is rather ‘ to yoke’ than ‘ to 
join,’ this verb truly represents the German ent-jochen. 
Inter-quieso- (Cato, Cic.) ‘rest after labour’ = ava- 
trav- (r.). Inter-dic- ‘forbid,’ inter-mina- (r.) (Plant., 
Cic.) ‘ forbid by threats,’ may be placed beside ent- 
sagen ‘ to renounce’ and unter-sagen ‘ to forbid, to 


Digitized by Google 



48 


ON THE REPEESENTATIVES OF 


interdict and with the same we may perhaps class 
inter-pella-. As the German ent- often signifies 
‘ escaping, disappearance,’ so we find inter-mor- ‘ die 
off, die out, swoon away’ (Cato, Plin., Cels.) ; inter- 
neca- ‘ kill off so that none are left’ (Plaut.) ; inter- 
fnng- ‘ break off’ (Cato, 44,^ but not Pliny ns an 
independent authority, for in xvii. 18 or 30 he is only 
quoting Cato) ; inter-aresc- (Cic., Vitr.) ‘ dry up’ (comp. 
ava-^paiv-) ; inter-bib- ‘ drink up’ (Plaut.); inter-mitt-^ 

‘ leave off’ (comp, unter-lassen, Dutch ont-leten, &c.) ; 
inter-rump- ‘ break off ’ (comp, unter-brechen) ; inter- 
stinffu- (Lucr.) lit ‘ stamp out,’ ‘ extinguish ;’ inter- 
ter- ? ‘ destroy by rubbing,’ a verb not itself producible, 
but implied in its derivatives inter-tr-igon-, inter- 
tr-imento-,^ inter-tr-ilura- ; inter-cid- ‘ fall away, slip 
away, escape,’ about which there can be less doubt, 
seeing it is so frequently used of ‘ slipping out of the 
memory, being forgotten,’ and thus exhibits a pecu- 
liarity common to the German verb ent-fallen ; inter- 
frigesG- (Vat. Fragm. § 155) lit. ‘die of cold,’ and so 
‘ become obsolete or forgotten.’ This metaphor brings 
to mind such passages as ; ‘ Crimen de nummis caluit 
re recenti, nunc in caussa refrixit,’ Cic. p. Plane. ; ‘ illi 
rumores Cumarum tenus caluerunt, Cael. ad Cic. For 
a time a word is warm with life, in the end it dies of 


’ Speaking of tlie boughs of the olive-tree. So Ovid has ‘ in- 
fringere lilio,’ Cic. ‘ infriiigero florem dignitatis,' while Iloinsius and 
Bentley would read in Horace ‘ teneros caulee alieni inficegerit horti.’ 
All this seems to prove that infringtre has an in = avn. 

’ Intermittere ignem ‘ to let the fire out,’ Cato. 

® The Bemhine Scholiast, quoted by Faemus ad Ter. Haut. iii. 
1, 39, saw part of the truth, when he wrote : 'Inter et De tantunden- 
significant, ad augmeutum ostondendum. lliuc dicitur inlerfectui.’ 


Digitized by Google 



THE GREEK PREPOSITION ava. 


49 


coldness and neglect. Inter-im- ' take oflF/ i.e. * kill’ 
(comp, av-aipe- and ab-sum-) ; inter-Jic- ‘ make away 
with/ ‘ put out of the way/ t.e. ‘kUl inter-i- 'pass 
away/ t.e. ‘ die also the expressions, ‘ he is gone,’ 
‘ decessit.’ About the Latin verb interi-re I had for a 
time much doubt, which was raised by a consideration 
of the German unter-gehen, lit. ‘ to go down, sink,’ 
and hence applied to the ‘ setting of the sun,’ &c. and 
by an easy metaphor to ‘ dying.’ Had the Bomans 
ever used inter-ire as they do occidere of the ‘ sun 
going down,’ I should scarcely have doubted that it 
attained the sense of dying in tliis way ; and then I 
must have admitted its substantial identity with the 
verb untergeh^n. But this German verb has a separ- 
able prefix with the accent on it, so that I could lay 
no claim to it 

In this enumeration I have omitted many com- 
pounds with inter, though fully satisfied that they 
belong to our preposition, as inter-clud- ‘shut off,’ 
inter-nosc- ‘know one from another’ = Sta-yt7y«B<rK-, 
inter-sepi- ‘ fence off,’ inter-cid- ‘ cut off,’ inter-vert- 
‘ divert’ inter-pung- ‘ point off or separate by a point ;’ 
inter-scind- ‘ cut off.’ At the same time I feel that 
such words admit of an interpretation by means of 
the ordinary inter, so that they should rather wait for 
a decision upon the words previously quoted than be 
adduced in proof of my doctrine. The same argument 
applies to many German verbs, as unter-scheiden. Still 
I am satisfied that the inseparable unter is always a 
secondary form of the German ent and the Greek ava. 

The sense of ‘ again/ so familiar in compounds with 
the Greek ava, serves also to explain the strange verb 
inter-polare, and the adjective inter-polus (or inter- 

E 


Digitized by Google 



50 ON THE REPBESENTATIVES OP 

polls), from which it ia evidently derived. Forcellini 
is no doubt right when he saya, ‘ proprium artia fuUo- 
nicae,' which ia fully aupported by the phraae, togam 
praetextam qiiotannis interpolare, Cic. — and probably 
he ia alao right in connecting it with polire, for thia 
word also belongs to the same business, being the 
equivalent in form and meaning of our own verb to 
full (cloth). Thus inter-polus, strictly used, should 
signify, ‘ fulled anew,’ and accordingly it ia so used by 
Cicero’s friend, the lawyer Trebatius : ‘ Si vestimenta 
interpola pro novis emerit’ (Dig. xviii. i. 45). Again 
when Cicero (in Verr. ii. 1, 61) uses the word of one 
who having made an erasure in his books subsequently 
polishes up the rough surface in order to hide the fact 
of erasure and substitute of new words, the verb in itself 
denotes only the repolishing, and not the interposition 
of new matter. It is only in later times that the 
notion of inter ‘ between,’ was able to bias the inter- 
pretation. In Plautus, at any rate, the word, used 
metaphorically, is simply ‘ to vamp up anew, to fur- 
bish up old things and give them a new shape.’ Pliny 
perhaps may have felt the wrong bias when he uses 
the word miscetur in the passage about the plant 
broom (spartum) : ‘ Est quidem eius natura interpolis, 
rursusque quamlibeat (or quamlibet) vetustum novo 
miscetur.’ 

I next quote inter^oga-. This word is commonly 
translated ‘ to ask,' but this is to ignore the prefix ; a 
neglect the less pardonable, as no family of words 
exhibit in their prefixes a more distinct power than 
the other compounds of roga-, e-roga-, pro-roga-, in- 
rogor, suh-roga-, ob-roga-, ab-roga-, ar-roga-, de-roga-, 
prae-roga-tiva. The present theory on the other hand 


Digitized by Google 



THE GREEK PREPOSITION ava. 


51 


secures to this inter a very clear meaning of its own, 
if we class it with such verbs as ava-Kpip-, unter-suchen. 
Nay, we find its representative in the Old German int- 
phrag-en ‘requircre,' where the int is in immediate 
relation to int-er ; and the German verb frag-en has 
probably the same root as roga-re. Moreover the 
meaning thus claimed for interrogor exactly accords 
with its use in legal language, viz. the searching exami- 
nation of witnesses and suspected persons. See the 
Digests, Livy and Tacitus ; and Forcellini, sub v. inter- 
rogatio. Among the Bomans legal terms often passed 
into the language of common life, and of course with 
much carelessness, so that interrogor is often found 
usurping the place of the simple verb. Inter-vis~ 
(Plant.) admits of similar explanation. As vis- means 
‘ go and see,’ so inter-vis- means ‘ go and himt up, 
go and see thoroughly into.’ The idea of ‘through,’ 
which is expressed by the prefix of ava-ri.rpa-, is often 
found with inter in Lucretius, as inter-fod- ‘dig a 
passage through’ (iv. 716), inter-fug- ‘fly through’ 
(vL 332), and inter-datus ‘distributed through’ (iv. 
868). For the last compare avarBiSeapi. So also inter- 
spiror ‘breathe through’ (Cato), inter-luce- ‘shine 
through’ (Verg.), inter-lucor ‘let the light through’ 
(Plin.), inter-fulge- ‘ shine through’ (Liv.). 

Having thus been brought back to the region of the 
Latin language, and endeavoured to re-estabbsh the 
long-ejected inter ‘ up,’ &c. in the possession of its 
rights, one is naturally led to cast an eye back to what 
has been said of Latin prefixes in the earlier part 
of this inquiry ; and the retrospect will repay us in 
some measure for the trouble. If my views have been 
right, it follows that our prefix inter- is but a compara- 

E 2 


Digitized by Google 



52 


ON THE REPRESENTATIVES OP 


tive of the prefix ad- or in- ‘ up,' and thus their com- 
pounds may possibly exhibit instances parallel to the 
Anglo-Saxon on-gitan ‘ to perceive,’ under-gitan ‘ to 
perceive on-secan ‘ to inquire,’ under-secan ‘ to 
inquire;’ German ‘to let off,’ unter-lasscn 

‘ to leave off ent-sagen ‘ to renounce,’ unter-sagen ' to 
interdict where, as the forms differ solely in the 
unimportant addition of a comparatival suffix, so the 
meanings are nearly identical. 

Such are found in Latin also. I refer not merely 
to the theoretic verbs imprendere and interprenderc, 
to which our English nouns emprise and enterprise 
conducted me, but to pairs of words well established 
in the Latin vocabulary : ad-misce- ‘ mix up,’ in- 
ier-misce- ‘ mix up;’ ac<juiesc- ‘ rest after labour,’ inter- 
quiesc- ‘ rest after labour ;’ ad-aresc- ‘ dry up,’ inter- 
aresc- ‘dry up ;’ ad-bib- ‘drink up,’ inter-bib- ‘drink 
up;’ ad-im- ‘take away;’ inter-im- ‘take away;’^ 
and perhaps also to in-cid- ‘ cut off,’ inter-cid- ‘ cut 
off ;’ in-fring- ‘ break off,’ inter-fiing- ‘ break off.’ 

It will have been observed that the instances of 
compounds with inter have been drawn in a great 
measure from the older writers, — Cato, Plautus, and 
Lucretius. This is to be accounted for on the reason- 
able ground that the more familiar preposition inter- 
was gradually intruding itself upon the minds of the 
Eomans to the detriment of our inter-, A preposition 
which has a separate existence, and may be used before 
nouns as well as in composition with verbs, has a great 


* Tho latter verb is only useil in tlie sense of death taking a 
person off, but even here compare Horaoe's two expressions, ‘ Mystm 
tulempium,’ and ‘ AtdrubaU interempto.’ 


Digitized by Google 



TUE (iREEK PREPOSITION oio. 


53 


advantage in such an encounter over one which occurs 
only as an inseparable prefix to verbs. Hence our 
inter gradually lost much of its vitality, so that it 
was no longer competent to form new compounds with 
it ; and those existing, one after another, disappeared. 
Under these circumstances the old authors naturally 
contain a larger supply of such compounds than those 
of later date. The same state of things exists in the 
German language, where it is now much more prac- 
ticable to establish a new compound with unter, sig- 
nifying ‘ under,’ than with the inseparable unter, which 
leaves the accent for the following syllable. 

On reviewing what has been here written, the fear 
suggests itself that the mind may revolt against a 
theory which involves the doctrine that prepositions 
of different origin and power frequently assume an 
identity of form. For example we have — 

Latin a<2 = to, Eng. another ad — ara. 

in = in or on, Eng. tn = orcu 

itiier from Lat tn infer akin to aro. 

Ang.-Sax. . . on = our on on = aya. 

under = our under under akin to ora. 


OEt = our at act = ora. 

Eng un = or privative un = ora. 

Germ, «i< in ent-zwei = in ent = aya. 


unter = our under unter akin to oro. 

on = our on an = oro. 

Nay, the Greek ava itself seems to represent two in- 
dependent particles ; for, besides the ordinary prepo- 
sition, we have something very like the Gothic ana 
(= our on and in) in such phrases as ava errofjM exetv 
' in ore habere,’ ava Bvfiov ex^tv ‘ in ammo habere,’ ava 
Tovt irpuTovs tivat ‘ in primis esse,’ examples I take 


Digitized by Google 



54 


ON THE REPRESENTATIVES OP 


from Liddell and Scott’s Lexicon, but with my own 
translation. A similar power exists in ai/a-Ko\\a- ‘ glue 
on or to,’ &c. But if such confusion he startling, an 
examination of other prepositions would lead to similar 
results. For example, the Latin di or dis, Greek Sia, 
German zer, Old Norse tor, appears in Anglo-Saxon 
and old Frisian as to, and thus encounters that other 
preposition to, which we still possess, corresponding to 
the German zu. Both are used in these languages as 
prefixes to verbs. Thus in old Frisian we have to- 
delva = * zu-graben,’ ‘ dig up (earth) and throw it 
against (an object),’ and to-delva = ‘ zer-graben,’ ‘ dig 
to pieces in Anglo-Saxon to-dalan ‘ attribuere,’ with 
to-dcslan ‘ disjungere ;’ to-weorpan ‘ adjicere,’ with to- 
weorpan ‘ disjicere ;’ to-clevan ‘ adhaerere,’ ‘ cleave to,’ 
with to-clevan ‘ diffindere,’ ‘ cleave in two.’ In this 
last example the confusion is increased by equivocal 
prefixes falling in with verbs no less equivocal. Anglo- 
Saxon scholars may perhaps be able to say whether 
there was a difference of accent to distinguish such 
verbs. Still in written prose the only security against 
error was in the context. Such a state of things must 
have been highly inconvenient ; and the struggle in 
Anglo-Saxon between the two prefixes appears to have 
ended in the utter annihilation of both sets of com- 
pounds, for we no longer possess a single verb com- 
pounded with either the one to or the other, at least 
as a prefix. Yet to — dis was still a living prefix for 
Chaucer, Shakspere, and the translators of the Bible. 

Another marked example occurs in the Irish lan- 
guage. Here two prepositions originally distinct in 
form, and directly opposite in power, de ‘ from,’ and 
do ‘ to,’ have for the most part (Kilkenny excepted) 


Digitized by Google 



THE OREEK PREPOSITION ava. 


55 


fallen into an awkward identity of form, do ; so that 
nothing but the variety of accent and the sense of 
the adjoining words are left to distinguish them (see 
Leo, Ferienschriften, 1852, p. 195). 

In the Latin language the prefixes de ' down,’ and 
di or dis ‘ in two,’ are constantly interchanging their 
forms, so that often the sense alone is a guide to the 
etymology of a compound. Nay, the poor word di- 
scribere * to distribute in writing,’ has utterly escaped 
the notice of all our lexicographers, the form describere 
‘ to copy,’ being allowed to usurp its place. Similarly, 
the prefix in (‘ not ’) of insanus is in form imdistin- 
guishable from the in of inire. Thus infectus 
represents two different words, as also invocatus, and 
according to our lexicons insepultus also, but this last 
assertion is the result of a mere blimder. 

In the same way the Greek ava and av- privative 
become one externally when prefixed to a word with 
an initial vowel, so that avurom might d priori signify 
either ‘ I render unequal,’ or ‘ I equalize again.’ 

Another fear which weighs upon me is lest it should 
be supposed that I would derive all the particles I 
have dealt with directly from the Greek ava. 'The 
habit of treating one language as deduced from another 
has been carried, I think, to a most unreasonable length. 
Sometimes we are told that the Latin is derived from 
the Greek; at another, that it is made up of two 
elements, one Greek and one Keltic. No doubt it is 
easy in such cases to produce a large stock of words 
more or less similar in the compared languages ; but 
this proves only a connexion between them, not that 
one stands in the relation of daughter to the other. 


Digitized by Google 



56 


ON THE REPRESENTATIVES OP ava. 


To call them ‘ sisters’ would be a better metaphor, 
though even this is somewhat objectionable, for in the 
life of a language there is no such breach of continuity 
as between a parent and a child. The Greek, Latin, 
Sanskrit, Keltic, and Teutonic races, not to speak of 
others, have a large amount of common property in 
language, which with small exceptions they no way 
owe to each other, but have received from their ances- 
tors. Were it possible to trace up each variety of lan- 
guage spoken by these races, we should probably see 
the similarity gradually increasing and at last merging 
in identity. 

In conclusion, I would observe that a consideration 
of the arguments put forward in this paper will show 
that they ought not to be considered as a chain, where 
weakness in one link would endanger the continuity of 
the whole, and so invalidate all the results. Their 
nature is such that they constitute rather a close net- 
work, and the presence of a rotten thread here and 
there no way threatens disunion, the adjoining meshes 
compensating for the deficiency. Or, in plainer English, 
I would request any one who may have had his doubts 
about isolated points of the argument, to ask himself 
whether these doubts are not removed by other parts 
of the paper, for each branch of the discussion has its 
bearings upon the other branches. 


Digitized by Google 



57 


II. 

ON THE PKEPOSITIONS €vt, in, AND RELATED 
WORDS. 

The inquiry into the representatives of the Greek pre- 
position ava in allied languages brought me into re- 
peated contact with the Latin preposition in, and its 
derivative inter ‘ between.’ The consideration then of 
this preposition in, and its allied forms in other lan- 
guages, may next be taken up. 

Grimm has more than once noticed the tendency of 
prepositions to appear at one time with only an initial 
vowel, at another with only a final vowel, an older 
form in his view having once possessed botL Thus, 
as he observes (D. G. iii. p. 252), the Gothic ana, 
whence the ordinary German preposition an and our 
071, takes in the Slavic languages the shape of na. 
This prefix na seems indeed to perform a double office, 
and at times to represent the Greek ava in its various 
senses of ‘ up,' &c. ; as from the Bussian dut’ ‘ to 
blow,’ nadut’ ‘to blow up,’ ‘inflate;’ from 7'uit’ ‘to 
dig,’ naruit’ ‘to dig up.’ Again in p. 254 Grimm 
throws out very doubtingly a suggestion that the 
Gothic du, Germ, zu, Eng. to, may be one in origin 
with the Gothic at, Old Germ, aj, Eng. at, and so of 
couise with the Latin ad, on the assumption that there 


Digitized by Google 



68 ON THE PREPOSITIONS m, in, 

once existed an original preposition adu. A close 
connexion in meaning, and the possession of a dental 
consonant in common, seem by themselves to be an 
insufficient foundation for such a theory ; and yet I 
believe tlie theory to be true, for the evidence wanted 
may l>e supplied, I think, from the Keltic tongues. 
In the Gaelic Grammar of the Highland Society, p. 27, 
appears the following : — 

‘ The preposition “ do” loses the o before a vowel, 
and the consonant is aspirated ; thus, “ dh’ Albainn” 
to Scotland. It is also preceded sometimes by the 
vowel a when it follows a final consonant ; as, “ dol a 
dh’ Eirin” going to Ireland. “ Do,” as has been already 
observed, often loses the d altogether, and is written 
a : as, “ dol a Dhun^idin” going to Edinburgh.' 

It will be here seen that the writer treats the a thus 
alleged to be inserted as a matter too unimportant to 
call for explanation ; but the strictness of modem phi- 
lology will not allow any such assumption of intrasive 
letters, and we may safely assume that the a was fully 
entitled to its position in the phrase, and not a mere 
euphonic insertion. If we assume an old preposition 
ado, all the three varieties above seen are explained. 
Moreover, the assumption that ado is an original tjrpe 
which suffers more or less mutilation, according as the 
particle happens to come into contact with vowels or 
consonants in the adjoining words, is in exact agree- 
ment with the fate of the preposition ag in the same 
language. The use of this preposition in the forma- 
tion of imperfect tenses in the Gaelic verb precisely 
corresponds with our own use of the equivalent par- 
ticle a (= in) for the same purpose. Thus ; — 

1. Preceded by a consonant and followed by a 


Digitized by Google 



AND RELATED WORDS. 


59 


vowel, the proposition is entire : as ‘ ta iad ag 
6iadeachd’ they are a-listening. 

2. Between two consonants ag loses the g, and is 
written a ; as, ‘ tha iad a dfeanamh’ they are a-doing. 

3. Between two vowels the a is dropped and the g 
retained ; as, ‘ ta mi 'g ^isdeachd’ / am a-listening. 

4. Preceded by a vowel and followed by a consonant, 
it is often suppressed altogether ; as, ‘ ta mi dbanamh’ 
I am a-doing. 

Indeed this very .preposition ag of the Gaelic seems 
to supply another example of the same principle, for 
we find standing beside each other ‘ ag’ at and ‘ gu’ 
to, which I am strongly disposed to regard as dedu- 
cible from a common source, agu. Nay, it is highly 
probable that this agu is but a variety of the Gothic 
ado, for the interchange of the guttural and dental 
medials is not rare in the Keltic tongues. Thus, while 
the Gaelic has a preposition gu or gus ‘ to’ or ‘ tUl,’ the 
Manx commonly writes gys, but at times replaces this 
by dys ; and, as- Leo observes, the identity of the Manx 
gys and dys is proved not merely by their identity of 
meaning, but also by the appearance of the same 
letter-change in gyn ‘without,’ and dyn ‘without’ 
(Ferienschriften, Halle, 1847). We may even go far- 
ther, for it seems not impossible that in the German 
bis ‘ till’ wo have a third variety of the initial con- 
sonant. Compare the relation which exists between 
the Latin bis and the Greek Bi» ‘ twice.’ 

What has been said in favour of a close connexion 
in form between the prepositions at and to, receives 
strong support in the equally close connexion as to 
meaning. It is true that now-a-days there are but 
few phrases in which an Englishman can indifferently 


Digitized by Google 



60 ON THE PREPOSITIONS tvi, in, 

use at and to. But that such distinctions are in origin 
quite arbitrary is proved by many argumenta It is 
considered more correet to say, ‘ I live at Oxford,’ yet 
in parts of England the preposition to has preserved 
its footing in this form of words, as ‘ I live to Ply- 
mouth.’ Tlie same variety prevails in some parts of 
the United States, where ‘ I live to Boston’ is in eom- 
mon use ; and it may be observed, that nearly all 
those terms and phrases which are supposed to be cor- 
ruptions, and of recent formation in that eountry, are 
genuine portions of the language which early emigrants 
carried out with them from the old country. I once 
heard ex-President Jefferson say that be had himself 
traeed a very large number of such peculiarities to 
their provincial site in England. Again, where we 
say at home, the German says to house {zu hause). 
But perhaps the most marked example of their equiva- 
lent use is seen in the employment of the prepositions 
before an infinitive, where the Swedish att toga and 
Danish at tage correspond to our phrase to take. In 
the present day at is commonly preferred where rest 
is implied, and to in order to denote motion. Yet we 
say, ‘ arrive at a town,’ ‘ throw a stone at a pig,’ and, 
on the other hand, ‘ he lives close to the church,’ ‘ he 
sat next to me.’ Thus we may fairly conclude that at 
and to are substantially one in sense and probably one 
in origin. 

If Grimm be right in identifying the Gothic hi, Old 
Germ, pi, hi, Mod. Germ, hey, and Eng. hy, with the 
Greek ent, then, as there can be no doubt that the 
Latin oh represents this Greek preposition, it will 
follow that our hy and the Latin oh are identical. But 


Digitized by Google 



AND RELATED WORDS. 


G1 


my doubt about the truth of the first of these proposi- 
tions prevents my assenting, as yet, to the conclusion. 

A clearer example of two prepositions concealing 
their affinity by the varied position of the consonant 
is seen in the German um ‘ round,’ and the Gaelic mu 
‘ about,’ two words closely akin, if not identical in 
sense, and the latter deduced from a fuller form, umu. 
Thus I am inclined to regard the Old German umpi, 
umbi, and Greek as secondary prepositions ; while 
the old Norse um and Latin am, as well as the German 
um, Dutch, Danish, and Swedish am, exhibit the pre- 
position in its simpler form. It is somewhat strange 
that Grimm should have failed to quote the Latin am, 
which is well seen in the compound verbs am-icio and 
am-plector, as well as in the adjective an-ceps ; and 
though the Oscan abl. amnud bo no longer regarded 
as = anno, the notion of a circle explains the forms 
annus, annulus, anus, solemnis, peremnis, while the 
interchange of m and n in these words needs no ex- 
ternal support. Even amare ‘ to love’ may first have 
signified ‘ to embrace,’ and so come eventually from 
am ‘ round.’ 

Grimm has no doubt truly explained the Swedish 
pd (and Danish paa) as an abbreviation of uppd, the 
equivalent of our upon (which also takes at times with 
us the reduced form ’pon). Thus the Greek into on 
the one hand is identical with our ‘ up,’ and on the 
other has its representative, so far as the consonant 
is concerned, in the first element of the Swedish pa. 

A similar relation probably exists between the 
English preposition of and the Geiman von, Dutch 
van. To what Grimm has said on this subject (p. 262 ) 
I would add that the form with n is not altogether 


Digitized by Google 



62 


ON THE PREPOSITIONS epi, in, 


wanting, as he says, to the English language. Our 
vulgar, but not on that account to be neglected, on, as 
used in the forms — ‘ six on us,’ ‘ two on 'em,' ‘I wasn’t 
a hurting on 'im,' for ‘ six of us,’ &c. — ^represents the 
derived preposition von or Old Germ, fona, itself repre- 
senting, as Grimm says, a fuller form af-ana, from the 
Gothic af — our off and of, the Greek airo, Lat. ab. 
But in regard to this af-ana there is no more necessity 
for holding the last letters to represent the Gothic 
ana ‘ on,’ than for assigning the same origin to the 
termination of the Gothic ilt-ana and hind-ana; so 
that Grimm’s scruple on this head seems groundless. 

The inference to be drawn from these considerations 
is, that whenever a preposition appears in a biliteral 
form, consisting of a vowel followed by a consonant, 
we should always look around for a second form in 
which the said consonant has won an initial position, 
and should also ask ourselves whether an earlier form 
of language does not present a triliteral preposition 
consisting of a consonant between two vowels. 

Now, if I understand Grimm rightly, he has com- 
mitted an error in speaking of the preposition in. 
After comparing the Gothic ana ‘ on,’ with the Greek 
ava and Slavic na, he proceeds to say that, although 
the preposition in is closely connected in signification 
with the Gothic ana, yet there is a marked external 
distinction, inasmuch as ana in its original form has 
always a final vowel, whereas in never exhibits such a 
vowel. Whatever be the case with the Gothic lan- 
guages, he should not have passed over the Homeric 
evi. In the following investigation, therefore, it will 
not be surprisuig if we find the preposition evi and 
its derivatives appearing at times with, and at times 


Digitized by Google 



AND RELATED WORDS. 


63 


without, an initial vowel ; and, indeed, already in the 
Italian nelh, mlla, nei we have an example of this. 

Again, the Greek evtpot * those below,’ and its deri- 
vatives, eveprepo-, evepraro-, evepdev ' from below,’ are 
with reason referred to the preposition ev as their source; 
but we are here brought to a variety of meaning, not 
so distinctly belonging to the preposition. The Latin 
superlative imo- stands in a similar position. By form 
it may well claim connexion with the Latin in, for a 
more regular superlative in-imo-, which the analogy of 
‘pro-imo- --- primo- and sub-imo- = summo- would 
suggest, would naturally be compressed down to imo-, 
just as the substantive animo- seems in the comic 
writers to have had a disyllabic pronunciation, some- 
thing like dmo- ; and such compression is confirmed 
by the shape which this word has taken in the French 
dme. Then as to meaning, although ‘ inmost’ wiU 
suit not a few passages in which imo- occurs, yet the 
notion ojf ‘ lowest' seems more commonly implied. 
Nay, even the adverb imo or immo may have had per- 
haps for its original meaning ‘ at the bottom,’ for the 
iise of the particle is to correct those who give only a 
part of the truth, not going to the bottom of things.^ 
StUl the two senses of ‘ in’ and ‘ down’ have a natural 
connexion. As prepositions generally are employed to 
denote the relations of place, and as the earth itself is 
the great object to which all motions and all positions 
are naturally referred, the ideas of ‘ further in’ and 
‘ further down’ have a natural coincidence. Again, the 
Latin in before an accusative, and the Greek «« for tvs, 
add the notion of ‘ into.’ But where many meanings 

* Mr. Parry, in liis Terence, has erroneously ascribed to me an 
etymology of imo which was never mine. 


Digitized by Google 



64 ON THE PREPOSITIONS ew, MI, 

Ixslong to a word, the right course is probably (see p. 46) 
always to give a preference to one which implies move- 
ment. Now if we accept downward motion as the 
primitive idea, when such descent is checked, as for 
instance by the earth, we arrive at the idea denoted 
by on ; but if the motion be not so checked, then we 
come to ‘ into,’ and that soon followed by the result of 
being ‘ in.’ Wliat is here said is quite parallel to the 
case of the Latin sub. In the preceding paper (p. 44) 
I assigned to that preposition as its primitive meaning 
that of ‘ up,’ or rather ‘ upward movement,’ and con- 
tended that it was when such upward movement was 
terminated that the preposition acquired the sense of 
‘ under,’ with rest 

Another instance of a word in which in carries with 
it distinctly the notion of ‘ down’ is incurvus, which 
Forccllini was contented to translate idem quod curvus 
or wtlde curvus, but which really means ‘ bent down.’ 
Hence in “The Eunuch” Archidemides, whom Chaerea 
speaks of as of the same age with his father, patris 
aequalem, and. who is therefore an old man, is subse- 
quently described as incurvus tremidus, &c. Again 
Cicero has the phrase Stesichori poetae statua senilis 
incwva, and the same writer quotes from a poet, Ramos 
haccarum uhertate incurviscere (Or. iii. 38). So again 
Pacuvius (Varr. L. L. v., p. 19 of Spengel’s edition) used 
the phrase incm'vicervicum pecus, corresponding to 
Sallust’s pecora quae natura prona fecit. The verb 
inflecto shows less distinctness in its sense ; but even 
this we find united with incurvus, as Incurvum et 
leviter a sumino inflexum bacillum. 

Inclinare is another word which, duly examined, 
will lead to the same result. In many ca'^cs mere 


Digitized by Google 



AND RELATED WORDS. 


65 


‘ bending ’ will satisfy the uses of the word ; yet still it 
may be presumed that the preposition at first was not 
added without a purpose. We may safely assume, 
then, that the notion of ‘ down ' belonged to the word 
when used of the declining sun, as in Juvenal’s ‘ Sed 
iumenta vocant et sol inclinat or of the heeling of 
a ship, ‘ Merso navigio inclinatione lateris unius’ (Plin. 
viiL 208) ; or of a tree laden with fruit, ‘ Palladis arbor 
Inclinat varias pondere nigra comas’ (Mart. i. Ixxvi. 8) ; 
or metaphorically of a declining condition, as, ‘ Incli- 
nata fortuna et prope iacens’ (Cic. Fain. ii. 16) ; ‘ In- 
clinatis iam moribus’ (Plin. xxv. 162); ‘Is primus 
inclinasse eloquentiam dicitur’ (Quint, x. 1), Nay, 
even the ordinary use of the word to denote a moral 
inclination to any object is in harmony with the 
notion of descent, for down-hill action is of course 
the easier; and indeed this accounts for the forma- 
tion of the words pronus, propensus, proclivis so 
used. (See the subsequent paper on pro ) 

That instances of the prefix in with the sense of 
‘down’ are after all but few, is a fact which finds its 
explanation on two sides. When the Latin language 
had once established the variety of in for an or ava 
‘ up,’ the particle in = ev was liable to confusion. On 
the other hand, the form de was no way ambiguous. 
But even this de will presently be claimed as a deriva- 
tive from in ‘ down.’ 

As regards the forms infra, inferi, inferior, injimtis, 
the best course is to compare them with the opposed 
family of words, supra, superi, superior, summus; 
and then we are led by an irresistible necessity to the 
conclusion that, as the latter series have their root in 
the first three letters, so inf must contain the more 

F 


Digitized by Google 



6G ON THE PREPOSITIONS fW, ill, 

radical portion of the former series. But inf being 
almost an unpronounceable combination of letters, we 
are further led to the assumption of an older form 
ene/ra, &c., following therein the analogy of many 
similar compressions. Thus umhra may be considered 
as a compression of on-uh-era, and so connected with 
the Latin nube- ‘ cloud,’ the verb nuh-ere ‘ to veil 
oneself ; ’ while nuhi-la-re and nnbil-um bring us 
directly to the forms veif>e\-r) and German nebel 
‘ mist.’ If inftila signify really a veil, and so stand 
for enef-ula, we have a case thoroughly parallel to 
that of inf-ra. Again, ofj.<f>a\o-, if it represent, as it 
well may, a fuller ovv<j}a\-o, and the Latin umbil- 
ico-, standing for onubil-ico-, bring us to Germ, nabel, 
Eng. navel; and ungui- for onugui-, by the side of 
OPVX-, to Germ, nag-el, Eng. nail. 

Following these analogies then, we may conclude 
that inferi stands for en-ef-eri, a coiuparatival form 
which should have been preceded by a positive enefus 
or nefus. This has a somewhat strange appearance, 
but is in reality identical with the Greek ve^os and 
Latin novus, for the interchange of the sounds / and 
w is no way rare, and indeed our own language sup- 
plies an apposite example in two varieties for the name 
of the same reptile, a newt and an eft. Similarly the 
Greek atnos EvpnriHijs are now by modem Greeks (and 
as regards the v were perhaps in ancient times also) 
pronounced aftos JSm'ipSthes. 

I next consider the foims in which the vowel e or t 
being dropped, the liquid v occupies the initial position. 
Nep^e, veprepot arc coexistent with the Greek adverb 
and adjective already quoted : but besides these there 
exists a superlative ve-mos, which at any rate by its 


Digitized by Google 



AND RELATED WORDS. 


67 


ordinary signification of ‘ lowest’ seems in a vciy 
decided manner to claim kindred with the root before 
us ; and the use of the feminine vemri or vryrif for the 
‘lowest string of a musical instrument’ (lowest in 
position, but highest in note) confirms this view. But 
I have here to contend with what appears to be a rival 
etymology, for vearot bears to the adjective veos ' novus,’ 
precisely the same relation that fietraros does to /teaof ; 
and this argument receives much encouragement from 
the fact that vearos, like the Latin novissimus, also 
signifies the ‘ last or most recent.’ I shall presently 
give reasons for the belief that this new notion is not 
at variance with the idea of ‘ lowest.’ 

But it will first be convenient to look in other lan- 
guages for the representatives of our root. Now the 
Sanskrit has a particle ni, used as a prefix to verbs 
with the sense of ‘ down,’ as from ni -|- dint, ni-dlui 
‘ deponere and from ni + a^, ny-as ‘ dejicere.’ Here 
again the two notions of ‘in’ or ‘on’ and ‘down’ 
belong to the preposition ni, so that from ni -t- gcmn we 
have ni-gam, ‘ to go into,’ ‘ inire.’ The Ossetic of the 
Caucasus, a language of the greater interest because 
it is one of the most outlying members of the Indo- 
European family, has also, according to Sjogren, a 
prefix ny of the same power, as ny-fyssyn, ‘ to write 
down,’ ny-vcBryn ‘ to lay down.’ Of the Slavonic 
languages it will be sufficient to take examples 
from the Russian, where we find ntV an inseparable 
preposition, denoting ‘ down,’ as niz-lozhit’ ‘ to lay 
down,’ from lozliit’ ‘ to lay ;’ niz-padat’ ‘ to fall down,’ 
from padal’ ‘ to fall,’ &c. ; besides the verb nizit’ 
‘ to lower.’ 

Again, the Lithuanian has a prefix nu ‘ down,’ of 

F 2 


Digitized by Coogle 



C8 ON THE PHEPOSITIONS evi, in, 

very frequent occurrence, aa nu-degti ‘ bum down,’ 
nu-tekH ‘ flow down’ (see Nesselmann’s Lexicon passim, 
and especially under the word nit). But this n« is a 
shortened form of an older nug ‘ down.' 

But it is not merely in the humble character of a 
prefix or particle that thb root occurs. Thus, to the 
Russian niz-it’ we may add the Chinese ni ‘ to sink, de- 
scend ;’ while in the Greek vev-<a and Latin nu-o we have 
verbs still carrying with them the notion of downward 
movement. In practice these two words are pretty 
well limited to the motion of the head ; but in the 
Greek phrase vevevKtat tijv xeipaXijv ‘ holding the head 
down,' ‘ demisso Canute,’ the very fact that K€<j>a\T)v is 
expressed proves that the verb itself did not imply 
this idea ; which is as much as to say that the verbs 
veveiv and nuere meant merely ‘ to lower.’ 

The Teutonic languages also abound in examples 
which contain the root under discussion. Here we 
usually find a dental consonant attaching itself to 
the particle. Thus the Danish has ned ‘ down,’ used 
commonly as a prefix to verbs, e.g. ned-shyve ‘ write 
down,’ ned-blcese ‘ blow down,’ besides an adverb 
nede ‘ below.’ The English language possesses still, at 
least in poetry, the simple neath, whence on the one 
hand the preposition be-neath, and on the other the de- 
rived words nether, nethermost. But the forms with 
the sufiix containing the letter r (no doubt compara- 
tival in origin) are of most frequent occurrence in the 
Teutonic dialects. Thus the German has nieder and 
the Icelandic nidr ‘ down.’ This latter language has 
also a substantive nid, to denote the time when there 
is no visible moon, although the idea of ‘ down’ is all 
that the word strictly denotes. (See Holmboe’s Ord- 


Digitized by Google 



AND RELATED WORDS. 


69 


forraad.) On the same principle no doubt the Latin 
noct- Greek wkt- and vvx- (as seen in vv)(a, vvj^tot, &c.), 
originally meant ‘ sun-down.' It was natural for a 
Roman to think more of the sun ; but an Icelander, 
less happily placed, owes a very large part of his com- 
fort to the light of the moon. 

Thus we have seen the simple en or in taking to 
itself a suffix or suffixes with a varying consonant, 
as — 1. A guttural in the Lithuanian niig, the Greek 
pvx of vvxa ; 2. a dental in the Danish ned, Icelandic 
nid, German nied-er, English neath, neth-er, Russian 
niz ; 3. a labial in the Latin inf-ra, &c. for enef-era ; 
while 4. no consonant shows itself in the Chinese and 
Sanskrit ni, the Ossetic ny, the Italian ne, the Lithu- 
anian nil or 7iu, and the Latin nii-, Greek vev-. 

I have just used the words suffix or suffixes ; but I 
am satisfied in my own mind that all these suffixes 
are of one origin, and I believe that the Lithuanian 
ntig and Greek wx, standing for 07i-^ and ov-vx, have 
preserved the suffix in its purest form. But I am here 
influenced by considerations which will be stated more 
fully in the subsequent paper on re and pro. Our 
own preposition on can no way be separated from the 
Latin in, Greek ev ; but I do not pretend to decide 
between the claims of these three forms, and should 
be equally pleased to find a variety en-ek, en-eg, or 
en-ech. This suffix, ug vx, &c., I believe to be of 
diminutival power. Just as Dr. Johnson speaks of 
the suffix le of our verbs sparkle, trickle, as diminu- 
tival in origin, yet bringing with it to these verbs the 
notion of iteration, so I think that nu-ere and vex>-eip 
have in the same way obtained the power of denoting 
a repetition of small acts. 


Digitized by Google 



70 ON THE PKEPOS1TION3 evi, in. 

But the liquiil n habitually throws out an excrescent 
t or d, e.g. in tegument-um, from tegumen {legmen), 
avbp-os for auep-os. Such a rf T find in the Greek «i»8- 
ov (Dor. evSot), in the Latin indu-perator, and our 
own und-er (from oji). On the other hand a t presents 
itself in the Latin int-iis,^ int-er} int-ro, and the 
German unt-er. But here again the notion of ‘ down’ 
is felt, most clearly indeed in the prepositions under, 
unter ; but also in the phrase ei>Sop yeypairrai ‘ it is 
written below,’ and the adverb evborepm similarly used ; 
as also in the Latin interula (sc. vestis) ‘ an under- 
garment,’ and the phrase agua intercus (ie. ‘ under 
the skin’). 

Inde is another instance of an excrescent d. But 
here caution is necessary, as scholars seem to have 
confounded together two independent words. Inde 
‘ from this,’ or ‘ hence,’ is of coui-se connected with the 
pronoun is, ea, id, of which, however, in rather than i is 
the base, as shown l)y the old nominative ‘ Is’ of an 
Inscription (Rhein. Mus. n. f. xiv. 380, note), for the 
tall ‘ r of this form goes far to confirm the doctrine I 
have contended for elsewhere (Philolog. Soc. Proc. III. 
57), that all demonstrative pronouns once had a 
final n. Thus the derived ind-e is one with the Greek 
ev6-ev, the d and 6 being alike excrescent. 

But the preposition in also formed a similar ind-e = 
Greek evbov, or evbos, with the notion of down. I 
refer to the familiar phrases tarn inde ah initio, &c., 
in which the usual practice is to ignore the inde ; but 
‘ dpwn from the beginning’ is so thoroughly intelligible 


' In my paper on Excrescent Consonants (see below), I have 
{■iven my reasons for so placing the hyphen. 


Digitized by Google 



AND ItKLATKD WORDS. 7l 

that I liopc it will be aeeepted aa a more precise 
translation. 

But this inde also enters into the formation of sub- 
inde, which must on no account be classed with tZe- 
inde, pro-inde, &c., for in these the inde is the genitival 
mde ‘ from this,’ corresponding to the Greek ev6tp. 
The literal translation of suhinde is ‘up and down’ 
{i.e. ' ever and anon’). The non-appearance of a par- 
ticle to denote ‘ and’ is in agreement Avith the habit of 
the Latin language, which preferred hi?ic illinc, pedibm 
manibus, to hinc et illinc, pedibus et manibus. Then 
as regards meaning Ave have Avhat is very similar in our 
OAvn combination ‘ off and on.’ This subinde has of 
course led to the Italiiui sovente and the French souvent. 

I now venture to claim theoretic varieties, ond-uk 
and end-ek, as standing by the theoretic on-uk, en-ec ; 
and then by decapitation the Greek verb Bv-, which 
under this view may well unite the tAvo meanings com- 
monly assigned to it of ‘ go in’ and ‘ go down,’ as used 
of sun-down, of diving, or in the phrase AiBao 
Bv(ja<T0at. Our own ‘ duck,’ used often like ‘ dive' in ' 
reference to water, but also in the sense of ‘ ducking’ 
or lowering the head, as in passing under a gateway, is 
truer as to form than Bv- and nu-, both of which it 
represents; Avhile our ‘dive’ is another variety of the 
same Avord, the guttural and labial interchanging as in 
nix, nivis. The Greek Sutttoj has also substituted a 
labial, as is usual in that language, for a guttural. 
On the other hand, by a similar decapitation eiidek 
leads to de, the long vowel of the latter corresponding 
to that of the preposition e for ek. That our own 
dovm is of the same stock can scarcely be doubted. 
Perhaps as our preposition ab-ove led to a secondarv 


Digitized by Google 



72 ON THE PREPOSITIONS evt, in, 

form abov-en or abow-en, which afterwards was con- 
tracted to aboon (see Jamieson’s Dictionary), so dotm 
may be for dow-en. 

I now go back to the adjective veo?. That this word 
must at one time have signified ‘ low’ follows at once 
from the use of the superlative vearo? as ‘ lowest 
and the sense of ‘low’ is more likely to have been 
original in the word than that of ‘ new.’ A relation 
of place is often found to coexist in the same word 
with a relation of time, but few will hesitate to give 
to the locative idea the priority of title. Thus ubi 
and ibi denoted ‘ where' and ‘ there’ before they were 
used for ‘ when’ and ‘ then.’ Again, in the familiar 
phrase interea loci, the latter word appears in a sense 
which is not primitive. Still the question remains, 
how we are to connect the ideas of ‘low’ and ‘new.’ 
The explanation I would suggest is that a consider- 
able duration of time is commonly expressed by the 
simile of a river. Thus we ascend the stream of time 
to the past, and on the other hand we come dotim to 
recent times. But there is also another view that 
may be taken. Youth and lowness of stature arc 
coincident, and every inch of growth is an evidence 
■ of increasing age. Thus veas might pass through 
the meanings ‘ low, young, new.’ 

Perhaps on the same principle we may be permitted 
to explain the German adjective alt, which is repre- 
sented among ourselves by old. This German word 
bears a tempting resemblance to the Latin altus, but 
a resemblance not nearer than that of the German 
adjective neu to the Greek veF-os. Is it possible, then, 
that ‘ high’ may have been the original sense of the 
German alt ? 


Digitized by Google 



AND BELATED WOBDS. 


73 


And if this be true, we are brought to the Latin 
al-ere ‘ to raise,’ and the Greek atp-eiv of like power, 
the root-syllable of which I assume to be ap, for the 
fuller form aeip-etv seems to be the result of redupli- 
cation. The Latin adjective ard-uus differs only in 
having an excrescent d. Possibly, too, the preposition 
ava may be an offspring of the same root ; and if so, 
both ava and ew will be deduced from verbs. 


Digitized by Google 



74 


III. 

ON TKE LATIN PREPOSITIONS re AND pro. 

I BEGIN this inquiiy by once more quoting from 
tlie “Deutsche Grammatik” (ii. 8G5), the following 
passage : — 

‘ The doctrine which holds true generally of par- 
ticles, that they become obscure in signification and 
disguised in form, is specially applicable to the in- 
separable particles.’ 

The little word re of the Latin language belongs 
to this class, as it is never found doing duty as an 
independent preposition, but occurs only in compound 
verbs or adjectives, and words deduced from them. It 
further deserves attention, in that it is difficult in the 
sister languages to find its representative. But it is 
precisely in short forms of this kind that the destruc- 
tive habit of language is found to have been most 
violently at work. Already the longer fonn red {red- 
eo, red~do, red-igo) exhibits a final consonant that 
once belonged to the particle. We must also claim 
ret as a variety, for ret-ro is a more trustworthy 
division of this adverb than re-tro. I'liis appears from 
the coiTcsponding adverb por-ro, the first syllable of 
wliich exhibits the simpler fonn of the Latin preposi- 
tion which led to the derived preposition pro, i.e. 
poi'-v. Indeed the simple preposition luis been 


Digitized by Google 



ON THE LATIN PUEPOSITIONS VC AND TS 

preserved in the verb 2 >or~ri(j-o, afterwards compressed 
to porgo and pergo ; and virtually in polliceor and 
2 )ollingo ‘ I lay out (a corpse).’ But to this preposition 
pro I shall have to recur again. 

But even ret is not the oldest form of the particle. 
The dental is in all probability a corruption from a 
guttural. Such a change is common to the last degree 
in language, and especially in the Latin language. In 
our own, for example, the diminutival suffix et has 
grown out of an older ick: thus, emmet and gimlet 
are known to have superseded emmick and gevilick. 
So again the Latin ahiet- stands in place of ahiec- 
(witness ahieg-no-) ; and as I have elsewhere noted, 
the frequentative verbs, vell-ic-are, fod-ic-are, mors- 
ic-are, have the suffix in a purer form than ag-it-are, 
quaer-it-are, clam-it-are. It is true that this latter 
variety outnumbers the former in the proportion of 
about a hundred to one; but it is not by numbers 
that such questions should be decided. The change 
from a guttural to a dental is a far more familiar 
matter than the converse ; and in the case of Latin 
frequentative verbs this particular change was en- 
couraged in a large number of instances by the prece- 
dence of a guttural in the simple verb, as for instance 
in the three verbs just quoted. But in the instance 
of ret, we have a confirmation in the fact that rec 
has been preserved in recu-pera-re, ‘ get back,’ a com- 
pound of parare, and in the adjective reci- 2 ^roco-, 
‘backward and forward.’ The verb recu-perare has 
met with much ill-treatment among philologers. It 
was once the practice to regard it as a derivative from 
recip-ere, and even Varro (L. L. vii. 5, p. 358) sanctions 
this view ; but this leaves the era without explanation ; 


Digitized by Google 



76 


ON THE LATIN PREPOSITIONS 


for volnerare, onerare, derive the syllable er from the 
nouns minus volner-is, onus oner-is. Again for many 
years there was to be found in the “ Gradus ad Par- 
nassum ” and similar works a statement that the u of 
re^uperare was a long vowel, and a line ascribed to 
Plautus used to be quoted in support of the assertion. 
But the said line did not come from Plautus, whose 
writings on the contrary contain many passages to 
prove the reverse, the metres of this poet invariably 
demanding a short pronunciation, such as ricuperdre, 
riedperdtor. Of reciproco-, more presently. 

But still further evidence in favour of the guttural 
presents itself in some of the allied languages. In 
the Greek pa^i-s paxerpov ‘ back or spine,’ we have 
evidently words of the same stock. So again in the 
German prefix riick {riickwdrts, &c.), and the sub- 
stantive riicken ‘ the back.’ These again bring us to 
the Anglo-Saxon hri// ‘back,’ the Scotch and Old 
English riff, and the Modern English ridffe. 

On this evidence then I claim rec as an older form 
than ret or red or re. 

But the particle has suffered more or less on the 
other side too. With myself the appearance of an 
initial r always raises the suspicion of a decapitation, 
and the Anglo-Saxon hriff is a witness in this par- 
ticular case to the same effect ; but I shall not bo 
satisfied with claiming some initial consonant. A 
vowel also is missing ; and in the selection of a par- 
ticular vowel I am guided here, as in other similar 
instances, by the law of vowel-assimilation. As in 
the case of the Latin pro I was led to claim a vowel 
0 as lost, the word standing for por-o, so for re I 
would suggest a preceding c, making ere, or rather 


Digitized by Google 



re AND pro. 


77 


er-ec. Still closer is the parallelism when for pro 
itself I find fuller forms, — first, (prod-ire, jyrod- 

esse) ; secondly, prot in the Greek vpor-epo- (my 
reasons for denying the t to the suffix have been 
given elsewhere) ; and thirdly, a still older proc in 
the very adjective already quoted, reci-rproco-. For 
the letter-change it may be useful to compare the 
varieties re, red, ret, and rec, as well as pro, prod, 
prot, and proc, with the negative particle hau, hand, 
haut, representing the Greek ovk and ov. Thus re and 
pro appear to me to be corruptions from disyllabic 
forms er-ec and por-oc. If I am asked what this 
guttural suffix denoted, what its power was, I answer 
that it is the diminutival suffix which, in my view, 
plays so important a part in language ; and I point 
to a parallel case in the Teutonic family. The Old 
German durah, written also duruh and duroh (Grimm, 
D. G. ii. 770), corresponding to our own throtigh, seems 
to claim connexion in its first syllable with the sub- 
stantive, which we write door, and a German thiir, 
w hi le the second syllable has aU the appearance of 
being the suffix of diminution. I might perhaps put 
forward as examples of the simple preposition the 
adjectives dur-liuhtic, dur-nehtic, dur-sihtic, quoted 
by Grimm from the Middle German. Of course in 
our tongue the word through is but an abbreviation 
. of thorough. Thus the Anglo-Saxon thurh-fare be- 
comes, in Chaucer, thurg-fare, and in Shakspere 
(Merchant of Venice, ii. 8) through-fare, where we 
write thoroughfare. Again Shakspere in the same 
play (iv. 1) has throughly in the sense of our 
thoroughly; and in the Midsummer Night’s Dream 
we find : “ Thorough the distemperature we see the 


Digitized by Google 



78 


ON THE LATIN PUEPOSITtONS 


seasons alter ; ” and “ Over hill over dale, thorough 
hush thorough briar, over park over pale, thorough 
flood thorough fire.” 

But if ec of the theoretic erec be a suffix, we have 
for the base of the word er , — that is, a prefix well- 
known in German. In the examination of this prefix, 
a fii-st duty is to consult the “Deutsche Grammatik.” 
Accordingly I have read with some care what is con- 
tained in the article on the subject in the second 
volume of Grimm’s work (pp. 818-832). What he 
says on the Gothic vocabulary I have checked with 
the lexicon of this language attached by Massmann 
to his edition of Ulphihus. But though all linguistic 
imjniries should include an examination of the oldest 
forms of language, this should not be to the exclusion 
of later varieties, and this for two reasons, that remains 
of the oldest forms of a language are for the most 
part very fragmentary, and not unfreqnently difficult 
of interpretation. With a language still spoken these 
two evils are less to be feared. Thus I should deem 
it most unwise to throw out of view what lies before 
us in Modern German. Under this impression I have 
tabulated to a great extent the German verbs com- 
pounded with er according to the meanings assigned 
to them in Meissner’s Worterbuch, taking this work 
because it happens to be at hand. The result of my 
examination has been to assign to the preposition the 
following meanings : — 

1. Up . — In support of this, I might quote nearly 
forty examples, including both physical notions, and 
those of a secondary or metaphorical character ; but 
am satisfied with erstehen ‘ stand up,’ erhalten ‘ sus- 
tain,’ erspriessen ‘ shoot up,’ ersteigen ‘ climb up,’ 


Digitized by Google 



re AND jivo. 


79 


erhehen ‘ heave up,’ ersclmdlen ‘ swell up,’ erhauen 
‘build up,’ erbrausen ‘surge up,’ ertragen ‘support,’ 
erdulden ‘ sufier,’ erncihren ‘ nourish,’ erziehen ‘ bring 
up,’ to which should be added the adjective, or rather 
participle, erhahen ‘ elevated.’ 

2. Back. — This sense naturally grows out of the 
preceding, inasmuch as the downward movement of 
substantial bodies, through the action of gravity, is 
more conspicuous, and thus apparently more natural, 
than the cori’csponding ascent of what is often in- 
visible, and so the upward action is regarded as a 
reversal of the fii-st. Examples are — erlassen ‘ remit,’ 
erkaufen ‘ransom,’ erschalleid ‘resound,’ erhallen 
‘ resound,’ ertbnen ‘ resound,’ erklingen ‘ resound.’ 

3. Again is a meaning which flows from, or rather 
is scarcely separable from, the preceding. This mean- 
ing occurs in erkennen ‘recognise,’ eimeuen ‘renew,’ 
ersetzen ‘ replace,’ erquicken ‘ revive,’ erfrischen 
‘ refresh,’ erinnem ‘ remind,’ erlosen ‘ release,’ er- 
hhen ‘refresh,’ erholen ‘respire,’ and ersinnen {sick) 
‘ remember.’ 

4. Reversal of the act expressed in the simple verb. 
— In p. 830, sub-section 8, Grimm deals with instances 
that faU under this head. In the Gothic indeed he 
finds no example, but gives not a few from the Old 
German ; as ur-erh-an ‘ exlieredare,’ ir-hals-an ‘ de- 
coUare,’ ir-hirn-an ‘ excerebrare,’ ir-k'cz-an ‘ oldivisci ’ 
(where the root-syllable coiTesponds to our get in 
Jor-gel), ar-meinsam-on ‘ excommunicare,’ ur-wir-an 


* In this verb ertcluillen and those following, the notion of loud- 
ness expressed by the idea of ‘ up ’ may perhaps be preferable. See 
remarks below on the Latin verb recita-. 


Digitized by Google 



80 


ON THE LATIN PREPOSITIONS 


‘castrare/ Again for the Middle German he quotes 
er-kim-en ‘enuclearc,’ but for the Modern German 
he expressly says there is no example. We may per- 
haps venture to doubt the correctness of the writer 
both as regards the Gothic and the existing language 
of Germany, when we find uslukan ‘unlock,’ and 
usluknan ‘ open oneself,’ in Maasmann’s Gothic Voca-. 
bulary, as also erldsen ‘ unloose,’ and erschliessen 
‘unshut’ (to borrow a good Old English word), in 
Modern German, a verb the more interesting as corre- 
sponding most precisely to the Latin recludere. 

5. Reaching. — The efibrt to reach an object may 
be exercised in all directions, as downward, to get at 
water in a well, or horizontally, as in one of Hood’s 
comic poems, where a child, shut in by the bar of its 
little chair, stretches out its arms to get at some fruit, 
and, unable to effect its purpose, adopts the ordinary 
revenge of crying, so as to justify the witticism written 
below the picture : ‘ Squall at Long Reach.’ But in 
ordinary life the difficulty is more commonly to reach 
what is above us, as with the Fox and Grapes. Thus 
the combination ‘ up to ’ readily expresses the idea of 
reaching, and the to is virtually expressed in the 
accusatival form of the accompanying noun. Thus 
wo have ereilen ‘ overtake, fetch up,’ erfahren ‘ over- 
take by driving, come up with,’ erfassen ‘ lay hold 
on (suddenly),’ erfinden ‘ find out ’ (literally, let me 
add, by feeling), ergehen ‘ overtake, reach,’ erkmgen 
‘ reach,’ erlaufen ‘ overtake by running,’ erleben 
‘ live to see,’ erpacken ‘ seize,’ erraffen ‘ snatch,’ 
erreichen ‘reach,’ erreiten ‘overtake on horseback,’ 
errudern ‘reach by paddling or rowing,’ errufen 
‘ reach with calling,’ erschlcudem ‘ reach with a 


Digitized by Google 


re AND pro. 


81 


sling-stone/ erschnappen ‘catch with open mouth,' 
erschreien ‘ reach with crying,’ erschreiten ‘ reach 
with a step,’ erschwehen ‘ reach by Hying,’ er- 
schwimnven ‘ reach by swimming,’ erschwingen ‘ soar 
up to,’ erspannen ‘reach by the span,’ erspringen 
‘reach in leaping,’ erstrecken (sick) ‘reach,’ ertappen 
‘ overtake,’ ertasten ‘ reach by feeling.’ Note in the 
translations the repeated use of our two prepositions, 
up and over (over-take). 

6. Up to in daring. — There may be added, as a 
sort of corollary to the preceding section, the reflec- 
tive verbs, erdreisten ‘ to be bold enough,’ erfrechen, 
erkecken, erkiihnen ‘ to dare, presume.’ 

7. Getting, by the act of the verb. — A meaning 
closely allied to those of § § 5, 6, but eontaining less 
of the physical action. It would be idle to enumerate 
instances, when in Meissner’s Lexicon I find over one 
hundred examples. But the construction will be per- 
haps better understood if it be first pointed out that 
with compound verbs the accusative may be dependent 
either on the verb or on the preposition. The best 
proof of this is seen in such Latin sentences as Iberum 
copias traiecit, where we have Iberum attaching itself 
to the preposition, copias to the simple verb iecit. In 
the preceding sections (5 and 6) the accusative be- 
longs to the preposition. Indeed in many of the 
examples the verb itself is clearly of an intransitive 
character. 

8. Making (a) aiul becoming (b). — Wliere the pre- 
position is compounded with adjectives to constitute 
a verb. The idea is closely akin to the preceding 
sections 6 and 7 and the two immediately following. 
Examples of this sense are tolerably numerous : (a) er- 

G 


Digitized by Google 



82 


ON THE LATIN PREPOSITIONS 


bitten, erfreuen, erhitzen, emiedem,erschweren; (b) erv 
blassen, ei-bleichen, erblinden, ergrauen, erlahmen. 

9. Opening. — This at first may appear strange, but 
the close connexion of the idea is evident from the 
very etymology of open, which in its root-syllable op 
is one with the preposition up. The examples seem 
but few : erbrechen ‘ break open,’ erbeissen ‘ bite open,’ 
eroffnen ‘open.’ 

10. Beginning. — This grows easily out of the last, 
or it might perhaps as easily be deduced from the 
sections headed ‘ up to,’ and ‘ making,’ for these are 
substantially one with the idea of commencement. 
For examples may be taken ; erbrausen ‘ begin roar- 
ing (of the storm),’ erdonnern ‘begin to thunder,’ 
erdrohnen ‘begin to sound,’ ereifern (sich) ‘fall into 
a passion,^ ergldnzen ‘ begin to shine,’ erglimmen 
‘ begin to glow,’ ergrausen ‘ shudder,’ erkcilten ‘ catch 
cold,' erkracken ‘ begin to crack,’ erkranken ‘ be taken 
ill,’ errbihen ‘ blush,’ erschaudem ‘ shudder, be seized 
with horror,’ erschrecken ‘be struck with fear,’ er- 
staunen ‘ be astonished,’ ertosen ‘ begin to roar,’ er- 
zittem ‘ begin to tremble.’ 

11. Thoroughly or up to the sourees (in a search), 
as : erforachen ‘ investigate,’ erhunden ‘ explore,’ er- 
kundigen ‘ inquire after,’ erproben ‘ test.’ 

12. Removal, disappeurance. — This meaning may 

be explained in two ways. In removing a thing the 
first act is to lift it up. Again a thing in vanishing 
generally rises, vanescit in auras. It is especially in 
reference to disappearance by death that the German 
compounds with er are so used. Thus — erbeissen 

‘ bite to death,’ erbleichen ‘ turn pale and so die,’ 
erdriicken ‘ press to death,’ enlvQsseln ‘ strangle,’ er- 


Digiiized by Google 



re AND i>ro. 


83 


dursten (prov.) ‘die with thirst,’ erfrieren ‘be frozen 
to death,’ erhdngen {sich) ‘ hang oneself,’ erlegen ‘ slay,’ 
erldschen ‘ go out as a fire,’ ermorden ‘ murder,’ er- 
saufen ‘be drowned,’ erchiessen ‘shoot (to death),’ 
erschlagen ‘ slay, kill,’ erschopfen ‘ drain, exhaust,’ 
erspiessen ‘ kill with a spear,’ erstechen ‘ run through 
with a sword,’ ersterben ‘ die (out), become extinct,’ 
ersticken ‘ smother, sufibcate,’ ertodten ‘ kill,’ ertrUn- 
ken ‘drown,’ ertreten ‘trample to death,’ erwUrgen 
‘ strangle.’ 

It has been assumed above that the Gothic prefix 
tis and the German er are one. This is generally 
admitted ; and in truth as the Gothic habitually has 
a sibilant, where an r appears in German, it is no 
matter for surprise that the Gothic form of our par- 
ticle should be us, which then only takes the form of 
ur, when an r commences the simple verb. 

Thus in the very limited vocabulary of the Gothic 
we find eight examples where Massmann translates 
the Gothic by the corresponding German verb with a 
prefix er,viz. : usfullian ‘ erfliUen,’ ushafjan ‘erheben,’ 
ushahan {sik) ‘sich erhangen,’ ushauhjan ‘erhohen,’ 
usldubjan ‘ erlauben,’ usldusjan ‘ erloscn,' ussteigan 
‘ ersteigen,’ usvakyan ‘ erweeken ; ’ while eighteen 
other verbs which in Gothic began with us or ur, are 
represented in the same book by German verbs com- 
pounded with er, auf, or wieder. Hence no one need 
hesitate in identifying the Gothic us and German er. 

In Old German the vowel varies so that we have ur, 
ar, ir, and er. As regards the « it is perhaps safe to 
assume that this vowel or an o had precedence over 
the weaker vowels, because a change from a strong 
vowel to a wenker is more in accordance with the 

o 2 


Digitized by Google 



84 


ON THE LATIN PREPOSITIONS 


habit of language. Add to this that the Gennan 
ruck ‘a jolt,’ and rilck ‘back,’ give support to the 
view that the vowel before the liquid was either u or 
0 . My reason for including the sb. ruck will appear 
presently. 

If we may rely upon Grimm, yet another corruption 
of our particle is found in the Old-Saxon and Anglo- . 
Saxon, where according to him it takes the simple 
shape a, as a corruption of as (ii. 819). But his view 
may be doubted, as this prefix, at any rate in Anglo- 
Saxon, seems better explained as a representative of 
the Greek ova. Indeed in the Greek language itself, 
as has been already noticed, the iEolic and Doric 
dialects, ordinarily employing the shorter form av in 
place of Ava, under certain circumstances cut this 
down to a. Then as regards the Anglo-Saxon, although 
on is the ordinary representative of the Greek ava, , 
there was a marked tendency in this language to 
exchange on for a, seeing that the ordinary preposi- 
tion on got reduced to a, as still seen in our own 
aboard for on hoard, afoot for on foot, &c. I feel the 
more at liberty to question Grimm’s theory that the 
prefix a is a corruption of as or us, because, in a few 
pages before (699), when he first puts forth the idea, 
he implies a doubt in his own mind by affixing a 
query. 

The mention of the Greek preposition ava reminds 
me that in my paper on that little word I was led by 
a similar investigation to assign to it the successive 
meanings: — 1, up; 2, back; 3, again; 4, reversal; 
6, 6, loosening, opening ; 7, commencing ; 8, 9, re- 
moval ; 10, 11, 12, thorough, thoroughly, including 
the special idea of searching up to the sources. " And, 


Digitized by Google 



re AND 'pro. 


86 


as I have already said, the meanings so assigned by 
me to avok have since received the sanction of Pott 
in his recent work on Prepositions, who expressly 
refers to, and so far adopts, what I had written. I 
would beg then inquirers to contrast the meanings 
now assigned to the German er with the meanings 
assigned to avoy the parallelism being complete. Nay, 
I was at first led to the belief that ava and ere were 
but varieties of the same word, knowing, as I did, 
that a German, like our countrymen in Kent (a word 
which itself represents Cantium), habitually has e 
where others have a — as Albts ‘ Elbe,’ Amisia ‘ Ems,’ 
Catti ‘Hesse’ — and further knowing that an interchange 
between the liquids n and r, when not initial, is of com- 
mon occurrence. But I was checked in this view by 
two considerations : one that the sibilant of the Gothic 
us seems entitled to precedence over the r of er, and 
secondly by the fact that in the ‘ Oberdeutsch’ dialects, 
as Grimm informs us (p. 819, line 15), the prefix appears 
with an initial d, as der-warp, der-heizte, der-haben. 

The loss of an initial d is not very rare ; and in the 
present case the authenticity of the d is confirmed by 
the Latin dorsicm ‘ back,’ with which there must have 
been a co-existing variety dossum, as shown by Varro’s 
adjeetive in aselli dossuarii and iumenta dossuaria, 
to say nothing of the Italian dosso and French dos. 
Moreover, the Latin dorsum, like our ridge, is often 
applied to a continuous moimtain elevation, just as 
we have om: ‘ Hog’s-back’ in Surrey. It may as well 
bo observed that the combination rs, as seen in dorsum, 
is liable to several changes. At times an r in such a 
position vanishes altogether in the Latin language. 
Thus it is now a fiuniliar fact that the Latin adverbs 


Digitized by Google 



86 


ON THE LATIN PREPOSITIONS 


rursum, surmim, prorsim, deorsum, took severally the 
forms rusum, susum, prosum, iusum ; the last of which, 
found in the pages of St. Augustine, accounts for the 
modern Italian giuso, ‘ down.’ Another change is that 
of rs into rr. This is most common in Greek, as in 
Xfp<^ot \eppos, aptnjv apprjv. But the Latin language 
has many half concealed instances of the same. Thus 
the nom. pat&r had previously passed through the 
changes, paters, paterr, pater, the last of which is 
justified by the use not only of Plautus and the older 
poets, but also of Virgil. This premised, I may safely 
assume as equivalent varieties, dossum, dorsum, dosum, 
dorrum, and dorum. Then, as regards the final letters 
um, I have long ago given my reasons for the belief 
that the neuter suffix um of the second declension has 
grown out of an older form ug or tic, corresponding to 
our own suffix ock ; and this when I had not arrived 
at any idea of a connexion between the Latin dorsum 
and the German ruck, or riick. I am therefore now 
prepared to give my full consent to the doctrine that 
dorsum and ruck are substantially one, the intermediate 
links being dorug, dome, and druc, or omc. 

I now return to the preposition in its Latin form, 
to deal with a question which naturally suggests itself. 
It has been seen in the examination of the German 
inseparable er, that it has its meanings best explained 
on the theory that the first meaning is ‘ up.’ Shall I 
be justified in assigning this as the first meaning of 
the Latin re f My answer is in the aflSrmative ; but, 
though in my view it is the original meaning, it must 
readily be confessed that the instances are few com- 
pared with the other meanings. Still this is no way 
fatal to the argument ; nay, it is to l»c expected in 


Digitized by Google 


re AND pi'o. 


87 


the constant changes to which words are subjected in 
both fonn and meaning, that the older a meaning be, 
the fewer are the examples preserved. In the first 
place I find what I am looking for in the adjective 
recurvo-. Our lexicographers are satisfied to translate 
this word as well as incurvo- without much attention 
to the prefixes. Now it has already been pointed out 
that incurvo- is not fully translated by any phrase 
short of ‘ bent down,’ as when speaking of a branch 
' weighed down by fruit, or a man bent down by age. 
In like manner the full power of recui'vo- is only 
given by ‘turned up,’ — that is, it speaks of bending 
where the concavity is upward. It is thus well 
applied to the back of the dolphin, which Arion must 
have found to be so far a more comfortable, or at any 
rate a safer seat amid the troubled waves. Ovid was 
right then in his choice of an epithet when he wrote, 
Tergo deljincu recurvo Se memorant onen supposuisse 
novo (Fast. ii. 113). Even when the crow goes off 
with the gilt howl, in its flight it would carry its feet 
behind it, and thus there is still a propriety of language 
when the same poet writes : Corvus inauratum pedi- 
bus cratera recurvis Tollit (ii 251) ; for the bird’s 
claws, in their natural position incurvi, in this altered 
state of things woidd have the concavity upwards. 
Both Virgil and Pliny use this epithet of the bucina, 
where again the eye has before it a concavity with an 
upward presentation. Repando- is another available 
witness in my favour. Here Forcellini speaks with 
some accuracy when he gives as its equivalent, ‘ retro 
et sursum reflexus ;’ in which, however, he would have 
done better to drop the retro et. He adds, too, the 
expressive words, ‘ qualia sunt dorsa ct ora delphinum.’ 


Digitized by Google 



88 


ON THE LATIN PREPOSITIONS 


Indeed the word is most commonly used of the dol- 
phin, and the compound repandirostro- of Pacuvius 
was correct in form, though laughed at as an un- 
wieldy superfluity by Quinctilian. Again, when 
Cicero describes the attire of Juno Sospita in one of 
the Italian temples, we are at no loss to understand 
the phrase cum calceolis repandis. llie verb reouba-re, 
as distinguished from the simple cuhorre, is intelli- 
gible, if we translate it ‘ lying with the head and back 
raised.’ Such a position is well suited for Tityrus 
when playing on his oaten pipe ; and indeed recum- 
here (like the Greek avaxetadai) with all accuracy 
denotes the attitude at meals. Accordingly, it seems 
to have been preferred for this sense in the later 
writers, as in Phsedrus, the younger Pliny, and Justin, 
to accumbere; perhaps because the Romans of that 
day had lost the perception of the true meaning of 
tlie prefix in the latter verb, connecting it with ad, 
‘to,’ rather than with an, ‘up.’ Just as recubare 
means ‘ to lie with the back raised,’ so Celsus, 
speaking of a bedridden patient (ii. 4), uses the term 
residere, ‘to sit up with the back raised,’ or as we 
have it, ‘ to sit up in bed.’ His words are : ‘ Contra 
graAds morbi periculum est, ubi supinus iacet porrectis 
manibus et cruribus, ubi residere vult in ipso acuti 
morbi impetu praecipueque pulmonibus laborantibus.’ 
The Latin verb recitare, ‘ to read aloud,’ finds no 
satisfactory explanation in the meanings commonly 
assigned to re. Now in discussing the powers of 
the Greek ava, I had occasion to refer to some thirty 
or forty examples in Liddell and Scott’s Lexicon, 
where verbs of more or less noise, when compounded 
with this preposition, denote a loud noise : for example. 


Digitized by Google 



re AND j>^o- 


89 


ava-^oa-to and ava-Kpa^-<o ", and I quoted three phrases 
of our own where the idea of ‘ up ’ is expressive of 
loudness : speak up, raise your voice, you speak too low 
to he heard. An application of the same principle 
accounts for the peculiar meaning of recita^. In Horace’s 
Dissolve frigus ligna super foco large reponens, the 
best way, it seems to me, of giving due force to the re 
is by translating the participle ‘ piling up.’ 

Another word which invites consideration is the 
so-called adjective recenti-, which has at least the 
external appearance of a participle, and contains in 
what is clearly its first syllable precisely the form 
which I have been led to assign to the prefix re as 
once belonging to it. Nor am I startled at finding 
the syllable performing the ofiBce of a verb. I have 
long thought that prepositions are many of them verbs 
in origin ; and some years ago, when one of the 
most valued members of the Philological Society of 
London, Mr. .Garnet, opposed to the doctrine that all 
words are in origin verbs, the argument that not a few 
verbs themselves were deduced from prepositions, as 
to utter from out, to intimate from in, I was led to 
think that he had not gone to the bottom of the 
matter. Again it is under the feeling that so-called 
prepositions originally were expressive of change of 
place or motion, that I have claimed for the Latin in 
(and Greek «») the active idea of ‘ down,’ as preceding 
the resulting position of on, and of into, as preceding 
that of in. So again with suh, I placed first among 
its meanings ‘ up/ and regarded the resulting position 
‘under' as secondary. That prepositions are actually 
used with' the power of verbs is clearly seen in the 
well-known phrase, ‘ Up, Guards, and at them.’ I 


Digitized by Google 



90 


ON THE LATIN PREPOSITIONS 


find then no difficulty in connecting a participial form 
recenti- with our so-called preposition rec as to form. 
But how as to meaning? Here also a fairly satis- 
factory solution presents itself when we look at such 
phrases as recens a lecto, a somno, a cena, ‘only 
just risen from bed,’ &c. I say, ‘ only just risen,’ 
rather than ‘ rising,' on the authority of such a con- 
struction as ; Romam veniens comitia edixit (Liv. 
xxiv. 7), ‘immediately on his arrival he,’ &c. A still 
stronger argument is found in the phrase of Persius, 
sub sole recenti, to represent Eastern climes, for ‘imder 
the rising sun’ is exactly what we want, and is in 
accordance with most, if not all, the terms employed 
to designate the East, as Orient of the French, and 
the familiar geographical names of Anatolia and 
Levant. 

But there stands in the way of my argument the 
adjective reciproco-} which 1 have already translated 
as ‘ backward and forward ; ’ yet even this word will 
turn out to be not altogether refractory, for it cannot 
bo separated from the compound rig and fur of the 
Scotch or from the ridge and furrow of our own 
Southern dialect. Rig and ridge I have already 
claimed as representatives of rec, and it is no less 
certain that the Latin porca {= poroca) is one in 
meaning and substantially in form with our furrow. 
Thus Festus interprets porcae by the phrase “rari 
sulci.” It is true that Varro (R. R. i. 29, and L. L. 


• My origin of this adjective, if indeed it need any support, 
receives it in an unmistakable manner from the line in 
Ennius : — 

“ rusus prosus rcciprocsit Huctus foram.” 


Digitized by Google 



re AND jwt>. 


91 


iv. 4) makes 2 >orca the ‘ elevation ’ in a ploughed field, 
damaging his authority however by a foolish etymo- 
logy. In the nature of things ridges and depressions 
go together, and so have a tendency to confound the 
words that denote them. Not unlike this change of 
meaning is that which has befallen the Latin verm-, 
for here also the ploughed field is the origin, but 
the word originally meant the ‘turning’ of the oxen 
between the end of one furrow and the commence- 
ment of another, yet after a time it came to denote 
the line of ploughing, for every new tvum marked 
one line finished and another to begin. A far more 
pertinent example occurs in the Latin noim lira, 
which by one writer is used in the sense of ‘ridge,’ 
by another in that of ‘furrow’ (see Forcellini). 
Again, in our own language the words dylce and 
ditch, which in origin are the same word, are used at 
one time to denote the hollow made by the removal 
of earth, and at another the bank made by the earth 
removed. Thus an Irishman talks of hiding behind 
a ditch, which to the English ear sounds somewhat 
strange. Nay, the German furche itself has suffered 
from this confusion. Thus Campe, in his Lexicon, 
says : “ Die von der PJlugschar aufgeworfene Erde 
die eine eben so lange Erhbhung ausmacht, als die 
Furche einer Vertiefung ist, wird von Einigen auch 
die Furche genannt." 

Again, when we look at the idea expressed by 
‘back,’ we come across what confirms the view. It 
is true that the human back generally presents only 
a vertical direction, but in nearly all other animals 
we see what is elevated, and more or less horizontal. 
A ridge of hills has the some character, and so also 


Digitized by Google 



92 


ON THE LATIN PREPOSITIONS 


a hay-rtci. In the German noun ruck, ‘a jolt/ we 
also find a short upward movement. Similarly in the 
roof of the Crystal Palace with its ‘ ridge and furrow ’ 
of glass we clearly denote the elevation by ‘ ridge/ the 
depression by * furrow.’ 

Still the ground is not clear unless I can establish 
in favour of the Latin preposition pro the notion of 
‘down.’ Here again I cannot dispute the ordinary 
doctrine that forward movement very generally be- 
longs to this word ; but, as I have already said, first 
meanings are apt to disappear ; and I shall think it 
enough to produce some clear examples in support 
of my views. Now the adjective prono- is the precise 
opposite in power of the adjective supino-, the two 
meaning respectively ‘looking downward’ and ‘look- 
ing upward.’ Thus Cicero (Div. i. 53) speaks of the 
three directions of motion, under the terms prono-, 
obliquo-, supino-, that is, straight down, oblique, and 
straight up ; and with this authority in our favour 
we need not be stopped by the fact that for Caesar 
the word had already attained in part to the notion 
of forward, so as to be used in the sense of ‘ obliquity ’ 
or ‘slanting/ as is clearly the case in the passage 
(B. G. iv. 17) ; Tigna non derecta ad perpendiculum, 
sed prona ac fastigiata. The idea of ‘ downward ’ is 
also seen in Sallust (Jug. 98), Ilex paulum modo 
prona, dein jlexa atque aucta in altitudinem. The 
adjective proclivi- is habitually translated ‘ downhill.’ 
The adverbs prorsum and prorsus seem not to obtain 
their full sense until represented by our English ‘ down- 
right.’ It has been long taught that in treating pre- 
positions we should start from the relations to the 
earth. Now what is thrown perpendicularly downward 


Digitized by Google 


re AND pi'o. 


93 


strikes the horizontal plane with all ita momentum, 
whereas in an oblique blow a portion is of course 
wasted. But the first idea being so fixed by reference 
to the earth, the word is afterwards applied to a 
movement perpendiciilar on any plane, and there 
results what we may express by ‘ completely,' ‘ abso- 
lutely.’ Among verbs I find no inconsiderable number 
where the motion of ‘ down ’ predominates, as procel- 
lere, procidere, procumbere, proclinare, proculcare, 
projligare, proicere, prolabi, proruere, prostemere 
{prostrato-), proterere ; nor is it unknown to promit- 
tere, witness the phrases p. capillum, barham, comas, 
sues ventre promisso, palearihus ad genua promissis. 
In the common phrase promere vinum, and Horace’s 
depromere Caecvbum, the idea is seen to prevail when 
we call to mind the Eoman habit of storing wine in 
the uppermost part of a house. To explain the sub- 
stantive propagon-, and verb propagare, Forcellini 
emplo3nB the phrase ‘depresso ramo.’ Propendere is 
a word more than once used by Cicero for the descent 
of the heavier scale in a balance. ’The verbs pro- 
scindere, provolvi, proturbare, and the participle pro- 
pexo-, seem often to require the translation ‘ down ;’ 
and lastly prodere in two of its uses involves the 
same idea : first, in prodere memoria or memoriae, ‘ to 
hand down a tradition ;’ and secondly, when it denotes 
‘betrayal or abandonment.’ This will be seen when 
we compare it with the verbs deserere, destituere, used 
in a similar sense. He who is aljout to abandon 
what has been entrusted to his care (say an infant) 
sets it down in some exposed place and then goes 
away. Where language has to speak of a series of 
acts, it often expresses the first and leaves the others 


Digitized by Google 



94 


ON THE LATIN PKEP08ITI0NS 


to be inferred, as, for example, in the military phrase 
colligere vasa, the first of many acts in abandoning 
a camp. 

But it -will no doubt be objected to much that I 
have here said that the notion of ‘ down’ is so closely 
allied to the notion of ‘forward’ that it is readily 
derived from it. When a person throws a stone, for 
instance, the forward motion is soon followed by 
descent. This is quite true, but the argument is 
applicable in both directions. ^Vlien I want to throw 
a thing down, I naturally give it something of a for- 
ward impetus, rather than drop it on my own toes. 

But I pass to another pair of words in which what 
we find in our dictionaries is not altogether satisfactory, 
profano- and profundo-. As fa-num seems to be the 
neuter of au adjective, and to signify ‘consecrated’ 
(ground), so we have a negative notion in the prefix 
of profano-. Again, 2rro/undo- should, I think, lie 
translated ‘ without bottom,’ so that CIcoto’s mare 
pro/undum et immensum contains something of ex- 
aggeration in both epithets. Now if the original 
meaning of pro be ‘ down,’ we have a use of the 
prefix parallel to what is felt in demens, dedecorus, 
decolor. 

There remains for me to say a few words on the 
verbs which I would connect with the particles re and 
pro. Calling to mind that re according to the theory 
is a corruption of a disyllabic form er-ec or or-uc, I 
am disposed to connect with it the verb ori-, which 
stands for or-ig-, as shown I think by the noun orig- 
an-. The meaning of course suits. I am also dis- 
posed to claim as of the same kin the Greek ope>^- 
e<r6ai, ‘to reach,’ when I call to mind the German 


Digitized by Google 



re AND ]yro. 95 

verbs, which obtained this idea through the prefix er. 
But from the Greek opey-etrffai I think it impossible to 
separate the Latin verb r^gere, to which the sense of 
‘ rule’ is generally without reason ascribed, in the 
belief no doubt that it is connected with the noun 
reg-, ' a king,’ whereas this word is probably of a dis- 
tinct origin and represents the Eastern raj or rajah, 
while rig- means simply ‘stretch,’ and so ‘make 
straight;’ nor indeed would I object to the translar 
tion ‘ rule,’ if the idea were limited to ruling straight 
lines. The Latin noun regio {regionis), which of 
course comes from it, is correctly translated ‘ direc- 
tion.’ A Latin g, by Bask’s law, is commonly repre- 
sented in English by a i sound ; but the law needs 
so far modification that in the south of England at 
least the k is supplanted by the palatal ch. Thus 
/rang- of the Latin corresponds to our words break and 
hi’each, broke or broken and broach. We must there- 
fore include in our family both the English verb reach 
and the English noun reach used in speaking of so 
much of a river as retains the same direction. But I 
would also claim the verbs arise and arouse, if it be 
true, as is thought by some of our best English philo- 
logers, that these do not contain any preposition, but 
have passed into rise and rov.se by decapitation. To 
these I add the Greek op-m, op-wfii, and op0-of, with 
an excrescent 0. 

I next take the particle pro, and with it the sub- 
stantive porca. The latter seems to point to the idea 
of digging ; but this in its primary form is no more 
than scratching, which as applied to the earth is rather 
a horizontal than a vertical movement. This however 
is of no moment, as downward action soon becomes the 


Digitized by Google 



96 


ON THE LATIN PREPOSITIONS 


prominent idea. The verbs of digging accordingly 
seem to have been originally imitations of the sound 
that accompanies scratching. Hence ypa^-a, 

i.e. yap-aif>-io, a-Kop-i(j>a-(i>, g(a)r-ciben of the German, 
s-c{a)r-atch and s-c{a)r-ape of our own language, and 
then with the loss of the initial guttural ap-o-<o, and 
ar{-a)-o of the Latin, together with the Greek op-vaa-to. 
But the rough liquid r is often superseded by the soft 1. 
Thus we have y{o)X-vif>-to, and s-cal-p-o, and in a 
simpler form col-o, in which the title of ‘ digging ’ to 
the first place in meanings is established, not only by 
Latin usage, but by the fact that its Scotch analogue, 
holl, had no other meaning. But let us ask what 
shape would col- of Rome take in the mouth of 
rustics. The answer is that por might well be the 
prevalent form, it being the provincial habit of the 
country outside of Rome to present a p where the 
polished dialect had a c. Thus pitpit is the Oscan 
form of quicquid ; palumbe-, ‘ the wild pigeon,’ corre- 
sponds to columbor, ‘ the tame dove,’ and the country 
people coming to Rome gave the cookshop which they 
frequented the name popina, when the city dialect 
would have preferred coquina. The doctrine that 
dig^ng WRS the first idea represented by the Latin 
particles por and pro accounts for the power of the 
preposition per, ‘ through,’ and at the same time for 
the fact that the Roman ear cared little for the dis- 
tinction between per and por, writing at one time 
porgere, at another pergere, perinde as well as 
proinde. The Greek language too, in its verb ireip-a, 
‘ I pierce,’ and iropo-s, ‘ a passage,’ presents us with 
words of the digging family, which have the precise 
form we should desire. I conclude then that reci- 


Digitized by Google 



VP AND pro. 


97 


proco- may well be translated by ‘ up and down,’ and 
80 correspond precisely to the Scotch phrase ‘rig 
and fur.’ 

In this paper I have not dwelt at any great length 
on the various secondary meanings of the Latin re ; 
but there is little diflSculty here. Besides the primary 
notion of up, and the ordinary meanings of back and 
again, there is of course the not unfrequent meaning 
of reversing an act, which j strange to say. Professor 
Ritschl seems to deny in the Prolegomena to his 
Trinummus (p. Ixxv.), where, in reference to the line 

“ Proin tu to, itidem ut Charmidstus es, rursura recharmida,” 

he ventures, in defiance of the MSS., to substitute 
decharmida on the ground that such a compound with 
re can only mean ‘ rursus indue Charmidis personam.’ 
As his authority is deservedly high, I deem it right to 
place the matter beyond dispute by a liberal quotation 
of examples to the contrary, viz. red-argu- ‘ disprove, 
refute,’ re-calcea-=excalcea-, re-can- (Plin. Xxviii. 19), 
and re-cin- (App.) ‘ reverse a charm, disenchant,’ re- 
canta- (Ov.) the same, re-cid-, implied in the adjective 
reddiuo- ‘ getting up again after a faU,’ re-cing- ‘ un- 
gird,’ re-clud- ‘unshut’ (so to say), re-auti- implied 
in recutito-, re-fell- ‘ undeceive, refute,’ re-feru- (Cic. 
Brut. 91) ‘become cool again,’ re-fibula- ‘unbuckle,’ 
re-fig- ‘unfix,’ re-fod- ‘dig up again what has been 
buried,’ re-frena- ‘ unbridle,’ re-gela- ‘ thaw again 
what has been frozen,’ re-glutina- ‘ unglue,’ re-laxa- 
‘tmloose,’ re-liga- (Catul. Lucan.) ‘unbind,’ re- 
lin- ‘imwax’ (so to say), re-ne- ‘unspin,’ re-nuda- 
‘ unbare,’ red-ordi-{r.) ‘unweave,’ re-pect- ‘un- 
comb (so to say), dishevel,’ re-pignera- ‘ take out of 
pawn,’ re-plumha- ‘ unsolder,’ re-secra- ‘ undo what is 

H 


Digitized by Google 



98 


ON THE LATIN PREPOSITIONS 


expressed by ohsecra-,' re-sera- ‘ unbolt,’ re-signa- 
‘ unseal,’ re-solu- ‘ unloose,’ re-string- (Plant.) ‘ open,’ 
re-su- ‘ unsew,’ re-teg- ‘ uncover,’ re-tend- (arcum) 

‘ unstring,’ re-tex- ‘ unweave,’ re-torque- as in ‘mentem 
laetata retorsit' (Juno in Verg.) ‘ smoothed again a 
soul so long by passion wrung,’ re-tura- (compared 
with obturor) ‘uncork,’ re-uela- ‘unveil,’ re-uinci- 
(Colum.) ‘ unbind,’ i-e-uolu- ‘ unroll,’ re-uorr- ‘ un- 
sweep (so to say) what has been swept.’ Here are 
already over forty examples ; and I have yet to add 
another, which cannot be dealt with in so summary 
a manner, for both editors and dictionary-compilers 
have done their best cither to destroy it altogether or 
to falsify its meaning. I refer to the sulistantive 
recubitu-. ’riiis appeai-s to be an ava^ Xtyofitvov. It 
is found in Pliny (xxiv. 13, 7), and to make the matter 
clear I must quote part of the passage : — ‘ Baculum 
ex ea’ (he is speaking of the aquifolia or holly) 
‘factum, in quoduis animal emissum, etiamsi citra 
ceciderit defectu mittentis, ipsum per se recubitu 
propius adlabi.’ Unhappily Sillig, abandoning the 
Paris MS. a, which he assigns to the eighth or ninth 
century, for a reading of the Paris MS. d of the 
thirteenth, gives, what is altogether senseless, per sese 
cubitu in place of per se recubitu. He probably did 
not know that recubitu would signify ‘ by ricochet.’ 
On the other hand our current dictionaries are utterly 
fiilse guides. Thus we find the word translated by 
Freund ‘das Niederfallen ’ and by his copyists of 
coume ‘falling down,’ One of these indeed assures 
us that while he took Dr. Andrews’ (American) Dic- 
tionary as ‘ the basis ’ of his labours, ‘ each article 
was compared with the corresponding word in For- 


Digitized by Google 


re AND "pro. 


99 


cellini but it is difficult to reconcile this statement 
with what is to be seen under recuhitus in the great 
Italian work, for here the true translation is distinctly 
given. After the general remark tliat recuhitus means 
‘actus recumbendi’ Forcellini adds the definite inter- 
pretation, * subsultus ille quern faciunt corpora in 
solum durum incidentia,’ that is, what in shorter 
language we call ‘ ricochet.’ This word has of course 
come to us from the French ; but that it was pre- 
viously imported into France from Italy may perhaps 
be inferred from the form of the prefix, ri in place of 
re. A genuine French noun would probably- have 
been recouchet. But over and above identity of 
meaning, there is a close connexion in form between 
the Latin re^uhitu- and tlie modern ricochet. But a 
friend suggests that I must here assume a secondary 
verb recuhicorre from which a substantive recubicatu- 
woffid readily flow, and then from a contracted 
recuh’catu- we should be led to a French re-couch-et ; 
and this view is confirmed by a similar series cubare, 
cubicare, coucher. Thus, while the obsolete cumb-ere 
meant ‘ to fall,’ the compound recurribere, reversing 
the meaning of the simple verb, must have come to 
signify ‘ to get up again,’ ‘ rise again,’ before the 
derived substantive recubitu- could bear the interpre- 
tation ‘ ricochet much as from cadere, ‘ to faU,’ we 
obtain first recidere, ‘ to rebound,’ and then the adjec- 
tive recidiuo-, ‘ rising again.’ Of course after what I 
have said above (p. 75) on Latin frequentatives, I am 
very ready to give admission to such theoretic forms as 
cubicare, and the very idea of ricochet is thoroughly 
consistent with repeated action, whether we have in 
view the movement of an oyster-shell on the surface 

II 2 


Digitized by Google 



100 ON THE LATIN PREPOSITIONS re AND prO. 

of a pond, or of a cannon-ball on the siuface of 
the sea. 

One word more in apology for the wide space I 
have given to the consideration of the German er. I 
have long thought that insufficient attention is paid 
both in grammars and dictionaries to the power which 
such little words bring with them to the composition 
of verbs; and the want is nowhere more felt than 
in dealing with the German prefixes ent and er. I 
discussed the former in my paper on ava. 1 have now 
spent not a few lines on the corresponding problem 
for er. 


Digitized by GoogI 


101 


IV. 

ON THE GERMAN PREFIX ver AND 
ALLIED FORM& 

In the paper on ava and its analogues I had to dwell 
upon the fact that inseparable prefixes are apt, as 
Grimm says, to undergo violent changes both as to 
form and power ; and upon the finther fact, that in 
consequence of this liability, particles of totally dif- 
ferent origin not unfrequently pass into an identity of 
form, thus bringing about a confusion, which ends in 
the disuse of both prefixes, and the employment of 
fresh forms of speech to make up the consequent 
deficiency. It is probably in this way that our own 
language has pretty well ceased to employ prefixes 
in the formation of compound verbs, finding it more 
convenient to take the simple verb and place after it 
an independent preposition ; as, ‘ he put upon me, he 
put me quite out, he put me up to something, he put 
this bad practice down, he put offi\ie meeting, he put 
the door to, he put his hat on, he would not put in 
more than sixpence.’ Similarly we have take up, take 
off, take in, take to, with peculiar meanings, which 
give much trouble to the foreigner, as they are often 
left unexplained in our dictionaries. We have now 
no verbs compounded with a prefix to, although the 
Anglo-Saxon had many such, including both those in 
which to was an equivalent for the Latin ad, and 


Digitized by Google 



102 


ON THE GERMAN PREFIX ver 


others with to = Greek Sta, dis of the Latin, tor of 
Old Norse, and zer of German, as to-brek, to-breat, to- 
hew, to-race, to-rend, to-shred, to-skatir, to-smnk, of 
Chaucer ; to-cleve, to-dele, to-drag, to-po.rt, of the 
‘ Ayenbite of Inwyt’ (a.d. 1340). Even the Bible 
(Judges ix. 53) has the words, ‘ A certain woman cast 
a piece of millstone upon Abimelech’s head, and all 
tobrake his scull ; ’ — and so also in Shakspere we 
find : — 

“ Whore (>.e. whereaa) these two Christian armies might combine 
The blood of malice in a vein of league. 

And not totpend it so unmannerly.” 

King John, v. 2. 

“ Then let them all encircle him about, 

And, fairy-like, topinch the unclean knighi” 

Merry Wivei of Windwr, iv. 4. 

A simUiir case of the accidental confluence of pre- 
fixes originally distinct, and a consequent disappear- 
ance, is to be seen in the particle of. Here three 
independent words have fallen into an identity of 
shape. Thus in the older forms of our language, 
whether called Anglo-Saxon or Old English, we have •. 
this little word representing in turn what appears in 
Latin and Greek as ab otto, ob em, sub xnro. Examples 
of the first abound in Gothic in the form af, and also 
in Anglo-Saxon, both in the form ofa, as o/a-dri/an, 

‘ to drive off,' ofa-heavmn, ‘ to cut off,’ and of-ferian, 

‘ to cany off,’ of-iman, ‘ to run off’ The Danish and 
Swedish also have numberless instances ; but here, as 
in Gothic, the original vowel was preserved, af, not of 
With the word tm, I have already claimed to con- 
nect our own aft, the t being excrescent ; as it is also 
in aft-ana of the same language, compared with ut-ana. 


Digitized by Google 



AND ALLIED FORMS. 103 

our own aft-er, and the old superlative aft-uma. But 
ftn has yet another representative in the prefix of 
the Anglo-Saxon of-axian ‘ ask after/ of-ridan, ‘ ride 
after/ of -spy ran ‘search after/ and of-sitan, ‘besiege’ 
(obsidere) ; as also of the Old English of-seche ' seek 
for/ of -sends ‘ send for’ (‘ Ayenbite of Inwyt/ Mr. 
Morris’s Preface, p. Ixvi.), and perhaps of the Anglo- 
Saxon of-licgan ‘ to lie upon.’ 

Taking suh and xnro next, I quote from the paper 
on ava, p. 41, the forms in our northern dialects which 
I hold to represent these ; viz. Goth, xif Old and 
Mid. Germ, uf Mod. Germ, aif. Old Fris. op or up, 
Dutch op, Norse and Swed. up>p, Dan. up, Eng. up, 
but Old Eng. also of Now in the examination of 
the German auf I find much that reminds me of what 
I came across in the study both of ava and its repre- 
sentatives and of the Latin re. Thus for German, 
leaving out of view the numberless instances where 
the idea of up is distinctly retained, I find (1) above 
sixty where the idea of ‘ opening’ appears ; (2) some 
eight of ‘beginning/ auf-bluhen, auf-brausen, auf- 
jammem, auf-krdhen, auf-kreischen, auf-lachen, avf- 
seufzen, auf-toben ; (3) of ‘ loud noise/ six : auf-lachen, 
auf-rbcheln, auf-sagen = recitjire, auf-snarchen, auf- 
sclmauben, auf-stbhen ; (4) full fifty where ‘completion’ 
is denoted ; (5) with the notion of ‘back mif-behalten, 
or auf-beivahren ‘reserve/ auf-halten ‘hinder’ (— inhi- 
bere), auf-krdmjjeln, auf-streifeln ; (6) not less than 
fifty meamng ‘ again / and (7), what is of much in- 
terest, nearly fifty in which the idea expressed by our 
English prefix un, i.e. the reversal of a former act, 
shows itself, viz. : — auf-binden, auf-<lecken, auf-drehen, 
auf-driesoln, auf-eisen, auf-fadeln, auf-faltcn, auf-fasen. 


Digitized by Google 



104 


ON THE GERMAN PREFIX ver 


auf-fitzen, auf-flechten, auf-gUrten, auf-Mkeln, auf- 
haken, auf-hefteln, auf-hefben, auf-hUllen, auf-klinken, 
auf-knebeln, auf-knopfen, auf-kniipfen, auf-koppeln, 
auf-kiinden, auf-lassen, auf-leimen, auf-losen, auf- 
lothen, auf-nesteln, auf-packen, auf-rebbeln, auf-riegeln, 
auf-ringeln, auf-rollen, auf-sagen, auf-scbliessen, auf- 
flchnallen, auf-sclmUren, auf-schnurren, auf-schrauben, 
auf-schtirzen, auf-spunden, auf-thauen, auf-weben, 
auf-weicben, auf-wickeln, auf-winden, auf-wirken, auf- 
wirren, auf-zaubem. 

To these I add two corresponding Swedish verbs, 
upp-tdcka = auf-decken and upp-losa — auf-Hisen. 

The Anglo-Saxon, besides its many verbs compounded 
with of = ‘ otf ' (Latin ab), retained a few, as we have 
seen, where q/" = ob or m-t ; and again it has a small 
group in which of — the Germ, auf as of-slandan = 
auf-stehen, of-gifan = auf-geben, of-delfan ‘ dig-up,’ 
offrettan = auffressen, of-haebban, ‘ retain,’ like auf- 
halten; with which we should no doubt include of- 
blindan ‘ make blind,’ of-munan ‘ remember,’ of-lician 
‘ dislike,’ of-thincan ‘ repent,’ of-unnan ‘ refuse,’ from 
unnan ‘ give.’ And I further quote again from Mr. 
Morris’s Preface to his edition of the ‘Ayenbitc of 
Inwyt,’ or from the body of the work : of-thincke — 
‘forihinh,’ ‘ repent,’ of-guo ‘ forgo,’ of-healde ‘ with- 
hold,’ of-take ‘ overtake,’ of-serve ‘ deserve.’ The last 
two verbs seem to call for a little explanation. If of- 
take and over-take mean, as by etymology they should 
mean, ‘ catch up’ we have what is still a familiar 
phrase for the same idea. Then of-sei've may well 
mean, like the Latin emeritus, ‘serve out one’s full time.’ 
The same old English work which supplied these ex- 
amples has also of-acksed for ‘ thoroughly questioned,’ 


Digitized by Google 



AND ALLIED FORMS. 


105 


of-dret for ‘thoroughly frightened,’ and of-tyened 
for ‘ thoroughly enraged.’ Of-guo and of-guoinge also 
occur in the index of the same book, with the trans- 
lation ‘ meriting or deserving,’ but how this meaning 
is to be assigned to them I do not see. ‘ Overgoing,’ 
like the Latin ‘ transgredi,’ might mean transgression, 
passing the border of what is right, and so correspond 
to the German vergehen ; but we cannot assign to the 
simple of ‘ up,’ the notion of the comparative over. 

But as the words tmth, hind, as has been just noticed, 
lead to secondary forms, vnd-er, hint-er (hinder), and 
the simple preposition in to int-er in Latin, und-er 
Eng., unt-er Germ., so two at least of our prepositions 
also assume a comparatival suffix. Thus, to take first 
the forms allied to ctt*, we have, as has been already 
said, aft and after in English, together with the Gothic 
afar without the excrescent t ; and in the Ang.-Sax. 
ovemoon (I take it from Bosworth’s Dictionary) we 
have probably a variety of our afternoon and no com- 
pound fix>m the more familiar preposition over. 

Whether far and farthert stand in the relation of 
comparatives to otto ah and the Ang-Sax. cf ‘ off’ I 
will leave for future consideration. But on the other 
hand, as the Latin has alongside of sub both super 
and subt-er with an excrescent t (pronounced supt-er 
and frequently so spelt in good MSS.), and as the 
Greek too has irtp from inr of inro, so we find both 
upper and over in English and idier in German. 

But when a preposition has given birth to derived 
forms, it is very common for the initial vowel in 
such derivatives to disappear. This is a matter which 
has been considered at some length in the second 
paper on evt ; and in accordance with what was there 


Digitized by Google 



106 


ON THE GERMAN PREFIX Ver 


said I now venture to claim the familiar Germ, prefix 
ver as a corrupted comparative of auf, in other words, 
as a decapitated variety of Uher ‘ over.’ The corre- 
sponding Ang.-Sax. prefix of-er has in its f the very 
sound which is heard in the initial consonant of ver. 
But the best proof of the substantial identity of the 
two forms will be found in the meanings, as seen, first 
in the following individual words : ver-hriicken ‘ to 
bridge over,’ Ang.-Sax. ofer-hrycgean ; ver-jdhren and 
ver-alten ‘to become superannuated,’ compared with 
Ang.-Sax. ofer-geare ‘antiquated,’ and o/er-eald the 
same ; ver-kehren ‘ overturn,’ ver-fahren and ver- 
fiihren ‘ transport ;’ ver-schlafen (sick) ‘ oversleep ’ 
oneself, ver-schiessen ‘ overshoot,’ ver-schlagen ‘ strike 
(a ball) out of bounds,’ ver-sprengen ‘ strike a billiard 
ball off the table ;’ ver-walten ‘ administer,’ and ver- 
lueser ‘manager,’ i.e. ‘ one set over others;’ ver-hsen 
‘ call (names) over :’ ver-hehlen ‘ cover over’ (conceal) 
= Ang.-Sax. ofer-helan ; ver-sehen ‘ over-look,’ i.e. 
‘ neglect,’ ver<tchten ‘ overlook,’ i.e. treat with con- 
tempt, ver-dunkeln ‘darken over.’ Ver-nehmen I 
would place alongside of the Lat intellegere, and as I 
assign to this for its literal translation ‘ pick up (know- 
ledge),’ so ver-nehmen may well mean ‘ to take up,’ 
like the Scotch uptake. (See paper on ava, p. 28.) 

Secondly, I set down a whole class of words in 
which the notion of ‘ over’ (‘ covering’) in its physical 
sense is undeniable : verblechen, verbleien, verdachen, 
verdielen, vereisen, vergittem, verglasen, vergolden, 
verkleiden, verlacken, verlarven, verlatten, verledem, 
vermanteln, veimoosen, vermorteln, verpanzern, ver- 
pichen, verquecken, verrasen, verreisern, verrinden, 
versanden, verschalcn, verschalmen, verschienen, ver- 


Digiiized by Google 



AND ALLIED FOKMS. 


107 


schilfen, verschindeln, verscblammen, verschleiern, ver- 
schmutzen, vcrsclmeien, versilbem (Corap. Ang.-Sax. 
ofer-sylfriaii), versteinen, veratobern, vertafeln, vcr- 
zaunen, verzinken, verzinnen, verzuckern, together 
with verdecken, verhUllen, verraalen. 

Thirdly, a class of words with the meaning of ‘ over- 
much, excess verbluten ‘ bleed to exhaustion,' ver- 
Jliegen ‘ fly too far,’ verhitzen ‘ overheat,’ verklettern 
‘ climb too high,’ verkochen ‘ overboU,’ verpfeffem 
‘ pepper too much,’ verrennen (sich) ‘ run too far,’ ver- 
salzen ‘ oversaJt,’ versauem ‘ make too sour,’ verschnei- 
dem ‘ cut too short,’ verschwarmen ‘ swarm too much,’ 
verspdten {sich) ‘ come too late,’ versteigen {sick) 
‘ climb too high,’ versOssen ‘ oversweeten,’ verv)U7'zen 
‘ spice too much,’ vei'zdrteln ‘ s]X)il (a child) by too 
much tenderness’ (with auf-zdrteln the same), 'ver- 
zuckern ‘ sugar too much.’ 

Fourthly, with the notion of transferring, and so 
changing, bartering, selling, paying : verandern, ver- 
deutschen (comp, the general term Ubersetzen), vergrie- 
chen, verfahren ; vcrtauschen, verkaufen, verwechseln. 

Fifthly, the notion of ‘ passing over,’ ‘ getting to the 
end of,’ and so ‘ consuming all,’ of which the examples 
are too numerous to quote. 

Sixthly, the notion of excess is akin to that of mis- 
doing, as verdeuten ‘ misinteiqiret,’ verdrehen ‘dis- 
tort,’ vei’drucken ‘ misprint,’ verheben ‘ lift in the wrong 
way,’ verkalben and verlammen (cf. our 7uiscarry), 
verkennen ‘ mistake,’ verleitcn ‘ mislead,’ ven’athen 
‘ betray,’ verrech'nen ‘ misreckon,’ ven'iicken ‘ derange,’ 
verchieben ‘ misplace,’ verschleppen ‘ misplace,’ ver- 
wiegen (sich) ‘ make a mistake in weighing,’ verwbhnen 
‘ spoil (a child),’ vei'zdhlen ‘ misreckon,’ verziehen ‘ draw 


Digitized by Google 



108 


ON THE GERMAN PREFIX ver 


wrong,' verzielen (sich) ' miss one’s aim.’ And with 
these may be included the notion of destruction or 
negation in the words verhieten ‘ forbid,’ vergessen 
‘ forget’ = Ang.-Sax. o/er-gitan), verhoren (= iiber- 
horen ‘ not to hear,’ and Ang.-Sax. qfer-herran or 
ofer-hyran) ; verlemen ‘ unlearn,’ verthun (Ang.-Sax. 
ofer-don, our fordo) ; verschwbren ‘ forswear,’ with 
which compare the Ang.-Sax. ofer-cymn. 

So far only the fuller forms (in ofer) of the Ang.- 
Sax. have been quoted ; but of course this language 
has for as the ordinary form of the prefix, like the 
Danish for and the Swedish for. And the fact that 
Ang.-Sax. possessed both forms adds greatly to the 
argument which treats the German ver as the result 
of decapitation. 

It must no doubt be admitted that it is difficult at 
times to find in the prefix an explanation of the mean- 
ing which it conveys to a verb. But the verb ver- 
dienen may well mean ‘ serve all one’s time,’ and so 
correspond to the old English of -serve above-men- 
tioned, as having the meaning of ‘ deserve.’ But the 
best proof that the prefix ver is but a decapitated 
iiher is found in the German vocabulary itself, as v^l 
at once be seen by prefixing first ver and then iiher to 
the following, and so testing their substantial identity ; 
-blechen, -bleien, -brlicken, -dachen, -decken, -dunkeln, 
-fahren, -gattem, -glasen, -golden, -heben (sich), -hbren, 
-kochen, -lassen, -ledern, -moosen, -olen, -pfeAfern, 
-pichen, -salzen, -schiessen, -schleien, -schneien, 
-schnUren, -sehen, -springen, -tafeln, -zinnen. 

In Latin it has been for a long time the habit of 
scholai's to identify with our own for of forswear, 
fordo, forlorn, foiget, forgwe, the per of 2)eriurus, 


Digitized by Google 



AND ALLIED FORMS. 


109 


perdo, pereo, perimo, perjidus ; and no doubt with 
reason : and to these may be added peruideo, as found 
in Horace’s 

“ Quom tua peruideas oculia male lippus inunctis, 

Cur in amicorum uitiis tarn cemia acutum t ” 

for it seems to have been an unnecessary proceeding 
on the part of Bentley to give up the reading of all 
the other MSS. in favour of one. It is scarcely a 
grave difficulty that the Latin language also used 
peruidere in another sense, seeing that we give two 
different meanings to our overlook, and the Germans 
to their versehen and Ubersehen: But the Latin seems 
to have examples where the prefix per has the original 
meaning of vher, over — viz. percell- ‘ knock over,’ 

‘ upset,’ as seen in the use of this word in the very 
oldest writers, e.g. with j^ldustruni in Cato, quercm in 
Ennius, to say nothing of Plautus and Terence, both 
of whom have the word in its true physical sense. 
Peruert- again means ‘ overturn,’ ‘ upset,’ as with 
pinus proceras (Enn.), aulas and turrim (Plant.), and 
especially in ‘si r6x obstabit dbviam, regem ipsum 
prius peru&rtito’ (Stic. ii. 1, 14). The meaning ‘ over’ 
gives the best interpretation to the compounds per- 
fund-, perlin-, and perung-. A further claim must 
be put in for the intensive per of adjectives, like 
permagnus, permultus, especially when we compare 
this with xnrep as a prefix to the adjectives -aaOevrfs, 
-OTititos, -e\a<f>pof, -koKos, ~\afiirpos, -iriKpot, -iroKvt, 
-<rotf>ot. The assumption here made is that super was 
cut down to per, and the loss of two letters may 
offend ; but such loss may well have been gradual, 
first one letter disappearing, and then the other. Thus 


Digitized by Google 



110 ON THE GERMAN PREFIX ver 

the Lat. suh (pronounced sup) seems to have lost its 
sibilant in one derivative, viz. : apeno apertus, the 
root syllable of which is the same as op of our op-en 
and our preposition up. This will be more readily 
accepted if contrasted with op-erio op-ertus. That 
ap-eri- and op-eri- (with their participles ap-er-to- 
op-er-to-) have their origin in prepositions is a matter 
on which I have little doubt, the en of these words 
corresponding to the eli of sep-eli (sep-^d-to-), the root 
of which is one with 6air of Bawr-oa ; and indeed a 
suffix er or el is well known in our own language as 
in quiver, shiver from quake, shake, gambol and 
gamble from game. But as to the root-syllables ap 
and op, my first thought was directed towards ab 
‘ from ’ and ob ' to,’ for ‘ opening ' is separation and 
‘ shutting ’ is re-union, and indeed we ourselves have 
the phrase ‘ put to,’ in the sense of ‘ shut.’ The other 
alternative was to look out for prepositions signifying 
‘ up ’ and ‘ down,’ which would correspond to our 
phrases ‘ put the window up ’ and ‘ put it down.’ 
Now our own verbs ope and open, the Germ, o^en 
and bffnen, the Dutch open and openen connect them- 
selves beyond a doubt with the several prepositions 
up, auf, and op, to say nothing of such a word as the 
German aufmachen ‘to open.’ But where am I to 
find op ‘down’ for ‘ op-eri-’ 1 I answer, in the ob 
of occid- ‘ fall down,’ ‘ die,’ ‘ set as the sun,’ occid- 
‘ cut down,’ oppet- and obi- ‘ go down,’ i.e. ‘ die,’ 
occuba- ‘ lie dead,’ obter- ‘ tread down,’ opprim- 
‘ press down.’ The Sanskrit gives this prefix in the 
shape ava ‘ down,’ and the Latin also has it in the 
ab ‘ down’ of abici- (abiecto-) ‘ throw down,’ ah-sorbe- 
‘ suck down,’ aJfUg- ‘ dash down,’ appos- (appon-) 


Digitized by Google 



AND ALLIED FOKMS. 


Ill 


‘ set down.’ The German too has this very form with 
the sense of down in several words, especially in 
he.r-ah, hin-al), and ah-wdps ‘ downwai-d.’ I am not 
blind to the fact that on this theory op-eri iniglit as 
well have signified ‘ to open,’ and aperi- * to shut’ 
But language is somewhat arbitrary and imcertain in 
such matters. 

The adverb perendie is sometimes spoken of as a 
liybrid word, made up of the Latin die and the 
Greek mpav, or Sansk. param ; but here again, with- 
out denying the identity of the three words, irepav, 
param, and peren, I would claim the last as a 
native and a corruption of superen, whence the 
adverb supem~e ‘ from above ’ (wdth a suffix like ind-e, 
und-e, and the Greek oiriaO-e, &c.) so that peren- 
die shall correspond in its first element to the Uber 
in Uber-moi'gen of precisely the same meaning. This 
theoretic peren or peran exists, though a little dis- 
guised, in the contracted form Iran. Before an r the 
Latin language was much given to a change of con- 
sonant. Thus trem-ere, as has been often noticed, 
is a corruption of cremere (Fr. craindre and Old Fr. 
cremir). A thoroughly parallel example to our theory 
about trans is seen in the adjective tranquillus for 
planquillus, a double dim. of plancus, as that again 
is a dim. of planus. The verbs trddo, traduco, trdicio 
seem formed from Iran rather than trans; and the 
co-existence of two forms, one with, one without an s, 
is seen in many of the prepositions, as ah, sub, ob, ec, 
di.' It may be noticed too tliat this theory brings the 
French particle tris ‘ very,’ so commonly used with 
adjectives, into immediate connexion with the Latin 
per and Greek «5»re/3 of the same habit Even within 


Digitized by Google 



112 


ON THE GERMAN PREFIX ver 


the limits of the Latin language we have what is 
really identity in the two forms trans-fUga and y>er- 
/uga ■ deserter,’ or more literally in German, Uber~ 
Idufer. 

In the Greek language I would first observe that 
certain compounds with the full form vvep present 
peculiar meanings which go far to support the doc- 
trine that ver is a shorter form of ilber, viz. inrepopaa 
xnrepiSeip ‘ overlook,’ and inrepovrot ‘ slighted ’ compared 
with versehen, overlook, and verachten. Indeed, the 
simple verb achten ‘ to look ’ is probably the analogue of 
the Greek oirr-ofuu, for the ir of this word appears as a 
c in the Latin oc-ulus, while ac itself is seen in the 
Lith. ak-i-s, as also in ac-tutu-m ‘in the twinkling 
of an eye,’ ‘ in einem augenblicke,’ to adopt Dr. Ebel’s 
explanation of the word (Kuhn’s Zeitschrift, iv. 320) ; 
and then a c in Latin should be represented by ch in 
German, while the o again should give place to an a. 
In both acht and om- the t I regard as excrescent. 
Again, in tnrepaj<ovi^o/iat, inreppMyppMi, inrepaXyew, 
itrepaypvTrveto ‘ fight, &c. for ’ — , we have an explana- 
tion of the ver in verfechten and vertheidigen ; and 
above all xnrep-airoKpiv-op^t ‘ answer for ’ corresponds 
with aU accuracy to ver-antworten. Then again, the 
negative power of ver, though in itself it means ‘ id)er,' 
is in agreement with the use of vrrepKoipot ‘over or 
beyond the time,’ hence ‘ at wrong times,’ like axaipot 
— to quote the words of the Lexicon. 

But the Greek also seems to have a decapitated 
variety ; irap for inrep, or irapa for inrapa ; for I cannot 
but think that language was in possession of two dis- 
tinct words irapa, which have accidentally taken the 
same form. From the words in which irapa means 


Digitized by Google 


AND ALLIED FUKMS. 


113 


‘ presence,’ or ‘ by the side of,’ I would separate those 
in which is to be found the idea of ‘ over,’ or ‘ what 
is wrong,’ or ‘negation.’ No doubt there are cases 
where ‘ passing by ’ and ‘ passing over ’ afford equally 
good explanations of the excessive. But to irapainjSa- 
‘ leap over’ and so ‘transgress,’ and irap-opi^-, alike 
in the sense of ‘ outstepping one’s own boundaries,’ or 
‘driving another over them into banishment,’ the 
notion of over seems alone applicable. So with irapaipe- 
when used either of ‘transferring a curse,’ or ‘drawing 
over to one’s own side.’ The verb irapaipTipi may well 
mean ‘persuade,’ if it have for- its literal translation 
‘ to talk over,’ ‘ iiher-reden! The same applies to 
irapaTreida). In wapoBiStupi and irapaXa/tfiavo}, irapa 
‘ over ’ is suited to denote a transference of property, 
like trans in the Latin trado. Then in not a few 
words the notion of covering is added by trapa, which 
suits well, if this be a variety of \mep, as irapaKaXvirr-, 
and, what is probably the same word at bottom, 
TrapaKptnrr-, irapaXei^, 7rapa/»7re^-, irapaireiaXo- ‘ covered 
with plates (of silver, &c.),’ TrapaveTawv-, irapairrfKtaTo- 
‘ besmeared with mud,’ irapairtapa^- ‘ cover with a lid,’ 
TrapaaKfjvo- irapairTri^-, 'irapaj(pt-. So, too, irapaivKaaa- 
will bear the translation ‘ watch over,’ irapovpo- that of 
‘ one who watches over.’ Again I cannot but give the 
preference to mapa — iiber over irapa ‘ by the side of,’ 
in such cases as irapav8e-= ‘ verbluhen,’ irapayrtpa-, 
irapoKpM^-, iraprfpa-, irapijXiK-, compared with veiyahren, 
veralten, the Ang.-Sax. ofer-geare, o/er-eald, and our 
own superannuated. So also napa^Xeir- 'irapopar must 
go with Horace’s peruidc-, with the German iibersehen 
and versehen of like meaning, Ang.-Sax. for-seon 
‘ despise ’ and our own overlook. Here indeed some- 

I 


Digitized by Google 



114 


ON THE GERMAN PREFIX VC7' 


thing is to be said for those who find the explanation 
in ‘ looking by the side of ; ’ but when we come to 
the sense of hearing the explanation fads. Still there 
can be no doubt that we must class irapaKov- and 
irapaKpoa- (r.) ‘ hear wrong,’ or ‘ fail to listen to,’ with 
the German verbs iiher-horen, verhoreyi {sich), and 
the Ang.-Sax. o/er-heoran or ofer-hyran, all of like 
meaning. Then the adjectives irapaOeppo- ‘ over- 
hot,’ ‘TraparoXfio- ‘ foolhardy,’ irapavffTrjpo- ‘ exceedingly 
austere ’ (I take the translations from L. and S.), are 
at one with the German iiberlang and Greek {nrepXap,- 
irpo- &c. while Ttapapo- corresponds to ivepKaipo-, To 
these I add two rare words, which alike seem to show 
that irapa is a corruption of xmapa, viz. ‘ vapairayot 
(or vapTrayos) the upper bolt of a door (Hesych),’ 
and irapaaeiop ‘ a <oy)sail ’ = Lat. supparum. As to 
<Tuov of vapaaetov, I would suggest the possibility of its 
having supplanted a fuller aex-tov, which as a dimin. 
of a foi-m aex, would correspond to the German seg-el, 
our sail ; and, on the other hand, the word sujyparum 
(there is also a reading siparum with a single p) seems 
to confirm the argument that •irapa of rrapaatiov is, as 
elsewhere, a corruption of a fuller imapa. 

A form a^x is commonly assumed as the base of 
6X-W ; and, again, the Latin ueh-o must be one with 
e^-w, as shown by the meaning of the derivatives 
ox-e<T- (cf. the Homeric o;^eo-(tn/), and oxo- ‘ a vehicle.’ 
But a sail is the chief instrument in the movement of 
a ship. Hence uelum, i.e. ueh-el-um (cf. for the guttural 
uexillum, oxiA pauci, paulum, pauxillum), the German 
seg-el and the theoretic aex-iov will be all diminutives 
from a common root. The appearance of a <7 by the 
side of a is not to be explained by any interchange 


Digitized by Google 


AND ALLIED FORMS. 


115 


between two sounds so utterly unlike. We should 
rather start from a base <rox, which, when followed by 
a weak vowel (t or e), would give way to the ‘um- 
gelautet’ swech, and so to wech or ueh. 

With this evidence before me of the close connexion 
between Trapa and the prepositions virep, iiher, and ver, 
I am strongly of opinion that we should in all the 
cases where the notion of wrong or negation lies in 
the prefix wapa, claim it as a totally different word 
from vapa ‘ by the side of.' 

But I must return for a short time to the Teutonic 
family of languages, to say a few words on a Norse 
prefix which seems to be allied to those wdiich we have 
been discussing, viz. of, in the sense of too much, as 
given in Rask’s Grammar (§ 302), viz. in of-mikill 
‘ too much,’ of-gamall ‘ too old,’ of-snemma ‘ too 
soon,’ of-seint ‘too late,’ of-dt ‘gluttony,’ of-dryckja 
‘ drunkenness.’ Here aU would have been intelligible, 
if instead of the simple of we had had some compara- 
tive as ofer, for then we should have had forms cor- 
responding to the Ang.-Sax. ofer-eatan ‘to overeat’ 
and ofer-etol ‘ a glutton,’ &c. Possibly the explanation 
may be that, in accordance with the law that irregular 
comparatives, just because of that irregularity, are per- 
mitted to drop the final suffix, for example bet, mo, 
leng, less, of our own language, in place of better, more, 
longer, lesser ; ma (mdlo), sat, aut, in Latin, in place 
of magis, satis, alteram, so this Norse of may be a 
curtailed variety of ofer. This would serve to justify 
the use of of-tycned ‘ very angry,’ and ofguoing ‘ trans- 
gression’ ('The ‘Ayenbite of Inwyt,’ pp. 66 and 215). 
But for the case of of -take (p. 104), I have already 
provided in another way. 

1 2 


Digitized by GoogI 


116 ON THE GERMAN PREFIX Ver AND ALLIED FORMS. 


In dealing with a German particle, I have the 
feeling that I am an intruder, and the more so as my 
knowledge of the language is a very loose one. At 
the same time an outsider is often alive to difficulties 
which never present themselves to a native, for the 
simple reason that a mother tongue is acquired with- 
out much exercise of the reasoning powers. It is 
probably due to this cause that there are so few good 
grammars of any modem language. At any rate in 
my own case it was the inability to find satisfactory 
explanations of the inseparable prefixes of German 
that induced me at different times to look with some 
care into the facts as presented by dictionaries, and 
then to connect such words with the equivalent forms 
of the kindred languages. It is in this way that I 
have dealt with the inseparable prefixes ent, unter, 
er, zer, and ver, and their representatives. I do not 
expect my views to be blindly accepted. It will be 
enough if I stir up an intelligent inquiry into the 
subject on the part of those who are better qualified 
to deal with it. 


Digitized by Google 



117 


V. 

Post AND after TUE SAME WORD. 

My first sketch of this paper, as it was drawn 
to bo read before the Philological Society, I wrote 
without any knowledge of what Ritschl (Rhein. Mus. 
n.f. vii. 573) and G. Curtius (Kuhn’s Zeitschrift, i. 268) 
had said before me. Where our views agree I find 
much in their testimony to strengthen my argument ; 
and where we differ I will give my reasons for the 
difference. 

I begin with the assertion that in both these words, 
post and after, the letter t is excrescent (see my paper 
on this subject below) ; in other words, that the more 
genuine, or, as the Germans say, the more organic 
forms are pos and afer. 

As regards the assumed pos, I first pointed to the 
old dactylic inscription (Mommsen’s oil. 1454) : ‘Qiir 
petis posUmpus ^ consilium ; qudd rogas ndn est ; ’ to 
another inscription (Orelli-Henzcn, 6561) ‘ . . . lib. 
libertabus posrisq. ( = posterisque) eorum ; ’ to the 
Umbrian pustertiu of the Iguvian Table (1 b. 40, 
Huschke) ; and to the phrase pus-ueres of the same 


* In tho instances postemput, pustertiu, posterganeus, postu, pos- 
templum, tho t might have been supposed to bo performing double 
duty but fur the abundant examples where the word following ^>os 
begins with other letters. 


Digitized by Google 



118 Post AND after the same word. 

(la. lines 7, 14, and 24), opposed to pre-ueres (lines 
1 1 and 20), which are respc(!tively translated ‘ post 
portas or rauros ’ and ‘ ante p. or m. ; ’ to the adjec- 
tive paste rganeus of late writers; to what Velius 
Longus (2237, 13 P.) quotes from Cicero’s Orator (43): 

‘ posmeridianas quoque quadrigas libentius dixerim 
quam postmeridianas' (for which last word Ritschl, I 
now find, suggests the substitution of pomeridianas) ; 
to i^omerium and pomeridianus, as naturally gi'owing 
out of posmerium and posmeridianus, just as reimis 
stands for resmus, and Camena for Casmena; to 
posquam, as given by one of the most trustworthy 
MSS. of Catullus (xi. 23), and adopted by Bergk ; 
to ^os legem and pos te, said by Diez (Gr. iii 215) to 
occur in the Agrimensores ; and lastly, to pone, as 
standing for pos-ne, and so receiving its best expla- 
nation. Cf. poflo for posno, the root pos alone appearing 
in posui, qwsitiis, while the n of the imperfect tenses 
performs the aime office as in sterno, cerno, spemo ; 
for I unhesitatingly reject the theory which would 
make pono a compound of sino. The old form posiui 
seems at first to support this view ; but compare 
quaero quaesiui, peto petiui. In fact it is common 
for consonant-verbs and i. -verbs to coexist, and then 
interchange their tenses, as in uenio ueni,/arcio farsi. 

To this evidence I now add from the MSS. of 
Plautus, as pointed out by Ritschl (1. c.), 1, posquam, 
Glor. 124, CDa ; 2, posquam, Bac. 277, Ba; 3, pes- 
quam {fox posquam), Poen. Pr. 104, BCD; i, poshac, 
Pocn. i. 2, 66 ; 5, postu {— post tii), Trin. 975, BC ; 
and then in abbreviated form with the symbol p’, 
which he with reason inteiqirets as = pos (or perhaps 
p«s?) ; 6, qTi’quom, Glor. 121, B ; 7, p’lci (for j)OS id). 


Digitized by Google 



Post AND after THE SAME WORD. 119 

Trin. 529, D ; 8, p’hunc, Men. i. 2, 3, C ; 9, p’qui 
(where other MSS. have postquam), Glor. 1331, Da ; 
10, Pquam (with the mai’k ’ accidentally omitted, he 
thinks). Pseud. 1,269, C ; 11, p . . ilia (with two letters 
erased). Men. v. 9, 58, Ba {p'illao Bb) ; and then from 
other sources, for which also I am indebted to Ritschl, 
12, postemplum, and 13, poscolu[mnani], Iscr. Marini 
Atti, &c., 182 and 258; 14, pusmeridianae, Cic. 
Att xii. 53, Med. MS. ; 1 5, posquavi res Asiae, ap. 
Marium Victorinum, 2,467, Putsch. 

That pos rather than post is the older form is 
further confirmed by the Sanskrit pap-cat ‘ post,’ and 
the S. adj. pap-cha ‘ after-coming,’ with its super!, pap- 
chrima ; as also by the Lith. ‘ postea,’ and 

Old Prus. pam-dau of like meaning (Pott E. F. i. 88, 
and Bopp’s Gl.). Part of this I take from Curtius {1. c.). 

Hence I confidently assume that the t of 2 ^ost is 
excrescent, as in the Germ, morast, our morass, Fr. 
marais. Germ, palast, Fr. palais ; or, to draw examples 
from the classical languages, ost-ium by the side of 
os oris, and oar-eov by the side of os ossis. 

Our preposition ajter, and the Gothic aftm, Bopp 
(V. G. § 295, vol. ii. p. 28 of the second edition) would 
divide so as to attach the t to the sufSx, as af-ter ; and 
on the same principle he places the hyphen before the 
dental in the Old Germ, prepositions or adverbs, ivi-dar, 
ni-dar, hin-lar, sun-dar, for-dar. But that such a 
proceeding is erroneous is proved by the form of our 
simpler words aft {ah-aft), with, neath (be-nealh), hind 
{be-hind), and forth. Had our adjective yonder fallen 
under his view, he would no doubt in aU consistency 
have divided it yon-der ; but here again we once liad 
a simple yond as well as yon, and still have be-yond. 


Digitized by Google 



120 Post AND after the same word. 

Of course he is thus driven to erroneous results. For 
example, his division of the Gothic wi-dar. Germ. 
wi-der, calls for some explanation of the first syllable ; 
and accordingly he finds this in the Sanskrit particle 
vi, which denotes, he says, ‘separation’ (‘trennung’). 
Had he thought of the identity of the English verb 
withstand, and the German understehen, widerstand, 
he would probably have admitted that wid, Eng. with, 
is the root-syllable, with the meaning of ‘ union’ instead 
of ‘ separation ;’ for though union often includes the 
idea of peace and friendship, a meeting may also be 
preparatory to hostile proceedings. Tecum coniungi 
generally implies an amicable union, but then we have 
also tecum pugnare as well as congredi, and conferre 
signa. A thoroughly parallel case is seen in the Latin 
contra and its primitive con or cum, two words which 
render it impossible to doubt the connexion of voider 
with our ^oith ; and though contra very often carries 
with it the idea of hostility, yet such is certainly not 
the case with the Fr. contre-danse, and our English 
equivalent but misspelt countiy-dance, where the 
parties are brought indeed face to face, but not 
for war. 

The Gothic aft-ana, too, compared with ut-ana, &c. 
goes far to prove that the t belongs to the first syllable. 
Accordingly, as Bopp himself remarks, Grimm gave 
a preference to the division aft-uma for the Gothie 
superlative. In a subsequent paper I hope to show 
that generally er rather than ter or ther is the suffix 
of comparatives, thus giving a preference to what is 
seen in ev-tpoi, sup-eri, inf-en, and making the division 
ao^T-epos, fSeXr-epos ; ij-epos, alt-er, Oth-er ; iror-epos, 
ut-er, %vheth-er. 


Digitized by Googl 


121 


Post AND after the same word. 

For the present purpose I have still more valuable 
evidence : first, in the Gothie forms afar and afara; 
secondly, in the Sanskrit apara ‘posterior,’ and the 
Vedic apama ‘postremua’ 

But I am also called upon to justify the assumption 
that an / in our Teutonic family has a tendency to 
throw out a t ; and I give as examples, left by the side 
of the Latin laeuus, the Germ, soft compared with our 
sap, the English sift in connexion with the noun sieve, 
and our adverbial alofi and the Germ, noun lufl, so 
closely allied to the nautical verb luff or loof The 
adverb often also is so commonly pronounced without 
any regard to the t, that one is tempted to conclude 
that an earlier form was of-en. As s of the Latin 
suh is lost in our up, so subinde (see p. 71), or rather 
supinde, may be one with our often, as it is one with 
the Fr. souvent. 

It was from a belief in the excrescent character of 
the t in aft that I was led to identify the root-syllable 
of with the Latin ob, and the Greek em. The b of ob, 
as of sub and ab, is proved to have supplanted a ^ by 
the Greek equivalents, and also by such derived forms 
as superi, supra. Thus the f in aft obeys Rask’s law, 
while the vowel change between ob and q/’ corresponds 
with what is seen in dorrvare and our tame. Germ. 
zdhm-en ; in rogare and Germ, fragen ; collum and 
kals ; rota and rad ; folles and balgen. 

I next revert to the doctrine so often put forward 
already (p. 57, &c.), that prepositions which begin 
with a consonant have often attained this form by the 
loss of a preceding vowel, and that such lost vowel is 
either one with or akin to that which follows the said 
consonant. On this principle pos suggests the form 


Digitized by Google 



122 Post AND after the same word. 

opos ; and we have a parallel case in the noun pomum 
as having grown out of an older opomum. (See paper 
oil ‘ a privative.’) This assumption of an initial o 
seems fully justified by the form tmarw {oTnaaa) with 
the very meaning one would desire ; the more so that 
the ojiposcd word Trpoo-® (irpoa-ato), i.e. nop-oa-a), tells 
us that OTT alone belongs to the root, while ap-o> and 
KaT-a> also compel us to mark off the tu of owiv-to and 
irpoa-o}, as in itself a suffix. 

Now the form op-os thus placed beyond doubt beai’S so 
strong a likeness to our English theoretic af-er, allow- 
ing for the usual letter-changes, that it may well raise 
the suspicion that they are virtually the same word. 
The ordinary suffix of the Latin comparative is of 
course ior, but this we know grew out of an older ios, 
the neuter, as melius, and the diminutive, as melius- 
culus, still retaining the original s. But of this suffix 
ios one or other vowel is apt to disappear (cf. Bopp’s 
V. G. § 303). In minor, minus (for minior, &c.), and 
secus (for secius ‘ other’), the weak vowel is lost, but 
the 0 in magis and peyiaros (for p.eytoaTos), in nimis 
and satis, and in pris of pristinus. In this last word 
we have again an excrescent just as we have in crast- 
inus from eras ; and the pris stands for prius, being 
one with the Greek irplv (of Homer) for vpiov. Looking 
then from this point of view, we find a comparative 
which has lost the weak vowel in our theoretic op-os 
of the Latin, and one which has lost the strong vowel 
in oir-i<r- of the Greek. Nay, the preposition irpos 
itself, i.e. vop-os, must also be of comparatival form. 

The doctrine that the os of pos (op-os) is of com- 
paratival character seems confirmed by the old Prussian 
jmns-dau, for ans is so far nearer to the Sanskrit suffix 


Digitized by Google 



Post AND nftav THE same word. 123 

yans of coraparativea It is also to be observed that 
the 0 of the form pos is long, as in a comparatival 
form it is entitled to be. This appears from two of 
the passages to which Ritschl refers, viz. — 

‘ H6nc equidem Vener6m uenerabor, mi5 [ut] amet pO.sliac 
pr6pitia.’ — Poen. i. 2, 66. 

And the line of Cretics : — 

Pr&eterhac si mihi t£le pOs h(inc diem.’ — Men. L 2, 3. 

And I venture to add yet a third from among those 
he brings forward, viz. : — 

' Ndque patrem umquam pOsilla (umjoam t) M. Quid 

uoa t6m patri,’ dec. 

Here the first four words {posilla or i>ostillq) are given 
in the order of all the MSS., and hence we may safely 
infer that the metrical accent fell on pos. I have in- 
serted usquam on conjecture, as the lost word must 
have begun with a vowel. Ritschl’s correction is more 
violent when he transfers umquam to the fourth place, 
and changes postilla into postiUac. 

The appearance of a short vowel in the Greek form 
IToffToi;^(o» for Postumius, which seems to have influ- 
enced Ritschl in his view that pos has a short o, de- 
serves, I think, to have no more weight than Plutarch’s 
transliteration of the Latin decies {dcciens) by Btstes. 

But the appearance sometimes of is, sometimes of 
os, in the Latin comparative, has its counterpart in 
the Teutonic family. Thus the Gothic has compara- 
tives ald-iz-a ‘ older,’ minn-iz-a ‘ less,’ &c., and also 
frum-oz-a ‘ former,’ frod-bz-a ‘ prudentior,’ &c. So 
again in Old German we find in abundance such forms 
as alt-ir-o oi alt-er-o, menn-ir-o or menn-er-o; and 
on the .other hand jung-br-o ‘ younger,’ and frol-br-o 
corresponding to the Gothic frod-bz-a, &c. And, to 


Digitized by Google 



124 Post AND ajier THE SAME WORD. 

complete the resemblance to the Latin forms, the com- 
paratives in { (e) have a short vowel, while the o is 
always long. Of course the er of the Modem German, 
as dlt-er, must be of the same origin of the corre- 
sponding syllable in alt-ir-o or alt-er-o. Hence if we 
apply to the Teutonic family what has been said of 
Latin, the iz and oz of ald-iz-a frum-oz-a, have in all 
probability grown out of an older ioz ; but the form iz 
is substantially the same as the ir or er of Old German 
and the er of Modern German. Consequently the er 
of after is not merely of the same power, but also one 
in origin with the os oipos (op-os). 

Thus the proposition with which I started, I venture 
to say has been established. But a few more last words 
may be permitted. 

It may be as well to note once more the habit of 
prepositions to take a comparatival suffix, in which 
case the secondary form not unfrequently supplants 
the original simpler word. Thus with ourselves near 
(for nigh-er) is in more frequent use than nigh. In 
this way I would account for the fact that ob lost the 
signification of ‘ after,’ which its Greek representative 
€7Tt long retained, as for example in evtyouof. 

I gather from G. Curtius (1. c.) that Aufrecht con- 
siders the Sanskrit pas ‘post’ as decapitated from 
apas, to which I so far of course assent ; but when 
they connect this assumed opos with the Sanskrit apa 
and Greek awo I cannot but differ from them ; and I 
am no way surprised that Curtius, holding this view, 
hesitates to connect with this family of words the 
Lithuanian pas ‘prope.* But all difficulty on this 
head dis<appear8 when eiri and ob are substituted for 
OTTO and ab. 


Digitized by Google 



Post AND after the same wokd. 125 

It may be observed as not improbable that the Fr. 
puis, It. poi. Span, pues, together with the compounds 
depuis, dopo, despues, owe the non-appearance of the 
t, not to any aphaeresis of that letter, but to the fact 
that they came directly from the true form pos. The 
Italian has also the compound forms ^sdomani, pos- 
porre, ^ostergare, jposvedere. 

I would further remark that Eitschl’s theory which 
treats post as a curtailment of poste, and which regards 
the d of postidea and antidea as inserted for the pur- 
pose of avoiding hiatus {hiatus-tilgende) as in both 
respects questionable. I should rather be disposed to 
look upon post-id and ant-id as derivatives from 
simpler prepositions, which passed ultimately into 
the forms poste and ante. This suffix id may perhaps 
be one with the ed of red (for er-ed), of which I spoke 
in the paper on re and pro. Thus post-id would 
correspond to r-ed, post-e to re (er-e). Of course in 
this view post is no longer a curtailed poste. 

In speaking above of the law of letter-change which 
holds between the mute consonants of the classical and 
Teutonic languages, I have thought it right to use 
the name of Bask rather than that of Grimm, having 
before me Bopp’s note (§ 87, or vol. i. p. 119 of the 
ed. 1857), which it will be well to translate at length, 
as the whole credit of the discovery is still for the 
most part unduly assigned to Grimm. Bopp’s words 
run thus : — 

‘In my former treatment of this matter (1st ed. 
p. 78, fol.) it had escaped my notice that Bask, in his 
prize essay, “ Undersogelse om det gamle Nordiske 
eller Islandske Sprogs Oprindelse” (Kopenhagen,1818), 
had clearly and conclusively put forward the law here 


Digitized by Google 


126 Post AND after the same word. 

given, which indeed it would have been difficult to 
overlook. A translation of the most interesting por- 
tion of Rask’s paper was also given by Vater in his 
“ Comparative Tables of the European Family of Lan- 
guages.” Rask’s discovery, however, must be so far 
qualified that he dcids only with the relation between 
the northern and classical languages, so as to take no 
notice of the second law of interchange as exhibited 
in German, which was first demonstrated by Grimm. 
Rask’s law (p. 12 of Vater’s work) is that among the 
mute consonants the following changes are especially 
common : — 

‘ 7T to y, as : irari^p, fadir. 

‘ T to as : tpsh, thnr; tego, eg thek ; tv, tu, thi'i. 

‘ « to h, as : Kpeag, hree (a corpse) ; cornu, horn ; 
cutis, hud. 

‘ is often retained : fi\a(rrdva> (sprout), hind ; ^pvo> 
(well), hrunnr (a spring of water) ; bullare, at India. 

‘ S to < ; hap.du>, tamr (tame). 

‘ y to k: yvvri, kona ; yhos, kyn or kin ; gena, 
kinn ; ay pot, akr. , 

' <f) to b: <f>nyot, hog (beech), fiber hifr ; ^lpo>, fero, 
eg her. 

* 8 to d; 8vptf, dyr. 

‘ X to <7 •• Vyder (gush) ; 

Xo^n, gali.' 


Digitized by Google 



127 


VI. 

021 THE SO-CALLED a PEIVATIVE. 

The old doctrine which treats a mere vowel a as the 
original form of this prefix, the v being an epen- 
thetic consonant, still maintains itself in some quarters. 
Thus in a Sanskrit Grammar published at Oxford not 
long ago, the form a is assigned to the prefix with the 
qualifying remark that ‘ when a word begins with a 
vowel, an is usually substituted.’ Perhaps it is to such 
words as aeiK7)s ‘ unseemly,’ axnrvos ‘ sleepless,’ awpos 
‘ untimely,’ that the false explanation owes its vitality ; 
but the lost initial digamma or <r, or asperate, accounts 
for these anomalies, the older forms having been aPetOT/i, 
atrwvos, a-mpot (see Proc. Philolog. Soc. iii. 52, <kc.). 
Again, the original av accounts satisfactorily for the 
long vowel of a-davaros, and for the p, of ap-^poros. 
But the strongest argument in favour of av, as against 
the claim of a mere a, is seen in the prevalence for the 
most part of a nasal in the corresponding prefixes of 
allied languages ; as — 

Lat. Bansk. Gael. Welsh. Gothic, &c. butch. Old Norse. Swed. Dan. 

in an ana, an, am an un on o o u 

It may be as well to add that the ‘&c.’ attached to 
the heading ‘ Gothic ’ must be interpreted as including 


Digitized by Google 



128 


ON THE SO-CALLED a PRIVATIVE. 


German of all ages, Old Saxon, Anglo-Saxon, and of 
course our own language. 

So much for the final letter of the prefix. The 
question whether the word once possessed a still fuller 
form will be considered presently. The next question 
is as to the meaning of the little particle. Our Greek 
lexicons assign to it not a few meanings. But no one 
probably will claim as of one origin the prefix with 
negative power, and that which signifies unity, or some- 
thing like it. KXo‘j(p», axotTn, aa-KeXfis = uroaKe\i)f, 
&c. have in all probability a common prefix with 
dirXovf ; in other words, have lost an initial asperate. 
Those, then, may be thrown aside. Then again, it is 
perhaps nearer the truth to regard the forms avrepovv, 
ao-Ta^(», (urraxvt, as more genuine than the familiar 
(rrepoirq, arcufuf, ffra^vt, instead of giving precedence 
to the shorter form, and calling the a a euphonic 
addition. But in either case, the a of such words 
has no connexion with the prefix wliich is under 
consideration. 

I have next before me the claim of the so-called a 
eiriTOTiKov, or intensive alpha. It has often been said 
that the Greek grammarians gave an undue extension 
to this particle, and indeed it has been objected to 
them that some of the examples which they quote arc 
but inventions of their own. That they were guilty 
of such a deliberate ofience is altogether unlikely. I 
hope presently to show that a prefix with the sense of 
intensity, one in form with the negative particle, and, 
as I believe, one in origin with it, was in extensive 
use in some members of the Indo-European family 
of languages ; and if this be admitted, then it will be 
rather matter of surprise that the Greek language has 


Digitized by Google 



ON THE SO-CALLED a PBIVATIVB. 


129 


SO few claimants for the meaning. No douht it was 
found to be a serious inconvenience that a language 
should have comiwunds with the same prefix bearing 
two meanings at first sight so inconsistent as negation 
and intensity. An ambiguity of this kind is pretty 
certain to be got rid of by the disappearance of one 
or both meanings. StiU I must contend that in the 
Homeric forms atrice\e» and aairtp'yet the a may well 
have added to the words the notion of intensity. Not 
so however with the adjective arew/j, which maintains 
its position down to a late period in Grecian literature. 
This word seems to me to be only a variety of tineyi}!, 
and so immediately connected with the familiar verb 
eineiPtt ; and I would justify the change of form by 
the tendency of the Greek tongue to drop an p, espe- 
cially before a dental, at the same time changing a 
preceding « to a. Thus the very verb reipu (rep) 
exhibits the change in its tenses eraBtiP, reraxa, rtrafiai, 
so ^ep0ot, irepffos coexist with fiaOos, vadot ; and gene- 
rally men, the termination of Latin substantives, is 
represented in the Greek vocabulary by /jmt or fia. 

As it was in the Gaelic language that I was first 
led to the conclusions which will appear in this paper, 
and as that language stiU furnishes, I believe, the 
most abundant as well as the most decisive evidence 
in the matter, I propose to give in some detail what is 
there found bearing on the subject. But before doing 
so, it win be convenient that I should state the theory 
by which the two apparently irreconcilable ideas of 
negation and intensity are brought into harmony. I 
would assign then male as the primitive idea of the pre- 
fix, the influence of which is most opposite, according 
as it is attached to an idea desirable or not desirable. 

K 


Digitized by Google 



13U 


ON THE SO-CALLED a PRIVATIVE. 


While male sanus can only mean ‘unsound,’ male 
turpis is an equivalent to Uirpissimus. This latter 
use of male has been well noted by Orelli (Hor. Carm. 
i. 17, 25), in the phrase, Ne male dispart incon- 
tinentes iniciat manus, where he quotes the parallel 
cases, male laxus mlceits, Hor. Ep. i. 3, 31 ; oculis 
male lippics inunctis. Sat. i. 3, 25 ; insulsa male et 
molesta, Catul. x. 33 ; male inepta, Tibul. iv. 10, 2 ; 
adding the just qualification, Ilomonymum est w. 
‘valde, admodiim’ cum vocahido ingratae qualitatis. 
There are of course words which in themselves are 
neither eulogistic nor dyslogistic. These however take 
their colour from the context. Thus, as Orelli again 
writes, male pertinax has the negative power in the 
digito male pertinaci of Horace (Carm. i. 9, 24), be- 
cause fii-mness was then to have been desired, but not 
so in Prudentius (Cathem. Praef. 14), Male pertinax 
Vincendi studium suhiacuil casibus asperis. 

The fact that the so-called privative particle some- 
times implied blame has of course attracted notice ; 
but the explanation commonly given, though in itself 
thoroughly intelligible, seems to me to be ill founded ; 
‘ AfiovXia — BvafiovXia, iW-COUnsel, and dirpOirmirof, ill- 
faced, ugly,’ say Liddell and Scott, are ‘strictly a 
hyperbole, counsel that is no counsel, Le. bad, a face 
no better than none, i.e. ugly.' The issue, I think, 
will be that this explanation is untenable. 

I now proceed to the quotation of examples from 
the Gaelic Dictionary of the Highland Society of 
Scotland, omitting for brevity those words where the 
negative notion prevails. It should be noticed, how- 
ever, that in the words now to be given the editor 
sometimes gives to the prefix the epithet ‘ intensive ; ’ 


Digitized by Google 



ON THE SO-CALLED a PRIVATIVE. 


131 


sometimes, pursuing a course which he finds more con- 
venient, omits all epithets, and confines himself to the 
safe phrase, ‘ a prefix.’ The compound words then, in 
the order in which the dictionary presents them, are — 

• aimfheoil ‘proud flesh’ {fedU ‘flesh’). 
aimhreit ‘ discord’ {rSit ‘ harmony’). 
aimhriochd ‘disguise’ {riochd ‘form’); also 

dinriochd ‘ pitiful, or imseemly appearance.’ 

• airibheus ‘ immorality’ {beus ‘habit’), 
t ainhhfheirg ‘ rage’ {fearg ‘ anger’). 

aincheist ‘ doubt’ {ceist ‘ anxiety’). 
ahidealbh ‘ unseemly figure’ {dealbh ‘ form’). 
aindligJie ‘unjust law, trespass’ {dlighe ‘law’). 

• aineachd ‘ misapplied prowess’ {eUckd ‘ feat’). 

10 t aineogail ‘astonishment’ {eagal ‘ fear’). 

+ ainghean ‘ excessive love’ (gean ‘ love’). 

ainghearrahd ‘ a short cut’ {gearradh ‘ a cut’), 
t ainiarmartach ‘ most furious’ {iarmartach ‘ fu- 
rious’). 

t ainiomad ‘ too much’ (iomad ‘ much’). 
ainlean ‘ to persecute’ {lean ‘ to follow’). 
ainmheas ‘ostentation’ {meas ‘valuation’), 
t ainneart ‘ violence’ {neart ‘ strength’). 
ainnis ‘ poverty ’ {6is ‘ want’). 

• ainsgean ‘ bad temper’ (gean ‘ mood’). 

20 ainteann ‘ constrictus’ {teann ‘ tense, stiff’). 

• ainteist ‘false witness, bad character’ {teist 

‘ testimony, character’), 
t ainteas ‘ excessive heat’ (^eew ‘ heat’). 

• aintigheam ‘ tyrant’ {tigheam ‘ lord’). 

+ aintreun ‘ ungovernable’ {treun ‘ brave’). 

• amhfhortan ‘misfortune’ {for Ian ‘ fortune’). 

K 2 


Digitized by Google 



132 


ON THE SO-CALLED a PRIVATIVE. 


amhsgaoileadh ‘ diarrhoea’ {sgaoileadh ‘ scatter- 
ing')- 

amlubach ' curling’ (lilb ' curve’). 
anabarr ‘ excess’ (hdrr ‘ excess’). 
anabeachdail ‘ haughty’ {beachdail ‘observant’). 
30 • anablas ‘ bad taste' {bias ‘ taste’). 

anabraise ‘ immoderate keenness’ {brats ‘ keen- 
ness’). 

anabuirt ‘madness’ {hurt ‘ridicule’). 

• anacainnt ‘ill language’ {cainnt ‘speech’). 

• anacaith ‘misspend’ {caith ‘spend’). 

(maceist ‘ difficulty’ {ceist ‘ anxiety’). Another 

variety of this word occurs below. 

• anacleachdUidh ‘bad custom’ {cleachdadh ‘cus- 

' tom’). 

• anacleas ‘ a bad deed’ {cleas ‘ a deed’). 

• anacradh ‘object of pity' {crMh ‘pity’). 

• anacriosd' ‘ antichrist’ {Criosd ‘ Christ’). 

40 t anacruas ‘ avarice’ {cruas ‘ hardness’). 

anacuibheas ‘immensity’ {cuibheas ‘enough’). 
atuxcuimse ‘immensity’ {cuimse ‘measure’), 
t anacilram ‘ excessive care’ {cUram ‘ care’). 

anaghlas ‘ milk and water’ {glas ‘ grey’), 
t anaghlaodh ‘loud shout’ {(jlaodh ‘call’). 
anaghletis ‘ disorder’ {gleus ‘ order’). 

• anaghldir ‘ill language’ {gldir ‘speech’). 
anaghloiinach ‘ renowned for valour ' {glonn 

‘ deed of valour’). 

• anaghndth ‘ an ill habit’ {gndlh ‘ custom’). 

.50 * anagrach ‘litigious’ {agarrach ‘claiming’). 

t anagrMh ‘doating love’ {grddh ‘love’). 
anaimsir ‘ unmeet time’ {aiinsir ‘ time’). 
anairc ‘ necessity’ {airc ‘ want’). 


Digitized by GoogI 


ON TBUE SO-CALLED a PRIVATIVE. 


133 


andm * unseasonable time' {dm ‘ time’). 

• anamharus ‘wrong suspicion’ {amharus ‘doubt’). 

• anamhiann ‘ lust’ {mvann ‘ desire’). 
andnZ ‘ very high’ {drd ‘ high’). 
anhhas ‘ a sudden death’ {bds ‘ death’). 
anbhdthadh ‘ a deluge’ {bdthadh ‘ drowning’). 

60 t atibhorb ‘ furious’ {borb ‘ fierce’). 

anbhroid ‘ tyranny’ {bruid ‘ a thorn’). 

, • anddn ‘ foolhardy’ (e^n ‘ bold’). 

andSistinn ‘ squeamishness’ {(Uistinn ‘ disgust’). 

t anddchasach ‘ presumptuous’ {d6chascu:h ‘ hope- 
ful’). 

• anddigh ‘ bad state’ {ddigh ‘ condition’). 

t anddhcts ‘ excessive sadness’ {ddlm ‘ woe’). 
anduine ‘wicked man’ {duine ‘man’). 
aneanraisd ‘a storm’ {aonrais ‘tempest’). 

t anfhad ‘ too long’ (/oda ‘ long’). 

70 ttT^Ziann ‘weak, feeble’ (yan7i ‘weak’). 

• anfhlath ‘ tyrant’ (Jlath ‘ prince’). 

• anfhocal ‘ reproach’ {focal ‘ word*). 
anfhosgladh ‘ chasm’ {fosghidh ‘ opening’). 

t anfhucvchd ‘ excessive cold’ {fuachd ‘ cold’). 

• aniarrttis ‘ wrong desire’ {iarrtus ‘ petition’). 
antd2 ‘ bad guidance’ (iti? ‘ guidance’). 
anlaoch ‘ exasperated warrior’ (fowcA ‘ hero’). 

t anluchdaich ‘ overload’ {Ittchd ‘ load’). 
anmhurrach ‘valiant’ {murrach ‘able’). 

80 * annspioradh ‘ a devil’ {spiorad ‘spirit’). 
anoQiir ‘ idle work* {obair ‘ work’). 

+ anrachd ‘ violent weeping’ {rachd ‘ tears’). 

• anriadh ‘ usury’ {nadh ‘ interest’). 

• anriar ‘a wrong gratification’ (War ‘pleasure’). 

• ansannt ‘avarice’ {sannt ‘ desire’). 


Digitized by Google 



134 . ON THE SO-CALLED a PRIVATIVE. 

anmoghoMa ‘worldly’ {saoghalta * worldly'). 
ansgdineadh ‘ chasm’ {sgdineadh ‘ bursting’). 
ansgaivt ‘ loud cry’ (sgairt ‘ loud cry’). 
antarruing ‘ strife’ {tarruing ‘ drawing’). 

90 * antogradh ‘ criminal propensity’ {togradh ‘ de- 
sire’). 

a7itoil ‘ self-will’ {toil ‘ will’). 
antrdlh ‘ wrong season’ {trdth ‘season’) 
t anlrom ‘grievous’ {trom, ‘heavy’). 

• anuair ‘ evil hour, bad weather’ {uair ‘ hour*). 

If we look to the meaning of these words, it readily 
appears that in those which are marked with an 
asterisk, neither negation nor mere intensity supplies 
what is required. It is true, that anaghleus ‘ disorder,’ 
may be considered as the negation of gleus ‘order,’ 
and aimhreit ‘ discord,’ as the negation of r6it ‘ har- 
mony.’ But no such interpretation will account for 
ainhheiis ‘immorality,’ beside beus ‘habit,’ or for 
anddigh ‘bad state,’ beside doigh ‘condition,’ or 
anfhocal ‘reproach,’ beside focal ‘word.’ On the 
other hand, intensity seems rarely if ever to' charac- 
terise the compounds, except where that intensity is 
in fact excess, in other words an evil, as in anfhad 
‘top long,’ from fada ‘long,’ anfhuachd ‘excessive 
cold,’ frojn fuachd ‘cold,’ and generally in those 
example^ to which f has been prefixed. There are 
indeed apiong the quoted examples some in which the 
assigned translation does not bring out the notion of 
badness, but these exceptions are probably to be re- 
ferred to the inaccuracy of the translator, as aincheist 
‘doubt,’ from ceist ‘anxiety,’ ainnis ‘poverty,’ from 
^is ‘ want,’ anaghlonnach ‘ renowned for valour,’ from 


Digitized by Google 



ON THE SO-CALLED a PltlVATIVE. 1H5 

glonn ‘deed of valour,’ As regards tlie last, a repu- 
tation for valorous deeds, though acceptable enough to 
those in whose behalf those deeds are exhibited, excites 
a very different feeling in the sufferers. Thus, the 
Hindoo and Mahratta had more fear than love for 
one whom in the last century they designated the 
‘Daring in war.’ Again, andrd ‘very high,’ from 
drd ‘ high,’ is not easy of interpretation to a member 
of the Alpine Club. But the Highlander was not of 
so romantic a disposition. He thought rather of the 
labour of ascent, and so to him everj' addition to the 
height of a place was an evil In the same way, a 
mere fissure in the ground was of little moment, so 
long as an easy leap would clear it ; but when it was 
both wide and deep, it was either dangerous to cross 
directly, or required a somewhat laborious circuit to 
turn it Hence probably the sufiix seen in anfhosgladh 
and ansgdineadh, both translated ‘ a chasm.’ On the 
whole, then, it may perhaps be safely aflSrmed that few 
win read through the list of ninety-four words without 
coming to the conclusion that the notion of badness is 
distinctly marked in a large proportion, and that the 
same notion gives a thoroughly satisfactory solution 
of the cases where intensity is the favoured explana- 
tion ; and thirdly, that even in the few cases where 
the idea of negation would also supply a reasonable 
explanation, the idea of badness is, to say the least, 
no less applicable. This being so, the only sound con- 
.clusion is, that the one idea which will explain all the 
cases is to be preferred ; in other words, that the prefix 
an carries with it the notion of malus. 

. Hence we must invert the order of the meanings 
which, imder the heading ‘ An, the Gaelic Dic- 


Digitized by Google 



136 


ON THE SO-CALLED a PEIVATIVE. 


tionary above mentioned puts forward, viz. : ‘1. Priva- 
tive. ... 2. Intensive 3. It is frequently found ’ (says 

the lexicographer) ‘ having the same acceptation as the 
adjective “ole” or "droch,” placed before its adjunct : 
pravitatem nonnunquam designat.’ We have here 
indeed an error, not uncommon in lexicons, and in 
one view pardonable. No doubt ultimately the priva- 
tive notion Avas the prevalent one ; and what is most 
common seems at first sight to have the best claim to 
precedence. It is in this way that our Latin Grammars 
place the secondary verbs in are at the head of the 
series of conjugations, in disregard of the claims of the 
simpler conjugation called the third. The spirit of 
modern philology however requires that the order of 
time should be observed here as much as in geology. 

On the varying forms of the prefix in the Gaelic 
tongue some notice will be taken below. It may bo 
observed however that among them is amh, which also 
occurs as an adjective in the same language, and one 
of the meanings assigned to it is ‘ bad, naughty, pravus.' 
This meaning is given on the authority of the well 
known Gaelic scholar, Shaw. Now the Latin malus 
is at present, I believe, an isolated word. It begins 
too with a letter which always incurs my suspicion, as 
occup}dng an initial place to which a liquid is not 
entitled. Thus, if I also assume the law of similar 
vowels for this adjective, an older form would bo 
am-alus, of which am alone would be radical, alus 
being a suffix just as in the Greek /*€>ya\7), 6fi-a7Ms, 
xOafi-a\o?, oTT-oXo?. This view I first threw out as a 
loose conjecture ; but it seemed even then to receive 
some confirmation from our own adjective evil. Germ. 
uhel, which may well represent a Latin amal. 


Digitized by GoogI 



ON THE SO-CALLED a PRIVATIVE. 


137 


But what was so far doubtful I regard as transferred 
to the region of certainty, when it appears that an 
adjective aval, in the sense of * bad,' was once knovTi to 
the Provenjal, Catalonian, Spanish, and Portuguese 
languages. It is to Diez’s Dictionary of the Eomance 
languages that I owe the knowledge of this. But, as 
my explanation of the v^ord differs •wholly from his, I 
think it due to him to state in English what he says : 
‘Aval, Prov. (adj. of one termination) “bad, wretched,” 
sb. avoleza. The word also occurs in Old Catal. Span, 
and Port. ; but is so rare that the statements as to its 
meaning fluctuate. Sanchez translates aval ome, in 
Berceo, by ‘ ladron,’ but this with hesitation ; Moraes 
translates the Port, word in Nobiliario (where more- 
over he exhibits a various reading avil) by “ mdo,” t.e. 
“bad." In Prov. it is of very common occurrence, 
though now, as in the other languages, obsolete, and 
signifies the opposite of pros, Fr. preux. That the 
first syllable has the accent is sho-wn by the contracted 
form dul, which stands to dvol just as frSul to frSvol : 
Scckendorf therefore is wrong when he writes av6l. 
As regards derivation, a guess has been made at the 
Gr. afiovX^s, “ disagreeable,” but this does not satisfy 
the meaning. Ducange, on the authority of a docu- 
ment of the year 1411, notices a form advolus — ad- 
vena, which is literally the Eomance word. As the 
Span, cuerdo is abbreviated from cordatus, and the 
Prov. din from dinatus, so advolus, avol might be 
abbre'viatcd from advolatus, avoU. The fundamental 
notion was “ hergeflogen,” t.e. “ homeless, foreign,” and 
the complete word was often so used : Ceux qui es- 
toient ainsi hannis ... Us appdloit-on avoUz (Du- 
cangc, V. advoli ) ; garce avoUe, Th^&tr. Fr. p.p. Michel 


Digitized by Google 



138 ON THE SO-CALLED a PRIVATIVE. 

449. From this notion that above mentioned might 
easily proceed, just as in our eleTid 1, “ peregrinus,” 2, 
“ miser.” It is true that in this way an adj. of two 
terminations was to have been expected ; but the 
word met with the same fate as frivol = frivolus, 
frivola.’ 

This far-fetched etymology I think Diez himself 
would have been the first to reject, had it occurred 
to him that malus was a decapitated variety of an 
obsolete amctlus, especially as the change from amol 
to avol agrees with the law which he himself lays 
down (Gr. i. p. 200), that a Prov. v corresponds to a 
Latin m. The same change is seen in evil, which in 
this respect stands to amal just as amn-is to Avon, 
as Damn-onii to Devon. Even the difference in the 
two vowels of evil is accounted for as soon as we call 
to mind that, as a\ of /leyaXt) is a suffix of diminution, 
so the same office is performed in English by syllables 
which have a weak vowel attached to the I, viz. h, as 
in mick-le, litt-le ; and of course a weak vowel in a 
suffix generally produces an ‘umlaut' in a strong 
vowel of a root-syllable, evil rather than avil or avle. 
A further argument for connecting the two words is 
that, as in Prov. dvol is reduced to dul, so our evil 
takes the form ill. 

It may be objected to this view, as to the connexion 
of the prefix av or amh with the theoretic amalus, that 
the more common form of the prefix is an, rather than 
am. This is true, but it is a special characteristic of 
the Latin language that it prefers the labial to the 
dental nasal. Still, in many roots which exhibit an 
interchange of m and n, it is difficult to say which 
form has the better claim to originality, as when we find 


Digitized by Google 


ON THE SO-CALLED a PIIIVATIVE. 139 

xOov- of the Greek standing beside and 
and the Latin humus, humilis, and again the Greek 
^av-ep- (ainjp) with Latin hom-on-, and Italian xiomo. 
Mem-or indeed of the Latin has in its first or root 
syllabic what seems to have been originally men, as in 
the Greek p,ep-os, p,e-p,v-iipai. So again an n seems to 
have the better claim as between om of om-it- (pmitto) 
and av of avifjpn. 

The examples so abundantly quoted from the Gaelic 
might be supported by no little evidence from the 
other members of the Keltic stock ; but I will confine 
myself to a few instances drawn from the Welsh, 
anngwres ‘ full of heat,’ from gwres ‘ heat,’ anngwyth 
‘ wrathful,’ from gwyth ‘ wrath,’ where the intensive 
power seems to predominate ; and anhap ‘ mischance, 
mishap,’ from hap ‘ chance,’ anlliw ‘ a stain,’ from 
lliw ‘ colour,’ anfod ‘ ailment,’ from bod ‘ being.' 
Here the notion of badness is beyond doubt, and in 
the first of the three, the English representative by 
the prefix mis confirms the theory. From the 
Cornish I take one example, for which I am indebted 
to Pott (E. F. i. 382), ananhel ‘procclla,’ from anhel 
‘ aura.’ 

In the Teutonic family, to take first the German, as 
the most familiar member of it, I find unart ‘bad 
behaviour,’ unhild (provincial) ‘ disgusting figure,’ 
unbot (j)rov.) ‘improper bidding,’ undienst ‘bad ser- 
vice,’ unding ‘monster,’ unfall ‘mischance,’ unfing 
‘ misdemeanor,’ ungeld (prov.) ‘ a tax,’ ungemach 
‘trouble,’ ungethier ‘monster, hobgoblin,’ ungewitter 
‘ thunderstorm,’ ungezogen ‘ ill-bred,’ unglUck ‘ mis- 
chance,’ ungott (obsol.) ‘ idol,’ unkraut ‘ a weed’ (Lat. 
mala heiba), unmensch ‘inhuman being, monster,’ 


Digitized by Google 



140 


ON THE SO-CALLED a PRIVATIVE. 


unmuth ‘ bad spirits/ unrath ‘ dirt/ unsitte ‘ bad 
habit,’ unthat ‘misdeed/ unthier ‘wild beast/ unweg 
‘ bad road,’ unwetter ‘ stormy weather,’ unzeug 
‘ nuisance.’ 

The German untiefe I must deal with apart from 
the rest, as I find the most opposite translations as- 
signed to it. In dictionaries, among which I include 
those of Sanders, Adelung, and Campe, as well as 
Meissner, the only meaning is that of shallow water, 
and this in Sanders on the authority of passages 
quoted from Humboldt and Niebuhr. On the other 
hand, I am assured by two German friends, who are 
enabled to speak with the highest authority on such 
matters, that in society they only know the word as 
signifying very great depth of water. But Pott 
(E. F. i. 387) speaks of the twofold meaning of the 
word, and to myself this ambiguity is most acceptable, 
for the doctrine that the German un == male in power 
explains alike the negative and intensive meaning of 
the word. To the mariner shoal water is the gravest 
of dangers ; and I may observe that it is in connexion 
with the sea that this notion is found to prevail, as 
for example in the passage from Niebuhr, to which 
reference has been made. On the other hand, with 
the landsman, or at any rate with the bad swimmer, 
it is deep water that is to be avoided. In the same 
page of his book, Pott quotes from Swiss dialects, the 
forms ungross (= sehr gross), Unkuh, Unmaul, as 
‘ positive Steigerungen des Begriffes.’ 

In the Norse our prefix drops the nasal, and takes 
o in place of the German u. Here we find the fol- 
lowing examples bearing testimony in favour of the 
power rmle ^ — 


Digitized by Googl 


ON THE SO-CALLED a PRIVATIVE. 


141 


6dr ‘ annonae difficultatea ’ {dr * annus).’ 

6lxrn ‘ exsecratio ’ {h<en ‘ precatio).' 

6dud * nefas ’ {ddd ’ virtus or perhaps * factum ’). 
6ddtnr ‘foetor’ (ddmr ‘sapor”). 

6daun ‘odor foedus’ {daun ‘odor’). 

6happ ‘infortunium’ (jMpp ‘bona sors v. fortuna 
inopinata ’) 

6kynd ' monstrum’ {kynd ‘genus’). 

6kdr (n. pi.) ‘ sors adversa ’ (fe>r ‘ sors ’). 

6land ‘ terra infelix ’ {Jand ' terra continens ’). 

6lestr ‘ mala fama ’ {hstr ‘ calumnia ’). 

6lund ‘ indoles prava ’ {land ‘ indoles ’). 

6madr ‘ nequam, nebulo ’ {madr ‘ homo ’). 

6rdd ‘ imprudens consilium ’ {rdd ‘ consilium’). 

6^fr ‘ foetor, odor ingratus ’ {yefr ‘ odor ’). 

Lastly, Haldorson, from whose work the above arc 
selected, has a general article : ‘ 0, litera praefixa 
plurimis dictionibus, vim habet ncgandi ct scnsum 
invertendi, item interdum in malam partem trahendi,’ 
where, as usual, the primary meaning is made to give 
place to that which is more common. 

I turn next to the Dutch, not so much to find 
parallel examples in ondaad, ondier, onding, onkruid, 
corresponding to the German unthat, unthier, unding, 
unhraut, as to point to another variety of the prefix, 
viz. the form wan so often found in the Dutch 
vocabulary with a power the same as that we claim 
for the German un and Greek ap. It will bo con- 
venient to give the meanings in German, as the power 
of the prefix will be then self-evident. 

wandaxxd ‘ missethat, un- wandank ‘ undank.’ 
that’ wangehruick ‘missbrauch.’ 



142 


ON THE SO-CALLED a PEIVATXVE. 


wangedrocht 'missgeburt.’ 
wangelaat ‘ iible mine.’ 
wangeloof ‘ missglaube,’ 
wangeluid ‘ misslaut.' 
wangeschikt ‘ungeschickt.’ 
vxmgevoelen ‘ falache 
meinung.’ 

wangevolg ‘ irrschluss.’ 
toangunnen ‘ missgonnen.’ 
wangunst ‘missgunst.’ 
wanhebbelijk ‘ unreinlich.’ 
vxmhoop ‘ verzweiflung.’ 
wanhout ‘verdorbenes 
holz.’ 

wankleurig ‘missfUrbig.’ 
vxtnlmt ‘ verkehrte lust’ 
wanorde ‘unordnung.’ 


wanraad ‘ schlechte wirth- 
schaft, unrath.’ 
tvanschapen ‘ missgestal- 
tig.’ 

vnnschepsel ‘missge- 
schopf.’ 

wansh ik ‘ unschichliclikeit.’ 
VKtnsmaak ‘ iibelgesch- 
mack.’ 

wanspraak ‘ falscbe 
sprache.’ 

wanstal ' missstand.’ 
warmjdig ‘ ungleichaeitig,’ 
wantaal ‘ sprachfehler.’ 
wantroostig ‘ untrdstlich.’ 
wantrouw ‘ misstrauen.’ 
wanvrucht ‘missgeburt.’ 


The Scandinavian branch, too, is familiar with a 
prefix van of the same power. The High German 
also employs wahn in much the same way. So also 
in Old English we have not merely unlust, untyme, 
unthank, untrust, unrest, unfaith ; but also wanhope, 
ivantrust, and vxmton, i.e. voan-towen = unrgezogen. 

But in spite of the oneness of meaning in the two 
prefixes un {on, Ac.) and wan {van, Ac.), the question 
of their identity involves matter for controversy. It 
is true that words beginning with u and o are precisely 
those in which the loss of a digamma is to be sus- 
pected, as in the Danish tdd, ulv, under, urt, and ol, 
ord, orm, compared with our own wool, wolf, wonder, 
wort, and wall, word, worm. On the other hand, the 
prefix van {vxin, wahn) has been referred with mud 


Digitized by Google 



ON THE SO-CALLED a PRIVATIVE. 


143 


reason to the family of words which denote emptiness 
or defect, as the Norse adj. van-r ‘ empty and the 
Gothic vans ‘ wanting,’ — to say nothing of our own 
verb wane, and the sb. want. These again claim kin 
with the Latin uanus ; but if so, the nasal is no longer 
radical, seeing that woe of uac-are, vac-iuus, or rather 
uoc of uocare, uoc-iuus, exhibits the root from which 
uanus is deduced, much as plenus from the obsolete 
ple-re {explere, &c.). 

Leaving this point open, I would next draw atten- 
tion to the prefix ue, uae or perhaps rather ueh, which 
presents itself in a small number of Latin adjectives. 
Here we have the very same difiiculty which we had 
with the prefix an. In uepallidus the ue is said to 
have an intensive power, whereas it seems to represent 
a negative in uesanus, uegrandis, uehemens. I pro- 
pose then the same solution, viz. that the word really 
meant male, so that uepallidus might well be equiva- 
lent to misere pallidus and uegrandis to male grandis ; 
and I put this forward with the more confidence when 
I call to mind the Ovidian — 

‘ uegrandia forra colonae 
Quae male creuenmt,’ &c. 

That grandis in the best writers is especially used 
of growth will be admitted ; and indeed it is probably 
of the same stock with our word grow, so that quae 
male creuerunt seems to be an absolutely literal trans- 
lation of uegrandis ; and such probably was Ovid’s 
meaning when he added these words. 

That male is the more precise power of this prefix 
seems confirmed by the use of the so-called interjec- 
tion uae in uae tihi ‘ill betide you;’ and then we 


Digitized by Google 



144 


ON THE SO-CALLED a PRIVATIVE. 


have the same word in the German substantive weh, 
and in our own woe. In the Latin uehemens we find 
the asperate preserved ; and indeed in the passage just 
quoted from Ovid (Fast. iii. 445), many good MSS. 
give uehegrandia. 

One of my colleagues at University College, when 
I communicated to him in words the substance of this 
paper, pointed out to me that the theory gave a satis- 
factory explanation of the name of the god Veiouis as 
‘ the had Jupiter.’ Aulus Gellius (v. 12) includes this 
god among the Uieua numina, as one in la^dendo magis 
quam in iuuando 'potentem. In the same chapter he 
tells us that he had a temple at Eome between the 
Arx and the Capitolium, and further that the statue 
in that temple was armed with arrows, Sagittas tenet 
quae sunt uidelicet paratae ad nocendum. The old 
form of the name appears to have been Vediouis. So 
Vedtouei patrei, Mommsen’s oil 1. 807 ; but the MSS. 
of Ovid, in the Fasti iii. 430 and 447, have Veiouis. 
Still in either case Ve is the prefixed syllable, not Ved, 
for louis is connected with dies, and had at first an 
initial d. Hence also the d in the Greek oblique cases 
Atof, Ac. 

It will be no violent assumption that this ue is but 
a curtailed variety of uan. A parallel case is to 
be seen in the root uan ‘blow,’ whence the Latin 
nouns uannus, uent-us, Ac. In Sanskrit we find this 
root taking the two forms va ‘blow,’ and an ‘blow;’ 
and the Greek aij/u exliibits the root, first as F« and 
then as a mere vowel a or ti, thus again coinciding with 
the short form of a privative. 

In the preceding investigation I have passed over 
the Latin language. Let me now briefly supply the 


Digitized by Google 



ON THE SO-CALLED a PRIVATIVE. 


145 


omission. The Latin informis, which corresponds 
precisely to the Germ, ungestaltet, I claim to be an 
equivalent of inale formata, and this on the direct 
authority of Priscian (1, iii. 10, Krehl) : Infot'mis 
dicitur mulier, non quae caret forma, sed quae male 
est formata; and this he says without any theory 
to bias him, for he is not dealing with etymology. 
Infamis again agrees with the notion ‘ having a bad 
character and intemperies corresponds with the 
German unwetter. Ignominia also implies an adj. 
ignomin-i-s corresponding to the adj. cognomin-i-s; and 
the prefix (iw) of i-gnomin-i-s must have carried with 
it the notion of ‘ bad.’ The word impotens is usually 
interpreted by scholars as an abbreviation of impotens 
sui ‘unable to restrain oneself, ungovernable’ — a theory 
somewhat too violent. But a scholiast on the phrase 
Aquilo impotens in Horace makes the adjective an 
equivalent of uaMe potens. This view leads me to 
suggest that the fuU meaning of the word is brought 
out by the phrase male potens ‘ using power badly ; ’ 
so that ‘ furious’ is a tolerably satisfactory translation 
of the adjective. 

But a still more decisive instance is seen in the 
adjective inuidus, which is very unduly considered to 
be a derivative from inuideo. The stream runs the 
other way, for it is contrary to the habit of the lan- 
guage to deduce adjectives in o- {inuido-) from com- 
pound verbs. Rather then let us treat inuidus as an 
equivalent to a theoretic maliuidm ‘ having the evil 
eye and from inuidus let us deduce inuide-re. This 
verb cannot have been formed from the ordinary pre* 
position in, as it means far more than ‘ to look at.’ 
I should have been prepared to regard it as a com- 

L 


Digitized by Google 



14G 


ON THE SO-CALLED a PRIVATIVE. 


pound with in = male, if the Latin language had so 
compounded verbs. We ourselves indeed have no 
difficulty in creating compounds, such as mistake, mis- 
spend, misunderstand ; and the Gaelic, as seen above, 
gives us ain-lean ‘to persecute,’ from lean ‘to follow,’ 
and ana-caith ‘ misspend,’ from caith ‘ spend.’ Hence 
it is very possible that the Latin insequi ‘ to pursue as 
an enemy,’ may have the in = male. The adjective 
insignis stands apart from the other adjectives com- 
mencing with in. We have here probably the ordinary 
preposition, so that the word corresponds to the Greek 

eiTiat}ftos. 

There is a question of form which has been passed 
over. It was probably noticed that some twenty of the 
Gaelic compounds had ana as the prefix rather than an. 
Here we have a parallelism Avith the Greek avaeXvrot. 
I might also have quoted avaeBvot, but that the better 
form seems to be aveFeSj/os (cf eFeSra of the Odyssey). 

I go back to the forms uan and amalus, to point 
out that these suggest a fuller uam-alus ‘bad’ of which 
nam alone belongs to the root, and this in English 
should take the form wav, the comparative of which 
should be wav-er} which is all but one with our pro- 
vincial ivaur ‘ worse.’ It is here assumed that the 
suffix al of the positive has no right to enter into the 
formation of the comparatives and superlatives, and 
this is a point which has long been established (see 
Bopp’s V. G. § 298 a). Thus in Sanskrit kship-ra 

* quick,’ leads to ksMp-tyas ‘ quicker,’ kshep-ishtha 

* quickest ;’ M<rx,-po- to ai<rx,-arro-', fiey-a\-r) to 

/tetfoi/- (for fiey-tov-), (ify-iaro- ; mag-no- to maior (i.e. 

’ So Waverirtt near Liverpool is called Wa’rtire. 


Digitized by Google 



ON THE SO-CALLED a PRIVATIVE. 


147 


mag-ior ) ; litt-le to less-er (for lett-er, cf. Germ, hesser, 
Eng. better), and least (for let-est, cf. best for bet-est, 
last for lat-est). 

In order to strengthen the argument that malus is a 
corruption of amalus, let me point to the fact that 
mdlus ‘ an apple tree,’ seems also to have supplanted a 
fuller amalus, seeing that the Welsh write the word 
as afal, which is of course one with the German apfel, 
and our apple. In some parts of England (Mr. Morris, 
in the ‘Ayenbite of Inwyt,’ Introd. p. 4, says Wilt- 
shire, Gloucestershire, and Somersetshire) the form 
opple prevails, and this opens a new vista. A German 
friend to whom I had communicated these ideas writes 
to me as follows : ‘ That before the a of {a)mdlus “ an 
apple tree,” an original consonant, not unlikely a 
digamma, has been lost, I should venture to conclude, 
on the evidence of the Eussian word jahloko and the 
Bohemian gahlko. Grimm recognises in apf-el the 
same root as in ob-st, and indeed evidence might be 
given that the a in this word is by no means original. 
Even in the modern dialects we hear sometimes the 
plural dpfel, comp. Lith. obolys. Obst again, or as the 
original form is opaz, obez, seems to be the same as 
Ang.-Sax. ofdt. hlay we compare the Greek oir-wpa 
i.e. the season when “ obst” is ripe V 

In giving an aflBrmative answer to this query, I may 
notice that the Latin opes ‘ wealth, power,’ may well 
liave had for its primitive meaning the fruits of the 
earth, and that Ops, as the Goddess of Fertility, con- 
tains the same idea. Then again as malum in this 
view stands for amdlum, so iwmuvi is probably a 
shortened form of op-omum. Lastly, this interchange 
of p and m seems also to explain the appearance of 

L 2 


Digitized by Google 



148 


ON THE SO-CALLED a PRIVATIVE. 


the p in peior and pessimus by the side of malus. 
Possibly malus first exchanged its Z for a c? (cf. our 
had), and then pedior pedsimus would easily have 
passed into peior pessimus. 

I ought perhaps to add that Pott (E. F. of 1859, 
i. 174) gives a very different origin both to the av 
privative and to the ue of uesanus. His words are : 

‘ Insanus (in-, Gr. av-, Sskr. an- eig. das. Pron. ana, 
jener) und v^sanus (Sskr. vi-vom Zahlw. dvi ; jedoch 
nach Anderen aus vahis, aus).’ From the same work 
(p. 386) I borrow also a passage of Simplicius, in 
which he speaks of the view which Chrysippus took 
of the prefix av. After showing that the use of the 
particle exhibits much confusion, he says ; avfi^aivei 
wore /lev rats airoifiaaeari {negationibus), irore Se Tott 
evavTiois av/i^vpeaOai ’, and soon after, Kai to kukov 
Be Bt/Xovrai iroWaKit, a<f>o>vov eXeyo/iev rpagalBov rov 
KaKo^tavov. Thus we have a direct confirmation of the 
chief points contended for in the present paper. 

P.S. A friend draws my attention to the following 
note of Davis on Cic. Tusc. ii. 8 : ‘ Vecors Oenei partu 
edita] quae Ciceroni uecors, ea Sophocli Track 1061 
est Bokmmt. Apposite Festus : “ uecors est turbati et 
mali cordis.” Vide et eundem in uegrande. Non 
priuationem, sed malitiam seu prauitatem particula 
(ue) denotat, quemadmodum etiam in Veioue ; licet 
earn uocem aliter interpretetur A. Gell. N. A. v. 12.’ 


Digitized by Google 



149 


VII. 

THE LATIN et, que, atque {ac), AND THE GKEEK 
K<u, re, ALL OF ONE ORIGIN. 

It has probably struck many philologers as somewhat 
strange, that the Latin language possesses three par- 
ticles to express the idea of ‘and.’ Such a super- 
abundance is at any rate an unusual phenomenon ; but 
it has conferred on Latin writers an advantage of 
which they have not been slow to avaU themselves. 
I refer to the power it gives of grouping the parts of 
a complicated sentence, so as to enable the mind to 
take in all the subordinate clauses without confusing 
them. This is a point to which I drew attention many 
years ago in a review of Mr. Henry E. Allen's valuable 
treatise entitled ‘ Doctrina copularum linguae Latinae,’ 
in the ‘ Quarterly Journal of Education,’ of the Useful 
Knowledge Society (vol. iv, p. 135). Thus in the 
passage (Cic. in Cat. iii. 8, 19), — ‘Caedes atque 
incendia, et legum interitum, et helium civile ac 
domesticum, et totius urbis atque imperi occasum — 
appropinquare dixerunt’ — it will at once be perceived 
that et is employed to unite the longer clauses, 
while atque {ac), filling a more subordinate office, 
connects words within each clause. But if we 
translate both et and atque alike by our ordinary 
conjunction ‘ and,’ the repetition at once offends the 
ear and confuses the mind. A better course is simply 


Digitized by Google 



150 


THE LATIN et, quc, tttque (ac), and 

to leave the et untranslated and to supply the loss 
by a pause. Thus : — 

‘Massacres and conflagrations, the annihilation of 
law, civil and domestic war, the downfall of the 
city and the empire — all these were approaching, 
they said.’ 

The insertion of the words ‘ all these ’ serves in fact 
as a compensation for the several omissions of et. 

We see a similar fitness in the use of the conjunc- 
tions in such a phrase as (Liv. xxviL 18); ‘Equites 
Numidas, leviuraque armorum Baliares et Afros de- 
misit,’ where troops of the same class are united by 
et, those of different classes by que. 

So far but two conjunctions are called upon to 
serve. In the following, all three are turned to 
account (Caes. B. G. vii. 79) : — ‘ Itaque productis 
copiis ante oppidum considunt ; et proximam fossam 
cratibus integunt atque aggere explent, seque ad erup- 
tionem atque omnis casus comparant.’ 

In this sentence we have first the taking a position, 
and secondly the active measures that ensued. These 
general ideas are connected by the particle et. But 
the active measures are again subdivided. On the 
one side we have a step towards action on the offensive 
in the dealings with the ditch ; on the other, what is 
for the purpose of defence, in the precautions a^inst 
a surprise of any kind. To mark this distinction que 
is employed. But these two ideas also admit of 
bifurcation. The obstruction of the ditch to an 
advance may be got over in two ways, by bridging 
it with hurdles, or by filling it up. So too of the 
threatened dangers, the most prominent, that of a 
sally, may well be selected for special notice. In 


Digitized by Google 



THE OllEEK Kal, re, ALL OF ONE OltlOIN. 151 

these subordinate divisions atque is available. Thus 
in Latin the particles et, que, and atque are employed 
to mark those distinctions, which in English we can 
often only denote for the eye by a variety of stops. 

No doubt at times et alone is employed throughout 
a long period to connect all the single words and 
phrases and clauses; especially when the object is 
rather to deluge a hearer’s mind with a torrent of 
ideas than to place them in due subordination before 
him , confusion for once being preferable to distinctness. 

But if the Romans, having the three conjunctions 
at their disposal, made an intelligent use of their 
wealth, it still remains to account for the existence 
of that wealth. Now of the three particles, the one 
most open to suspicion is atque, and that on account 
of its greater length ; for it is the habit of language 
to use for such an inferior office only short words. 
Some ycara ago I had placed before me an interpre- 
tation of a Lycian inscription, in which the interpreter 
had assumed that a certain repeated word of not less 
than foim syllables meant ‘ and,’ a suggestion against 
which my mind revolted. But even a disyllabic word 
has in it what is slightly suspicious ; and this feeling 
is encouraged by the very form of the word, which 
may well bo looked upon as made up of the ordinary 
preposition ad and que. Such at any rate was the 
view of Scaliger ; and if this view be right, then the 
translation ought to be, not ‘ and,’ but ‘ and what is 
more.’ With this idea before me, I have been led of 
late, while reading any Latin author, to feel my way 
whether such a translation accords with the use of the 
word ; and 1 am strongly inclined to answer the query 
in the affirmative, so far as a very large proportion of 


Digitized by Google 



152 THE LATIN et, que, atque (ac), and 

the examples is concerned, at the same time readily 
admitting that there are instances where the word 
seems to have been used with the power of a simple 
et or que. Of course the non-translation of the ad of 
atque will still leave an intelligible sentence in which 
but little is lost through the omission. Thus a reader 
is apt to be satisfied with the ordinary translation of 
atque as a mere ‘ and.’ But my own conviction has 
been strengthened by what recently occurred to me. 
Having made known my feeling on this subject to an 
accomplished scholar, who happened at the moment 
to have the ‘De Amicitia’ in his hand, I found that 
he entertained a strong doubt on the subject, and, in 
support of this, pointed to two passages in the last 
chapter of that treatise ; viz. : ‘ Nemo unquam animo 
aut spe maiora suscipiet, qui non sibi illius (Scipionis) 
memoriam atque imaginem proponendam putet ; ’ and 
soon after : ‘ Nam quid ego de studiis dicam cogno- 
scendi semper aliquid atque discendi ? ' In these cases 
he was disposed to regard memoriam and imaginem, 
cognoseendi and discendi, as practically synonyma 
But I could not help feeling that in the first passage 
the more complete translation would tell us that the 
aspirant after glorious thoughts and deeds would think 
it a duty to place before himself the memory of the 
great Scipio, ay, and if possible, to have his bodily 
form in his mental view, for his statue or bust must 
have been familiar to the citizens of Rome. Again, in 
the second passage cognoscere, at any rate in the im- 
perfect tenses, means strictly only ‘ to look thoroughly 
into,' ‘to study with all care but after all such study 
may be profitless ; discere, however, is ‘ to learn,’ 
denoting successful study. It is true that the Latin 


Digitized by Google 



THE GREEK «ca(, re, ALL OF ONE ORIGIN. 153 

Nosce teipsum,* and the Greek yptaOt veavrov are 
usually translated ‘know thyself/ yet a more exact 
rendering would be, ‘ study thyself.' I hope then still 
to win over my friend to my opinion. At any rate, 
I have to thank him for drawing my attention to the 
use of rrpos he in Homer and Herodotus, and of teat 
•trpos in many Greek writers, where irpos hke ad of 
atque is used without a substantive, or, as the phrase 
is, ‘absolutely,’ so that we have a precise equivalent 
to a^que as understood by me. 

Let’ me notice, too, that in such constructions as 
Est id quidem magnum, atque hau scio an maxi- 
mum (Cic. Fam. ix. 15, 1) the atque fully supports the 
part I would assign to it, and to substitute et or que 
would be wholly inadmissible. Again, in Horace’s 
Vocatus atque non uocatus audit, how incomparably 
more forcible is the atque than a mere et f 

I am not sorry to find some confirmation of my 
view in what Wagner has written in his ‘Quaestiones 
Vergilianae,’ as first (q. xxxv. p. 663) : ‘ Haec quoque 
exempla confirmant, id quod supra indicavi, ac gravius 
esse copula et;‘ and again (567): ‘Singularem huic 
particulae {atque) esse gravitatem, quum alia mibi 
indicare videntur, turn haec,’ &c. where he goes on 
to quote a number of passages in proof, to which I can 
only refer. 

In the case of the familiar phrase atque adeo, ‘ and 
what is more,’ we have what may be used alike for 
and against the present theory. On the one hand, 
the use of atque rather than et is consistent with the 
power here claimed for atque ; but it may be urged, 
that, as the second particle already contains the pre- 
position ad, we have a tautology that has no justifica- 


Digitized by Google 



154 THE LATIN ct, quc, atquc {ac), and 

tion. It may perhaps be enough to ifeply that in the 
Latin as in other languages such tautologies are of 
frequent occurrence. Thus phrases like ad Caesarem 
accedere, incurrere in colurnnas, with a repeated pre^ 
position, are met with everywhere ; and, what is more 
to the point, tautology is one of the means employed 
in language to mark emphasis. Thus a verb of the 
first person ending in o has already in that final letter 
a compression of ego, and yet whenever the idea is to 
be made specially prominent, another uncompressed 
ego is attached : ego scribo in preference to scribo. In 
Spanish again, although tigo, migo are ab-eady full 
representatives of tecum, mecum, it is found more 
intelligible to say contigo, commigo ; no doubt because 
the go had ceased to carry with it its proper meaning. 

In the two formulae atque utinam and ac ueluti 
there seems to be some reason for suspecting that the 
atque {ac) is but a deceitful imitation of our con- 
junction. To some extent this view receives support 
from two of the most distinguished scholars of Ger- 
many. Thus Lachmann, speaking of atque utinam, 
in a line of Propertius (iii. 15, 51), says that in this, 
construction ‘ delitescere copulativam ac particulae sig- 
nificationem.’ So Haupt again tells us : ‘ In optandi 
formula atque utinam prior particula noimunquam 
non connectit orationem, sed cum altera artissime 
cohaeret.’ (See Haupt’s ‘ Observationes Criticae,’ of 
the year 1841, p. 38.) In the same pamphlet (pp. 
46, 47), four passages are quoted where atque utinam 
occurs in a position which seems at variance with the 
usual habit and meanmg of the conjunction atque. 
First from Caesar, in the verses where he addresses 
Terence : — 


Digitized by Google 



THE GBEEK km, re, ALL OP ONE ORIGIN. 155 

‘ Lenibus atque utiDam scriptiB adinncta foret uis, 

Comica ut aequato uirtus polleret honorc 

Cum Graecis, neque in hac deepectus parte iaceres.’ 

Then from Valerius Cato ; 

' Istius atque utinam facti mea culpa magistra Frima foret.' 
Thirdly, from Valerius Flaccus (vi. 599) : 

‘ Eat atque utinam euperetque labores.’ 

And lastly, a passage from Appuleius (lib. vii. p. 199, 
Elm.), where atque commences a sentence in such a 
manner that the idea of connexion, commonly belong- 
ing to the particle, seems out of place, viz. ; — 

‘Atque utinam ipse asinus, inquit, quern nunquam profecto 
nidiseem, nocem quiret biunanam dare meaeque testimonium inno- 
centiae pethibere poBset.’ 

I have quoted the passages at length because the 
treatise of Haupt, like most of those occasional ad- 
dresses which are published in Germany, is not very 
accessible to English scholars. 

In the case of atqtie utinam, what appears to me 
to be a satisfactory explanation may be given. That 
utinam stands to quisnam in the same relation as uti 
to quis, will I think, be readily admitted ; but in our 
own language the particle ‘that’ needs a preceding 
‘oh,’ before the idea of a wish or prayer is fuUy 
expressed. Now, the interjection ah is well suited 
for introducing a wish, as in the Fasti (iv. 240) : 
‘Ah pereant partes quae nocucre mihi.’ But this 
inteijection on the best authority should be written 
as a simple vowel a. Thus Wagner in his ‘Ortho- 
graphia Vergiliana’ has: ‘a interjectio ubique in 
Mediceo Romano aliisque optimis libris sine aspi- 
rationc scribitur . '. . . Idem volunt vetcrcs gram- 


Digitized by Google 



156 


THE LATIN et, que, atque (ac), and 

maticL' In the second place, as vhi, unde, uter are 
now admitted to have had originally an initial c, as 
cubi, cunde, cuter, so for ut we may claim an older 
variety cut, making it in fact a mere neuter of the 
relative, — that is, an equivalent to quod ; and so for 
utiiuxm we are bound to insist on an older variety, 
cutinam. Now it is precisely where a combination 
with a previous vowel-ending word occurs that the 
guttural might be expected to mamtain its ground. 
It is thus that in an inscription of the Augustan age, 
we find ne-cuter, which afterwards gave way to neuter. 
So again in si-cut and hu-cusque the c may well be- 
long to the second element, for si ‘so’ is older than 
the compound si-c (for si-ce) : witness the phrase si 
dis placet ‘such is the pleasure of heaven.’ Thus 
Mommsen in his interpretation of his Inscription 
1447 unnecessarily assumes the loss of a c, where 
the recorded letters run sei si fecerit, which may well 
represent si sic fecerit of the later language. So 
again, ho ‘ hither,’ as seen in horsum (for hd-vorsum), 
is older than hoc or hue, which arose from a compound 
ho-ce. This theoretic ho would correspond to isto 
(= istuc), illo (= illuc), for the forms isto illo are 
of far more frequent occurrence than our editions of 
Latin writers would lead us to believe. 

•Putting then the two points together, that a is 
more correct than ah, and that cutinam must have 
been an older form of utinam, we have in a cutinam 
a good phrase for the expression of the idea ‘ oh that ;’ 
and, as the words are closely combined in pronun- 
ciation, they readily pass first into ac utinam, and 
then, under the ordinary doctrine that atque rather 
than oc should be preferred before a vowel, into atque 


Digitized by Google 



THE GREEK (tat, tc, ALL OF ONE ORIGIN. 157 

utinam, which in sound would still be identical with 
ac utinam. 

As regards oc ueluti it is not easy to find so satis- 
factory a solution ; but stiU as the adverb sem-el is 
admitted to have for its first syllable what denotes 
‘one,’ as also sim-plici-, sim-plo-, sing-ulo-, &c., we 
can scarcely refuse to treat the second syllable of 
semel as that sufilx of diminution which is so fami- 
liar in the Latin language, but is commonly con- 
verted into ul. Thus we have oc-ul-o- and ocello-, 
the latter standing for oc-el-el-o-. Similarly, semel, 
semol, and simul are now regarded as equivalents in 
form. Again, proc-ul may well be formed from proo 
as an older form of pro (see p. 77), by addition of the 
same suflBx. Following these clues, I would suggest 
as a possible adverb from the same stock as the 
adjective aequo- a form aequel; and then the com- 
bined formula aequel uti ‘just as,’ would readily 
slip into ae ueluti. Be this as it may, I venture to 
deny that in the phrase ac ueluti wc have any repre- 
sentative of the ordinary conjunction atque. 

I next proceed to the main purpose of the paper, 
the identification of the particles et, que, km, and re. 
That T6 is really one with que has, I believe, been 
long an admitted truth. The use and power of the 
two little words are in aU respects identical ; just as 
the pronouns ns of Greek and quis of Latin are the 
same. But of the two forms we cannot hesitate to 
regard the guttural as the earlier occupant of the 
ground, for the passage of a guttural to a dental is 
of familiar occurrence. But if re has supplanted an 
earlier «e, we have in the two forms km and /ce no 
great difference. Indeed in some alphabets the cora- 


Digitized by Google 



158 


THE LATIN et, que, atque (ac), and 

bination ai is the only mode of designating an e. 
Still there remains, or seems to remain, a difference 
of quantity. I say ‘ seems,’ because the Homeric 
hexameter abounds in examples of a lengthened re, 
as (II. ii. 495), 

ApK€atXaot rt flpoOotiyup rt KXoi'toc re. 

It is true that the Homeric examples generally have 
two initial consonants or the suspicion of two initial 
consonants in the word which follows t«; for not a 
few words commencing with a liquid have lost a pre- 
ceding consonant, and such derived forms as eaveva, 
eiravfievos, imply that aevto itself has undergone some 
such change. Thus we cannot altogether rely on such 
a case as 

T aopi Tt ftiyaXouTi re \eppiaZiuiiri», 

although Mr. Brandreth’s form PteerfoXoitrt seems un- 
satisfactory, if only because it is unpronounceable. 

Nay, even the tenth line of II. xi. affords no sure 
ground — 

E^Oa «TO»’ rptae Bea peya re Beiroy re — 

for several of the secondary forms of BetSa (with 
which, of course, Beivot is closely connected), as 
tBSeiera, vrroSBeta-at, raise a suspicion that this family 
of words commenced with something more than a 
simple mute consonant In confirmation of this view, 
one of my colleagues observes that the perfects 
BeBotxa and BeiSoiKa cannot justify their possession of 
the diphthong ot by such a form as \e\oirra, for this 
Ijelongs to the class of so-called second perfects, the ir 
forming part of the root, whereas the syllable tea of 
BeBoeKa, as of 7reif)iXr]Ka, belongs to the tense-ending. 
Thus the best explanation in his view of BtSoixa is on 


Digitized by Google 



THE GREEK km, re, ALL OF ONE ORIGIN. 159 

the supposition that the o is part of the base of the 
verb, so that So sounds as the dw of our dwell. 

Further, it may be observed that the adjective Suvos 
has the power of giving length to other final vowels 
than that of re, for example, in 

Ai^ioc rc fiot t<rm ftXi txvpi Sco'oc r< — 

where sKvpe, as the same scholar points out, must have 
supplanted a fuller form aFsKvpe, corresponding to the 
Sanskrit svasru, the Gothic svaihra, and the German 
Schvxiger and Schwieger. Thus the final of 
becomes for the time long before the combination <rF 
of the following word. 

In the Latin language, however, the examples of 
a lengthened que before a single initial consonant are 
more indisputable, for already Attius (Fest. p. 146) 
has 

‘ Calones famuliquo metalliquQ caculaeque.’ 

In Vii’gil indeed, as in Homer, the examples have for 
the most part two consonants, as 

‘ Aestusque pluuiasque et agentis frigora uentos. 

TcrrasquS tractusque maria caelumque piofundum 

or else a liquid, as 

‘ LimmaquS laurusqne del totuaque mouerL’ 

But as the I of laurus is but a substitute for a d, 
as shown by its analogue the Greek Sa<j>irrj, we have no 
ground for suspecting the loss of a consonant before 
the 1. Yet even Virgil has (xii. 363) 

‘ CliloreoquS Sybarimque Daretaque Thersilochumque.’ 

Ovid again, who is generally more strict in metrical 
matters than Virgil, was not afraid to write : 

‘ Faunique Satyrique et monticolae Siluani’ (Met. i. 193). 


Digitized by Google 



160 


THE LATIN et, quc, atque {ac), and 

‘TelasquS calathosque infectaque pensa reponunt’ (iv. 10). 

‘ Sideraque uentique nocent auidaeque uolacrea’ (v. 484). 

‘ Otbrysque Pinduaque et Pindo maior Olympus’ (vii. 225). 

‘LiliaquS pictaaque pilas et ab arbore lapses’ (x. 262). 

‘ Feleusque comitesque rogant ; quibus ille profatur’ (xi. 290). 

While later writcra, who were much more scrupulous 
in these respects than is commonly thought, have 
occasional instances of a similar liberty, as : — 

'Taxique pinusque Altinatesque genestae’ (Grat. 130). 

‘Electra Alcinoeque Celaenoque Meropeque’ (German. 262). 

‘Laeuaque dextraque acies astaie uideres’ (Corip. Laud. Just. iiL 
177).' 

On the whole then we must not reject the theory 
that re and que had once a long vowel, though of 
course the short vowel in the end thoroughly estab- 
lished its position, and this was to be expected when 
we consider the enclitic character of the words, 

I have not stopped to discuss the favourite and con- 
venient doctrine that the quantity of re and que in 
such lines is to be ascribed to the influence of what is 
called ceesura or arsis, because I believe this doctrine 
to be merely a screen for the concealment of ignorance. 
I hold it to be a more just explanation that the two 
little words have lost a final consonant, a former 
possession of which would remove all the difficulties. 
For this theory I find a parallel in the case of uel ‘ or,’ 
which as an enclitic takes the shorter form of ue, as 
uel mater, or else materue. 

This uel is in origin probably an abbreviation of 
uele, i.e. an old imperative of the verb uol- ‘wish,’ 

' TheM three examples, together with that quoted from Attius, 
were suggested to me by Lucian Muller’s elaborate work on Latin 
metres. 


Digitized by Google 



THE GREEK Kai, re, ALL OF ONE ORIGIN. IGl 

where the root vowel has passed from d U> S, under 
the influence of the final e, in accordance with the 
law of ‘ umlaut.’ Another example of such a modified 
vowel is to be seen in heus ‘harkee,’ an abridgment 
of an imperative hense, from a theoretic verb haus- 
‘ hear/ a verb which would stand to the Latin sb. aim- 
or aun- much as our own vb. hear to our sb. ear. 
Again, the assumed loss of an e in tiele would be in 
accordance with the formation of the imperatives es, 
fer, die, due, and indeed ama, doce, audi also. Again, 
from an obsolete verb gon-, or con- ‘ look’ (the parent 
of the secondary gn-osc-, i.e. gonrosc- ‘ learn’), I assume 
an imperative gene, or cene {kene in sound), which first 
cut down to cen prepares us for two other varieties, 
viz. by decapitation, en ‘look,’ ‘behold,’ and, by loss 
of the final, ce, the familiar suffix of demonstratives, 
and demonstratives alone, as hie, istic, illic, sic, nunc. 
Here, too, let it be noted that it is only when doing 
duty as an enclitic that it discards the final n. Nay, 
con and cen themselves are perhaps truncated words, 
for oTTr-ofiat and okko- of the Greek, and oc-ulo- of the 
Latin point to a stem ok (ott) or okk, whence ecce 
would be a good imperative ; so that the verb con 
would be a truncated derivative for oc-on. It may be 
noted, too, that the original symbol for the vowel o 
was a picture of an eye, and the Hebrew name for the 
letter meant ‘ an eye.’ This view accounts also for the 
e of exeivo-. Another instance of a word losing a final 
consonant when employed enclitically is seen in the 
family of words Kev, *«, Dor. tea, and the more familiar 
av. Why the degraded form Ate should be selected as 
that under which Lexicons deal with these particles, 
it is difficult to say. The more legitimate course 

M 


Digitized by Google 



162 


THE LATIN et, quc, atque, (oc), and 

would be to start from Ktv, for few will now defend 
the doctrine of a paragogic v. Still the error is a 
common one. Our English grammars, for example, 
still speak of an indefinite article a, which assumes, 
they say, an n before vowels ; and in the same way 
Greek grammars persist in the folly of talking of 
a privativum, when the more genuine form is av. 
What however is important for our present purpose is 
admitted, that the form ice is only used as an enclitic. 
But I may also call attention to the Doric «a as show- 
ing that here too a long vowel was once known, and 
secondly to the disappearance' of the initial guttural 
in av, for this also is a matter which will throw light 
upon what is about to be said. It will be well how- 
ever to note that, as the several forms of icev, *e, «5, 
and av may be well deduced from a form kov, it is 
highly probable that our own language still possesses 
the verb from which all may have been deduced, I 
mean the verb can, which by its meaning is thoroughly 
fitted to supply the root of a ‘ potential’ word ; and 
further, the verb was known to the Latin language in 
the form que~o, for here also a final n once existed, as 
is proved by the archaic ne-quin-ont. 

But the connexion between icac and que may next 
receive illustration. First of all the u in Latin words, 
which divides a preceding q from a vowel, must, as 
still in French, have been silent. This is shown by 
the shortness of the preceding vowel in such words as 
aliquis, nSque, dqua, Idquor. 

It still remains to consider the passage of the diph- 
thong ai first into e and then into 6. Now a parallel 
case presents itself, as it seems to me, in a comparison 
of a certain class of Greek infinitives and the ordinary 


Digitized by Google 



THE GREEK Kai, re, ALL OF ONE ORIGIN. 163 

Latin infinitive. In Greek, as in Welsh, we find a 
great variety of forms for the infinitive, as rvirrev, rinr- 
re/iev, whence with the loss of the ft, rwrecv for rvjrreev ; 
also Tvirrevat and rvinefievai, to take these as types, 
rather than as all representing actual forms. With the 
disyllabic suffix of rvirr-evai I compare the suffix of 
the Latin scrib-ere. That a Greek v should be repre- 
sented in the first place by a Latin a, and then by a 
Latin r,^ is always to be expected. Thus the plural 
Tvirroiiev ‘ we strike,’ goes with a Doric runrofies and a 
Latin tundimus. Again the comparatival suffix u>v of 
the Greek has for its Latin analogue an archaic ios, 
melios, and a later ior, melior ; and even the change 
in the quantity of the vowel of the Latin comparative 
follows the law, which gives us scriptdi'es in Latin by 
the side of the Greek ptjropes. In the Latin infinitive 
esse, and the archaic passive dasi, for dan, we have 
the earlier sibilant retained. There remain then for 
comparison the final diphthong ai of Timrepai and the 
final e of scribere. Now a final m in Greek soon 
lost much of its diphthongal power. Even Buttmann, 
a most zealous advocate of the prevalent accentual 
theory, lays it down, with others, that a final ot or ot, 
though long for metrical purposes, must for the most 
part be considered as short in the rules of accentuation 
(Ausfuhrl. Gr. Gr. Spr. § 11, 7). ‘ Thus,’ says he, ‘ the 

plural nommatives rplaivai, &c., the passives in o«, as 
rvvropai, &c., and the infinitives iroiijffai, &c., are all 
accentuated in a manner that is inconsistent with the 
usual law for words with a long final and he adds 
the remark, ‘It is therefore clear that in these very 


' We have already an example in the Greek catftyr), Latin lauru-. 

M 2 


Digitized by Google 



164 


THE LATIN et, que, atque (ac), and 

common suffixes these diphthongs had been so for worn 
away that in the ordinary language they sounded to 
the ear as short, and that it was only in the sustained 
language of poetry that the long quantity was main- 
tained.’ So much for the Greek a». Much the same 
occurred in the final e of the Latin infinitive, for this 
also was once long in the old language. Some in- 
stances of this occur in Plautus, as : 

‘Atque ugento c6mpaTando flngerB falldciam.* (Aain. ii. 1, 2.) 
‘Quid biicciomt Ulut dicerfi uolui femur.’ (Glor. i. 1, 27.) 

‘ Nunquam 4depol uidi pnimerS. Verum h<5c erat’ (Glor. iiL 2, 34.) 
‘T4 salutem m4 iuaseront dicerS. Saluie sient.’ (Glor. iv. 8, 6.) 

And also in Terence, as : 

‘ Potjn 4s mihi verum dicere 1 Nil fdcilius.’ (Andr. ii 6, 6.) 

‘ Ausciilta. Pergin credere t Quid ego dbsecro.’ (Phono, v. 9, 7.) 

In the ‘Rheinisches Museum’ (xxii. 118) Di*. W. 
Wagner lias added to this list, from Plautus ; 

‘Egd scelestuB mine argentum prdmere posBdm domo.’ (Pseud. 
356.) 

‘ Nam c^rtumst Bine dote h4u dare. Quin tu i moda' (Trin, 584.) 
‘Eum opdrtet amnem quiereie comit4m sibi.’ (Poen. iii 3,, 15.) 

‘ Non 4udea aliquod* mibi dare mnniiacaium.’ (True, ii 4, 74.) 

And from Terence : 

f Male dfcerS^ male ficta ne noBcint eua.' (Andr. Prol. 23.) 

To say nothing of the cases where the e in question 
closes the first dimeter, as in Plautus : 

‘ Abac4de ac sine me p4rdere qui Bumper me ira inc4ndii* (Aa. 
420.) 

‘Quid r41icuomt aibat r4ddere qnom ext4mplo redditdm esset.’ 
(As. 442.) « 


f Or ‘ali(]ui(l . . mumisculi.’ ’ Add fore (Most, i 3, 67.) 


Digitized by Google 



THE HREEK <cat, t«, ALL OF ONE ORIGIN. 165 

‘ Vix h6c Tiidemar cr^dete : magis qu{ crodatis dicam.' (Poen. v. 
4,94.) 

‘At 4ccum e iano rdcipeie aided ae Suncer&atuin.’ (Poen. iv. 1, 5.) 

These from septenarii, or comic metre. Instances from 
complete tetrameters are : 

‘Studeo hiino leonem pdrdere qui mdum erom misere mdceiat.’ 
(Poen. 4, 1, 2.) 

‘Peril, 4nimam nequeo udrtere: nimia nili tibicdn aiem.’ (Merc. 
125.) 

‘ Qui aiim pollicitoa diiceie t qua audAcia id iaceie Andeam.’ (Ter. 
Andr. 613.) 

The passage quoted from the Gloriosus (L 1, 27), 
though it has the full sanction of the MSS., Ritschl 
already condemned in his Prolegomena (p.ecxxLx.), and 
again in his text of the play. In the BL Mus. (vii. 
312) he discusses the question at some length, arguing, 
on the authority of what he deems parallel cases, that 
the order of words, illut dicere uolui femur, is against 
the habit of Plautus. But in fact the cases he quotes 
are not parallel ; and I venture to assert that when 
illut is used, as here, to draw attention to a coming 
word or words, in opposition to what precedes, it is 
a law of the language that the word or words so 
referred to should lie at a distance from the pronoun, 
as seen in the examples which I have quoted in my 
Grammar (§ 1106). 

All this, then, tends to justify the doctrine that a 
Greek km may well have for its analogue in Latin 
both que and qui. 

But. if KM, re, and que be admitted to be one in 
origin, there still remains the Latin et. This some 
have thought to explain as only a metathesis of re. 
Such a doctrine I of course put aside as untenable. 


Digitized by Google 



166 THE LATIN et, qtie, atque {ac), and 

My view is that as the Latin particles en ‘behold,’ 
and ce ‘ look,’ arc corrupted varieties of a fuller ken, 
so Aot, re, and que have all lost a final consonant, 
while et has lost an initial, viz. a guttural, or A-sound. 
This theory, that et and que grew out of a fuller quet, 
is confirmed by the fact that que of the Latin quando- 
que uterque is pid in Oscan and pe or pei in Umbrian 
(Corssen's Aussprache, 1 337). But I am not wedded 
to a 2 as the original final. I think it not unlikely that 
the earlier letter -was an n. Indeed a Greek particle 
could not have ended in a I am led to a preference 
of an n over a f by the form of our own and and the 
German und, for these virtually end with an n, a final 
d after an n being a common outgrowth in these two 
languages ; and indeed in not a few combinations we 
ourselves practically drop the d, as for example in the 
phrase, ‘ four an twenty blackbirds,’ &c. ; and this not 
merely when a consonant follows, for we also habitually 
say, ‘ five an eight make thirteen,’ dropping the d of 
and. I am the more tempted to identify the Latin et 
and English and, when I find the Greek erepos taking 
in German the form ander; and it may also be observed 
that the syllable «t of krepot represents the h of the 
numeral et», thus furnishing an instructive example of 
the interchange of v and t. But if et belongs to the 
same stock with Kai and que, it must have lost an 
initial guttural. Of the loss of an initial consonant 
numerous examples have already been noticed in this 
paper, and the loss of a final v in Greek is the great 
characteristic of forms in that language, a fact which 
has commonly been concealed under the theory of the 
V eipeXKVcrriKov or irapaycoyiKov. The Latin, too, shares 
the habit : thus while the Greek vTote indifferently 


Digitized by Google 



THE GREEK Kai, re, ALL OP ONE ORIGIN. 167 

rrpoa-dev or trpoaOe, xev or xe, &c. the Latin has inde 
in place of evOev, though it is not unlikely that indti/f, 
or indts, rather than inden, prevailed in the older 
Latin, for we have here virtually a genitive with the 
power of ‘from.' Exactly in the same way in our 
own island there co-existed forms henn-en, heth-en, 
henn-es, as well as our still current hence. Nay, over 
and above these Chaucer in the Knightes Tale (2,3S8) 
has the abbreviated hen — 

' The fyr^ which that on mine auter bren 
Shuln the declard or that thou go hen.’ 

Thus we have a form scarcely distinguishable froiri 
the French en, which is the representative of the Latin 
inde, to say nothing of hin as it appears in hin-c. 

But, to return to the little family of copulative par- 
ticles, let me ask whether they may have grown out of 
the demonstrative family. The adverb item, signify- 
ing ‘ likewise,’ has what is very near the meaning of 
our little word, but it is itself a compression of itidem, 
which stands to the adjective idem much as ita ‘ so’ 
to is {ea, id). If this be admitted, a form ken, which 
as before stated I have long regarded as the primitive 
form of the family of third-person pronouns, is in its 
exterior well suited to have been the origin whence 
came the particles xai, re, et, and que, as well as re and 
que, with long finals. 

I was first led to the train of thoughts out of which 
this paper has grown by the consideration that xat and 
re, on the one hand, could not well have been cor- 
relative particles unless they had been one in origin. 
But que and et also serve together ; at any rate 
in short phrases. 'Thus, Livy has seque et cohortem 
(xxv. 14), et singulis universisque (iv. 2) ; and Sallust, 


Digitized by Google 



168 THE LATIN et, que, atque {ac), and 

seque et oppidum (Jug. 26), seque et exercitum (ib. 
55). 

It was of course reasonable that the Greek language 
should use in correlation a repeated t«, and the Latin 
in like manner both a repeated et and a repeated que. 
Thus in exactly the same way the latter language has 
aut . . aut . . , vel . .vel . . , sive . . sive . . , simul . . 
simul . . , qua . . qua . . , turn . . turn . . , nunc . . 
nunc . . , modo . . modo . . So in English we at times 
use or .. or , nor . . nor . . But here the more 
prevalent forms are neither . . nor . . , either .. or . 
in which the principle seems to be violated. The ex- 
planation however is not far to seek. Our either, so 
used, of course corresponds to the Germ, entweder. Old 
Germ, ein-weder (Grimm, D. G. iii. 38), where the 
ein is the mere numeral and weder a comparative of 
the relative. Hence it is virtually the same with the 
Latin alter-uter ‘ one of the two (no matter which) ’ ; 
and this has for its positive ali-quis ‘ any one of any 
number.’ In the same way neither seems to have 
grown out of a form ne-whether, corresponding to the 
old Latin ne-cuter, aft. neuter. Hence the just ex- 
planation of the combinations above quoted, is that 
originally a pause occurred after the words either and 
neither, as : ‘ either (of them), A. or B.,’ ‘ neither (of 
them), A. nor B.’ In the second of these cases the 
omission of the negative before A. has its parallel in 
the old construction, still admissible for poetry, which 
is seen in Shakespere, as (Antony and Cleopatra) : ‘ For 
Antony, I have no ears to his request The queen of 
audience nor desire shall fail and again in Gray : 
‘ Helm nor hauberk’s twisted mail, nor e’en thy virtues, 
tyrant, shall avail’ Indeed we find the same in Greek 


Digitized by Google 



THE GREEK km, re, ALL OF ONE ORIGIN. 1G9 

poetry also, as (Aesch. Agam. 532) : riapt; >)iap ovt€ 
avirreXT/t iroXig to Spafui rov iraOovs irXeov. 

Nor is there any real difficulty or ambiguity in such 
phrases. The negative which precedes the second 
member makes its appearance in time to affect the 
following verb, and through this to influence the first 
of the two members. The same principle is at work 
in those Latin sentences where non mode was once 
said to stand for non modo non. Thus in such a 
sentence as : ‘ Assentatio non modo amico sed ne libero 
quidem dignast,’ the ne of ne libero quidem converts 
digna into indigna, and so acts upon the preceding 
amico. I may add that this explanation of neither 
and either is also applicable in such constructions as : 
‘ both (of them), A. and B.,’ ‘whether (of them), A. or 
B.’ A strong confirmation of this argument is seen in 
the occasional use of two interrogative particles after 
the Latin utrum, as in Ter. (Ad. iii. 3, 28), ‘ Utrum, 
studione id sibi habet an laudi putat fore, si,’ &c. 
‘ Which of the two is the just explanation — does he 
look upon it as an amusement^ or does he t hink it will 
be a credit to him, if’ Ac.? Thus the particles which 
really correlate with each other are ne and an ; and 
these may well be of the same origin, the two being 
connected by the disyllabic anne, which instead of 
being a compound I believe to be the original word 
whence both an and ne proceed.' Thus, as already 
noticed, «/ of Greek, and ni of Sanskrit, find them- 
selves co-existing in the Greek ew ; av of Greek' and 
the Sclavonic na in the Greek ava ; to say nothing of 
the other cases quoted above. 

* See tlio followng paper. 


Digitized by Google 



170 


VIII. 

ON THE LATIN PARTICLES aut, an, ni. 

As I have been led to connect these little words 
with the adjectives alio- and altero-, Sanskrit anya- 
and antara-, I must commence by considering the 
origin of the latter ; and in doing so my first duty 
is to put aside some derivations to which others have 
lent their sanction. Thus Bopp (V. G. § 19, voL i. 
p. 33), and Pott (E. F. of 1859, pp. 301,' 381, 393), 
are disposed to treat alius as a derivative of the San- 
skrit ana, Latin ille ; and the former connects ullus 
with ille and ultra. Dr. Donaldson in his Latin 
Grammar is so enamoured with the first of these 
two views, that he puts it forward three times, as 
p. 45, ‘ alius (like ille “ that other," of which it is a 
by-form),’ &c.; p. 74, ‘alius “another," is in constant 
use as a by-form of ilh:’ p. 386, ‘alius, which is 
merely another form of ille — ollus’ That ille and 
ultra are of one stock is past douht ; but ullus is of 
course the diminutive of units, as uillum is of uinum, 
as bellus of bonus (cf. bene). 

Again, the doctrine of the Indian grammarians that 
the Sanskrit antara- {altero-) is formed from anta ‘end,’ 
and a verb ra ‘ reach or attain,’ may be accepted as 
an example of the way in which native Sanskritists, 


Digitized by Google 



ON THE LATIN PAETICLES aut, an, 7lS. I7l 

satisfied with external similarity, deem it superfluous 
to consider the meaning of words ; and the same one- 
sided examination of etymological problems is not 
unknown among European Sanskritists. 

That n of the Sanskrit anya-, antara- is more 
genuine than the I of alio-, altero- is rendered pro- 
bable by the prevalence of the n in the Teutonic 
family, as Germ, and-er, Norse anrtrar ; as also by 
the fact that the Latin language had a special love 
for the soft liquid, which often led it to substitute an 
I for other consonants. But besides alio- the Latin 
also possessed a short form ali- (whence alis, alid of 
Lucretius, and aliter). Tlie ratio then of alio- to ali- 
suggests for the Sanskrit an equal ratio, anya- to any- ; 
and this theoretic any is for Englishmen an actual 
word. But oiu* any is one with the German einig, 
two words which ore in fact diminutives of the 
numeral an Eng., ein German ; just as ullo- ‘ any’ is a 
d imin utive of uno-. Hence, reserving for the moment 
all question as to the connexion of ideas, the an of 
the Sanskrit an-ya-, ant-ara- seems to be identical with 
our numeral an, and consequently with our one and 
the Latin uno-. But the g of einig also claims at- 
tention, and this suggests the idea that alio- is only 
a variety of unico-, the guttural having disappeared. 
This explanation seems preferable to Bopp’s explanation 
(§ 292) that the ya of anya- is the stem of the relative, 
for the two Latin forms ali- (glis) and alio- (alius) 
bear evidence that the y and a of anyor are two 
independent suffixes. 

Some support to the doctrine that al of alius, &c. 
originally carried with it the notion of ‘ one,’ is to be 
found in the identity as to meaning of the Greek 


Digitized by Google 



172 ON THE LATIN PARTICLES aut, an, ng. 

aWtiXot (evidently consisting of a repeated oXXo-) and 
the German ein-ander and our own one another. 

In alio- and altero- it is commonly held that ‘ dif- 
ference’ is the primary meaning of the first element ; 
but this in no way suits the compound forms aliqui-, 
aliquot, aliquanto-, aliqtcando ; nor indeed all the 
uses of alio- and altero- themselves. The doubled 
alter and the doubled alius render it necessary to 
give to the adjective on its first occurrence the trans- 
lation ‘ one’ (pi. ‘ some’) ; and even the following clause 
makes no objection to the same translation, though the 
word ‘other’ is then admissible. Thus aliud est 
maledicere, aliud accusare, ‘ it is one thing to abuse, 
one to accuse.’ So again alter exercitum perdidit, 
alter vendidit, ‘one of the two lost, one sold an 
army.’ 

Although it seems at first a strange result that a 
word formed from one, itself so often employed to 
denote identity, should eventually attain to the sense 
of difference, cases nearly parallel may be adduced. 
Thus when Ovid, describing the half-military character 
of the farmer in his place of exile, says, ‘Hac arat 
infelix, hac tenet arma manu,’ the repeated pronoun 
evidently refers to difierent objects ; and so we may, 
in place of the literal translation ‘ this,’ substitute the 
words ‘the one,’ ‘the other.’ This repetition of hie 
has its counterpart in-'a similar repetition of ille, as 
(Ter. Ph. iii. 2, 16) : 

‘G. Qui istuc? Ph. Quia non r^to accipitri t^nditur, neque mfluo, 

Quf male faciunt ndbia : illis qui niliil faciunt tiinditur ; 

Quia enim iu illis friictus est, in Ulis opera Idditur.’ 

We may quote, too, as an illustration what Bopp says 


Digitized by Google 



ON THE LATIN PARTICLES OUt, an, 11 ^. 173 


in his V. G. (§ 37l) : ‘ That which in Sanskrit signifies 
“this” means also for the most part “that,” the mind' 
(he should have said the finger) * supplying the place 
whether near or remote.’ Hence there is nothing very 
strange when we find in our oldest writers such a line 
as that which occurs in the Life of St. Edmund the 
King (Trans. Philolog. Soc. 186^), v. 9 : 

' Hubba was |>ofer ihote : & )H>])er bet H3rngar.' 

Just as the finger serves to distinguish ‘this’ and 
‘ this’ when they are to be referred to diflerent objects, 
so no real confusion occurs when Davus in the Andria 
(ii. 2, 12) addresses first PamphUus and then Cha- 
rinus as a tu — 

‘ id paves ne diicas tn illam ; tii autem ut ducas.’ 

Again in Ovid’s Fasti (ii. 676) a consideration of 
this simple kind would have led to the correction in 
the easiest way of what in the received texts, even that 
of Merkel, is mere nonsense. 

The passage is one in which the poet addresses the 
god Terminus ; and, as both Merkel and Paley give 
it, runs — 

' Et sea uomeribus, sea ta polsabere lastris, 

Clamato, suos est hie ager, ille tao%’ 

while others have, ‘ Metis est hie ager, ille sum.’ 

Now meus and suus are clearly wrong, because with 
meus Terminus would ^ claiming the land as his 
own ; while suus would’^mean that the land belongs 
to itself, that is, if the phrase has any meaning at all, 
that the land is without-' an owner. Common sense 
requires *tuus est hie ager, ille turn,’ the god ad- 
dr^ing first one person, and then another. Strangely 
enough, ‘ tuns est hie ager’ is the reading of nearly all 


Digitized by Google 



174 ON THE LATIN PARTICLES aut, an, nS. 

the MSS. ; and thus the substitute of mens or smis in . 
place of tuus is, on the score of authority and on the 
score of meaning alike, utterly indefensible. 

In the compounds aliqui- aliquot &c. the notion of 
‘some’ or ‘any’ prevails; but this is a meaning that 
constantly connects itself with words of numerical 
origin, as for instance in our own an-y. Germ, eiurig, 
Lat ullo- already quoted, and this with reason ; for a 
diminutive of ‘one’ still leaves the idea of ‘some.’ 

But our own term oth-er is itself only a compara- 
tival form of one, standing for on-er. I was first led 
to this view by the recollection that our language, 
while it shares with the Greek and the Norse a strong 
love for the asperate th, also habitually interchanges this 
letter with an n. Thus the 6 of fiey-e0-o», tuc-a$-tw, 
corresponds to the v of re/i-ev-ot, mag-n-us, pig-n-us, 
\afi^-av-eiv, sper-n-ere ; the 0 of iTa0-o» to’ the v of 
irtp-ofiai and of course if ^ be convertible with v, 
a fortiori with v0 : so that the forms iSpw0i}i>, 

ap.irvvv0t)v, from lipva, avavveo, and irev0-of, fiev0-o», by 
the side of ira0os, ^a0os, have nothing in them that is 
very strange. 

In Anglo-Saxon again, the plural of the indicative 
present ends in as, but that of the subjunctive present 
has on or an, and the past tenses alscL prefer on. 
Similarly a Norse nom. ann-ar (= alter) forms a 
fern. ac. ojera, a dat. s. &iru, a dat. pi. &srum, Ac. ; 
and a nom. mas-r ‘ man’ stands by a gen. mann-s. 

But it is not only in the Norse that our ‘ other’ is 
represented by two forms, one with a liquid, annar, 

^ L and S. deny tlio connexion of these two words, liolding that 
iraO-n{ belongs to wa(T)(u. But why may not all thi-eo be of one 
stock! 


Digitized by Google 



ON THE LATIN PARTICLES aut, an, nS. 175 

the other with an asperate, axra or &fru, &c. In the 
provincial utterance of Lincolnshire the original n has 
been preserved. Thus a friend from that part of the 
country supplies me with the following phrases, which 
may be heard, he says, any day ; — 

‘Was it A. or B. who told you?’ Ans. ‘I don’t 
know which, hut it ims toner.’ 

(Speaking of two pigs.) ‘ Toner a mun (I must) sell, 
but which on ’em a hardlins know! 

’It was toner {— either) Mrs. P. that I met, or 
toner (else) Mrs. O! 

Let me add, what it is not beneath the dignity 
of philology to record, that a youngster, F. S., aged 
two, seeing one day on the dinner-table a second 
pudding to his delight, exclaimed in my hearing, ‘ Oh, 
’nunner podn ! ’ while his elder brother, H. 8., at the 
same age had given a preference to another intelligible 
variety, ’nudder. It may further be noted that the 
theory which finds in the aU of aliqui-, &c. an equiva- 
lent of ‘an’ or ‘one,’ has its proof in the Norse form 
ein-hver, which Grimm himself (D. Q. iii. 38) trans- 
lates by the very word aliquis ; and of course the 
Sw. en-hvar and Dan. en-hver, though now signifying 
‘quisque,’ are the same word. The neuter form of 
the pronouns, ett-hvart, et-hvert, prepares us for the 
German et-was, which again = ali-quid. 

This brings me almost to the Greek h-epo-, which 
however stands apart from all its congeners, as having 
an asperate. But this very peculiarity furnishes the 
strongest confirmation of the present theory, for among 
the various forms of the first numeral the Greek ev 
stands alone in this particular. I thus at any rate 
escape fium the difiSculty which Grimm meets by 


Digitized by Google 



176 ON THE LATIN PARTICLES aut, an, nH. 

simply cutting the knot, telling us that ir€pos had 
originally in all probability no asperate (iii. 636). That 
the asperate in these two words very possibly super- 
seded a digamma I readily admit, seeing that the 
archaic Latin oeno-, the Lith. vnenor, and our own one 
virtufilly begin with this sound. It should also be 
noticed that both krepo- and Iv- agree in a common 
vowel, and that the interchange of a dental liquid and 
a dental tenuis is of the most ordinary occurrence. 
One result of this derivation is that epo- alone, not 
rtpo-, constitutes the comparatival suffix ; but this is 
what I gladly accept.^ If my explanation of hepo- 
be correct, the leading sense of the word is ‘ one of 
the two,’ which in our Greek lexicons is given indeed, 
but is commonly relegated to the last place. To test 
this little matter I run my eye over our best lexicon, 
and find that in twenty-three adjectives compounded 
with erepo- the word ‘ one’ is essential to their trans- 
lation : eTep-aXxea--, -ax&etr-, -^p,epo-, -^ape<r-, -yXavKo-, 
-jvaOo-, -517X0-, - 0 a\ea-, -drjiCTO-, -KXivea-, -Koxfto-, -ftaXXo-, 
-p,aaj(aXo-, -p,epe<r-, -fioXio-, -vXoo-, -iropiro-f-ppovo-f-trKut-, 
-oaropM-, -ovar- one-eared, -^oeo--, -o^OaX/to- one-eyed ; 
and I might add erepo-woS- ‘ one-footed,’ h‘epo-trKeXt<r- 
‘ one-legged,’ for a person who has an imperfect leg or 
foot may well be so called. Of course in all these adjec- 
tives the notion expressed, viz. * one,’ is ‘ one of two.* 
The Latin iterum seems to claim a place among the 
words which have been under discussion, and this 
claim is perhaps confirmed by the form of the German 
ioieder. At any rate those who would derive iterum 
from the pronoun i~ ‘ this’ (Dr. Donaldson for one), 

' See the eleventh of these papers. 


Digitized by Google 



ON THE LATIN PARTICLES aut, an, nS. 177 

have overlooked the fact that the signification of the 
words repudiates the theoiy. A derivative from such 
a pronoun would signify ‘ hither, citerior.' The logical 
connexion of iterum with hepov, Sansk. itara, is satis- 
factory, as well as that of form. Still a doubt hangs 
over the question when we find, devoid of all com- 
paratival suifix, the Old Germ, ita ‘ again,' the Anglo- 
Sax. prefix ed- ‘ again,' the Welsh prefix ad-, and the 
ad- of like power in the Latin ad-mone- ‘ re-mind,' 

. a-gnosc- ‘ re-cognise,' as well as the English a-cknow- 
ledge and archaic orcknow, for here we come across 
representatives of the Greek ava. The Danish otter 
and Swedish dter are simply corruptions of an older 
achter, a variety of our after. 

I conclude this part of the subject by collecting, 
chiefly from the D. G., the various forms that repre- 
sent the Latin altero - : — Sanskrit antaron and itaror. 
Old Prus. antar-s, Lith. antra-s, Lett, ohtr-s. Old 
Slav, utoryi, Greek trtpo-, Latin altero-, Goth. a»ij>ar. 
Old Fris. other. Old Sax. other, oear, odar, Ang.-Sax. 
oeer, Saterl. ar, or, Eng. other. Old Germ, andar. Mod. 
Germ, ander, Dutch ander, Norse annar, Swed. annan, 
Danish anden. 

But the form tother must not be passed over. When 
it means ‘ the other' it is not difficult to account for 
the passage of th into a mere t, as such change is 
only in harmony with the law in Greek, which ■writes 
0pt( and rptxof, but not dptxof- This tother (also 
foper in Old English, as quoted above) is exactly one 
with the Greek Oarepov for to arepov. But for the 
most part the form tother (Scot, tither) has an article 
preceding it; and then the t is due to what Mr. 
Whitley Stokes calls Provection, ha'vdng been trans- 

N 


Digitized by Google 



178 ON THE LATIN PARTICLES ClUt, an, 

ferred from the end of the preeeding word, just as in 
/or the nonce, in place of /or then once. In other 
words, t/ie tother would be more correctly divided thet 
other, precisely as the tone should give way to thet 
one. In fact, in the older writers tother is rarely 
found, I believe, except with a prefixed ‘ the.’ In this 
form, for example, Jamieson gives one quotation from 
P. Plowman, two from R. Brunne, and four from 
Scotch authorities. I have noted fourteen occurrences 
of the phrases that oon, that othur, one or both to- 
gether, in the metrical parts of the Canterbury Tales. 
Thus in the Knightes Tale, v. 477 : — 

‘ Of whiche two Arcita higte that oon, 

And he that othur liighte Palamon.’ 

And, again, the Life of St. Edmund the Confessor 
(Trans. Philolog. Soc. 1858) has in v. 477 : — 

‘ Nis ]>at on )i|>er youn^ ; |>e| heo ne loro ]>at o|)Cr also.' 

In Greek too to Oarepov probably originated in rod 
drepop, the 0 before the asperated vowel representing 
the final dental of the original pronoun. I feel the 
more entitled to defend the division rod drtpov,^ be- 
cause Greek MSS., like Latin MSS. in similar cases, 
write such words as the article in immediate con- 
nexion with their nouns ; the division, which is seen in 
our printed books, being due to editors alone. 

I here assume that thet or that is an older form 
than the, and so discard the common doctrine that 
we have in the final t of that a neuter suffix. Indeed 


' So Bopp (§ 156): ‘Aua dem Zeugniss der verwandten 

Sprachon erkennt man daas to UTspriinglich ror oder toS gelautet 
hahe.’ (Seo also note 13 to § 319.) 


Digitized by Google 



ON THE LATIN PARTICLES mU, an, tlS. 179 

such a theory is inconsistent with the fact that the 
pronoun t/iat (like what) is capable of being used in 
connexion with words which are distinctly not neuters, 
e.g. that man, that looman. The original form I 
believe to be rather then} or than, the n having sub- 
sequently passed, as it so often does, into a t. Thus 
in then-ce, when-ce, &c., or, as they were once written, 
them^es, whenn-es, the es is a genitival suffix signifying 
‘ from,' precisely as in voO-ev, for so I divide the word, 
just as I wrote rod-arepov above. The idea of any 
neuter suffix is, I think, to be rejected, among other 
reasons, because suffixes to imply negation are in them- 
selves improbable ; and again in ut-ero-, ‘ which of the 
two,’ TTOT-epo-, the root-syllable has a claim to the 
dental. The p of ayaOop and m of bonum, as I have 
elsewhere explained, are no exceptions to the general 
liw which rejects suffixes of mere negation. 

With this preface I next proceed to the adverbial 
forms which signify ‘ or,’ taking first those of the Teu- 
tonic family, as exhibited by Grimm (D. G. iii. 274, § c). 
But here I would suggest a caution against a prevalent 
error, that of attaching too much weight to antique 
as compared with later forms. A safer course is to 
give a preference to fuller forms over shorter, so long 
as one is sure that the greater length is not due to the 
addition of a new element. Thus while the Gothic 
dif\idu. Old Germ, edo, eddo, erdo, odo. See., and Ang.- 
Sax. oetse, Norse ern, and Latin aut exhibit no final r, 
we are justified in regai-ding, as so far purer forms, the 
Modern Germ, oder, and the Swed. and Dan. eller. 
The greater fulness alone is an argument in favour of 


* See tuy paper on Pronouns of the third person. 
N 2 



180 ON THE LATIN PARTICLES aut, an, nS. 

this view, but the question is at once decided by the 
necessity for a comparatival suffix in order to express 
the required idea. Similarly when we compare the 
varieties edo, eddo, erdo, we are bound to give a pre- 
ference to the last ; and of the Middle Germ, varieties 
ode, oder, aide., alder, while oder and aide have each 
their own superiority over ode, the highest claim 
belongs to alder, which differs but slightly from the 
Modem Germ, adjective ander. That aut is an abbre- 
viation of alter (Fr. autre), seems to be commonly 
admitted; nor need we be surprised at the abbreviation, 
when we find such abundant evidence of the gradual 
absorption of the comparatival suffix in general in all 
the Teutonic branches (D. G. iii. 589 — 596), whenever 
the irregularity, so-called, of the formation prevents 
any resulting confusion with the positiva Grimm 
indeed seems to limit this truncation to the adverbial 
comparatives, but our old English writers extend it to 
adjectives. Thus Shakspere (Othello, iv. 3) talks of 
‘ mo women,’ ‘ mo men,’ and Chaucer abounds in such 
phrases as (C. T. 9,293) ‘ Bet is quod he a pyke than 
a pikerell ;’ and we still use less in place of lesser} 
So the Latin, besides aut, exhibits an abbreviated com- 
parative in the first part of ma-velle, ma’lle, which 
corresponds to our Ang-Sax. md, and in mt for satis. 
I’he Keltic family takes the same liberty in its irregular 
comparatives, as for example the Breton in mdd ‘ good,' 
gwelloch or rather gwell ‘-better ;’ drouk ‘ bad,’ gtoasoch 


^ In the old language other instances are found, as leng in place 
of the fuller lenger (for longer ) ; for example in St. Edmund the 
Confessor (published in the Society’s volume for 1858), v. 366 : 
‘ )>er hit gan dasche adoun : hit nolde no leng abide ; ' and again, 
V. 610 ; ‘ He answerede him }« leng )>e wors. 


Digitized by Google 



ON THE LATIN PARTICLES aut, an, n?. 181 

or rather gwas ‘worse,' So too in Welsh the cor- 
responding forms are gwell and gwaeth rather than 
gwellach and gwaethach, and in this language indeed 
there are some twelve other comparatives that have 
undergone the like curtailment.' 

The etymology of the Gothic di^]4u, which Grimm 
places at the head of the series of words representing 
‘ or/ had been the subject of a previous discussion in 
p. 60 of the same volume, but the writer with good 
reason seems to attach no great value to his own solu- 
tion of the problem. The view taken in the present 
paper of course requires that the numeral ‘ one' shall 
constitute the first element, and accordingly it agrees 
closely with the Gothic form of this numeral, din, 
making allowance for the passage from the dental n 
to the dental th, which we have already seen in this 
word. The final du of di\\du corresponds no doubt 
to the final vowel of the old German eddo : and, as 
this appears to have lost an r belonging to the com- 
paratival suffix, so the Gothic may be presumed, like 
ovTw for ovrms, to have lost an s, which in that dialect 
represents the German r; and 6s is the very form 
which Grimm assigns to the comparative of Gothic 
adverbs (D. G. iii. 585, and with two examples 
596 B. i.). 

Our own particle or, as proved by its German 
equivalent oder, has sufiiered the same compression 
as gaf-fer for grandfather, as gam-mer for grand- 
mother, as where (= quo) from whither, as where 
in Somersetshire for whether, as Scotch smure for 
smother, as the Danish far-hroder {i.e. ‘patruus/ or 


* See also Grimm, D. G. 


Digitized by Google 



182 ON THE LATIN PARTICLES aut, an, n^. 

father's brother). But better evidence cannot be found 
than that which the Old Frisian forms ofFer, where the 
ordinary adjective other, besides the fuller forms, has 
a gen. or-a, a dat. or-em, an acc. or-ne (Richthofen's 
Altfriesisches Wdrterbuch, v. other), and ^ain or-hctXf, 
as well as other-Jutlf, corresponding to the German 
anderthalb ‘ 1 The word eith-er, so much used as 
the correlative of or, one is tempted at first to regard as 
a mere variety of other, especially as the first syllable 
eith coincides, as nearly as is to be desired, with the 
Gennan ein ‘ one.’ But the German entvoeder, which 
in use corresponds to our particle either, is no doubt, 
as Grimm suggests, deduced from ein-weder, the n of 
which, in my view, has thrown out an excrescent td 
If so, we have what is nearly an equivalent of the 
Latin aJter-uter, the sole difference being that, while 
the Latin attaches a suffix of comparison to both 
elements, the German with a wise frugality is satisfied 
with the presence of a single suffix of this nature. 
Care however should be taken not to confound the 
English either which corresponds to the Latin alteruter 
with that other either of our language which had the 
power of uterque, and in Anglo-Saxon was written 
aghvader (Grimm’s D. G. iii. 56, § c.), or in shorter 
form agder. From Modem English this latter form 
has disappeared ; and the German jeder which repre- 
sents it has given up its legitimate sense uterque for 
the more general quisque. An early example of our 
either = ‘both’ occurs in the Life of St. Kenelm, as 
published by the Philological Society (v. 355) : — 

‘ For rijt as hoo fe vers radde : out berste ai|>eTe hire eje.’ 

* See the Tenth of these Papers. 


Digitized by Google 



ON THE LATIN PARTICLES aut, an, nH. 183 

Tlie forms an and ne remain. Already Grimm 
claims an as a word which belongs to the class of 
alter, influenced no doubt by the forms of the Germ. 
ander and Norse annar. But if this be right, and 
for one I have no doubt about it, the process probably 
was this : starting from a form anner, out of which 
alter grew, first the r was lost, in accordance with 
the law which governs irregular comparatives, which 
gave anne, a form actually in use in interrogative 
clauses to denote ‘.or;’ and then this anne by the 
loss of its tail became an, and by decapitation ne, 
whereas the received doctrine has been that anne is 
compounded of an and ne. The only awkwardness in 
these results is that we are making aut, an, and ne the 
same word, whereas in use they must not be altogether 
confounded. Neither an nor ne can ever be allowed 
to act as substitutes for aut, nor the reverse. On the 
other hand, though ati and ne may at times bo inter- 
changed, there are idioms in which this licence would 
not be admitted. This theory, by which an and ne 
are. regarded as corruptions of a fuller anne, has its 
parallel in the theory (see above) which deduces both 
que and et from a fuller quet. 

I feel that what I have here written will scarcely 
find acceptance with one class of philologers, — I mean 
those purists who expect roots and the derivatives from 
roots to take one and one only form, whereas in truth 
no language is so strictly homogeneous. Practically 
one finds every language surrounded by a cluster 
of what are called dialects, out of which the written 
language has borrowed no small number of elements. 
Thus our own language exhibits root-words sometimes 
in a triple variety, as bag, hay, and how , — words in 


Digitized by Google 



184 ON THE LATIN PARTICLES aut, an, nS. 

origin all one, yet in uso far fix>m interchangeable, for 
it would require an interpreter if one came across such 
a statement as : ‘ He put a few clothes into his carpet 
bay, made a bag to his friends, and started for a voyage 
across the Botv of Biscay.’ It is matter for less won- 
der then, if, starting from a twofold numeral, an (a) 
and one, we find a variety in the root-syllable of its 
derivatives, as any, other (or), else (A.S, ell-es), eleve^i, 
{el-leveii), either ; while the German has ein-ig, and-er, 
od-er, et-ioas, ei-lf; and the Old Norse goes so far as 
in the same noun to give us a nom. sing, ann^ar and 
a dat. ph bth-r-um. 

Still I find myself supported by the authority of 
Bopp as more than once expressed, as § 19, vol. i. 
p. 33 : ‘ Die Spaltung einer Form in verschiedene mit 
grosscrem oder geringerem Unterschied in der Bedeu- 
tung, ist in der Sprachgeschichte nichts Seltenes and 
again in § 616, vol. ii. p. 389 : ‘Hierbei hatte man zu 
beriicksichtigen, dass in der Sprachgeschichte der Fall 
nicht selten vorkommt, dass cine und dieselbe Form 
sich im Laufe der Zeit in verschiedene zerspaltet, 
und dann die verschiedenen Formen vom Geist der 
Sprache zu verschiedenen Zwecken benutzt werden.’ 


Digitized by Google 



186 


IX. 

ON mmAL FOEMS IN LATIN -Wmi A SINGULAR 

MEANING, AND ESPECIALLY ON VIRGIL’S USE . 

OF menta. 

One of the most serious hindrances to a right under- 
standing of the Latin vocabulary is the doctrine, 
often propounded, that the poets by some strange 
licence might use a plural for a singular. But, when- 
ever such an assertion is made, the only safe con- 
clusion is, that the true meaning of the singular has 
been misimderstood. Castra occupies a prominent 
place among such words, but it is not an easy matter 
to decide what was that meaning of castrum which 
justifies the translation of castra as ‘a camp.’ Tra- 
dition supplies no evidence to guide us, and so we are 
driven to etymology. Now castrum in its final letters 
agrees closely with rostrum, rostrum, claustrum, 
plavstrum ; and there can be no -doubt that of these 
the first three are derivatives from the verbs rad-ere, 
rod-ere, claudere. Plavstrum as to foi-m stands in 
the same relation to plaudere, but the connexion 
of meaning is somewhat obscure. Perhaps the ex- 
planation is this. We know that the old roads of 
Italy were narrow ; and hence it was important that 
a large and heavy vehicle should so far as practicable 
give early evidence of its approach. Thus in the 
present day when a carriage of any kind enters a 


Digitized by Google 



186 


ON PLURAL FORMS IN LATIN 


long lane, too narrow for the passage of another 
vehicle in the opposite direction, it is found a useful 
practice on entering the lane to blow a horn ; and 
thus it becomes a sort of law of the road, that the 
giving such a signal carries with it for the time a 
right to the sole use of the passage. Again, especially 
at night time, it is found expedient for a waggon to 
be provided with a set of bells. Now a clapper or 
two boards incessantly striking against each other is 
a cheaper way of effecting the same object ; and 
plaustrum ought etymologically to signify ‘ a clapper.’ 
It is true that even then the clapper is not the waggon. 
Still the sound of the clapper would be good evidence 
of the approach of the waggon, and thus there is no 
wide jump from the one idea to the other. The 
vallum or palisade of a rampart, for example, is only 
part of tlic rampart, which includes the mound and 
the ditch ; but, to an advancing army, the vallum from 
its superior height was the first object seen, and so at 
last came to signify the whole of the rampart. 

If then we apply the preceding evidence to castrum 
we are brought to a syllable cad as the root; and 
here we come across what are too commonly regarded 
as independent words, cadere and caedere ; but these 
are in fact as closely allied as our rise and raise, our 
lie and lay, or, what is nearer to the purpose, our fall 
and fell. In fact, fal, the root-syllable of these two 
English verbs, is the analogue of the Latin cad, for a 
Latin c has often supplanted a labial ; and indeed the 
corresponding ver of vtina> has preserved the original 
consonant ; but a classical p under Eask’s law should 
be represented in our language by an f Other ex- 
amples of a Greek tt, a Latin k (</), and an English f 


Digitized by Google 



WITH A SINOULAll MEANING. 


187 


corresponding to one another, are seen in irurvpet, 
quattuor (Go. jrdmr), Eng. four, and doubly so in 
irefiirTos, quin{c)tus, with the proper name Quinctius, 
fifth. Again, a in Latin, Greek, and even English, 
is often interchangeable with an 1. Thus to take 
what in form is precisely parallel, the familiar noun 
calamitas was written, we are told, by Pompey, as 
kadamitas. I have said that this word thus quoted 
in illustration is identical in form. I may go further, 
for it is also of the same stock, as calamitas speaks of 
a something supposed to issue from the stars, a blight 
falling upon a crop. 

Nay, the change from cad to fal is also to be traced 
in Greek and Latin in this very root, for a^aXXeiv and 
faUere mean strictly ‘ to cause to fall, to trip up : ’ 
hence *the frequency of the combination fallere pedes. 
And again, in our own language, although the ortho- 
dox course is to make fall an intransitive verb, in- 
country life ‘to fall a tree’ is at least as, common as 
‘ to fell a trea’ 

My belief then is that in military language, and 
among the Romans military language was familiar to 
everyone, castra meant generally ‘ trenching tools,’ the 
ordinary axe, and besides these the pickaxe, spade, &c. 
This view is confirmed by the fact that the verb 
castrare ‘ to cut,’ has not merely the notion of eTnas- 
- culare, like our own verb as applied to horses, but is 
applicable in the general sense of the verb cutting, 
and so is used in connexion with such accusatives as 
arundineta, uites, arbusta, caudas catulorum. 

One advantage that results from this theory is that 
the phrases mouere castra and ponere castra receive 
an intelligible explanation, whereas with the transla- 


Digitized by Google 



188 


ON PLURAL FORMS IN LATIN 


tion castra ‘ a camp,’ we are reduced to an absurdity, 
for even the trees of an abattis, after serving the pur- 
poses of one camp, are never carried on to the next 
station to perform the same duty. The tools, how- 
ever, form an important part of a soldier’s plant; so 
to say ; and when an army arrived at the close of a 
day’s march, they would be the first things to be 
taken from the impedimenta. Yet after all there is 
a gap in the theory ; for although trenching tools are 
essential to the making of a camp, and although the 
phrases mouere castra and ponere castra already ob- 
tain in this way thoroughly satisfactory translations, 
yet there is a wide difference between the tools em- 
ployed and the resulting camp. This gap I propose 
to bridge over by the suggestion that the castrorum 
metator, in laying out the proposed form of a camp, 
marked the outline by having the tools themselves 
deposited as he went along where they would presently 
be needed. On the completion of tliis duty, the figure 
would be duly represented to tha eye by the series 
of tools. 

But the use of a plural form to denote a singular 
idea is so inconvenient, that when the use of the word 
in the singular with its original meaning has passed 
away, there is an irresistible tendency to call the 
singular again into service with the new meaning 
hitherto limited to the plural. Hence castrum *a 
fort,’ at last established itself, and still more the 
diminutival castellum ‘a little fort.’ It should be 
noted, however, that in the connexions castrum Inui 
(Verg. iEn. vi. 766), castrum Mineruae (Apulorum) of 
the Itineraries, and castrum Mineruae (Brutiorum) 
of Varro (ap. Probum ad Verg. Eel. 6), the word is of 


Digitized by Google 



WITH A SINGULAR MEANING. 


189 


a totally different origin and meaning. We have now 
a noun belonging to the same family with the so-called 
adjective but rather participle castus ‘pure,’ and the 
sb. castu- ‘ purifying.’ These evidently point to a 
verb, and the verb really exists in cdrire {lanam) ‘ to 
card wool,’ that is, ‘purify’ it, for Varro is no doubt 
right when he explains the term (L. L. vii. 92, p. 339, 
Spengel’s ed.) by purgare, and connects it with ciirere. 
In Greek the root is represented in the adjectives ica6- 
apo- and Ktv-o-. In this view castrum is ‘ a place of 
purification,’ ‘ a shrine,’ and so identical in power with 
deluhrum from lau-ere. Again this second castrum, 
has also its derived verb castrare ‘ to purify,’ whence 
castrare uina saccis ‘to strain’ wine, of Pliny, and 
perhaps castrare lihellos of Martial (i. 36). In the 
latter passage there may possibly be a double 
entendre. 

Nay, that pUxustrum itself did not in origin mean 
‘a waggon’ is shown by its use as a plural in not a 
few passages, where evidently a single waggon was 
before the writer’s mind, as : Modo longa coruscat 
Sarraco ueniente abies atque altera pinum plaustra 
uehunt (Juy. iii. 256) ; Ipse uides onerata ferox ut 
ducat lazyxper medias Histri plaustra bubulcus aquas 
(Ov. Pont. iv. 7, 9); Tardus in occasum sequitur sua 
plaustra Bootes (Germ. Arat. 139). In other passages 
the notion of a single waggon seems, if not decided, 
yet preferable, as in : Tardaque Eleusiniae matris 
uoluentia plaustra (Virg. Georg, i. 163) ; Dicitur 
et plaustris uexisse poemaia Thespis qui canerent 
agerentque (Hor. Ep. ii, 3,* 275) ; Buris opes paruae, 
pecus et stndentia plaustra (Ov. Tris. iiL 10, 59). Thus 
plaustra itself belongs to the class of words here under 


Digitized by Google 



190 


ON PLURAL FORMS IN LATIN 


consideration. Yet already in Plautus, Cato, and 
Cicero the singular plaustrum was in use with the 
meaning of a single waggon. 

Another word in which the true meaning of the 
singular is commonly missed is furca. This word is 
in fact a compression of a trisyllabic for-ic-a, the 
first syllable of which is seen in the verb for-Orre, 
and virtually in fod-ere, for the r and d are inter- 
changeable in these words, just as in auri- sb. ‘ the 
ear,’ and audi- vb. ‘hear.’ Our own language also 
shares the interchange, for the root, in obedience to 
Rask’s law, appears with a 6 in bore and bod- of 
bodkin, whether we use this noun with Shakspere in 
the sense of ‘a dagger,’ or in reference to the little 
instrument which belongs to a lady’s workbox. Then 
as regards the meaning of furca, there can be little 
doubt that we should translate it ‘a prong,’ seeing 
that bi-furco- and tri-furco- mean ‘ two-pronged ’ and 
‘ three-pronged.’ It was at first then only as a plural 
that it could be employed to denote ‘ a fork.’ Some 
of our dictionaries indeed venture to give as the 
original meaning of furca ‘ a two -pronged fork,’ 
quoting in proof Virgil’s furcasque bicomes, which 
however rather points the other way ; for if the noun 
already denoted a two-pronged instrument, the epithet 
bicornea would be superfluous. However, the phrases 
Furcae and Furcul/xe Caudinae for the fork in the 
road near Caudium were established at a time when 
it was still necessary to use a plural to denote ‘ a fork.’ 
So Plautus (Persa, ad fin) has, ‘ et post dabis (manus) 
sub Jurcis,’ where later writers would have said sub 
furca. It is true that in the Casina (ii. 6, 37) we 
find, ‘ ut (juidem tu hodie canem et furcam feras ; 


Digitized by Google 



WITH A SINGULAR MEANING. 


191 


but here we may well suspect that the poet wrote 
furccts, and that the singular was an adaptation to 
later usage introduced in after-time. Such changes 
may be proved to have taken place in the text of both 
Plautus and Terence, just as has happened to the plays 
of Shakspere. 

The nouns forceps, forpex, and forfex have suffered 
much in the hands of our modern lexicographers, who 
have followed the guidance of the author of the book 
entitled ‘ Varronianua’ The writer of that work 
thought he saw in the first part of these words the 
adverb foris, and he was disposed to deduce the final 
syllablfi from the several verbs cap-io, pect-o, and 
fac-io. But in truth the three forms are only dialectic 
varieties of the same word. From fore- of furc-a it 
was thought desirable to form a derivative by the 
addition of the diminutival suffix ec. I say diminu- 
tival, because Pott has clearly shown that the suffix 
UK of Greek substantives adds the notion of little ; 
while the identity of the Greek ax and Lat. ec is 
proved by the forms murex, sorex, pellex, podex, 
corresponding to fiva^, vpa^, waXKa^, irwBa^. But the 
power of the suffix is also sufficiently determined by 
the three words cimex, pulex, culex. Now in the case 
of furca, the addition of a suffix ec would have led to 
an unpleasing fona, forc-ec-, and hence, to soften the 
sound, a labial was substituted for one of the offensive 
gutturals; and so arose the three varieties, ybre-e^-, 
forp-ec-, forf-ec-. But as for cep-, standing for forcec-, 
could only mean ‘ a small prong,' it required a plural 
to denote the more complex instrument consisting of 
two claws. Thus forcipes, as ‘ a pair of pinchers’ for 
the extraction of teeth, is used by LucUius ; uncis 


Digitized by Google 



192 


ON PLURAL FORMS IN LATIN 


forcipibus dentes euellere (ap. Charis. L 74); but the 
later writer Celsus in the same sense habitually uses 
the singular. The word is also used as a plural for 
the ‘ blacksmith’s pinchers’ in Cato ; but here again 
both Virgil (Geo. iv. 175, and ^En. xii. 404) and 
Ovid (Met. xii. 277) have tenaci forcipe ferrum or 
ferrum forcipe curua. It was from the consideration 
of this special use of the pinchers that some etymo- 
logists would derive the word from the adj. formus 
‘ hot’ and cap-ere. But the connexion with /urea is 
confirmed by the fact that while Pliny (ix. 31, 51) 
ascribes to the crab hrachia denticulatis forcipibus 
(al. forfeibus), Apuleius (ApoL p. 297, 4) speaks of 
the Jurcae cancrorum. 

The plural uolseUae is used of a pair of tweezers by 
Varro in the proverbial phrase, ‘ pugnant uolsellis non 
gladio and also by Martial; but for Celsus the singular 
has supplanted the plural, so that the word follows the 
example of forceps, and is used in the same sense. 

Again the familiar noun rastro- (m. or n.) I may' 
safely assume to have meant originally ‘a, single 
tooth of a rake,’ or ‘ a scraper with but a single point 
or edge.’ Hence Terence, Virgil, and Ovid agree in 
the need of a plural to express the more complicated 
rake with many teeth. StUl, as these were perma- 
nently combined in one instrument, it was found in 
the end convenient to use the word in the singular, and 
as such it occurs in the later writers, Pliny and Seneca. 

Another example is bigae, which is of course a con- 
traction of bliugae, and so being an adjective requires 
a noun equae to complete the meaning, ‘two mares 
yoked together for the purpose of drawing a chariot ;’ 
and in this form it is employed by Varro, Catullus, 


Digitized by Google 


WITH A SINGULAR MEANING. 


193 


and Virgil ; but again the unity of the combination 
becoming fixed, eventually later writers, Tacitus 
Pliny Suetonius and Statius, exhibit biga as a 
singular. Precisely the same fate attended the use 
of quadrigae, ‘ four mares yoked together for drawing 
a carriage,’ for the word is a plural in Cicero and 
Virgil, but is exchanged for a singular quadriga in 
Propertius PHny Martial and Ulpiau. If it be here 
objected, that Virgil and Propertius being contem- 
porary might have been expected to use both of them 
either the singular or the plural, a legitimate answer 
seems to be found in the consideration, that the higher 
style of Virgil’s poetry would justify, if not require, 
the use of the older form. 

An eighth example is cassi-, the plural of which 
denotes ‘ a net,’ in Virgil (speaking of a spider’s web) 
and Ovid generally ; but the singular with the sam§ 
meaning is found in Ovid (A. A. iii. .'554) and Seneca. 
Hence it seems reasonable to suppose that the singular 
word originally meant ‘ a single mesh of a net’ At 
the same time it must be admitted that many little 
nets are at times imited to form one large net 

Ninthly, folles as a plural, like our own equivalent 
in form and meaning bellows, is the only shape known 
to Cicero Virgil and Horace, and this agrees with the 
fact that the instrument consists of two flaps ; but 
Livy (xxxviii. 7) Persius and Juvenal have in the 
same sense the mere singular. 

As litera originally meant but a single character of 
the alphabet, a plural was necessary to denote ‘ words 
or writings;’ yet Ovid and Martial have the word in 
the singular with the sense of a letter or epistle. 

It was once the fashion in school books to say that 

0 


Digitized by Google 



194 


ON PLURAL FORMS IN LATIN 


limina was used poetically for the singular, meaning 
‘ a threshold.’ This error however has long been 
thrust aside, as it is known that a door has two 
limina, the 1 . superius or ‘lintel,’ the 1 . inferim or 
‘ threshold,’ the word signifying what carpenters call 
‘a tie,’ and being derived, not indeed from liga-^e, 
which would have given ligamen, but from a lost 
lig-Sre which has also produced a noun lictor (not 
ligator). Still in not a few instances the singular is 
used to denote a gate or entrance. 

Currus is another word as to which our lexicons 
are unsatisfactory. It is clear that in not a few pas- 
sages the plural of this noun is used in speaking of a 
single carriage, as in Virgil (iEn. x. 574), ‘EflFundunt- 
que ducem rapiuntque ad litora currus;’ and Ovid 
(Met. ii. 6), of the chariot of the sun : ‘ Vasti quoque 
rector Olympi non regat hos currus.’ Again the same 
poet (Trist. iii. 8, 1) has: ‘Nunc ego Triptolemi 
cuperem conscendere currus.’ So in Lucan (vii. 570) ; 
‘ Manors agitans si uerbere saeuo Palladia stimulet 
turbatos aegide currus.’ Further, that currus did not 
'in itself mean a carriage, is shown by Virgil’s use of 
the word in speaking of the plough (Georg, i. 174) : 
‘ Stiuaque, quae currus a tergo torqueat imos.’ Now 
the phrase regere currus has a special fitness, if currus 
means strictly ‘ a wheel,’ for it is the wheel which a 
driver has to look to. Further, it is probable that mere 
rollers came into use before carriages. Moreover, the 
word roll is but a variety of whirl and hurl ; and the 
last word in Scotch is a synonym for wheel in the term 
hurl-harrow (Jamieson). Nay in Scotch hurler by itself 
means ‘ one wlio drives a wheelbarrow ; ’ and the 
simple verb hurl is applicable alike to the driving a 


Digitized by Google 



WITH A SINaULAR MEANING. 


195 


wheelbarrow and to a ride in a carriage (Ib. supple- 
ment). For the latter use I quote from the same; 
‘ If a frien’ hire a chaise and give me a hurl, am I to 
pay the hire ? I never heard of sic extortion.' Even 
when hurl has the sense of the Latin torquere 
(hastam), we have the notion of the circular move- 
ment which with the sling and Komau yacM^um pre- 
ceded the casting forward ; but cur of curro is the 
equivalent of the hur or hir of our hurry, hurl, and 
the Dorsetshire hir-n (A. S. yrn-an). Hence I do not 
hesitate to claim for the liatin curro the original 
notion of revolving rather than that of running. 
Such will well suit the repeated phrase in Catullus 
(64, 327, &c.) : ‘ Currite ducentes subtegmina, currite 
fusi and Virgil’s similar use of the verb (Eel. iv. 
46) : ‘ Talia saecla suis dixerunt Currite fnsis . . . 
Parcae.’ So also in those passages which speak of the 
potter’s wheel, as Horace’s (Ep. ii. 3, 22) : ‘ Currents 
rota cur urceus exit?’ No doubt the mere notion 
of running or quick forward movement is far more 
common ; but it is most unphilosophical to decide 
the question of priority by mere number. Nay, it is 
generally to be suspected that the older the meaning 
of a word the fewer should be the examples. From aU 
this I conclude that mere circular movement was first 
denoted by the root, and secondly that the onward 
circular movement as of a rolling stone was the idea 
which preceded that of simple running; so that we 
have here a mimetic word, an imitation of the sound 
heard in rapid whirling. 

The word septentriones at the outset could only 
have been used as a plural ; and such was still the 
form in favour with Cicero and Ctesar ; but Virgil 

0 2 


Digitized by Google 



19(5 


ON PLURAL FORMS IN LATIN 


Ovid Pliny and Vitruvius have the singular; and 
this variety was only the more requisite, when names 
were required alike for the Ursa major and Ursa 
minor, where Vitruvius employs the terms major and 
minor septentrio. It seems indeed a somewhat violent 
proceeding for Virgil to have retained the singular 
form, when by tmesis, as it is called, he gives an 
independent position to the numeral in Talis ITyper- 
horeo septem subiecta trioni Gens (G. iii. 381). As to 
the etymology of the word, two different accounts are 
recorded by Festus. That which would deduce it ‘ a 
septem bobus iunctis quos triones a terra rustici appel- 
lant^’ has little internal evidence to support it. I 
cannot but give a preference to his second statement, 

‘ Quidam a septem stellis,’ for tara is the Sanskrit for 
a star, and indeed is still preserved in several of 
the vernacular languages of India. For the Latin I 
would assume a form ter-iones with that masculine 
diminutival suffix ion, which is well known in 
matell-ion-, senec-ion-. An e rather than an a is sup- 
ported by the familiar Stella, i.e. ster-ula, or rather 
ster-el-a. 

I have thus dealt with castra, plaustra, furcae, 
fordpes, forpices, furjioes, uolsellae, bigae, quadrigae, 
casses, folles, literae, limina, currus, septentriones ; 
and these examples are sufficient to establish the prin- 
ciple that, when an object consists of two or more 
like parts, a word, in itself denoting one of these 
parts, is first employed as a plural to denote the com- 
pound, but eventually is supplanted by the singular, 
which then also denotes the compound. 

AVith this premised, I call attention to the use of a 
plural menta in the .^Eneid : Nosco crines incanaque 


Digitized by Google 



WITH A SINODLAH MEANING. 


197 


menta Regis Romani primdm qui hgibus urbem fun- 
dabit, &c. (.<En. vi. 810). Now the ordinary meaning 
of mentum, ‘ a chin/ will not avail here, for we need 
not stop at the English phrase ‘ a double chin.' My 
own conviction is that the first meaning of mentum is 
‘ a jaw,’ and thus the plural menta would denote ‘ both 
the jaws,’ that is ‘ the. mouth/ or rather in the present 
passage those parts on which the beard grows, both 
above and below the opening expressed by the word 
mouth. 

How readily words of the same stock are employed 
to denote ‘ the jaw/ whether upper or lower, the mouth 
made up of both jaws, the chin, the beard, the cheeks, 
the gums, is well seen in those which begin with the 
syllable yet/ or gen. Tims in Greek we have (1) yei/t^ 
sb. f. to which our lexicons assign the meaning of 
‘ under-jaw,’ and in the pL ‘ both jaws, the mouth with 
the teeth;’ (2) yeveiov ‘strictly the upper jaw, but 
usually the part covered by the beard, the chin, and 
later the jaw, the cheek ;’ (3) yei/etoS- sb. f. ‘ a beard/ 
and in pi. the ‘ cheeks ;’ (4) yvado-. (= yav-affo-) sb. f. 
‘the jaw, mouth;’ but strictly the ‘lower jaw;’ (5) 
yvaOfio- sb. m. ‘the jaw;' (6) Lat. gena- ‘cheek;’ 

(7) dens genuinus ‘ a cheek-tooth or double tooth ;’ 

(8) gingiva ‘ the gums ;’ (9) Welsh g^n ‘ mouth, jaw, 
and chin;’ (10) Fr. gran-oc/ze ‘ lower jaw.’ Next, with 
the changes of consonant to be expected, (11) Sanskrit 
kanu, ‘the jaw;’ (12) Gothic hinnu, ‘chin;’ (13) Old 
Germ, kinni ; (14) Eng. chin; (15) Lith. zanda-s 
‘ jaw ; ’ and then, with a labial in place of the n, 

(16) Germ, gaumen and Eng. gums, together with 

(17) Sanskr. jamba ‘the chin;’ (18) yan^Xat ‘jaws 
of lion/ &c. ‘ beak of bird.’ 


Digitized by Google 



198 


ON PLURAL FORMS IN LATIN 


But, as has long been pointed out, especially by 
Buttmann in his Lexilogus when treating of /tcXat 
and Kekatvot, a guttural often slips into a labial, pro-- 
bably by a passage first into gu or gw, then into 
w, and then into m. We must therefore connect with 
the preceding family of words (19) the Latin mentum, 

(20) the French menton of the like meaning, and also 

(21) Germ, mund, (22) English mouth, (23) the Gr. 
fivoTUK- sb. m. ‘ upper lip or moustache,’ as also 
(24) ftaffraic-, sb. f. ‘mouth, beak, upper lip.’ And 
this with the more confidence when we find (25) a 
Welsh mant ‘ a mandible,’ whence is-fant ‘ lower jaw,’ 
and gor-fant ‘ upper jaw.’ 

But it behoves the philologer never to be satisfied 
until he come to a verb as that from which the other 
parts of speech are deduced. Now the one among the 
many meanings attached to yews, montum, &c. which 
most reatlily connects itself with action is ‘ the jaw,’ 
as the instrument of mastication ; and the Latin verbs 
mand-ere and mand-tuxi-re at once present themselves 
with the desired meaning and a suitable form, for as 
the Latin scdla stands to the verb scanef-ere, precisely 
so mala to mand-ere. Indeed the combination nd 
and I are frequently convertible, and the Latin seems 
generally to have a predilection for the liquid I, Thus 
the verb mdl-ere ‘ to grind,’ has long been held to be 
of the same stock with mand-ere. Many too of the 
allied languages exhibit the form with an I, as Greek 
/ivXij and Latin mola ‘ mill,’ with pi. /n/Xa* as ‘ the 
grinders,’ or to use the Latin phrase the dentes molare-s. 
So we have Lith. verbs, mal-u, mal-in-u and mald-in-u, 
Russ, melju, Irish meil-im, Goth, mal-an. Germ, mahl- 
en. (See Bopp’s Glossarium Sanscritum, v. mrid.) 


Digitized by Google 



WITH A SlNHULAll MKANINU. 


19» 


But Buttmaim in his Lexilogus (§ 48, on ovXai, p. 198) 
justly observes that stamping or pounding was a pro- 
cess older than grinding, and so he finds the earlier 
sense in the Latin sb. mall-eo- m., and the Latin verb 
mulcorre. To these we must add the Sanskrit mridd 
or mardd ‘ dust,’ the Gothic malma ‘ sand,’ and mulda 
‘ dust,’ as also our own mould, whether applied to earth 
well broken up or to brown sugar, and the German 
malm-en, zermalm-en ‘to crush.’ 

The Greek verb /ta\-o<r<r-<a too is said to have sig- 
nified originally to beat and so make soft, as in 
dressing leather, or, we might add, in making a beef- 
steak tender. 

The same scholar treats the Greek verb a\e-a> and 
the nouns ovXat and aXevpov as of the same stock. 
And if this be just, we must add to the family the 
nouns oi;\a n. pL ‘ gums,’ and oXfio- m. ‘ a mortar, 
kneading-trough, the hoUow of a double tooth,’ &c. 
The doctrine that crushing preceded grinding is con- 
firmed by what we see in the familiar verb mordre- 
‘bite,’ of the Latin compared with the Sansk. mrid 
or mard, which is translated ‘ conter-ere.' Hence 
too we see that the Latin mort-ario- n. has been 
justly claimed as belonging to the family. Further, 
the Prakrit has mal in the place of the Sanskrit 
mard. 

Hence mordere and molere are of the same stock ; 
while as to form, we have a precise parallel to the 
connexion between ardere and alere, for the notion of 
elevation is the original idea in both these words, os 
also in the adjective arduus (see p. 173) ; and if ardere 
happens to be limited to the action of fiame, so also 
alere jlammam is a familiar combination. But I am 


Digitized by Google 



200 


ON PLUEAL FORMS IN LATIN 


here reminded of a doctrine which I put forward some 
thirteen years ago as to the etymology of the verb 
obsolesco. To the arguments I then adduced I have 
now to add other evidence, and as the matter is of 
some importance I will put together all that I have to 
say upon the subject. 

Those who would connect the verb with aholere and 
exolescere fail to explain the appearance of the s, for 
ohs in place of ob can only be defended when the 
simple verb begins with a thin consonant, p, c, or t, a 
principle which extends to the two other prepositions 
which end in 6, and sub. Accordingly, I at one 
time was tempted to connect the root of obsolesco 
with the sol oi solium, ‘a seat/ consulere (old form 
consol- or cosol-) ‘ to sit together (in deliberation), ’ so 
that obsolesco should contain in itself the same meta- 
phor as our verb ‘supersede.’ But I have now no 
doubt that the idea of dirt belongs to obsolesco, and 
that it is of the same stock with sordes. I am led to 
this conclusion, first, by the habitual union of the two 
words, as ‘Ut eum, cuius opera ipse multos annos 
esset in sordibus, paulo tamen obsoletius uestitum 
uideret’ (Cic. Verr. ii. 1, 58) ; ‘ Splendetque (uirtus), 
per sese semper neque alienis umqRam sordibus obsol- 
escit’ (Cic. Se&t 60); ‘In homine turpissimo obsole- 
fiebant dignitatis insignia’ (Cic. Phil. ii. 105) ; then in 
Hor. (Epod. xvii. 46), * 0 nec paternis obsoleta sordi- 
bus;’ and (Od. ii. 10, 5), ‘Obsoleti sordibus tecti;’ 
and in Val. M. (iii. 5, 1), ‘Candida toga turpitudinis 
maculis obsoleta.’ 

In two of the passages just given, the notion of 
defilement is supported by the appearance of the 
adjective turpis, or the abstract noun derived from it; 


Digitized by Google 



WITH A SINGULAR MEANING. 


201 


and these had nd doubt for their physical and original 
sense that of dirt, whence the idea of moral poUutidft 
readily flows. But the word obsoletus did not need 
the aid of sordibus for the expression of this idea, for 
we find uestis obsoleta in Liv. xxviL 34 ; uestitus obio^ 
letior in Cic. in Bull. ii. 5 ; obsoletus Thessalonicam 
uenisti in Cic. in Pis. 36 ; and uestitus obsoletus in 
Pseudo-Nep. Ages. 8. Again, the 'word is opposed to 
enituit in the younger Pliny (Pan. iv.) ; and in one of the 
tragedies which bear the name of Seneca (Agam. 9 “6), 
there occurs the marked phrase sanguine obsoletus. 

But the matter seems placed beyond all doubt when 
we come across a verb obsordesco, uniting in itself the 
same two distinct though connected meanings which 
belong to obsolesco, ‘ to pass oUt of use,’ and ‘ to be- 
come dirty on the surface ; ’ and this on the authority 
both of an old and of a late writer, so as to prove the 
great length of life which obsordesco enjoyed, vi«. 
* Obsdrduit him haec in me aerumna miseria ’ (Caecil. 
ap. Non. vii. 603, who himself translates the word by 
obsolescere), and ‘Ne coma fusa umeris fumo obsor- 
descat amaro’ (Prud. Apoth. 214). 

Of course the common result of non-use is the col- 
lection of dust, riist, mildew, mould, and aU those 
undesirable objects, which the Romans included under 
their term sittis, a word which etymologically ought 
only to mean ‘ putting down and leaving alone,’ as it 
of course comes from the verb sinere. 

Several of our modem languages exhibit the same 
root with the same meaning, as with ourselves in soil 
the verb, soil the sb., especially in the form night-soil, 
also in sullage, s{o)lush, sully; but in the last we have 
a word of French origin, representing souiller, by the 


Digitized by Google 



202 


ON PLURAL FORMS IN LATIN 


side of which exists the adj. sale. Perhaps after all 
we come back to the obsolete verb sol-&re (to sit down), 
whence consolere, for the natural place for sitting in 
the earliest states of society is the ground, solum; 
and this also is the leading source of what we call 
dirt. Even the notion of solere ‘ to be wont,’ results 
from that of permanence, which resides in the posture 
of sitting, as opposed to locomotion. Hence indeed 
suesco and consuetudo are probably akin to sol in the 
sense of ‘ sit ’ (cf. the Germ, sitte) ; and we have a 
parallel in the Latin mos morns, which is of the same 
stock with mora ‘ delay,' and so with the Greek /lovy 
and the verbs and maneo, as also with our 

own manner and Fr. manoir, &c. 

But to return to the word mentum, I find a little 
difficulty in three words, which by meaning and partly 
by form seem to claim connexion with the family of 
mol or mal ‘ crush, grind,’ &c. viz. : ftaa-a^o/uu ‘ chew,’ 
yaao-v-w ‘ knead’ (with its derivatives fuvy^ut), and 
maxilla ‘the jaw.’ Thus while mala serves beyond all 
dispute to connect mand-ere on the one side with • 
maxilla on the other; rrwjxiUa and fia<r<r-io imply a form 
nay rather than n^K. Yet X and y seem to be sounds 
utterly inconvertible, imless indeed we may say that 
the y sound forms an intermediate link between them. 
Such was my contention in a late paper read before 
the Philological Society, which compared Bvyar-ep- and 
Jilia- ; and the argument derives strength from the 
parallelism seen in the Greek /toyif and /xoXw. 

At any rate, the Latin nouns, which having a long 
vowel before an I, form diminutives in xillo (or sillo), 
seem to owe the long vowel of the simple noun and 
the X of the diminutive to an original guttural in 


Digitized by Google 



WITH A SINGULAR MEANING. 


203 


the syllable which precedes the 1. Thus aUt for ahala, 
paulo- for paucuh-, tdlo- beside a<rrpaya\o-, pdlo- be- 
side pango, and tela beside tex~ere, seem all to claim a 
lost guttural, which would accoimt for the forms axilla, 
pauxillum, taxillus, paxillus. The loss of the guttural 
would be exactly parallel to what we see in our own 
words, nail, hail, rail, sail, vxtin {waggon), rain, beside 
the German nagel, hagel, regel, segel, wagen, regen ; 
and indeed the Latin velum, whence uexillum, may 
have grown out of an older suegelum, and so be one 
with the German segel. See also the remarks on the 
Greek noun trapaaeiov ‘ upper sail,’ in the paper on the 
German prefix ver. 


Digitized by Google 



204 


X. 

EXCRESCENT CONSONANTS. 

Although generally averse to the introduction of new 
grammatical terms, I have thought it desirable to ask 
admission for one on the present occasion, because the 
ordinary term ‘ epenthesis’ seems to have been formed 
upon a false theory, and so to have misled, as it ap- 
pears to me, not a few philologers ; and among these 
several who hold a place in the front rank. I especially 
refer to the three German scholars, Grimm Bopp and 
Diez. Thus the words ‘ einschicbung,’ ‘ eingeschoben,’ 
‘ einschaltung,’ are with them in constant requisition ; 
and in my mind this assumption of an ‘ inshoving’ 
always raises a presumption that some error lies con- 
cealed beneath them. For example, in speaking of 
certain diminutives (iii. 668), Grimm has to deal with 
a syllable in, which, not seen in the nominative, ap- 
pears in the oblique cases, and so he is led to regard 
the n as intrusive, viz. in prentili * a small brand,’ g. 
prentilin-es, d, prentilin-e. See. ; where however it 
seems more reasonable to suppose that the nom. has 
lost an n, as is admittedly the case with the Latin 
ordo ordin-is, ratio ration-is, caro cam-is. The same 
doctrine is repeated by him twice in p. 672 and again 
in p. 678. 

In my paper entitled ‘ Quaeritur’ (see below), I 
refer to Bopp’s dealings with the Sanskrit genitives 


Digitized by Google 



EXCRESCENT CONSONANTS. 


205 


plural, dsvd-n-dm, tri-n-dm, sAnd-n-dm, the n of which 
he regards as euphonic, while it appears to me to be 
the genitival suffix, as in our own Frie,r-n Bamet, 
contrasted with Abbot’s Langley and King’s Langley, 
as also in Buck-en-ham (Norfolk), and, what is sub- 
stantially the same, the county town Buck-ing-ham, 
which originally was nothing more than ‘ Mr. Bucks 
home or house,’ for the largest town had its begin- 
ning, and this often in the residence of a single family. 
Again in Weinhold’s Alemannische Gr. (Berlin, 1853 ), 
I find (§ § 409 , 411 ) that the nouns fater, Karl, Hein- 
rich had two forms of the genitive, /ateres fateren, 
Karles Karlen, Heinrtches Heinrtchen, &c. Besides, 
if the n of dsvd-n-dm &c. be not a genitival suffix, there 
is nothing whatever to represent the idea of genitivity 
(excuse the word), since am, like the corresponding top 
of Greek nouns, is a mere symbol of plurality. 

Diez too (Gr. ii. 201) assigns to the old French 
perfect of dire a form d6imes (= disdmus) or ‘mit 
eingeschobeuem s, diismes.’ But in my paper on the 
Latin perfect (Plulolog. Trans. 1860 - 1 , p. 185 ), I was 
led to a very difierent view, viz. that deismes is the 
more genuine form, seeing that the Latin diximus 
itself grew out of a fuUer dix-ismus, corresponding to 
dix-istis. 

Again it was probably an impression that a conso- 
nant was required to prevent hiatus which led the 
French to sanction the division aime-t-il, as though 
the t were a foreign element ; but of course we have 
here what represents the Latin amat iUe. In il est 
the t is silent ; but in the inverted form est-il ? coming 
before a vowel, it is pronounced ; and the same applies 
to aimet-il, as it should have been written. 


Digitized by Google 



20fi 


EXCRESCENT CONSONANTS. 


Those who would insert consonants ‘ hiatus vitandi 
caussa,’ never stop to explain to us why one consonant 
rather than another is selected for this ignoble office. 
But in truth it may be doubted whether any real in- 
stance can be found, unless we are to accept such as 
‘ Maria Ranne ’ or ‘ the Law ran the Prophets’ of 
liOndon speech. At any rate in a large majority of 
the instances usually adduced it will be found that the 
so-called epenthetic consonant is no foreign matter, but 
either an original part of the word, or else a simple 
outgrowth from the consonant immediately preceding. 

In a paper by Mr. Weymouth (Philolog. Trans. 
1856, p. 21), and in Bindseil’s valuable, even though 
unfinished work, ‘ Abhandlungen zur Allgem. verg. 
Sprachlehre’ (Hamburg, 1838), the true theory, as it 
seems to me, is given as regards the difference in the 
position of the organs of speech for the production on 
the one hand of the nasals m, n, ng, and on the other 
of the mutes b, d, g. As those writers point out, it 
depends solely on the position of the velum palati 
whether the one set of sounds or the other is heard. 
When the velum is so placed as to leave a free passage 
■for the air through the nose, we have the nasal ; but the 
moment this passage is closed, the sormd passes at once 
to the allied mute ; so that what began as ah m may 
end as h, what began as ng may end as g {goose, hag), 
what began as n may end as d. Such secondary con- 
sonants then must be regarded as natural outgrowths, 
or, to use my new term, as excrescent, rather than 
intrusive, as intrinsic, not extrinsic. 

But it is not with the nasals alone that there is 
this tendency to pass from one consonant to another. 
Whenever the organ.s of si>cech which are employed 


Digitized by Google 


EXCRESCEKT CONSONANTS. 


207 


in the production of two consonants lie near one 
another, a passage from one to the other is apt to 
occur. But it is especially from the dental series that 
excrescent consonants proceed ; and this was perhaps 
to be expected, as this class of consonants occupy a 
middle place, and so have ah afiBnity for the labials on 
the one side, and gutturals on the other. Precisely as, 
when we throw great force into the sound of an n at 
the close of a syllable, — for instance, to take a vulgar 
example, but not the less valuable on that account, in 
pronouncing the words govm or drovm-ed, — there is a 
strong tendency to produce what would be written as 
govmd or droumd-ed ; so if we lay a stress upon an s 
there naturally results a following t, and hence a 
Roman intending to say pos found that he unin- 
tentionally uttered post. 

I propose then to take into successive consideration 
all the following combinations, in which for convenience 
the alphabetical order is preferred : ht (cht). It, 

nt, pt, rt, St : bd, gd, Id, nd, rd ; \0, v6, p0, <t 0, <f>0, 
cs, gs, ns ; Iz, rz ; pf; nib, mp, ng, m. In treating 
these combinations I must be brief; but throughout 
my view is that the second of the two consonants is 
excrescent. 

1. Ct: plecto, cf. 7r\e#c-a», pUco, simplex; nect-o, 
cf. necesse, necessarixis, &c., avayKr/ ; Jlect-o ; pect-o ; 
yaXa for yoKax, but gen. yoKaxT-os, with Lat. loc, also 
lacte as a nom., g. lact-is ; cf. Ang.-Sax. meolroc and 
our milk; ucr-epos, with iec-ur and r^tr-ap; xreiv-m, with 
xaiv-m, txavov, xexova ; xraopai xeKnipai beside ’iraopMt 
•7reirap,ai, for the consonants x and tt are here inter- 
changed as in xorepof, Trorepof, &C. ; iwxr-os, noct-lS 
compared with vvxa, vvxtot (see p. 69). 


Digitized by Google 



208 


EXCRESCENT CONSONANTS. 


The Latin nouns in etitm had an older form in ectum, 
as shown clearly in the case of uirechim, dumectum, 
aft. uireticm, dumetum. In these, however, um alone, 
in my belief, is the sufiBx, so that carect-um, frutect- 
um, salict-um, See. come from car-ec-, frut-ec-, salic- 
(n. carex, frutex, salix), throwing out at the same time 
an excrescent t. 

Indeed generally this passage from a c to ct is apt 
to be followed by the loss of the guttural, so that the 
c seems itself to have been transformed to a t; but 
whenever the preceding vowel is found to be long, it 
would probably be safer to assume the existence of an 
intermediate form with ct. Thus diutius by the side 
of diu leads mo to suspect that the base of the latter 
word was diuc- ; and, indeed, I have elsewhere (Tr. 
Phdolog. Soc. 1856, p. 320) given my reasons for 
believing that all nouns of the fourth declension once 
ended in uc, so that genuc-ulum is a regular dimi- 
nutive from genu{c-), and the adj. metuc-ulosus duly 
formed from metu{c-). Hence I hold diutius to stand 
in place of diuct-ius, which exhibits an excrescent t. 

Otium again, for this form (not ocium) has the ex- 
(dusiye support of inscriptions, I am strongly inclined 
to regard as a shortened form of an obsolete oct-ium, 
a^d .this again as a decapitated variety of uoct-ium, 
from the root of the verb xtde-are ‘ to be empty,' and 
the adj. uoc-iuo~ ‘ empty,’ — two words which, following 
the best authorities (Bergk, Zeitschrift f. Alterthums- 
W. 1848; Mommsen, Corp. Inscr. i. p. 70, b. ad Jin. 
&c.), I am bound to write with an o, not an a. As 
regards meaning, such an etymology has all in its 
favour ; and I cannot but prefer it to Prof. Aufrecht’s 
suggestion (put forward, however, with much hesi- 


Digitized by Google 



EXCRESCENT CONSONANTS. 209 

tation), that it may come from auere * to be happy/ 
an idea substantially the same as Corssen’s (Kritische 
Beitrage, p. 17), who would connect it with the 
Sanskrit root av of various meanings, * iuuare, tueri, 
ualere, gaudere,’ &c. out of which he seems to give a 
preference to ‘tueri,* so that his autium would bo 
directly opposed to helium, peace to war. 

The long e of sUius by the side of s&ctis and 
si&quius also receives its due explanation,, if it be 
considered as standing for sectius with an excrescent t. 
Much has been written on the origin of these words, 
as by Corssen (Beitrage, p. 5), Fleckeisen (Rhein. Mus; 
viiL 227), Schweizer (Euhn’s Zeitschrift, viii. 303). My 
own view is that we must start from a theoretic adj, 
sequis ‘ second ’ from the verb sequL This word I find 
entering into the formation of sesquis ' one and a half,' 
Germ, anderthalb, which we cannot but treat as a 
compression of semi-sequis, when we place it by the 
side of sestertius, i.e. semis-tertius ‘two and a half/ 
Germ, drittehalb. But the words alter and Bevrepof, 
the ordinary words signifying ‘second,’ are in form 
comparatives. Hence a neuter comparative sequins 
has nothing in it to offend ; and secvLS itself I also 
hold to be a variety of the same word, having lost its 
i, precisely as minor and minus have (see p. 122). On 
the other hand, the meaning of ‘ other or otherwise ’ 
accords well with the uses of the words in question, 
as in the common phrase hand secus, and such pas- 
sages as : — ' 

‘H&ec nihilo 'see mfhi oidintoi s^tius quam sdmnia’ (Flaut. 

Men. V. 7, 67) ; 

‘ Quid s^quinst aut quid interest dare te in manus . . . t’ (Trin. 

130); 

P 


Digitized by Google 



210 


EXCKE3CENT CONSONANTS. 


where I readily accept the simple emendation of 
sequiust for secutus est of the MSS. A very different 
but I think not very satisfactory etymology of setius 
is given by Corssen (Beitrage, p. H). 

As to the derivation of nitor nixus, which Corssen 
deals with in the same work (p. 20), but I think not 
very happily, all is smooth, if, following the guidance 
of the old Latin grammarians themselves, we start 
from the sb. genu- ‘ a knee,’ or rather from its older 
form genuc- as heard in the Greek yw^, the Latin 
genuc-ulum, afterwards genic-tdum, corresponding to 
our own hnuck-le, Germ, hiock-el-n, to say nothing of 
the Germ, knicks ‘a courtesy.’ Thus the original 
meaning of niti is ‘ to kneel,’ and its secondary sense 
of ‘straining' or ‘striving’ arises from the idea of 
employing the knee as a fulcrum in many muscular 
actions, as for instance in cording a trunk. The old 
form gnitor, given by the grammarians, and the par- 
ticiple nixus, together assure us that nitor has lost two 
gutturals, and must have been corrupted from a fuller 
gnictor, where we have almost the same form as in 
genic-ulum, ^d what strictly corresponds to the 
Germ, knick-s, excepting indeed as to the t, which I 
claim ^ an ordinary outgrowth. Forcellini for once 
seems to have gone wrong, and to have missed the 
original meaning of the verb niti, so that it may be 
worth while to give evidence on the subject, which 
frotunately is easy to find. Thus Festus has, ‘ Nixi 
di appellantur tria signa in Capitolio ante cellam 
Mineruae genibus nixa.’ So we have in Plautus 
(Rud. iii. 3, 33), ‘Ambae te obsecramus genibus nixae;’ 
in Virgil (xii. 303), ‘ Impressoque genu nitens terrae ;’ 
in Livy (xxvi.9), ‘Matronae nixae genibus orantesque;’ 


Digitized by Google 



EXCRESCENT CONSONANTS. 


211 


in Ovid (Her. xxL 100), ‘ Et de qua pariens arbore nixa 
deast;' in Seneca (Thyest 60), ‘Quern genu nixae 
tremuere gentes Germ. (Arat. 67), ‘ Dextro namque 
genu nixus diuersaque tendens Bracchia.! And if this 
be not enough we have the fact that the constellation 
called by the Greeks Evyovavtp has for its Roman 
name indiJSerently Ingmiculus and a simple Nixus. 
Thus Cicero (Arat. 373) has, ‘Flexo confidens corpora 
Nixus* and again (N. D. ii 42), * Engonasin uocitant 
genibus quia nixa feratur;’ while Ovid (Met. viii. 
182) writeSj ‘Qui medius nmque genu est anguem- 
que tenentisj’ and Manilius (v. 645), ‘Nixa genu 
species et Graio nomine dicta Engonasi ingenicla 
inuenis sub imagine constans.’ 

In closing this lung section on the combination ct, 
I deem it but right to add that Corssen asserts that 
an original c never disappears before a t, while ho 
admits that when a c has superseded an earlier g, as 
in autor autumnus from augeo, it is not so protected ; 
but a doctrine in itself so arbitrary has found little 
support with other scholars 

2. Ft : see paper on post and after (p. 121). To 
the examples there given add tufi by Fr. touffe, Scotch 
tuff; -schaft, the s uffix of German nouns, freund-scAa/15, 
fiend-scAct/'t, contrasted with our friend-s^i}?, &c. ; 
hxught-er fix>m laugh. 

3. Ht (cht, ght) ; Ang.-Sax. liht, Germ, licht, our 
light, compared with Lat. luc- ; so miht, macht, might 
with the Ang.-Sax. verb mag-an; niht, nacht, night 
(see noct- above) ; Germ, specht by the side of Lat. 
pico- and our wood-peckeri acht-en and veracht-en, 
the analogues of ovr-tix6ai, vjrep-OTrr-iadai ; fechten, 
fight, compared with pug of the Latin pugnus 

p 2 


Digitized by Google 



212 


EXCRESCENT CONSONANTS. 


pugna pugil and Trwf; geldcht-er by the side of 
lach-en lache; and the whole classes of German 
adjectives in ig and icht, as heinig and beinicht.. 

4. Lt : fie\T-tov- ^eXT-iaro-, compared with hello-,, 
mel-ior, and our well; alt-ero by the side of alio- 
ali-qui- ; and iilt-ra, ult-ro, ult-erior, ult-imo-, as 
contrasted with sup-ra, por-ro, sup-erior, min-imo-‘, 
but forms connected with comparatives and super- 
latives will be discussed more fully in the following 
paper. Add salt with Lat. sal; Germ, falte com- 
pared with pal of the Latin jmhna, palam, pl-ic-a, &c. 

5. Nt : as /Sevr-toro- by bono- and bene ; \eovr-of by 
leon-is ; S. ant-ara ‘ one of two’ by Norse ann-ar, our 
one, Sc. ane, and the so-called article an ; Lat. int-ra, 
int-ro, int-erior, int-imo-, int-er, int-xis, from in; 
cont-ra, cont-ro, from con, but see following paper; 
Germ, ent-zwei, ent-gegen, where ent stands for the 
prep, ein or rather en — Lat. in ; ent-weder ‘ either,’ 
formerly ein-weder, where ent stands for the numeral 
ein ‘ one ;’ ent the prefix, Lith. ant, the analogue of 
the Greek ava; the Germ, eigent-lich, nament-lich,, 
hescheident-lich, &c. ; Fr. loint-ain, from loin, cf. pro- 
chain, from proche ; Lat. tegument-um, cognoment- 
vm, &c. from tegumen, cognomen, &c. The t also of 
imperfect particles I hold to be an outgrowth from the 
preceding n, so that in the crude form scribent-i- 
scrihen is an old substantive like the Lat. unguen 
(whence unguent-um), and still more like the German 
so-called infinitive schreiben, while the final i is pro- 
bably a remnant of the preposition, or rather in this 
case the postposition, in. We ourselves, by the way, 
once wrote i for in, as in the Shaksperian phrase, ‘ By 
the second hour i’ the mom’ {Antony ami Cleopatra, 'w. 


Digitized by Google 



EXCRESCENT CONSONANTS. 


213 


7, 4). In this way the Latin participle will be brought 
into agreement with the Keltic forms, as the Welsh 
yn myned ‘ iens,’ more literally ‘ in itione,' and with 
our old phrase a-going for an going, of precisely the 
same power. To these I add from Mr. Weymouth’s 
paper tyrant. hy rvp-awo-, and ancient; the latter in 
both its senses, as an adj. = Fr. ancien, and as a sb. 
= our ensign. 

6. Pt : 9TToX«/*o-, TTToXi-, TTTveXo-, iTTepva-, compared 

with iroXtfio-, irve\o-, ‘/rtpvar-, and Lat. pema- ; 

iTTv-a compared with Lat. spu-o, Eng. spit ; twttt-w, 
piirr-w, 6airr-o}, xnrr-tos, Lat. suht-us, subt-er ; but as to 
these see next paper. 

7. Rt: heart by Lat. cor, and seap ; uespert-ino~ 
from uesper- ; fert-ili- from fer-; mort-i- from mor- 
‘ die,’ as well as sort-i-, part-i-, art-i- (but see next 
paper) ; Germ, juchert or juchart, representing Lat. 
iuger-. In such words as braggart, I have long been 
inclined to think that the t is excrescent, and that the 
syllable ar is a diminutival suffix, one with the er 
of fresh-er ‘a little frog,’ and of hamm-cr, dagg-er^ 
Jing-er, and the Germ, mess-er. The Latin in the 
same way was in the habit of forming contemptuous 
terms for men, by means of a diminutival suffix, as 
toc-ulioTtr, and the Greek too, as v\ovr-aK-, (f>ev-aK-, 
\a\-ay-. If such explanation be just, it must apply 
also to such words as slugg-ard, cow-ard, &c. 

8. St: crast-ino- from eras; prist-ino~ from prist 
i.e. prius ; rust-ico- from rus ; Ligust-ico- from Ligv^ ; 
Libyst-ico-, Libyst-ino-, and Libyst-id, from Libys ; 
o<rr-€ov by the side of os ossis ; ost-ium from os oris. 
Here let me add that the Greek «rropar- is but a deca- 
pitated ovr-oprar-, and so of the same stock with the 


Digitized by Google 


214 


EXCRESCENT CONSONANTS. 


Latin os oris. In ov-o/x-ar- the Greek has preserved 
the root-vowel o, which is lost in the Latin nomen 
and Germ, namen. Again an excrescent t is seen in 
the Latin post, as also in post-ero~, post-umo-, in ext- 
ero-, ext-erior-, ext-umo-, ext-ra-, magist-ero-, minist- 
ero-, dext-ero-, sinist-ero-, dext-imo, &c. (see the next 
paper) ; in ust-ula from ur- (tw-) ‘ bum and probably 
in agrest-i-, caelest-i-, terrest-ri, sihiest-ri-, domest-ico-, 
modest-o-, for in these words I am inclined to believe 
we have derivatives from lost words with a neuter 
suffix in es. Thus modesto- and the verb modera-ri 
seem to point to a noun modus moderis as once co- 
existing with the noun modus modi, just as glomus 
~eris stood by the side of globus-i, and so led to the 
formation of glomerare. Precisely in the same way 
Pott (E. F. i. 235, note, ed. 1859) contends, with 
reason, that /lepoeuci^s, which is commonly derived from 
the neuter noun /xhot, implies a form fxivof of the 
second declension ; and he treats in the same way 
elSoTTotoy and reixoiroiof. So too one must, I think, 
assume a masculine aino-9 to explain such forma as 
al/xofia<l>ris and aiixo-w, rather than, as is usually done, 
refer them to the neuter alfiar-. 

Add to the preceding list first arhust-um from arhos. 
Tffie common doctrine that this is an abbreviation from 
arboretum is clearly an error, for arboretum itself is 
for arbor- ect-um, and so contains the diminutival 
sufidx ec, of which there is no trace in arbust-um. 
It is under the same wrong view that some hold 
frutectum to stand for fruticetum, salictum for salic- 
etum. Such a doctrine would lead us into an endless 
series ; for if salictum is for salicetum, then, as salic- 
etum must have grown out of a form salicectum, we 


Digitized by Google 



EXCRESCENT CONSONANTS. 


215 


must again assume a fuller saUc-icetum, and then a 
ealiodcectum, and so ad infinitum. 

The names of female agents, tonstrix^ defemtrix, 
persuastrix, and the noun tonstrina, come from mas- 
culine nouns in or, tonsor, defensor, and persuasor, 
though the last is no longer to be found. The dis- 
appearance of the long o of tonsor -oris might have 
been a difficulty^ had we not the undoubted case of 
doctr-ina from doctor. Oiir sister. Germ, schwest-er, 
has the same suffix, er, as pater, mater, frdter, and 
that probably a diminutival suffix of affection ; while 
or has obtained a preference in sor-or (for sos-or) and 
ttx-or, solely through the influence of the vowel in the 
preceding syllable, o and u. The pronoim ist-o- I 
would divide so as to leave o alone to the suffix, as in 
ill-0- and e-o- {eum, earn, &c.) ; but my reasons I must 
reserve for a more convenient occasion, as the argu- 
ment would run to a great length. Vest-i- comes 
from a root uen or ues, as seen better in the Greek 
pev-vvfit, ; but t alone belongs to the 

suffix, as also in part-i-, where it is a corruption of 
ic, as seen in part-ic-ulc^. 

Cust-od- I also claim as one belonging to this class, 
giving to it for its original meaning ‘door-keeper.’ 
The first syllable I believe to be an earlier and truer 
' form of os {oris), except that, like ost-ium, it has 
thrown out a t. The disappearance of the initial 
guttural is what is already familiar in uhi, unde, uter, 
from cvhi, cunde, cuter. The second element dd I 
compare with the corresponding part of 0vp-wp-o- irv\- 
top-o-, which . Buttmann deduces from opa-m, and I 
f.bink with reason, so as to make them signify ‘ door- 
watcher, or door-warden.’ But a p is at times repre- 


Digitized by Google 



216 


EXCRESCENT CONSONANTS, 


seated by a Latin d, as in caditceo- by the side of the 
Doric xapvKeto-. Even opa-a itself is of the same 
stock ■with the Latin uide-o, as proved by its aorist 
FetSov, FtSetv. Nay, I must also claim as one with 
the root of opam, and so -with that of eiZov, uideo, 
pur own vxire, wary, and with an excrescent d, ward, 
VKird-en, and the French guard-er, &c. together 
with our own regard ‘ look back.’ The etymology of 
custod- here proposed corresponds to that of a^di- 
tumus or aedituus, from tu-eor, or, as I am inclined 
to assume, an older tumreor. 

In French we have also instances that belong to this 
class, in estre, now Mre, by the side of the Itah easere, 
in naitre for nascere, &c. ; as well as in the old French 
perfect distrent, by the side of disrent, now dirent. 
Possibly we owe to this principle the personal ending 
of our verbs, as lovest, where an s alone seems justified 
by the older branches of the Indo-European family. 
So in German we find morast, palast, axt, einst (see 
‘ German for the English,’ by Sonnenschein, &c.), as 
also obst supplanting an older opaz. 

9. Bd : fio\v^S-o-, which in this respect stands half- 
way between the Latin plumbo- and our o'wn lead. 
The Latin uerher- I believe to be a compression of an 
older wr-eh-ery, and again the Greek pa^Bo- to have 
pome by decapitation from a form Fap-afiB-o,^ so that 
the root-syllables are virtually the same, and the sufiSx 
which immediately follows it. 'Po^-e-a (for <rop-o<f>-e-w) 
is of course one with the Latin sorheo (for sor-ob-eo) ; 

^ Tliose wlio are alarmed at this theory of decapitation are in- 
vited to compare the Greek pal (pay-) ■with the Latin /(a)rc^-um 
‘ a (straw-) berry and indeed with our own herry, in which two 
weak vowels, « i, have supers^ided the strong vowels, a a. 


Digitized by Google 



EXCHESCENT CONSONANTS. 


217 


but in Greek by the side of the compound avappo^aa 
■we find also avappot^B-ee>. 'EySSo/io- when contrasted 
with the Germ, siebente seems to have an excrescent S. 
The Latin verb ped- might well appear in Greek as 
fieX- or 0€v- ; but we find ^Bev-vvfu, fiBeX-vtrau, and an 
adj. fiBeX-vpo- (see Rd below). The Fr, coude was 
written in the sixteenth century as couhde, and so 
establishes its identity with the Latin cubito-. 

10. Gd : pi/yBa by the side of f*iya ; and it seems 
likely that Tftp,r)Tqp passed through TBri/jttjrtip on its 
way to ^ijfirjTtjp. 

11. Ld : here Diez (i. 194) supplies most of the 
following examples : Span, valdri; Prov.^oWre (=old 
Fr. fouldre); toldre=tollere ; Old Fr. tnouldre, re- 
souldre, pouldre, now movdre, resoudre, povdre. The 
Xatin corulus passed first into coluims (r being sup- 
planted by an I, and in the following syllable the 
converse) ; then colunis produced couldre, which is 
now coudre, a word which is therefore the genuine 
analogue of oxa hazel; Germ. baldrian'=ualeriana: 
Du. helder= Germ, heller. Add to these our own 
aW-er = Lat. ah^us, where the English r represents 
the Latin n, just as is the case with order (Fr. ordre) 
compared with ordon-. So our old alder-Jirst and 
ftller-Jirst correspond to such German compounds as 
aller-beste. Lastly, dXB of aXB-aivu>, aXB-fiaxm some 
scholars identify with the Latin al-o. 

12. Nd : avB-pos for avtp-os ; evB-ov and evB-o» by 
the side of ei> ; in Latin mand-are, pre-hend-ere, both 
from man of man-u ‘ hand,’ for the m and h^ in these 
words represent each other, much as in Greek do /lep 
(whether the particle or the root of p,ov-os) and ev- 
(nom. els) ; tend-ere by the side of tene-re ; in, the 


Digitized by Google 



218 


EXCRESCENT CONSONANTS. 


Lat prep, leads to ind-e ‘ down ’ (see p. 70), and also 
appears with a c? in ind-igeo, indu-perator, endoter- 
cisus, &c. The pronoun is, ea, id had for its base in 
(see p. 70), and hence ind-e, of which e alone belongs 
to the suffix with the notion ‘ from ’ (see the following 
paper) ; so, hen or hun being the base of the relative, 
we have UTid-e (orig. cund-e), ‘whence.’ The Latin 
gerund scribend-um I hold to be in its first part one 
with the form scrihen already spoken of (p. 212) as equi- 
valent to the Germ, schreihen. In the French language 
instances abound, as cendre, tendre, Vendredi, vien- 
drai, tiendrai, moindre, gendre. So the Spanish has 
pondrS, tendrS, vendre; the German, niemand, abend, 
and-er, mind-er, hund, Mailand for Milan, <&c. ; Eng- 
lish, yond, beyond, mind, sound (sb. and adj.), thund-er, 
gand-er, hind, as well as kin {=gen of genus), com~ 
pound ; while our Henry {Henricus) appears in Scotch 
as Hendrich ; and conversely bind, mind, find appear 
without a cf in the Dorsetshire btn, min, fin. The 
Latin words cale-,palam and pahna, &c. polle- ‘weigh,’ 
praepolle- ‘ outweigh,’ exhibit an { in the root-syllable, 
and an I is often interchanged with an n ; but in Latin 
there is a tendency not to be satisfied with an n, but to 
add to it an excrescent d, the more so perhaps as the 
Latin is also fond of an interchange between I and d. 
Be this as it may, we find by the side of the words 
just mentioned cando, at least in compounds (ac- 
cend-o, &c.) and derivatives as candela, pando, pendo, 
with pondus. Again, m and n being freely convertible, 
the Latin gemere, tremere, and an obsolete abemere 
‘ to take down,’ have led to the French geindre as well 
as g^mir, craindre as well as crimir, and aveindre. 

As regards the Latin verbs which end in ngere, it is 


Digitized by Google 



EXCRESCENT CONSONANTS. 


219 


not altogether certain that the n is non-radical. Al- 
though the Latin iugum, coniugium, the German j'ocA, 
and English yoke plead strongly in behalf of a mere 
guttural as ending the root-syllable, yet the Sanskrit 
gives us a form yiin-aj-mi as the equivalent of the 
Latin iungo, which seems to imply that yun was one 
form of the root. This view seems to be confirmed by 
the fact that the French writes joind-re where the 
Latin has iung-ere ; and similarly a«<retW-re,yein(i-rej 
peind-re, ceind-re, oind-re. As &d interchanges alike 
with n and I, it is no strange matter that we find the 
Latin uad-ere leading on' the one hand to the French 
allrcr, and on the other to the South Italian arv-are ; 
but here again the ordinary dialect prefers and-are. 

13. Kd : an r and a d are often convertible, as is 
shown abundantly in Sanskrit and occasionally in Latin, 
for example, as was just observed, in caduceo- (m.) from 
the Doric tcapOxeio- (n.), in cust-dd- (see above) com- 
pared with irvX-ap-o, &c. Hence we should be the less 
surprised at a c? growing out of an r, as in cor cord-is 
and KapS-ia by the side of Keap Kijp- J otd-ior by or-ior ,* 
Latin mordeo-' by the side of mol- ‘ grind,’ Lith. mal - ; 
hurd-en from hear; murd-er from a root=Lat. mor- 
and Sanskrit mar or mri; gird by the side of the 
Latin giro- (written commonly gyro-) ; French tord-re 
by the side of Latin ter- ‘turn’ and torque-; our 
haggard = German hager. The Latin verb ped- may 
well have had pUd- for its base, just as scrib-, die-, 
nub- are lengthened firom the simpler forms scrib- 
(cf. conscrlf6-t7io, CatuL XXV. 10, andy/ja^-), die- {mali- 
dlco-) nUb- (pronUba, conUhio-, v€<f>eXri) ; and then this 
p^d- might have for its Greek analogue irep-, but we 
find irep&-. Again, I and r being convertible we have 


Digitized by Google 



220 


EXCRESCENT CONSONANTS. 


the equivalent forms al-ere and atpew (ap-) ; but from 
aUre come ard-uus and ard-ere. So also obsordesc- 
and obsolesc-. (See the following paper.) 

14. A0: a\d-e>, aKj0-r)<Tu, a\0-e<r<ra) are said to be of 
the same stock with the Latin al-o ; but should 
perhaps be connected with our heal. In either case 
the 0 is excrescent. 

15. N0: ap0-ev- from an old root ap- ‘blow' (cf. 

which the Sanskrit retains ; ep0-ep, of which 
the first ep alone belongs to the root, and the second ep 
alone to the suflix (see the following paper). As p 
and 0 are readily interchanged, for example, in e-pLa0-op 
compared with fiep-ecr-, fie-p.p-f)fun ; in our oth-er, the 
compar. of one, in the Norse ann-ar and its pi. dat. 
dlh-rum ; in the Old English plurals loveth and loven, 
find the Old English adverbs henn-en and heth-en, I 
hold that the right way of explaining such forms as 
nap0apco by the side of ep.a0op, Pep0-etT- by that of 
pa0-e(T-, Ttpt^ip0o- by that of Tpep.t0o-, is to consider 
that p before the 0 in p,ap0apa, &c. corresponds to the 
0 of epa0op, &c. ; and that the 0 which follows the p is 
excrescent. What I here say of p0 I say also, mutatis 
mutandis, of Xaii^apw, nepSm, ^c. compared with eXa/Sov, 
ped-o, &c. 

16. V0\ op0-o- and op0-po- ‘ dawn ' from op of op<o 
oppvfu ; Sap0-apo> by the side of dor-mio, our dr-eam 
and dr-ouhsy : irop0-p.o- root irop- as seen in irop-o-, 
irop-i^u, corresponding to our fare; ap0-po- (n.) and 
ap0-p.o- form ap- ‘join;’ a-Kap0-p,o- from axap- of aKatp-u 
‘skip rep0-po- (n.) ‘an end’ compared with rep-p,ar-\ 
epep0-€P from epepo-. 

17. 20: ea0-\o- = Doric ea-\o- ; fe<r0-r)T = Latin 
uest-i- from a root ev or «<r ; om.<T0-ep compared with 


Digitized by Google 



EXCRESCENT CONSONANTS. 


221 


oTTttr-o) ; evro<T0-t with evroi. Again, as 6 and <r are 
convertible, rtnno/ieOa has substituted a 0 for the <r 
of the Doric rvnrofu?, while 7tnrrofiea0a has preserved 
the <r, out of which a 0 has grown. So i0-fiar- ‘a 
road ' and K70-fio- co-exist ; fia0-a\XiB- ' a sort of cup ’ 
and fjM<r0-a\iB- ; ea0-uo by the side of esse es-ca. The 
Greek <T0ev-etr- too is probably of the same stock with 
our sin-ew, as also with the Greek noun iv- (nom. ts) 
of the same meaning, or rather iv-, for this noun and 
uir- of the Latin uires seem to owe their digamma to 
an old form aPiv-. 

1 8. ^0 : the 0 of <f>0eyyto, <f>0eip, ^0iva>, <f>0apa, <f>0ovos, 
as compared with irroXis, ■/rroXetiot, can scarcely be other 
than excrescent. In the adj. e\ev0epo- we virtually 
have €\ttf)0epo, which there can be little doubt is one 
with the Latin libero-, the b of which represents the 
^ of the Greek word as usual. Consequently the 0 is 
excrescent. As for the e of e\ev0epo- it is not a mere 
euphonic vowel, as is commonly taught. On the con- 
trary, libero- has lost what the Greek has retained. 
At the same time, both have suffered decapitation, 
loebero- (I prefer dealing with the older form) standing 
for sol-oeb-ero- or sol-ub-ero (for oe and u seem to 
have had the same sound, as they now have in 
Dutch). Thus we come to what is almost identical 
with sol-uh-ili - ; and e\ev0epo- may well have grown 
out of e\ev0fpo-, the asperate having eventually dis- 
appeared in consequence of the neighbouring 0. The 
Latin ub-er- originally ‘ a stream,' as is shown by the 
ubera mammarum of Lucretius, as well as by the' 
common phrase ubertim hxcrumare and the use of 
ubertas for ‘ a flow of words,’ is immediately related 
to the verbs um-esco uu-esco. Hence the genuineness 


Digitized by Google 



222 EXCRESCENT CONSONANTS. 

of the labial is established ; and consequently a pre- 
ference should bo given in Greek to ov^ap over ov6ap, 
the latter having probably grown out of a fuller form 
ovif>0ap. 

19. X0 : Sty^0a Tpi^Oa for Tptx» i ax^-ofttu by 
ax-ea- and the Sanskrit sah. The g of the Latin vb. 
rug-io, as in leones rugiunt, should have for its Greek 
analogue a xs accordingly we have /Spwx-® and 
^pvx-aofuu; but we also find pox0-o~ ‘roaring;' and 
again by the side of opey-a> there stands ope^^-eai. In 
iX^v- and exO-po- the d is probably excrescent. 

20. Cs, Sc, Gs, Sg ; the sibilant being often inter- 
changeable with a guttural, we find that the guttural 
on the one hand often throws out a sibilant, and con- 
versely a sibilant a guttural. Hence cum and 
avv ; Koipos and ^vvo » ; p.iywp,i puaya, misceo, mis-tus, 
and our mix; augeo and av^ava with our wax; ex 
and ef in both Greek and Latin ; ^vpo* ®Jid aKippos ; 
t^ot and uiscum. Thus the second of the two 
consonants in the four combinations seems to be 
excrescent 

21. Ns : by the side of xi]v the German has gans 
and the Latin has ans-er, which, like our own gand-er 
and the German gSns-er-ich, was in strictness ap- 
plicable only to the male bird, er being a male suffix, 
as in the German kat-er a ‘ tom-cat.' The prep, et* 
for evt has probably only an excrescent <r attached to 
the ordinary prep. ev. 

22. Kz, Lz, Tz: as in German Aerz =cor ; kurz-^ 
our short ; schmerz = smart ; salz = salt ; malz — malt ; 
katze = cat; ratze == rat; hitze => heat ;zu = to and too ; 
zwei = two. 

23. Pf: in German, as |?/acZ ; pfahl^pale ; 


Digitized by Google 



EXOBESCENT CONSONANTS. 


223 


pfand =pautm, &c.; ap/el — apple ; stdpfel -= stopper ; 
tropf-en == to drop. 

24. Mb: ftea-rififipia for fievTj/tepia ; p,ep,pKaiKa for 
fieftoXwKa; French chamhre from camera-; humble from 
humili-; nombre from numero-; combler from cvrnu- 
lare ; Spanish nombre from nomine ; hombre frona 
homine; hembra from femina. Again, as n and m 
interchange, we have in Greek avhev- = Latin imbu ; 
Italian and-are with Latin amb-ul-are, and with Ger- 
man vxmd-el-n and toand-er-n ,• Greek yaar-ep-, Latin 
uent-er-, but English womb, Scotch wemb, the last 
used of the belly generally ; English loins, German 
lende, but Latin lumbi. Add with a silent b, which 
however was probably once pronounced, our lamb=^ 
German lamm, our = German daum. 

25. Mp : Latin templum and extemplo with extem- 
puh by the side of rep,eve<r-, all from a root rep,- ‘ cut,’ 
whence repv-a ; temp-os-, temp-era-re, &c. from the 
same root. Other familiar examples from Latin are 
sumpsi, sumptus, ademptus, contempsi, hiemps ; while 
in English we have the equally familiar Thompson, 
Simpson, Hampton. To these add the English hump 
with a dim. humm-ock ; stump by the side of the Ger- 
man stumpf and a dim. stumm-el. Again, as m and n 
interchange, we have Latin tund- = our thump ; as m 
and the guttural n {ng) interchange, our stamp — 
stingu-, better known in ex-stmgu- ‘ stamp out ’ (fire), 
di-stingur ‘stamp differently.’ M seems to have a 
stronger affinity for b than for p. Thus if we find an 
excrescent p making its appearance in the middle of 
words, the preference given to it over a b seems in 
some measure due to the influence of a following s or 
t. At any rate, it is for the most part in such com- 


Digilized by Google 


224 


EXCRESCENT CONSONANTS. 


pany that the p presents itself. At the end of words 
a is the less strange, since medials in this position 
are habitually pronounced as tenues. 

26. Ng : in the simple strong, long, the final is but 
a nasal n, but takes to itself a distinct guttural sound 
in strong-er, strength, longer, length. 

27. Rn: if the n of the German fern ‘far’ had 
been a suffix, it would not have passed into the comp. 
j&m-er. I am disposed then to regard the » of stei'n 
also as excrescent. In the cases of mourn, hum vb., 
hum sb., turn, it is of course clear that the n is no 
part of the root, seeing that we have the Latin maere-, 
hur- of com-hur, and hustum, our dim. hr-ook, and 
ter- in Latin; but whether it be excrescent or the 
remnant of a suffix it is difficult to decide. In 
the case of the vb. hum, the German hrenn-en and 
our own brand seem to be evidence that the n is 
referable to a suffix. 

In putting together these examples I have omitted 
some classes of words which might well have been 
inserted, because the question involved some matters 
which required a full discussion. Many of these will 
appear in the next paper. On the other hand some 
apology may perhaps be thought due for inserting 
examples of processes so familiar as what is seen 
in avtpog and ' pe<rnp.Ppiai The purpose was com- 
pleteness. 


Digitized by Google 



225 


XL 

ON FALSE DIVISION OF SUFFIXES. 

The number of suffixes in verbs, nouns, and particles 
has been unduly multiplied, as it seems to me, through 
various errors, which I propose to consider under 
several heads. First of all many of those which are 
supposed to be independent of each other are simply 
varieties of the same. Thus there is no substantial 
distinction between the neuter substantives of the 
Latin and Greek languages which take for their con- 
sonant an s, r, n, or t with various vowels, as 
jep-av- repar-ot, ovopar- ovopar-ot, <r0ev-e<r- a0ev-e-o9, 

rjir-ap' ijTTOToj, iS-up, vSar-ot ; opes~ -eris, frigos^ ~oris, 
ub-er- -eris, roh-ur- -oris, fuly-ur- -uris, ungu-en- 
-inis, nom-en- -inis. Whether such a form as ovopar- 
grew out of an older ovopam-, itself deduced ^m an 
earlier ovopav- by the outgrowth of a t, is of little 
moment for the present question. The free interchange 
of the four consonants just enumerated appears partly 
from the words themselves, and is further confirmed 
by the appearance of ovopaivm by the side of ovopar-, as 
also by such changes as appear in am^pov- and ao><f)po<r- 
xnnj, and the Latin pi. femina by the side of femur. 

In the same manner there is no substantial difierence 
in the liquid suffixes of the German fess-el (f.), deg-en 
(m.), and mess-er (n.), for these liquids are apt to iuter- 

Q 


Digitized by Google 



?2G ON FAMK DIVISION OF SUFFIXES. 

cliange, so that we ourselves have as equivalents for 
two of them fetter and dagger. 

A second cause through which the number of suffixes 
is unreasonably increased is the confusion by which a 
compound suffix is taken for a simple one. Thus in 
fac-in-os-, and the oblique cases it-in-er-is, 
iec-in-or-is, two distinct suffixes have been united. The 
ei» or in of these words may have been due to a 
secondary verb, as seems probable in the case of the 
first, since the form Tefv-a> is in common use. The 
same is possible in the case of fac-in-os-, for the Latin 
language also has many secondary verbs in in, as 
xper'n-o, po{s)'n-o, &e. ; corresponding to our own 
reck-on, op-en, and the Greek ftav6-av-<o, \afi^-av-o>. On 
the other hand, the nouns ungu-en-, fem-en-, nomen- 
also jKissess such a suffix. But this is a matter which 
may be left open. When the vowel before the n dis- 
appears, as in uol-n-es- (from uello), pig-n-os- (from 
pango), there is a still greater tendency to consider nes 
{nos) as a simple suffix. So also I believe that the ion 
of opinion-, &c. is one' with the ig-on of vertigon-, 
origon-, and so the analogue as regards suffix of such 
German nouns as ver-ein-ig-ung, a word which might 
have been represented by a Latin per-un-dron. Indeed 
the Latin actually possesses the simple un-ion- ‘ a little 
one,’ sometimes applied to a single one in a necklace 
of pearls, sometimes to a single one in a rope of onions. 
Our own language too abounds in cases where a com- 
pound suffix is not seen to be a compound. Thus 
English grammars speak of ling, lock, let, kin as 
though they were simple, when in fact they are all 
shortened forms, standing severally for el-ing, el-ock, 
elret or rather el-ick, and ick-in. That el is itself a 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


227 


suffix of diminution is abundantly proved by such 
substantives as nozzle, speckle, thimble, as well as the 
adjectives litt-le, mick-le ; whUe ing alone is seen in 
farth-ing ‘ a little fourth/ tith-ing, loi'd-ing ; ock or 
ick in butt-ock, rudd-ock, or ridd-ick (Jennings) ‘ a red- 
breast/ Jist-ock ‘ a little fist/ as once used, mamm-ock, 
and no end of words in the Seotch dialect.’ I have 
purposely passed by hillock and bxdlock, because here 
an I might have been by some claimed for the suffix. 
The. suJffix et is seen in not a few ■words, as cygn-et, 
sign-et, giml-et, ernm-et ; but this suffix is probably a 
corruption of an older ock or ick. 

It is not however denied that nowadays let and 
lizig are often added to a word per saltum. 

Another fertile source of error lies in the habit of 
what Mr. Whitley Stokes calls ‘ Provcction,' a word 
which may well take a place in the nomenclature of 
PhUology. He applies this term to what occurs in 
such a phrase as for the nonce, where the n has been 
unduly transferred from the preceding word, the more 
correct form being for then once ‘ for this once.’ 
Similarly the tone, the tother grew out of that one, 
that other. What is seen here in distinct words also 
applies to the prevalent error of treating fun in Greek 
and men in Latin as simple suffixes, for ovofun-, nomen, 
tegumen, and tegumentum should be divided ov-ofirar- 
or ov-vfi-oT-, g{o)n-om-en, teg-um-en, teg-um- ent-um. 
That ttT in the Greek noun and en in the Latin are in 
themselves suffixes is shown by the words already 
given, ^TT-oT-ot and ungu-en-. On the other hand, um 
is sufficiently familiar in bell-um, &c. and virtually in 
the Greek epy-ov, &c. Again, that /* does not belong 
to the suffix of aifjMT seems to be proved by its appear- 

Q 2 


Digitized by Google 



228 


ON FALSE DIVISION OF SUFFIXES. 


ance in at/io-w, which implies a sb. a»/*o- (ra.). I hare 
elsewhere (Philolog. Trans. 1856, p. 341) given some of 
my reasons for believing that the suffix um of Latin 
neuters had in origin a guttural rather than an m. In 
our own language the guttural asperate ough is often 
pronouneed as a labial asperate, as rough, cough, &c. 
So in different parts of England we have at the pre- 
sent day the three terms shoch, shoof, and sheaf applied 
to the same objeet. On this view there is nothing 
strange in the fact that helium, apium, and Ilium 
should lead to adjectives bellic-us, apiac^s, Jliac-us ; 
or that apium and allium should in German take the 
forms eppich and lauch, in English leek; and con- 
versely that the Greek irvpBaK- and Latin podec-^ 
should in English be represented by hott-om as well 
as hutt-ock. 

It is strange to find Madvig in his * Bemerkungen’ 
(Brunswick, 1844) putting forward the doctrine that 
an accusative case in Latin has no true suffix, the final 
m being, he says, a mere euphonic addition, while the 
V of Greek accusatives he disposes of in the most 
summary manner by calling it i^iKKxxmKov. Thus he 
says, ‘The accusative is only the theme “euphoniously 
modified."' In this way he accounts for the identity of 
such forms as nom. lignum and acc. lignum. My own 
conviction, as I have said, is that helium has grown 
out of an older hell-ogh, which eventually passed 
also into hello, so as to enter the second declension. 


^ More strictly /un<2o-, but this I hold to stand for a fuller fund- 
OC-. How readily a suffix oc may pass into a mere o is seen in the 
Scotch mnnocJc and our more corrupt wind-ow, haddock the fish, 
also called haddow. Indeed the suffix ow of our language, which to 
tlie ear is but an o, has perhaps always grown out of a guttural. 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


229 


If what I have said he true, the adjective bellic-tis 
was in origin a mere genitive, ‘ of war,’ though in the 
end compelled to undergo inflection, precisely as hap- 
pened to the genitive cuius of the relative, which at 
one time the Bomans had the courage to decline as 
cuius a um, 

I proceed to other cases of what I regard as Pro- 
vection, in which the letter t plays a very important 
part, while c and b also occur ; and in order to suggest 
a doubt as to the propriety of the division usually put 
forward, I place by the side of each suffix examples 
which exhibit no t, c, or b. Thus to tion of lectio, 
aratio, I oppose ion of legio ; to tat of bonitat-, at of 
sat-iat- and uolunt-at-; to telaoi tutela,ela of querela; 
to itia of auaritia, ia of miso'ia ; to itie of canitie-, 
ie of desidie- ; to tudon of multitudon-, vdon of hebet- 
udorv-, and edon of dulcedon-; to ti of morti-, i of tor- 
qui- ; to tut] of seruitut-, iuuentut-, ut of salut- ; to 
itio of seruitio-, io of remigio- ; to tro of claustro-, 
aratro-, ro of fuh-ro- ; to ta of nauita, a of incola ; 
to tu of ‘partur, conuentu-, u of man-ti-, ac-u-, portic-u~. 
These for substantives ; and then for verbs : to or 
ita of ducta-, clamita-, a of sona-, tona- ; to tita of 
lectitOr, ita of clamita-; to tula of ustula-, ula of 
ambulor ; to tilki of cantilla-, ilia of sorbilla-. So for 
adjectives : to terno of sempiterno-, erno of hibemo- ; 
to tumo of diuturno-, urno of diurno- ; to tili of 
aquatili-, uolatili-, and bili of amabili-,Jlebili-, utibili-, 
nobili-, uolubili-. Hi of riuali-, facili- ; to bundo of 
saltabundo-, querebundo-, nitibundo-, and to cundo 
of iracundo-, uerecundo-, rubicundo-, the mere undo 
of sec-undo-, regundo-, onundo- ; to tiuo of captiuo-, 
iuo of uociuo-; to tico of aquatico-, ico of ciuico-; to 


Digitized by Google 



230 


ON FALSE DIVISION OF SUFFIXES. 


ceo of rosoceo-, eo of aureo- ; to cio or tio of trihu- 
nicio-, to of regio-; to ceri of alaceri-, uoluceri-, to 
beri of luguberi-, and to teri of compesteri-, the 
simpler eri of ac-eri-; to cero of ludicero-, ero of 
‘pig-ero-, rub-ero- ; to tero of dextero-, <ro<f>«>repo-, ero 
of svpero-, evepo- ; to timo and raro of superlatives, 
■imo and aro of minimo-, pecaro-. So also for diminu- 
tives, whether substantives or adjectives : to cuhis a 
vm of sermunculus, sororcula, corpusculupi, hreui- 
culus, ulus a urn of regulus, barbula, scutulum, 
hilarulm. 

Now in all these the error called Provection has 
been at work. In other words the c, b, t, assigned to 
the sufEx, belongs properly to the preceding syllable. 
The foundation of my argument is of so extensive a 
character that here I can do little more than refer to 
the two papers ‘ On the Representatives of the Keltic 
Suffix Agh or Ach “little,” in the Latin Vocabulary,’ in 
the Transactions of the Philological Society for 1856, 
pp. 295 — 354, and to the preceding paper, ‘ On English 
Diminutives,’ pp. 219 — 250. Of the former of these 
a summary was also given in the second appendix of 
my Latin Grammar. The result will be found to be, 
that, while I find the suffix ag in its full form in phg 
of plango, in frag of frango, in strag of strag-es, 
stragulus, whence straui, stratum, in uorag of uorag- 
on, &c., and but little changed in trah of traho traxi, 
the vowel is modified in frug- of fruge^ by the side 
of fruw and fructus, in fug- (for fug, cf. the Germ. 
flieh-en, flucht). On the other hand the medial g is 
exchanged for a tenuis in orac-ulum, lauac-rum, 
yerec-uiulus, ridic-tdus, uoluc-ris, inuoluc-rum; or 
retained, but with the loss of its vowel, in sparg-, 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


231 


terg-, merg-, uerg-. A second change I assume is 
the passage of a-^ into a<f>, as in y{a)p-a<f>-w by the 
side of {yapaatro), cf. rapay^ from rapaarrm). 

Our own laugh, with its written guttural but sounded 
labial, when contrasted with the German lach-en, is 
another distinct example, and one the more valuable 
as it is taken in the act of transition. But a Greek ^ 
has generally for its Latin analogue a 6. Hence 
scrih-o; and indeed the Greek language itself at 
times exhibits a /3, as in Oopv^os by the side of 
rapa'xri, and also in rpt/Soi from a root rep. Indeed, 
the Latin also must once have had a secondary verb, 
ter-ih- or ter-eb-, by the side of ter-, for so only can 
we account for the nouns tnh-ulum and (ereb’ra, as 
also for the forms tri’ui, tri’tus. In this way I would 
explain the Latin mir-ab-ili-, dol-ab-ra-, Jl-eb-ili-, ten- 
eb-ra-, illec-eb-ra-, cr-ib-ro-, (g)n-ob-ili-, uol-ub-ili-, 
sol-ub-ili-, and the vb. gl-ub-. 

The assumptions I am making may appear to be 
over-bold, as I claim not merely a change of the con- 
sonant, but a change without limit for the vowel. 
But it should be observed that the change of vowel 
obeys something like a law, inasmuch as the assumed 
vowel is, in a large majority of instances, that which 
is one with, or if not so, stiU in keeping with, the 
root-vowel ; and this doctrine of vowel-assimilation 
demands attention for perhaps aU languages. As an 
instructive example, I point to a family of substan^ 
tives in some of the chief Indo-European languaiges, 
where the leading idea is that of a living creature 
small or young, viz. irtoXot ‘a young horse,' pullus 
‘ a young horse or a chicken,' waXXaf ‘ a young man,' 
pellex literally ‘a young woman,’ but employed as a 


Digitized by Google 



232 


ON FALSE DIVISION OF SOFFIXE8. 


euphemism for ‘a concubine,’ fillie ‘a young mare/ 
pollock a Scotch term lor ‘ a young fish or crab.’ 

So far I have taken into consideration the' use of 
the suffix ag, &c. in Latin and Greek verbs only. But 
it also plays its part in the formation of nouns, both 
substantives and adjectives ; and here I was en- 
couraged in the outset of the inquiry by finding 
that Pott had demonstrated by a large induction that 
ox is a Greek sufiix of nouns signifying ‘ little,’ while 
I also found ec performing the same duty in Latin 
and the Slavonic language, and our own too has clear 
representatives of the same suffix, which take a great 
variety of forms, but all proceeding, I think, from' an 
original och, as bullock. The Latin nouns ctm-ec- ‘ a 
bug,’ pul-ec- ‘ a flea,’ cul-ec- ‘ a gnat,’ are unmistakeable 
diminutives, although the primitives have ceased to 
exist, But with us, as with the Greeks and Romans, 
the guttural passes into other sounda As they gave 
admission to yptufia and scribo, so the German has 
grab-en, and we both grub and grave. Hence I 
cannot but treat the f of calf, half, turf (for which 
the Scotch has a simple toor) as standing for of or oof, 
and so a diminutival sufiix. 

But not unfrequently with us a final guttural dis- 
appears altogether, though at times it leaves for the 
eye its ghost, in the shape of a silent y or w. Thus 
we have shad-ow for ‘ a bit of shade,’ haddow, window, 
by the side of haddock, and the Scotch winnock, way, 
day, say, any, honey, corresponding to the German 
weg, tag, sagen, einig, honig. 

With this evidence from our own island, I venture 
to put forward the startling doctrine, for those at least 
who now hear it for the first time, that all the vowel 


Digitized by Google 



ON FAL8E DIVISION OF SUFFIXES. 233 

conjugations of the Greek and Latin verbs, and all 
the vowel declensions of their nouns, have grown out 
of older forms, with the guttural suffix ac or ag more 
or less modified. Hence I account for the eighteen 
Latin neuters in aculo- or aero-, as sub-lig-ac-ulo-, 
laii-ac-ro-, for the ab in the four hundred adjectives 
(I give this number after duly counting them) in 
ab-ili-, as mirah-ili- ; also for the frequentative parti- 
ciples, over sixty in number, such as plorah-undo-, 
contionah-undo-, gemeb-undo-. Here the suffix is 
fitly employed, since we also, as Dr. Johnson points 
out in his ‘Grammar of the English Language,’ employ 
our diminutival el or le in the formation of iterative 
verbs, as sparkle, gamble, or gambol. 

On the principle here put forward, the vowel verbs 
of Latin should exhibit some traces of the same 
meaning ; and I see such traces first in such verbs as 
fricorre, laua-re, tonorre, all of which deal with 
actions which are eommonly repetitive, while tho 
simple verbs lauire and tonere are not unknown to 
the older language, and frictm fricui again implies 
a consonant verb fric-. Moreover, as in the Slavonic 
languages verbs fall into two classes, which their 
grammars call ‘ momentary ’ and ‘ continuative,’ so 
the Latin vowel verbs, where they fail to mark 
iteration, are distinctly employed for what is akin to 
this idea, that of continuity, as in stare (for set-a-i'e, 
as opposed to the simple set, seen in si-s{e)t-o, which 
denotes the momentary act of stopping) ; in sede-, iace-, 
pende-, as opposed to sid-, iac-, pend- ; in uide- ‘ see 
in s{e)c-i- ‘know,’ from a lost sec-, corresponding to 
seh of the German sehen ‘ to look at.’ Here, how- 
ever, the original meaning was probably of a physical 


Digitized by Google 



234 


ON FALSE DIVISION OF SUFFIXES. 


character ; and, if so, we should identify the root 
with sec of sec-are ‘cut,’ in which case we should 
have what is parallel to cemo ' I sift or separate,’ and 
to uideo as compared with diuid-o ‘ I separate.’ But 
both Greeks and Romans, though without the simple 
verb sec (ae«), have deduced from it, by the addition 
of the very suffix we are discussing, a secondary srk- 
ek; which, offensive by its repeated guttural, led to 
the substitution of a labial for one of them, and so 
supplied the Greek language with its aKev-, and the 
Latin with its s’pec-. 

As to nouns, the adjective rosac-eo-, to take this 
as representing a class, when compared with aureo-, 
receives its explanation so soon as we look upon 
ros-a- as having grown out of a fuller ros-ac-, whicli 
is nearly identical with the Greek diminutive poS-oK-. 
Similarly tribunic-io- may well have been deduced 
from an older tribun-oc-, or tribun-ic-. In the ad- 
jective aprug-no- from apero-, the guttural happens to 
have been preserved. Similarly ciuic-us and bellic-tts 
have, in the its alone, the suffix which constitutes 
them adjectives ; this suffix being'_ probably, as I have 
already hinted, one with the ordinary suffix of the 
genitive as seen in the Greek <rw/t«T-o», &c. 

I have not so far appealed to an argument which 
seems to me of much weight, that, in all languages, 
diminutives have the habit of supplanting the primi- 
tives from which they sprang. Thus fratello, sorella, 
in Italian, soldi, abeille, in French, sparr-ow in 
English, and sper-Ung in German, though evidently 
in origin diminutives, stand now alone in their re- 
spective languages ; and again the primitive stare, of 
which star-ling is the diminutive, is almost obsolete. 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


235 


What has been said of Latin substantives is equally 
applicable to Latin adjectives ; and strangely enough 
the whole Latin vocabulary fails to present us with a 
single* original adjective of monosyllabic form, all such 
simple adjectives having been superseded by words 
which have assumed a diminutival suffix. Conse- 
quently all are disyllabic, or still longer. There is 
one apparent exception to this assertion in trux 
trucis : but this is only a compression of tor-uc- with 
the guttural suffix in unusual purity, and even this 
had by its side the double diminutive tor-UrO- for 
tor-uc-o-. This general formation of diminutival 
adjectives seems to be the result of something like a 
feeling of modesty, a desire to keep within due bounds. 
The Romans would not say without qualification 
that a thing was absolutely long. It was ‘ somewhat 
long,’ ‘ rather long than not,’ or, as in familiar English 
we say, ‘long-ish.’ Thus in Latin breuic-ulo- is the 
more con’ect division, breui- itself being but a cur- 
tailment of breu^ic-. So we also have adjectives in 
our yell-ow, shall-ow, holl-ow, call-ow. 

But it is common for a c to give place to a t, as has 
been more tlian once noticed in these pages (see pp. 
75, 227) ; and this especially in the case of diminutives. 
Hence in breuit-er, er alone is strictly speaking the suf- 
fix, which probably grew out of an older es, if we may 
argue from the general habit of the language. Now, 
when Plautus wished to give to the Greek adverb 
avewxnii'^*^ £>■ Latin dress, he wrote ineusceme. But 
the Greek ovtus had also the shortened form ovrm, so 

* Except perhaps par, ■which however lias a pi. par-i-a. 

* Trin. iii. 1, 24 (see my paper on Hilschl’s Plautus, Philolog. 
Trans. 1860-1, p. 178). 


Digitized by Google 


236 


ON FALSE DIVISION OF SUFFIXES. 


tliat the loss of the s was no violent proceeding ; and 
I venture to suggest that as the Greek suffix was 
reduced to so my assumed Latin suffix es was to e, 
as seen in tneusceme, docte, &c. Nay, we may go one 
step fartlier, and identify the assumed suffix es with o>», 
the e with the w, since the Latin had certo, uero, &c. 
as adverbs by the side of certe, uere ; and the same 
intercliange of these long vowels is seen in avarup and 
ira-njp, oixtirtop and oncrjrifp, Ariio and Anienis. Thus 
I am disposed to contend for the identity of the three 
adverbial suffixes er (for es) of hreuiter, e of docte, 
0 of vero, both with each other, and with the Greek 
ws and o> of ovT<09, ovrea. 

The cases of canit-ie-, auarit-ia- are also cases where 
a t has superseded a c, and indeed the forms in ic, 
canicie-, auaricia-, have good manuscript authority, 
though not the best, and they are to the present day 
preserved in the Spanish vocabulary. Nay, the com- 
paratives tristic-ior, iustic-ior, laetic-ior so frequently 
present themselves in manuscripts of the second order 
in place of the orthodox tristior, iustior, laetior, that 
it seems to be a safe conclusion that such forms were 
in provincial use. (See the paper already referred to, 
p. 346.) Thus we have direct evidence in favour of 
the forms tristic-, iustic-, laetic-, fixim which tristi<yia, 
iustic-ia, laetic-ia would be duly formed. 

What has been said of these words applies of course 
to the classes represented by bonit-at-, rmdtit-udon-, 
seruit-io-, nauit-a-. So also as regards the frequenta- 
tives in ita, as clam-ita-, it may justly be contended 
that it-a is a double suffix of diminution standing for 
ic-a. Indeed the verbs uell-ic-a-, fod-ic-a-, mors- 
ic-a- have preserved the guttural. No doubt the fre- 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


237 


quentatives in ita form a very large majority ; but 
these are mtters, as I have said before, not to be 
decided by numbers ; and further, there was a special 
reason why many of them should prefer to substitute 
ita for ica, as having already a preceding guttural, as 
clamita- for instance, agita-, and quaerita- ; I say 
‘many,’ after counting over three hundred so con- 
stituted, and a combination of three hundred might 
well lead to something like a law for the others. 

The same applies to the so-called supines, or nouns 
of the u declension, which have a short vowel before 
the t, as fremitVr, gemitu-, crepitu- ; as also to such 
nouns as position-, expositioiir. Indeed in ration- wo 
know, from our own verbs reck and reck-on, that a 
guttural is the more genuine letter. So again with 
the noun sation- a guttural may be claimed as having 
preceded the t on the strength of our verb sow with 
its final w. For condition/- and dition- the case is clear, 
as the older forms are now known to have been con- 
dicion- and dicion-. Lastly in red/-it-u-, ad-it-u-, &c. 
the t is an original part of the root, as shown by the 
forms it-er, ex-it-io-ip), in-it-io-, com-it-, ped-it-, &c. ; 
and the old verb per-bit-ere, &c. 

In many cases, however, the appearance of a t, no 
way belonging to the suffix, seems to admit of its true 
explanation in the theory that it is excrescent. I refer 
to such forms as lect-ion-, cant-ion-, capt-ion-, assert- 
ion-, ust-ion-, gent-i-, inort-i-, uest-i ; cant-u-, iact-u-, 
que^t-u-,part-u, salt-u-; iuuent-ut-, uirt-ut-, senect-ut-; 
and the verbs lect-ita-, ust-ula-, cant-iUa-, But not 
so in capt-iuo-, &c. for such an adjective denotes the 
belonging to the class of men capti or things capta, 
so that a perfect participle is called for, which is not 


Digitized by Google 



238 


ON FALSE DIVISION OF SUFFIXES. 


the case with the nouns in ion and u, for these speak 
only of an act in progress, not of an act completed. 

But there remain for consideration the cases in 
which a t is preceded by a long vowel, as aration-, 
uolatili-, aquatili-, aquatico-. All these I would 
explain on the same principle which was applied (p. 
208) to otio-, diutius, setius, viz. that the older forms 
had a c before the t, and that the t itself was an out- 
growth from this c, the several words being deduced 
from ar-ac-, uol-ac~ (cf. uolu<yri-), and aqu-ac-. 

Again, I am disposed to transfer from the suffix to 
the verb the t which precedes or in the names of 
agents, as arat-or, ac(-or, duct-or, rapt-or, past-or, 
cant-or, sart-or, the or itself being only a variety 
of ^l^r, just as the Welsh has harf ' beard,' harf-wr 
‘ barber mor ‘ sea,’ mor-wr ‘ sailor pryn-u ‘ to buy,’ 
pryn-wr ‘a buyer ;’ pechod 'em,’ pechad-wr ‘sinner’ 
(one of course with the hat. peccat-or). And these 
Welsh forms are, I believe, generally regarded as con- 
taining the word gwr ‘ man.’ But if I thus treat t 
of the above Latin words as excrescent, what is to be 
done with the Greek names of agents, such as oucrirtup, 
otmiTijp 1 My answer is simply this, that they should 
be dealt with in precisely the same way, the t being 
attached to what precedes. One advantageous result 
of this view is that the French forms such as taill-eur, 
brass-eur become intelligible ; and further that the 
Teutonic branches fall into agreement with the classical 
languages as regards nouns in or, er, like our sail-or, 
dealer, and the Germ, hiif-er, geb-er, &c. 

I have yet to deal with comparatives and superla- 
tives, and to justify the division ao<f>eor-epo-, Betvor-epo-, 
cdt-ero-, svnist-ero-, 7r<n~epo, ao^ayr-aro-, Seivor-aro-, ult- 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


239 


mo-, int-imo-, &c. (p. 120), as contrasted with the pre- 
vailing habit of assigning the t to what follows. What 
I subsequently said in the same paper will more than 
prepare the way for my defence. It may be remembered 
that, starting from the suffix tov of Greek, ior (ios) of 
Latin comparatives, I called in aid the doctrine of 
Bopp, that these two suffixes were apt to lose one of 
the two vowels, an t being lost for minor and minus, 
and secus as superseding secius (cf, p. 122), to which I 
might have added the Latin prim-ores (for prim-iores) 
and the Greek ir\e-ov for rXe-wv, the pie-ores of the 
Carmen Arvale, and plus or ratlier pious itself, as a 
contraction of pUos, and lastly the theoretic op-os, 
afterwards post, of the same paper. On the other 
hand, we have but the i in magis, nimis, satis, prist- 
ino-: probably also in oir-Kr-o), more certainly in wp-tv, 
which, though short in later writers, has often a long 
vowel in Homer. It has been said indeed that the 
vowel is ‘ properly' short ; but surely the authority of 
Homer is for the present question more weighty than 
that of any number of later poets ; and indeed gene- 
rally it may be asserted that the passage from long to 
short vowels is more in accordance with reason and 
with the history of language. Thus irpiv may well 
have grown out of rp-tov, i. e. irop-iov. 

In the same paper I pointed to the fact that both in 
Gothic and the younger German languages the same 
appearance, sometimes of i (e), sometimes of o, is the 
universal characteristic of comparatives, as cUd-iz-a, 
minn-iz-a, together with frum-oz-a,frod-ozra in Gothic, 
alt-ir-o {alt-er-o), menn-ir-o {menn-er-o), with jung- 
or-o, frot-or-o in Old German ; and from this I drew 
what I deemed a reasonable inference, that these 


Digitized by Google 



240 ON FALSE DIVISION OF SUFFIXES, 

Teutonic suffixes grew out of an older suffix ios or 
102 {ior). The Greek and Latin suffixes differ Indeed 
in the one having a v, the other an s (r) ; but this 
is a distinction which must not be regarded as affect- 
ing their identity, seeing that it is in obedience to a 
law which subsists between the two languages (see p. 
IGll) ; and indeed the Sanskrit serves as a connecting 
link with its fuller suffix tydiis (Bopp, V. G. iL pp. 32, 
35), which at the same time accounts for the long i 
in the Greek com-parative. 

But, as I said in the former paper, no one can for a 
moment separate the suffix of Modem German and 
our English comparatives, as in dlt-er, old-er, wis-er, 
from the er of the Old German alt-er-o, &c. ; nor this 
again from the classical ev-ep-oi, sup-er-i, inf-er-i. The 
word 'r-ep-o- ‘ one of two,’ proclaims its intimate con- 
nexion with the numeral h- ‘ one,’ both by its meaning 
and by its asperate, for the Greek language stands 
apart from all its congeners in giving an asperate 
to these two words. Then as to the change of con- 
sonant we have the same in our old bet-est from a 
root hen, and, what is still more to the purpose, the 
Danish has alongside of each other an as m. and f., 
et as neuter for the indefinite article, which of course 
is but the numeral one. We are taught indeed that 
the t in this and other words is a neuter suffix; 
but in fact such a thing as a neuter suffix has no 
existence, all that characterises a neuter being the 
loss of some final consonant, and not any addition. 
A change of n and t is seen not merely in the Norse 
participle m. haldinn (for hakUns), f. haklin, n. hald- 
it, but also in our own hold-en, clov-en, as compared 
with the Scotch ahas-it, English clef-t, and with the 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


241 


Latin pos-it-o-. Of course a classical t according to 
Rask’s law should appear in our tongue as th, and 
accordingly the comparative of our one is otlv-er. 

Again, so far as regards the Teutonic family, it is 
the common doctrine, and one which I think cannot 
be disputed, that superlatives are formed from the 
comparative, and not directly from the positive. In 
the French le meilleur too this is self-evident ; as also 
in the Lapp dnek ‘ short,’ dnekub ‘ shorter,’ dnekumus 
‘ shortest,’ which throw light upon the Latin superlar- 
tive. The same theory accounts for the forms ^eytoTos, 
fi€\TiiTTos, where fteyis, /SeXrts, following the analogy of 
Latin comparatives, have given a preference to the 
<r over V. It matters little whether we divide these 
as /ieyi<r-rot, fie\Ti<r-Tos, or treat the r as excrescent, 
inasmuch as both tos and oy are weU fitted to repre- 
sent the Greek article, so that ^(Xnaros will exactly 
correspond to the French le meilleur. Such a posi- 
tion for the definite article is the ordinary con- 
struction in the Scandinavian family, and by way of 
example I give the Danish patriot-en ‘ the patriot,’ 
dag-hlad-et ‘ the day-leaf or joumab’ Further, it is 
by adding the definite article as a suflSx that the so- 
called definite declension of adjectives has attained 
its peculiar form, as Bopp has clearly shown. The 
Latin superlative, nouissimu^ for example, Bopp 
divides as nouis-timus; but I hold the true division 
to be nouist-imus. Indeed generally it seems to me 
that the ordinal numerals, as BeK-aros, evv-arof, e^B-o/tos, 
dec-umus, sept-umus, and the forms derived from pre- 
positions, vTT-aros, 'jrptoTos {= irpo-aros), /ittr-aTOs (from 
fieaos, which is probably akin to fiera), ecj^-aros, imv-f 
(= in-imus), primus (= pro-imus), summus (== suh- 

II 


Digitized by Google 



242 


ON FALSE DIVISION OF SUFFIXES. 


imus), inf-imus, are safer guides in etymology than 
the superlatives from ordinary adjectives ; and of 
course the same consideration applies to comparar 
tives. Hence I have no longer any doubt as to the 
division of alt-ero-, post-ero-, &c., post-umo-, ult-umo-, 
int-imo-, <fec., in all of which the t is for me excres- 
cent, while ao<fmtr-€po-, <ro0o)T-aTo-, I regard as more 
genuine than Setvor-epo-, Seivor-aro-, and standing for 
ao^-oKT-epo-, <Toif>-oKr-aTo-, aoif>-o- itself having super- 
seded an older aoff>-oK-. 

The three forms vor-epo-, ut-ero-, wheth-er, I defend 
in a wholly different way. In these I hold the dental 
to be part of the root syllable, representing probably 
an original n, as does also the d of id, quid, and quod 
in Latin, and t of our it, that, what, for again I pro- 
test against the doctrine that in these letters we have 
a mere suffix of neuters. I cannot here pretend to do 
justice to a question which covers so wide a field. My 
arguments were given many years ago in a paper read 
before the Philological Society (Proc. iii. 56), and were 
subsequently more fully developed in a paper attached 
to my little book entitled ‘ Alphabet,’ &c., where they 
oceupy twenty-three pages. There are also some brief 
allusions to these papers in the last of these essays. 
I have here only room for the outline. My argument 
contends, in the first place, that the third-person pro- 
nouns, which for Latin are represented by hie, iste, ille, 
is, and qui, together with the two prepositions cis, uls, 
and their derivatives, as also the demonstrative enclitic 
ce, seen in hie, istie, illie, with their adverbs, and in sie, 
nune, tune, were in origin all of demonstrative power, 
or, in other words, accompanied the physical act of 
pointing to an object ; secondly, that the root form of 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 243 

these words appears to have been ken, the guttural of 
which is preserved in cis and ce, also in kuvos and in the 
modern Italian chi, queslo, costi, quello, colui, while 
the final n also is frequently cropping up. The already 
quoted ksivos, with its analogues the German jemr and 
our yon, beyond, present it distinctly. Of the forms 
hinc, inde, evdev, hence, thence, whence, I shall have to 
speak again; and I have already referred (p. 70) to 
the occurrence of the phrase Is Locvs in an inscription, 
where the long i in the nominative seems to point to 
an old root in for the pronoun, as distinctly as in 
Greek the numeral ds and the preposition eis do to 
forms eV and ev. In isto- the n has given place to an 
s, and that s has thrown out an excrescent t, while 
illo-, or oh- (cf. Virgil’s olli and the old form aboloes 
for ab illis), or rather both ilh and olio together, point 
to a form iol as an older variety of the root ; just as 
mag-is and minrus suggest an older suffix ius. But 
iol stands to our yon much as the Latin sol to our 
sun ; and indeed the Slavic languages have the older 
ono in preference to oh. 

Then for the relative and interrogative the evidence 
is stiU stronger. I shall not here rely on the fact that 
in the Tatar languages ken is the ordinary form of 
the relative, as in Finn and Mongolian ; while in 
the first of these it has also the by-form cu precisely 
as the Latin, although at the same time I am satisfied 
that the German School of Philology is wrong in 
denying to these languages all affinity with the Indo- 
European family. But the evidence at hand is enough. 
The existence of an initial guttural will not be dis- 
puted, although it has been lost for unde, ubi, uter, 
ut. Nor is it difficult to establish the claim of the 

R 2 


Digitized by Google 



244 


ON FAI^E DIVISION OF SUFFIXES. 


concluding nasiil. In the Spanish quien it has been 
preserved, but more commonly it has passed into tlie 
allied m, and this to an extent but seldom noticed. 
First the Sanskrit has it in kim, a form which has 
much distressed some philologers. Dr. Guest (Proc. 
Phil. Soc. L p. 287) claims whom as an old English 
nominative, and he also refers to such forms as Swed. 
nom. hwem, gen. hwem-s, Dan. nom. hvem, Fries, gen. 
waems, Du. gen. wiens. Then our pronoun he is 
represented in Old Norse for the m. by nom. han-n, acc. 
han-n, gen. han-s, dat hon-um. So again our definite 
article the takes the two forms of den and en, and 
carries the n for both m. and f. into every oblique case 
of both numbers. Lastly, the enclitic ce, so familiar 
in hie (hici-ne), islic, iUic, and their adverbs, as also in 
nunc {nunev-ne), tunc, sic, has in Umbrian the form 
cen in the abl. of eiso, eizu-c or eisu-cen (A. K. p. 135). 

I cannot stop to discuss the rationale of the change 
of meanings by which a mere demonstrative became 
available for use as a relative or interrogative. The 
fact is what concerns us most, and no one will doubt 
that in Greek the so-called definite article was not 
unfreqiiently employed as a relative, as by Herodotus 
when he says, 6vov<rt rtf Tlapdev^ rovr re vavtfyovt km 
root av "KaPaai, ; while the German der and our own 
that perform both duties, and the same is true of 
most branches of the Indo-European family. 

But the question remains. What is the meaning and 
what the origin of the assumed root ken ? As a pre- 
paration for this inquiry, I would meet the question 
by another. What word is best fitted to accompany the 
act of pointing to an object ? and the answer can only 
be one, a word signifying ‘ look.’ Now ken in Scotcli 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 245^ 

and Old English is a verb with this very meaning, and, 
as I show in a subsequent paper, this word is no way 
limited to our own island or to the Teutonic family. 
It is the root of our verb know {kon-oiv), and so lies at 
the bottom of the Latin gnosco, the Greek ytyvoxx/vw, 
the Sanskrit jna, and indeed extends far beyond 
these limits, so as to include Chinese itself, for again 
I repeat my unwillingness to draw an absolute hue 
of distinction between our Indo-European languages 
and those of the Tatar stock. I go back, then, to the 
forms TTOT-epo-, ut-ero-, wheth-er, and claim the t and 
th as representing the n of the root xev. 

I conclude this paper by a matter for which this 
last inquiry was a necessary prelude, the just division 
of the adverbs signifying whence, as irodev, evOev, 
otriaOev, ovpavodev, inde, unde, hinc, illim and illinc, 
istim and istinc, caelitus, funditus, intus, &c., hence, 
thence, whence. I may observe in passing that the 
use of intus in Plautus in the sense of ‘ from within’ 
is not to be denied. Now the power of the genitival 
suffix, viz. ‘ from,’ is precisely that which belongs to 
all these words, and us (os) is a familhir form of the 
genitival suffix, although an older form Avas probably 
IMS, as in cu-ius, nullius (= nullo-ius), Xoyo-to(s). In 
the preceding paper I gave evidence that en of Hein- 
rich-en. Frier n Barnet, and the Sanskrit sun-u’n-am, 
was but a variety of this. I might at the same time 
have pointed to the so-called adjectives lin-en, gold-en, 
as genitives in origin ; as also to the in of ^v\-iv-o-, 
crast-in-o-, Roma’n-o-, as having the same origin and 
the same power, and even {pace Germany) to the 
Finn, where for the consonant declension wicras ‘ a 
gacBt,’ kirwes ‘a hatchet,’ caunis ‘beautiful,’ have for 


Digitized by Google 



‘246 


ON FALSE DIVISION OF SUFFIXES. 


genitives wierah-an, Hnoeh-en, caunihrin, and the 
Lapp, where toli ‘ a stool,' has a gen. toli-n. But if 
this suffix exist in the adverbs above given, we must 
divide them as iro0-ev, evd-ev, onria-B-ev, ovpavo0-ei>, 
caelU-us, fundit-us, int-us, where 0 of iro0-ev belongs 
to the root, as does the t of wor-epo-, the 0 of ev0-ev, 
oTTia0-ev is excrescent, that of ovpavo0-ev is a substitute 
for the X of ovpavox-, while caelit-us stands for caelic- 
us, and the t of int-us is again excrescent. As in 
oma0-ev, &c. the v is often dropped, leaving but owt<T0e, 
so ind-e and und-e probably represent an older ind-us, 
undr-us, for the loss of the s would be followed by a 
degradation of the vowel into S, as in ipsus ipse, 
magis mage, scripserunt scripsere. Precisely on the 
same principle the short final of superrdi and infernS 
is to be explained, for these words denote ‘ from above,' 
‘ from below,' so that they are to be regarded as repre- 
senting lost adverbs supernus, infemus, and are not to 
be mixed up with ordinary adverbs in e as recte, in 
which the suffix has a very different power. 'Then 
as exinde, proinde, &c. are often cut down to exim or 
exin, &c., illinde, which analogy suggests, would lead 
to what is actually found, illim, and that, with the 
enclitic added, to illin-c. So also with hinc and ist- 
inc ; and this view is confirmed by the co-existence of 
utrinde and utrimque, or utrinque. With us the 
genitival suffix, as in Latin, has a sibilant, so that the 
old forms henn-es, thenn-es, whenn-es, are but genitives 
of a stem henn, &c., and thus here too the n of the 
root reappears ; and to confirm the assumption that 
•jro0-€v and our wheth-er, oth-er have in the th a repre- 
sentative of an 71, we also find in Old English the 
variety heth-en, for what is now written hence. I con- 


Digitized by Google 



ON FALSE DIVISION OF SUFFIXES. 


247 


elude then that in ind-e, und-e, as in ev6~ev, the dental 
mute is excrescent, and the n a portion of the root. 

A dijficulty may suggest itself in the cases of the 
assumed ill-inde, ist-inde, in that ill and ist alone 
belong, on my own showing, to the root; but my 
defence is that in ill-o- and ist-o- we have reduplicate 
pronouns. In the Greek ov-ro-s this is commonly 
admitted. But the neuter forms ill-vd or ill-ut, ist-ud 
or ist-ut, suggest a s imil ar reduplication for them ; and 
this admitted a form ill-ind-e is no longer an anomaly. 

Lastly, let me notice the fact that the word look 
is used whether we point to a near or to a remote object, 
so that we here have an explanation of the difficulty 
which may weU have occurred to the reader, — I mean 
that I have been assigning a eommon origin to words 
so opposed in meaning as hie and ille, citra and ultra, 
this and that. Thus we have in Terence luciscit hoc 
iam (Hast, iii 1, 1), ‘ it is getting light, look, already 
while, on the other hand, when Virgil (Aen. v. 457) 
says : — 

* Fiaecipitemque Daren ardens agit aeqnore toto. 

Nano dextra ingeminans ictus, nano ille sinistra,’ 

we find an equally satisfactory exponent of ille if we 
translate the second line by the words : — 

‘ Now with his right redoubling blows ; now, look, look, 
with his left.’ 

This paper in a great measure grew out of a desire 
to deal with questions which presented themselves in 
the preceding paper ‘ On Excrescent Consonants,' and 
so may be regarded ns a sort of supplement to it. 


Digitized by Google 



248 


XII. 

QVAEEITVE. 

THE SANSKRIT LANGUAGE AS THE BASIS OF LINGUISTIC 

science; and the labours op the GERMAN 

SCHOOL IN THAT FIELD — ARE THEY NOT OVER- 
VALUED ? 

FinaT Part. 

I HAD thought at one time of placing at the head 
of this paper ‘ Doubts of a Non-Sanskritist’ But on 
reflection it seemed desirable that the title should be 
more definite. If the words I have actually used be 
thought by any one to savour of national ill-will, I 
must giye the assurance that nothing could be more 
remote from my purpose or from my feelings. Those 
who have to deal with the classical languages must 
be either blind or ungrateful if they fail to acknow- 
ledge the deepest obligations to the scholars of 
Germany. The editions of Greek and Latin authors 
that have appeared in England during the last half- 
century have not been numerous, but even of these a 
large proportion have been simply reprints of German 
works. Again, the Lexicons of the two languages that 
have issued from the English press during the same 
period are for the most part so thoroughly of German 
materia], that it would have been more creditable if 
the title-pages had carried the words, ‘ Translated 
from the Gennan of with some few changes 


Digitized by Google 



QVAERITVR. 


249 


and additions.’ Again, if we turn our thoughts to the 
opposite side of the English Channel, wo find no great 
activity in the sphere of classical, especially Greek, 
literature ; yet what progress is visible there is chiefly 
due to the energy of German, not French, scholarship, 
as witness the valuable collection of Greek authors 
that has proceeded from the press of Didot. Nay, 
the high and indisputable reputation that Germany 
has won in this field only renders the duty more 
imperative to watch lest failure or shortcomings on 
any side should be kept from notice owing to that 
very prestige. Further, I wish it to be observed that 
the term I have used is ‘ overvalued,’ which is quite 
compatible with an admission of great value ; and 
again, I put what I have said in the form, not of a 
proposition, but of a question. It is only when that 
question is answered in the affirmative, or when the 
arguments put forward in this paper remain un- 
answered, that the slightest damage can be done 
to the reputation of the philologers concerned. It 
would have been simply indecent, if the present 
writer had expressed his fears in the form of a direct 
proposition, conscious as he is that he comes to the 
inquiry wholly destitute of what may at first sight 
be deemed an essential requisite, a knowledge of the 
Sanskrit language. Nay, he cannot pretend even to 
that smattering which may be obtained by a three 
weeks’ study of the language, and which has before 
now served to float a big book in the English market, 
a little sprinkling of the Devanagari character, and a 
judicious use of the hard words Vriddhi, Anuswara, 
&c., passing for profundity in the eyes of the unini- 
tiated. Such little knowledge as I have is that only 


Digitized by Google 



250 


QVAEEITVB. 


which may be acquired in the perusal of grammars 
and glossaiies, and works of like nature. 

The question here naturally suggests itself, how it 
is that I have taken upon myself to enter into a 
contest for which I am confessedly so ill-equipped ; 
and my answer is that I find the same suspicions 
which have foirnd a way into my own min d enter- 
tained by many others, and those too gentlemen 
whose position as scholars gives great weight to their 
opinions, though, like myself, they are whoUy wanting 
in the special qualification, a knowledge of Sanskrit. 
In every point of view then it seems desirable that 
the question should be raised. If our fears are ill- 
founded, it is well that they should be removed, and 
the road more thoroughly cleared of all obstruction for 
the Sanskritist. If otherwise, it is surely good for the 
progress of philological science that the matter should 
be thoroughly sifted. 

I do not propose to enter into the domain of San- 
skrit history and cluronology, a task for which I am 
wholly unfitted, especially as those who have the best 
qualifications admit that it is involved in the greatest 
obscurity ; nor indeed could one expect easily to find 
materials for accurate investigation in such a literature 
as that of the YSdas. The ‘Mantras,’ on the one 
hand, dealing for the most part with the devotional, 
and the ‘Brdhmanas,’ on the other, with the cere- 
monial and dogmatic, can scarcely be available for 
such a purpose. As to the Upanishads or the short 
appended treatises, I will be satisfied with a second- 
hand quotation from a work of a learned Hindd, that 
they ‘contain some rude indications of philosophic 
thought, and, like the twinkling of the stars in a dark 


Digitized by Google 



QVAEEITVH. 


251 


night, may occasionally serve as guides in a history 
of Hindi! philosophy. They do not however exhibit 
any great attempt at method, arrangement, classifi- 
cation, or argument. Even there the poetry pre- 
dominates over the logic. Bold ideas abruptly strike 
your fancy, but you find no clue to the associations 
which called them forth in the author’s mind, and 
search in vain for the reasons on which they are based. 
Sublime thoughts are not wanting, but they resemble 
sudden flashes, at which you may gaze for a moment, 
but are immediately after left in deeper darkness than 
ever. Nor are they free from those irregular flights of 
the imagination in which poets with vitiated tastes 
delight to indulge, setting at defiance all rules of 
decency and morality.’ (Banergea, Westminster Review, 
new series, vol. xxii. p. 463.) 

An argument for the antiquity of the Sanskrit 
language has recently been founded (‘ Lectures on the 
Science of Language,’ by Prof. Max Muller, p. 204, 
third edition,) upon certain passages in the Book of 
Kings and the Book of Job, but it is an argument 
which, as it appears to me, withers to the touch. AU 
rests upon the statement that four articles imported 
into Judea in the days of Solomon, viz. the ape, the 
peacock, ivory, and sandal-wood, are called in the 
Hebrew text by names foreign to that language, but 
indigenous in Sanskrit. But it is not an easy matter 
to prove that a word is indigenous in a language, and 
the Sanskrit-speaking race on their first entrance into 
the Indian peninsula (for they are allowed on aU 
hands to have been immigrants), would naturally 
adopt the native — that is, non-Sanskrit — terms for 
those objects which are peculiar to the country. 


Digitized by Google 



252 


QVAERITVR. 


provided indeed tliey had not already adopted them 
in the previous intercourse of commerce. But, passing 
over this consideration, let us throw a glance at each 
of the four words on which this important super- 
structure has been erected. Koph, the Hebrew for 
‘ ape,’ is, we are told, ‘ without an etymology in the 
Semitic languages, but nearly identical in sound with 
the Sanskrit kapi.’ It is of course implied here, 
though not said, that the Sanskrit does furnish a 
satisfactory etymology for its kapi. To supply the 
omission, I turn to Bopp’s Glossary, and there find 
that kapi ‘ ape,’ has for its root the Sansk. verb kamp 
‘ tremble,' so that, for some reason denied to us, the 
ape was conceived by the Indian mind as * the 
trembler.’ Then ivory has for one of its Hebrew 
names shc7i habbim, where, as shen means ‘tooth,’ 
habbim might well speak of the ‘ elephant,’ and this, 
it is said, ‘ is most likely a corruption of the Sanskrit 
for elephant ihha, preceded by the Semitic article.’ 
If, as I suppose is the fact, Mui be a misprint for 
ibha, the resemblance is even then limited to the 
consonant, and we have nothing oflered in the way 
of proof that this name for the elephant is the 
original property of Sanskrit. Thirdly, tukhi-im, in 
Hebrew ‘peacocks,’ bears no doubt a tolerably close 
resemblance to the Malabar name tog'ei; and this ‘in 
turn has been derived from the Sanskrit sildiin ‘ fur- 
nished with a crest.’ Lastly, the Malabar and Sanskrit 
name for sandal-wood is valguka; and ‘this val- 
gu{ka),’ the Professor says, ‘is clearly the name which 
Jevdsh and Phoenician merchants corniptcd into 
algum, and which in Hebrew was still further changed 
into almug.’ I would submit that at any mte the 


Digitized by Google 



QVAERITVR. 


253 


word ‘ clearly’ is somewhat out of place in an etymon 
which involves four assumptions, the aphaeresis of 
V, the apocope of ka, a paragogic m, and the meta- 
thesis of gum to mug. Even if true, such derivations 
have scarcely strength enough to serve as the foun- 
dation of so large a theory.* 

But the same writer has elsewhere (‘History of 
Ancient Sanskrit Literature,’ p. 524), contended that 
the Vedas have an antiquity far older than the know- 
ledge of writing. ‘The collection of the (Vaidic) 
hymns, and the immense mass of the Brithmana 
literature, were preserved,’ he says, ‘ by means of oral 
tradition only.’ In another passage of the same work 
(p. 507) he teUs us that ‘before the time of P&nini, 
nay, even when he himself wrote {sic) his great work, 
writing for literary purposes was absolutely unknown.’ 
To understand the full force of this proposition, to 
form an adequate idea of the extent to which the 
Professor would tax the mnemonic powers of the 
Brahmans, we must remember that P4nini, according 
to his own authority, was preceded by whole gene- 
rations of grammarians. In his ‘Lectures on Language’ 
(p. 110) he says : ‘ Those valuable lists of words, 
irregular, or in any other way remarkable, the Ganas, 
supplied that solid basis on which successive gene- 
rations of scholars erected the astounding structure 


* I leave this as I wrote it, but I eubaequently found that Prof. 
Max Muller bad borrowed the whole argument from Laseen’a 
‘ indisebe Altertbumskunde,’ vol. i. p. 538, &c., so that the setting 
alone was bis own. It is true that he himself referred to this 
passage of Lassen ; but his reference was so placed, that a reader 
might well suppose the argument about ‘ivory’ alone to have been 
drawn from Lassen. 


Digitized by Google 



254 


QVAERITVR, 


that reached its perfection in the grammar of F&nini.* 
But if the structure be ‘astounding,’ and ‘the per- 
fection of a merely empirical analysis of language,' 
it seems not to be possessed of much that would be 
interesting to the mere European scholar, for the 
Professor concludes his panegyric with the words : 
‘Yet of the real nature and natural growth of lan- 
guage it teaches us nothing.’ 

As regards the V6das themselves, one can readily 
imagine that religious feeling and poetical feeling 
combined may do much to invigorate the powers of 
memory, while the mere rhythm of verse contributes 
to lighten the task ; but intense indeed must havei 
been the feeling of duty which could induce Brahmans 
to commit to memory, and there retain, a complete 
library of the driest grammariana 

The whole argument then carries with it, as it 
seems to me, its own refutation ; and in truth the 
challenge implied in the words, ‘I maintain that 
there is not a single word in Pdnini’s terminology 
which presupposes the existence of writing,’ has 
already received a twofold answer from my colleague. 
Professor Goldstiicker (‘ P4nini, his place in Sanskrit 
Literature,’ 1861) ; first a self-refutation, quoted from 
the Oxford Professor’s own words; ‘This last word 
lipihara (a writer or engraver) is an important word, 
for it is the only word in the Shtras of P4nini which 
can be legitimately adduced to prove that P4nini was 
acquainted with the art of writing;’ and as my 
colleague observes (p. 17) : ‘ It is obviously immaterial 
whether another similar word be discoverable in his 
grammar or not; one word is clearly sufficient to 
establish the fact.’ But he further produces from 


Digitized by Google 



QVAERITVR. 


255 


P^ini’s own work an abundant supply of terms 
which could have no meaning whatever when writing 
was unknown. Let me quote one more passage from 
the same work (p. 14): ‘ As, according to his (Max 
Muller’s) view, P4nini lived in the middle of the 
fourth century B.C. (pp. 245, 301, ff.), it would follow 
that, according to him, India was not yet in possession 
of the most useful of arts at the time when Plato died 
and Aristotle flourished.’ 

I have entered into these details to show the un- 
satisfactory condition of the chronology of Sanskrit 
literature, and at the same time I would suggest the 
question whether there should not be a little more 
caution in the acceptance of literary conclusions, even 
from those to whom the English public has been accus- 
tomed to look as authorities above all controversy. 

But if we cannot have the advantage of a reliance 
on literary history, we must be content to examine 
the internal evidence supplied by the language itself, 
and the dealings therewith alike of Indian and 
European authorities. As my own doubts, and I 
believe those of the friends to whom I have already 
aUuded, were flrst raised by what appeared to us as 
most strange, though generally sanctioned, etymo- 
logies, I win proceed to produce some of these, limiting 
myself for the most part to a single class. 

Already Max MUller (‘Lectures on the Science of 
Language,’ p. 370) himself quotes, as an example of 
Indian etymology, the derivation of the sb. kdka 
‘ crow,’ from apakdlayitavya, i.e. ‘ a bird that is to 
be driven away,’ but adds that Y4ska, another gram- 
marian, anterior to P4nini, considered kdka to be an 
imitation of the bird’s note. Whether the Professor 


Digitized by Google 



256 


QVAEKITVR, 


himself adopts or rejects this mimetic origin of Mka, 
his words do not enable one to say. But be this as it 
may, in another Sanskrit noun, kdrava, Lat- cor{p)vo~, 
‘raven,’ he steadily refuses to see what, for one, I 
must regard as a still more exact imitation of the 
bird's note, viz. cor cor. Had he included in his view 
the Greek icop-aic-,^ he might perhaps have assented to 
Pott’s doctrine (E. F. iu 506, 507), that ax in Greek 
substantives is a suffix of diminutival power, so that 
Kop alone would be the root. He himself, in his 
aversion to what he calls by way of disparagement 
the Bow-wow theory, strives to deduce the whole 
family, kdrava, Kopovt), raven, &c. from the Sanskrit 
verb ru, to which he ascribes ‘ a general predicative 
power’ as expressing sound, ‘ from the harshest to the 
softest,’ and so applicable ‘ to the nightingale as well 
as to the raven,’ nay, even to ‘ the barking of dogs’ 
and ‘the mooing of cows.’ In a note however he 
hesitates between this etymon and one from the 
Sansk. kdru ‘singer.’ To the special honour of this 
last derivation, the raven seems to be about as well 
entitled as the parrot or the peacock ; and the de- 
duction of kdrava from ru, a general term impl)dng 
‘sound,’ would probably be regarded by lawyers as 
‘ void for uncertainty.’ 

The same objection of excessive generality applies 
to the whole class of etyma with which I now propose 
to deal, viz. those of words ascribed to roots of various 
forms, but with the one meaning ‘ to go.’ Thus the 


> We must not suppose the ancients in their nomenclature to 
have distinguished with modern accuracy between the raven, the 
rook, and the crow. (Soe Mr. Wedgwood’s paper on tliat subject.) 


Digitized by Google 



QVAEEITVR. 


257 


S. go (gav), the equivalent in power and probably in 
form of Lat. bov-, Gr. /Sow-, as also of our own cow, is 
deduced by Sanskritists of all classes, Indian and 
European, from a S. vb. ga ‘go and that this ex- 
planation of the word may not suffer for want of 
company, I may add the S. ild ‘cow,’ referred by 
Bopp (Gloss.' B. V.) to the vb. il ‘ go.’ Now that ani- 
mals like the ‘hare’ or ‘stag’ should receive a name 
from their marked power of locomotion is, at any 
rate on the logical side, admissible, and thus we may 
perhaps be ready to assent to the current etymologies 
of Aare (Germ, hose), the Latin lepos-, and the Greek 
tXa^o-. But the cow is scarcely entitled to put in a 
claim for such distinctions as against any other living 
creature. Strangely enough the same pair of words, 
go and ila, also signify ‘earth,! and these again have 
the same origin ascribed to theui (Bopp, Gl. a w.). 
So also the Gr. yaia passes with Bopp (V. G. § 123) 
as standing for yaFta, and so an adjectival offspring 
of a sb. corresponding to the S. go ‘ earth,’ and 
eventually of such a vb. as ga ‘go.’ In the same 
section S. gmd ‘a name for the earth in the V5da- 
dialect,’ is deduced from the S. vb. gam ‘ go.’ Nay, 
our own earth, though it comes immediately from our 
old English vb. ear ‘ plough,’ represented in Sanskrit 
by ar, is traced ultimately to the S. r ‘go’ (Bopp, 
ibid., M. Muller, ‘ Lectures on Language,’ p. 256, and 
Pott, K F. L 218). It would be an interesting fact if 
such a series of at any rate consistent etymologies 


* I haye preferred to draw from this work, althongh now some- 
what ont of date and superseded by other works, simply because it 
comes from the founder of the science, 

8 


Digitized by Google 



258 


QVAERITVE. 


could be accepted as proofs that the Hindi! mind had 
already discovered the motion of the earth, whether 
about its own axis or about the sun. But as it seems 
more probable that then, as now, there existed an 
inveterate tendency to treat the earth as the one fixed 
object to which all the movements around us are 
conveniently referred, we must look for some other 
explanation of the theory ; and accordingly Bopp 
suggests that the movement of the earth must here 
be regarded as only ‘passive,’ — in other words, the 
earth {erde) is ‘the betrodden one' (‘die betretene’). 
Though it does not visibly move itself, man and beast 
would be in an awkward predicament for locomotion 
if there were no earth to move upon. Before leaving 
the earth, I ought to notice that Prof. M. Muller 
believes (p. 257) our word aroma to be another 
ramification of ar ‘plough’ and r ‘go,’ for does not 
Jacob say (Gen. xxvii. 28), ‘ The smell of my son is 
as the smell of a field which the Lord has blessed’ ? 

From land I pass to water, but the same etymology 
pursues us. Thus the Ganges itself is the Sanskrit 
GangA, literally the ‘ Go Go’ (M. Miiller, ibid. p. 384). 
So we have S. salila ‘water’ from S. vb. sal ‘go’ 
(Bopp, Gloss, s. V.) ; S. ay) ‘ water,’ the analogue of 
the Lat. aqu-a, from vb. ap ‘go;’ sarit ‘river,’ from vb. 
sr * go’ (M. M. ibid. p. 253) ; and saras ‘ water,’ fi:om 
the same vb. (ibid.). The last noun is by Bopp trans- 
lated by the Lat. lacus, and declared to be one with 
the Greek eXo» (cr. form «Xe<r-) ‘a marsh,’ in which 
case the motion of ‘going’ seems to disappear. Yet 
after all Sanskritists may contend that marsliland, 
being half water, half land, has a double claim to a 
derivation from roots which denote ‘going.’ 


Digitized by Google 



QVAERITVB. 


259 


Take next the class of worms and reptiles. Bopp, 
for example (V. G. § 86, 1), refers the Latin vermi~ 
{= quermi-) and S. himi to the S. vb, kram ‘go;’ 
the Germ, schlange ‘ snake’ to S. vb. h'ang ‘ go and 
the Lat serpens, S. sarpa, first to S. vb. srp ‘ go,’ and 
tdtimately to sr ‘ go.’ Had any of these verbs meant 
*go by little and little,’ the derivation would have 
been satisfactory, but the meanings given by Bopp in 
his Glossary to these verbs, as well as to aU their com- 
pounds, furnish no authority for such an assumption. 
No doubt in his comparison of the verb srp with 
kindred languages he dwells much on the idea of slow 
movement in those kindred languages ; and again Pro • 
fessor Wilson in his Grammar, though he adds the 
meanings ‘ creep or glide,’ gives precedence to the 
general term ‘ to go.’ 1 cannot but think however 
that the suflSx of as compared to sr and that of 
the Lat. ser{e)p- with the varieties of (p{e)ir- and Lat 
r-Sp- (r^ere), Eng. c{e)r-ep {crep-t, creep). Germ. 
hr-iech-en, and with an additional suffix of diminution 
cr-aw-l, represents the idea paulatim, as it seems to do 
in the Lat car-p- (see my paper ‘On the Suffix agh’ &c. 
Trans, for 1856, p. 336). When I wrote what is there 
seen, I expressly stated that I was at a loss for the root 
ser- ‘ go.’ This I now find in the S. sr, although I still 
believe a form her, as heard in cr-eep and kr-iech-en, 
to be more genuine than sr with the sibilant I am 
not deterred from regarding the two roots as sub- 
stantially one by the fact that as a role the guttural 
A: or c of Western Europe is usually represented by 
the palatal s of Sanskrit, not by that which occurs in 
sr and sr^. 

Another class of words, which Bopp is disposed 

s 2 


Digitized by Google 



260 


QVAEKlTVn. 


generally to refer to roots significant of simple move- 
ment, are those which denote ‘ time’ (‘ da Uberhaupt 
die Zeitbenennungen meistens von Wurzeln der Bewe- 
gung stammen,' V. G. § 69). For instance our word 
year, old Germ, jdr, together with what he regards as 
an equivalent in form, tlie Gr, wpa ‘ season,’ is referred 
by him to the S. vb. ya ‘ go,’ but by Lassen and Burnouf 
it seems to S. vb. tr ‘ go.’ Again the Goth, atvs (crude 
form aiva) as also its relatives, Lat. a^uum, Gr. atav, 
is deduced by Bopp, GrafiF, and Kuhn from S. vb. i 
‘go’ (ibid.). The i might perhaps not have passed 
with the ill-informed as forming the kernel of these 
words ; but all is accounted for ; the initial a, it seems, 
attains its position through ‘ Guna,’ and all that follows 
the i (or e) is to be regarded as a suffix. In spite of 
such a combination of authority I am still disposed to 
prefer my own etymology of aeuum from the Latin 
auge- (for the vowel-change compare the variety seen 
in the allied oef-«a and av^-ap-a>), with ‘ growth ’ for the 
original sense, as exhibited in the well-known line, 
Crescit occulto uelut arbor aeuo, ‘ grows like a tree 
with growth concealed.’ The Lat. saeculum is also 
referred by Bopp (V. G. § 248, Anm.) to a S. vb. 
sac or sale ‘ go, follow.’ Further, as he (Gloss, s. v, and 
V. G. ib.) considers S. omati ‘ time’ a derivative from 
am ‘ go,’ so he is also inclined to deduce from the same 
stem the Latin annm (as standing for amnus) as well 
as the Greek evo9. That amnus was in fact the older 
form of annus is proved by the derived sol-emni-s, 
but to the derivation from a verb ‘ to go’ I would 
oppose that other derivation which connects it with 
the Latin prep, am ‘ round,’ German um. The very 
idea of a year implies a circle, and the words annulus 


Digitized by Google 



QVAERITVK. 


261 


* a ring,’ and the noun dnus with a long vowel, seem 
to complete the proof. On the same principle the 
word year itself, like yar-d ‘ an enclosure,' and gar-d- 
en, &c., claims kindred with many words denoting a 
circle, as %opro-, horto-, X">P'0~> cor-o-na-, circ-o-. The 
initial change between a Gr. x> Lat. h, and a ^ (y) in 
German and English is in accOTdance with the usual 
law, as seen in heri, hestemo-, gestem, yesterday. 

So much for the alleged deduction of substantives 
from Sanskrit verbs signifying ‘to go.’ But in the 
formation of secondary verbs also the roots i ' go,’ and 
yd ‘go’ are thought by Bopp well fitted to play im- 
portant parts, as for example in furnishing suflixes by 
which verbs are converted into passives (§ 739) and 
causals {§ 740). As regards the former, if k6rd ydi, 
to take Bopp’s own example from the Bengali, have 
for a literal translation ‘ I am made’ (‘ ich werde ge- 
macht’), as given by himself, then gemaeht is by itself 
already a passive, just as verloren is in the Modern 
German gehen verloren, literally ‘ to go lost.’ We 
too may say ‘ become detested’ or ‘ become fascinating,’ 
where the distinction between the passive and the 
active idea turns upon the accompanying participle, not 
upon the word ‘become.’ Again, Bopp’s illustration 
from the Latin amatum iri is surely not applicable. 
If the principle for which he is contending be valid, 
we ought already to have a passive in the indicatival 
phrase amatum eo ‘ I am going to love,’ but this is a 
mere future of the active. The introduction of a pas- 
sive of eo, whether in the indicative as amatum itur, 
or in the infinitive as amatum iri, is only a convenient 
mode of exhibiting an impersonal verb, equivalent to 
the French on va aimer. The examples of ueneo and 


Digitized by Google 



262 


QVAERITVE. 


jpereo, quoted by Bopp, a,re at first sight more to the 
purpose, and he would have done well to strengthen 
his case by comparing them with uendo and perdo. 
Yet after all uentre, standing for uenum ire, means 
probably ‘ to go into the window,’ and so ‘ be exhibited 
for sale,’ which certainly is more truly the meaning of 
the phrase than ‘ to be sold.’ So also uenui est admits 
of the literal translation ‘ it is in the window,’ i.e. ‘ is 
oflered for sale.’ Again, jierire ‘ to come to an end,' 
like the English go to the dogs or the Greek eppe e» 
KopaKat, contains no doubt what is virtually a passive 
idea ; but this arises from the combination with the 
per and the e» Kopaica?, &c. That ‘ go’ does not carry 
in itself the idea of a passive is clear from our own 
phrases ‘go to the Bar,’ or ‘ into the Church,’ or ‘into 
business.' Curtius (Beitrhge, p. 329) goes still far- 
ther, and conjectures that the 0t) which appears in the 
aorist and future of Greek passives is connected with 
the S. vb. yd ‘ go,’ in which however all resemblance 
seems limited to the long vowel. I pass then from 
the passive. 

The causal mood of the Sanskrit verb, as well as the 
tenth conjugation in general, having for their distin- 
guishing character the syllable ay,^ Bopp’s mind is 
divided by a doubt whether this suffix should be re- 
ferred to the verb i ‘ go’ or t ‘ wisL' The latter one 
would think is far better fitted for the formation of a 
desiderative mood, which, it seems, is a general appen- 
dage to the Sanskrit verb. Nor does i ‘go’ at first 
sight appear a satisfactory element for the purpose of 


* Of causals some make ay the suffix, some ya. I believe the 
former to he right 


Digitized by Google 



QVAERITVR. 


263 


constituting a causal verb ; but we are assured by 
Bopp (§ 740) that several Sanskrit words which denote 
‘ motion' at the same time denote ‘ making.' Whether 
the particular verb i has this convenient privilege he 
does not stop to tell us. Assuming however that it 
has, we have before us a strange combination, that 
roots expressive of ‘going’ are alike fitted to form 
passives and to form actives. 

But fiirther, although the causative idea is declared 
to be the character of the tenth conjugation, I find 
little proof of this in the list of fifty-seven verbs quoted 
by Professor Wilson in his Grammar, for of all these 
at the utmost one in five can be explained as contain- 
ing the idea ‘ to make.’ Thus the first ten in the 
series are translated by the English verbs ‘ steal, dis- 
respect, hurt, send, wink, speak, play, be feeble, be 
able, sound.' I am not then surprised to find in § 772 
such a sentence as, ‘ It deserves however notice that in 
Sanskrit denominative verbs in ya^ occasionally avail 
themselves of the causal form without any causal 
meaning.’ My own feeling is that the original notion 
paulatim resides in ay, and that it is the Sanskrit 
variety of that suffix which I have discussed at length 
in my paper on agh or ag, the passage o£ & g between 
vowels {aydmi) into a y being a common occurrence. 
On this theory the meaning may well pass into that 
of frequentative or continuous. But leaving this 
question open, if we accept that one of Bopp’s two 
explanations which finds in the suffix of the so-called 
Sanskrit causals or tenth conjugation the root i ‘ go,’ 
we shall have to assign to this use of the word a 


* See note on the preceding page. 


Digitized by Google 



264 


QVAERITVB. 


somewhat vast domain in the classical and German 
languages, for Bopp connects with the same type all 
the vowel-verbs of the Latin, at any rate the first, 
second, and fourth conjugations of that language 
{§ 745 c.) ; all the Greek verbs in eat, a<a, oto, a^a>, tfw 
(§§ 109 a. b. 749, and 762), together with the par- 
ticular verbs fiaWu, laWa, and ; and lastly 

all the weak verbs of the German stock (§ 109). A 
few of these verbs specially noticed by Bopp himself 
may claim a few words. We are assured that the 
Latin facio = S. Mv-dydmi, literally ‘ I make to be 
iacio = ydp-dydmi ‘ I make to go doceo == gndp- 
dydmi ‘ I make to know rapyio = rdp-dydmi ‘ I 
make to give’ (§ 747). It seems somewhat damaging 
to this theory that the suffixes i or e of the Latin, 
which Bopp himself holds to be the representatives 
of the S. ay, contribute but little to the formation of 
the causative idea, seeing that fac-, iac-, doc-, rapt- in 
themselves already express the full notion of ‘ making, 
throwing, teaching, robbing as may be seen in the 
forms fac-erc, iac-ere, rap-ere, and in fac-tus, iac-tus, 
doc-tus, rap-tus. Yd-p-dydmi is thought to possess 
a second suffix of causation in its p, so that yd ‘ go’ 
is the real base of the verb ; and if this case be doubt- 
ful, a causal p is declared with greater certainty to be 
an element in ghd-p-dydmi ‘ I make to know,’ ^id (or 
in English characters ^’n<J) being what Bopp is pleased 
to call a root-verb, the equivalent of our hiow. But 
of thisynd more hereafter. To place Bopp's doctrine 
clearly before me, I throw aside the equivalent por- 
tions eo and dydmi, and there results the equation, 
Lat ddo == S. jndp. No doubt the palatal j of the 
Sanskrit is with reason assumed to be a corruption of 


Digitized by Google 



QVAEHITVR. 


265 


a medial guttural g or 7. The bueineBs then is to prove 
that d^c is equal to gnA'p. I make no difficulty about 
the final consonants, for a Lat. c habitually corresponds 
to a S. p. But there still remain three problems for 
solution, — to identify the d with g, the short 0 with the 
long a, and to account for the appearance of n in the 
Sanskrit or its disappearance from the Latin. For 
the first Bopp simply quotes the instance Ar)-(ii]Tr]p ■= 

; on the difference of vowel he says nothing. 
The difficulty as to the nasal is disposed of by the 
assurance that for gnd-nd-mi ‘ I know^ there occurs an 
actual gd-nd-mi, and that in Persian there exists the 
form dd-ne-m ‘ I know.’ But surely the asserted loss 
of an n from gnd-nd-mi, when followed so closely by a 
second n, is but a poor justification for the disappear- 
ance of an n in doc for dnoc. For one then I must 
regard the doc of doceo as better explained within the 
limits of the classical languages by dec of deico (= 
dico) and BeiK-vvpt, by Saic, the root of Si-Saa-xa>, Si- 
SaxTot, and SaxTvXof, by the die of di-dic-i and dig of 
dig-itus. But if I must look to the Sanskrit, here too 
I find a thoroughly admissible representative in the 
vb. dis ‘ show,’ with that palatal s which regularly cor- 
responds to a western i-sound ; and indeed Bopp him- 
self I find, in his ‘ Glossary,’ regards this root dis as 
one with the root of Seixwfn and the Lat. dico.^ 

^ Ab some friends well acquainted with Sanskrit were slow to 
believe that a writer like Bopp could have published such ‘ extrava- 
gancies,’ I quote his very words (§ 747) ; ‘Kan ich aber das c der 
genannten Form (facia) nicht mit dem skr. causalen p. vermitteln, 
so glaube ich doch dem Lateinischen noch ein anderes Causale 
nachweisen zu kbnnen, worin e die Stelle eines skr. p vertritt, 
namlich dooeo, welches ich im Sinne von ich mache wissen auffasse 
und ftir verwandt mit di-ico (eigentlich ich wiinsche r.u wissen) 


Digitized by Google 



260 


QVAERITVfi. 


The p of rap- or rapi-, as also that of the S. rdp~ 
aydmi, is again treated by Bopp as of causal power, 
and he finds in his root rd ‘give’ only a variety of dd 
‘give.’ Thus ‘to give’ and ‘to cause to give or rob’ 
owe their marked difference of meaning to the causal 
suffix ; not that this is an essential matter with him, 
for this same root dd or rd is thought by him to be 
identical with the S. vb. Id, to which simple form is 
ascribed the double meaning of ‘to give’ and ‘ to take,’ 
a mixture of ideas which, if carried out in life, might 
lead to inconvenient results.^ 

So much for the value to the Sanskritist of his roots 
signifying ‘ to go’ in the way of etymology ; and the 
Stock is no small one. Taking of the ten conjugations 
the first alone, and again limiting myself to the series 
which Professor Wilson quotes in his Grammar as ‘ the 
most useful verbs of this conjugation,’ I find just twenty, 
viz. 1. a/ ‘ to go ;’ 2. at ‘to go ;’ 3. i ‘ to go ;’ 4. c?i4 
‘ to go ;’ 5. ukh ‘ to go ;’ 6. r ‘ to go,’ ‘ to gain ;’ 7. rj 
‘ to be straight’ or ‘ honest,’ ‘ to gain,’ ‘ to go,’ ‘ to live 
8. kram ‘ to go,’ ‘ to walk ;’ 9. gam ‘ to go ;’ 13. vichckh 
‘ to go ;’ 11. char ‘ to go ;’ 12. dhauk ‘ to go ;’ 13. pat 

und dem gr. tSdtiv, iiSdano halte. 1st das d dieser Forman sns 
entstanden (vgl. ana so fdhrt doceo zum skr. p^p- 

dydmi, ich mache wissen gd-nd-mi ich weias filr gnd-nd-mi) and 
zmn para, dd-ne-m ioh weisa. Ala ein Baispial ainas lat. Canaale, 
worin das urapriinglicbe p unverandart gabUaben ware, erwiesQ 
aich rapio, im Fall ea dem skr. rdpdydmi ich mache gaben ent- 
spricht, Ton der Wz. rd gaben, die, wie mir achient, nichta anders 
ala eine Schwachung von dd ist Auch kommt, aovrie neben dd 
eine erweiterte Form dot bestehh neben rd im Y6da-Dialekt rds 
vor. Mit rd und dd acheint auch ibrem Uraprunge nach die Wz. 
Id identisch, welcher die Bedeutungen gaben und nehmen zuge^ 
schrieben warden.' 

‘ Ibid. 


Digitized by Google 



QVAEIUTVK. 267 

‘to go,’ ‘to fall;’ 14. sad ‘to wither’ or ‘decay,’ ‘ to 
go’ (with this appended : When the verb means ‘to 
go,’ the causal retains the final, — sddayati ‘he causes to 
go,’ or ‘drives’) ; 15. sad ‘ to decay,’ ‘ to be sad,’ ‘to go ;’ 
16. sasj ‘ to go ;’ 17. sidh ‘ to go ;’ 18. sr ‘ to go ;’ 19. 
srp ‘ to go,’ ‘ to creep’ or ‘ glide ;’ 20. slcand ‘ to go’ or 
‘ approach.’ I should have made some addition to thia 
list had I included those verbs which only express a 
more special or limited form of motion, as ‘ pervade, 
jump, hasten, run, gallop, approach, wander.’ 

With such an abundance of verbs to draw from, a 
philologer should the more hold himself bound to pro- 
ceed with caution, and so take care that the logical 
connexion between the root and the supposed deriva- 
tive should be well-marked. Whether the examples I 
have quoted exhibit such caution, I leave to others to 
decide. Lastly, I t hink it right to repeat that, by con- 
fining myself almost wholly to those instances of hold 
etymology which deal with verbs signifying ‘ to go,’ I 
avoid the charge of selecting instances favourable to 
my view. Indeed without some such limitation it 
would be an easy matter to pick holes in any of the 
most carefully elaborated philological works, for the 
most cautious et}Tuologer is apt to he carried away at 
times by tempting theories. In the next section of 
my paper I purpose more particularly to consider 
Bopp’s celebrated work, the ‘ Vergleichende Gram- 
matik,’ in its general system. 


In the short discussion which followed the reading 
of the above paper at the Philological Society it was 
replied on one side that the idea of ‘ to go’ was pre- 


Digitized by Google 



268 


QVAERITVR. 


cisely that which was well adapted to denote an active 
verb. To this I answer that a vb. ‘ to go’ was equally 
claimed for the special formation of passives ; but in 
truth the argument seems to me upset by its very 
generality. What is fitted to denote every form of 
action is for that reason unfitted to denote any form 
of action. The very essence of language is distinction 
or difference. Accordingly the other answer to the 
difficulties 1 had raised was that although simple 
‘ going’ is commonly assigned as the meaning of the 
verbs I have quoted, yet in truth each of them origi- 
nally denoted some special form of going. Such seems 
to be the feeling of Bopp also (V. G. § 515). I will 
only reply to this that I took the verbs with the mean- 
ing attached to them by the several authorities fi"om 
whom I was quoting. But over and above this, when 
the discussion was brought to the individual substan- 
tives, I found that the Sanskrit scholars who were pre- 
sent employed in the defence of the Indian etymolo- 
gies a vagueness as complete as that expressed in the 
general term ‘ going.’ Thus go and ild ‘ the cow,' and 
go and ild ‘ the earth,’ were said to be well entitled to 
such derivation, as being in the Indian mind the centres 
of activity most important to man. 

I take the opportunity of making a slight addition 
to the paper. As sr, according to Wilson’s Grammar 
(p. 200), at times signifies ‘to go quickly’ or ‘run,’ I 
am the more justified in attributing to the suffixed p 
of the power of paulatim. At any rate it has no 
causal power here. Further, if the Sanskrit vocabu- 
lary could deduce from a verb signifying ‘ to run’ by 
the addition of this suffix a secondary verb srp ‘ to 
creep,’ I am justified in connecting our own cr-ep 


Digitized by Google 



QVAERITVR. 


269 


(whence creep and crep-t), as regards its root, with the 
base of the Dorsetshire hir-n = Ang.-Sax. yrn-an ‘ to 
run and that base, hir, corresponds of course to tho 
Lat. cur- of curro. Again, if the S. verb sal ‘ go’ 
is one with the verb sr * go,’ we have the analogue of 
this sal in the Greek dWonat and Latin scdio, whence 
sal tu-s ‘ a sheep or cattle run! I am the more 
inclined to attach some value to this conjecture, be- 
cause as fal of fallere ‘ to cause to fall’ seems to 
furnish the only^ root for /ors fortis, so does sal- for 
sors sortis ‘ that which leaps from the um’ (situla), a 
noun from which has come the verb sortiri of the 
Latin and the verb sortir (with a very different power, 
more akin to the original root) of the French. Ijastly, 
let me observe that if the Sanskritists had been con- 
tented to derive sarit ‘ a river’ from a root sr ‘go’ or 
rather ‘run,’ there coxild have been little objection, our 
own terms ‘ current’ and * watercourse,’ Bull’s ‘ Run' 
and ‘ runlet,’ exhibiting a similar origin. Such terms 
as saras ‘ marsh or marshland’ and ap * water’ have 
not the same* justification. 


^ I haye been somewhat hasty in using tho woid ‘ only,’ for I find 
Mommsen (Inscr. p. 268) writing : ' Eecte omnino tarUm derivarunt 
grammatici quidam a terie et tertrtdo , ut fort venit a fertndo' But 
I still adhere to what is stated above. 


Digitized by Google 



270 


XIII. 

QVAERITVR. 

, Seookd Pabt. 

It would be to shrink from the task I have undertaken 
were I not to take into special consideration the great 
work of Bopp, who appears with something like gene- 
ral consent to be entitled the founder of C!omparative 
<rrammar as a science ; and the claim upon my atten- 
tion is only the stronger that his ‘ Vergleichende 
Grammatik,' the first portion of which was published 
in 1833, has been recently reprinted with some changes 
and considerable additions (1867-60). 

Here, as in what I have already said, I shall without 
further apology for my temerity proceed to state unre- 
servedly the objections that have presented themselves 
to my mind, not expecting those objections to be ac- 
cepted as valid, but desirous that they may attract the 
notice of scholars whose more intimate acquaintance 
‘with the subject will enable them to detect any errors 
I may have committed. The contest is happily one 
in which the victorious and the defeated must alike 
be gainers, the one object of both parties being to 
promote the cultivation of the science of language. 

First of all then I find in the very title of the com- 
mencing chapter (‘Schrift- und Laut-System’) what 
appears to me unphilosophical, viz. the precedence 
given to writing over sound. Over a large portion of 


Digitized by Google 



QVAERITVB. 


271 


our globe there exist whole races possessed of the faculty 
of speech, but without any knowledge of written sym- 
bols ; and indeed no small part of the population even 
of this country is in this position. But I should have 
passed over this matter if the error, so to call it, had 
not told unfavourably on the arguments that follow. 
The very first paragraph in the chapter gives to three 
of the vowels a special character, which, as it appears 
to me, is not due to them. Thus the title of original 
vowels (Urvocale) is assigned to a, i, u ; and this, I 
believe, on no other ground than that the Sanskrit 
alphabet had special characters for these when the 
sounds of e and o may have been denoted by combi- 
nations of the first three, much as the French language 
employs its dipthongs at and au as simple vowels. 
Had the school of philology founded by Bopp looked 
upon the materials for oral language as belonging to 
the domain of physical science, and wholly indepen- 
dent of those other forms of language which are 
addressed to the eye, such an error could not have 
occurred. In particular I must repeat the regret, to 
which I had already given expression in the year 1852, 
when I drew up the present paper (Proceedings, vol, 
V. p. 192), that the valuable paper on vowel-sounds 
which was read by Professor Willis before the Cam- 
bridge Philosophical Society (November 28, 1828, and 
March 16, 1829) seems to have been wholly unnoticed 
by the leading scholars of Germany. At any rate, 
when I entered upon the present inquiry, I had never ^ 

* 1 subsequently found that 1 had not done justice to German 
scholars in this remark. In Dr. Bindseil’s ‘Abhandlungen zur allge- 
meinen vergleichenden Sprachlehre’ (Hamburg, 1838), p. 84, refer- 
ence is made to Professor Willis’s paper, and from the appended 


Digitized by Google 



272 


QVAERITVR, 


come across the slightest allusion to this paper, or to the 
principles established in it, in any German writer ; yet 
had read much from this quarter that would never 
have been written by any one acquainted with the 
results of Mr. Willis’s experiments. Nay, I do not 
recollect to have seen at that time in any of their 
prominent works in the field of philology any refer- 
ence to that physiological organ which may literally be 
called the primum mobile of human speech, — I mean 
the two chordae vocales. Now that Professor Czermak 
of Prague by liis simple apparatus has enabled the 
inquirer to witness the action of these musical strings 
in the living man, we may hope that the study of oral 
language may be placed on its proper basis. It will 
then be laid down as the first dogma that as vowel- 
sounds constitute the substance of language (for brevity 
I drop the word ‘oral,’ which is the only form here 
under consideration), so the character of any vowel 
depends almost wholly on the distance for the time 
between the chordae vocales and the margin of the 
lips, — in other words, on the length of the vocal pipe, — 
the position of the tongue being of no moment so long 
as it does not close the passage of air. So thoroughly 
definite and mathematical is the character of the 
physical experiments on which Professor Willis’s 
results arc founded, that he has given numerical values 


note I leam that the paper itself was reproduced in the German 
language in Poggendorff’s ‘Annalen der Physik und Chemie.’ Still 
Dr. Bindseil himself seems to have been satisfied with a bare refer- 
ence, making little or no use of the principle, nor does his work 
appear to have met with much notice among his countrymen. It 
stopped abruptly with the first volume, although this contains only 
a general introduction and a treatise on gender. 


Digitized by Google 



QVAERITVR. 


273 


to the distances that belong to sucli of the vowels as 
are most familiar to English ears. At the same time 
as the number of points in a line is infinite, so the 
vowel-sounds pass by imperceptible gradations from 
the one extreme i (the sound in feet) to the other 
extreme u (or oo in boot). Thus it is wholly owing to 
the imperfection, yet necessary imperfection, of alpha- 
bets, that there is but a limited set of symbols for 
vowel-soimd. The number itself is essentially infinite ; 
and it was therefore a subject of amusement as well as 
regret to hear some few years ago that a conclave of 
learned philologers was then sitting in London to 
determine, among other high matters, what was the 
full number of vowels. 

But the vowel-order i, e, a, o, u (with the sounds 
which prevail on the Continent), as resulting from 
Professor Willis’s experiments, would have supplied 
the German philologers with a principle capable of 
solving pretty well all the problems that arise in 
connexion with the vowels, not merely of the Indo- 
Europeau family, but of language in general. In the 
paper already referred to (Proc. Philolog. Soc. vol. v. 
pp. 191 — 204) I have shown in some detail that it 
explains the umlaut and inick-umlaut so-called of 
German philology, the formation of plurals in Eng- 
lish, &c. by what Grimm calls ‘ motion,’ — that is, an 
alteration of the root-vowel, as in geese from goose, 
and generally the assimilation of adjoining vowels 
so familiar in all the Tatar languages and prevalent 
to a considerable extent in the Keltic, Teutonic, and 
Classical languages, to say nothing of others. In 
page 203 of the paper I gave, from my colleague. 
Professor Malden, a tabular view, showing the full 

T 


Digitized by Google 



274 


QVAERITVR. 


development of the principle in the changes of Greek 
vowels and diphthongs. And I have little doubt that 
the mysterious Guna and Vriddlii of Sanskrit are 
simply results of the same law. 

No doubt Bopp has allusions to the principle of 
vowel-assimilation, but these are altogether incidental. 
Thus it is only when he passes from the Sanskrit 
41, 42) to deal with the Zend, that he notices some 
cases where the presence of a y, i, or e affects the vowel 
of an adjoining syllable, and in § 46 mention is made 
of a similar euphonic influence belonging to a Zend 
V (w). But these are matters which should not be 
treated as peculiarities of the Zend. The philologer 
is bound to state the law of vowel-assimilation in its 
broad simplicity. 

But there is another point in which Sanskritists 
seem to have been misled by the habit of looking at 
language in its written aspect. They ascribe to the 
Sanskrit, in accordance no doubt with Indian authority, 
two vowels, r and Ir, which at any rate do not present 
themselves in the vowel-series of the Cambridge Pro- 
fessor. Moreover it is admitted that this vowel r is 
closely related to the ordinary liquid r. May I pro- 
pose as the probable solution of the whole difficulty 
the following ? — It is well known that the two liquids 
r and I often lead to the disappearance of an adjoining 
vowel •, most persons would say to a metathesis of the 
vowel, a doctrine which I hold to arise from an inac- 
curate view of the matter, though this for the present 
is not important. Our own thorough for example 
appeal’s in the two shapes through Eng. and durch 
Germ. Again, in our provinces the form brid is at 
times used, where the prevalent language prefers bird ; 


Digitized by Google 



QVAERITVK. 


275 


80 iireAty and perty coexist. The Latin too has truc- 
and toru-o-, and the Greek 6paaos and dapvos, with 
but little distinction of meaning and no distinction of 
origin. In such cases it is convenient to have a 
notation which will readily adapt itself to the two 
varieties of pronunciation ; and on this principle it 
would not have been unwise to employ such a form 
as hrd, prtyy to represent at once bird and hrid, perty 
and pretty. ' The Slavic languages are not less given 
to such varieties than others ; and accordingly words 
without any represented vowel occur in the Bohemian 
vocabulary, ea krt ‘mole,’ Tcrk ‘neck,’ hJh ‘blockhead,’ 
wlk ‘ wolf.’ Yet Dobrowsky does not on this account 
class r and I with the vowels of the language. Pos- 
sibly the habit of virtually dropping the letters r and 
I, as in the case of bird in the mouth of a Londoner 
{bod), and tdik, calm generally, as well as the Fr. 
meilleur, may halve had its counterpart in India, and 
‘ so have lent some encouragement to the doctrine that 
they are vowels. 

But to return' to the ordinary vowels : if a language 
is limited to” three s3rmbols for their representation, it 
is a matter of course that a shotdd have a first pre- 
ference, because, lying in the middle of the series, it is 
for that very reason the easiest to pronounce, and con- 
sequently the most common ; and after a the vowels i 
and u have the next claim, as occupying the two 
extremities. 

It has also been urged that the Sanskrit alphabet 
has a special claim to our consideration in its philo- 
sophic completeness. But this claim is open to grave 
doubt, seeing that it appears to have been without 
any character for the sound, if indeed it possessed the 

T 2 


Digitized by Google 



276 


(^^'AEE1TVK. 


sound itself, that is heard in the initial consonants of 
our English thin and thine, fat, vat, in the two con- 
sonants of the Fr. juge and the final of the German 
einfach. On the other hand, it appears superfluously 
rieh in its ten asperates, distributed through the 
so-called gutturals, palatals, cerebrals, dentals, and 
labials ; that is, if our informants be right in pro- 
nouncing these asperates as we pronounce the italic 
consonants of blocJt^ouse, lo^/iouse, cqacA/iouse, \mdge- 
Aouse, cartAouse, guardhouse, chopAouse, clufchouse. 

If such be the correct pronunciation, the non-asperate 
character together with the simple h might surely 
have sufficed.. 1 have also assumed that ^ (va of 
German Sanskritists) corresponded to an English w. 
But if it really be a v, then a to is wanting ; if it be * 
at one time a v, at another a w, then we have another 
defect in the alphabet, two uses of a single symbol. 
But these very difficulties about the pronunciation 
seem to be valid reasons why we should select our 
primary facts from the known sounds of living tongues, 
rather than draw from alphabets of ancient date, no 
matter how venerable, in which the problems of pro- 
nunciation must to a considerable extent be full of 
difficulty, if not insoluble. 

The second main-heading in Bopp’s work is 'On 
Roots’ (Von den Wurzeln). As regards the pre- 
liminary discussion which treats of the distribution 
of languages into classes, I will confine myself to the 
remark, that as in the preceding chapter, so here again 
the author appears to have been led astray by the 
consideration of written language. No doubt the 
Chinese is to the eye monosyllabic. To the ear not so ; 
for it is well known to those who have learnt to speak 


Digitized by Google 



QVAERITVK. 


277 


the language in China itaelf, that it abounds in disyl- 
labic and polysyllabic words, whose unity, as with us, 
is denoted by the possession of a single accent. Thus 
Bopp is simply wrong in his statement of facts about 
the Chinese language (§ 108, p. 201, note) ; and again 
his definition of the Semitic family as one having 
disyllabic roots is at variance with the doctrine, now 
maintained by many of the first Hebrew scholars, that 
these apparent roots are in truth secondary forms. 
And indeed the Hindostani furnishes an instructive 
parallel, for here too it seems the existing verbs cannot 
be reduced to forms of less than two syllables, until 
we pass from the limits of the Hindostani to the 
parent Sanskrit. 

I must also point to another instance of error simi- 
larly caused. The peculiar notation employed for 
Hebrew words, in which symbols for consonants play 
the most important part, and the habit of denoting 
variations of meaning to a great extent by mere varia- 
tion of vowels, as katul ‘ killed’ with a fern, ktul-ah, 
and kote.l ‘killing’ with a fem. kotl-ah (§ 107, p. 196), 
have together led Bopp and his followers to call the 
consonantal combination ktl the root of the verb in 
question, although this combination is for the ear on 
absolute nullity. Nor is he himself blind to this 
inference, for he expressly says : “ A Semitic root is 
unpronounceable.” As well might he, with the English 
words bind, hand, bond, bound, bundle before him, set 
down as the root of this English verb the letters bnd. 

But I pass to a graver matter, and one that aSects 
the whole texture of the book. The German philo- 
loger, departing from the course marked out by his 
Indian authorities, refuses to accept the doctrine that 


Digitized by Google 



278 


QVAERITVB. 


all words are traceable back to verbs. Accordingly he 
divides the roots of the Indo-European family into 
two classes. ‘The main principle of Word-buUding 
in this class/ says he (§ 109 a, p. 203), ‘appears to 
me to lie in the union of verbal and pronominal roots, 
which together constitute, as it were, the life and 
soul’ (of the language). Poetical escapades of this 
kind naturally excite a suspicion of weakness in a 
theory. I propose then to examine this doctrine of 
pronominal roots in some detail. It is one that is also 
maintained by Prof. Max Muller in his ‘ Lectures on 
Language’ (p. 272,' &c.). His nomenclature indeed is 
slightly different from that of Bopp’a To ‘ verbal’ he 
prefers the term ‘predicative,’ and instead of ‘pro- 
nominal’ he talks of ‘ demonstrative’ roots ; but sub- 
stantially the two writers agree. As Prof. Max Muller 
is somewhat more definite than his fellow-countryman 
in his statement on this subject, I will quote a few 
lines from him. ‘ If they (our primitive ancestors),’ 
says he, ‘ wanted to express here and there, who, what, 
this, that, thou, he, they would have found it impos- 
sible to find any predicative root that could be applied 
to this purpose.’ And hence he says soon after ; ‘We 
must admit a small class of independent radicals, not 
predicative in the usual sense of the word, but simply 
pointing, simply expressive of existence under certain 
.... prescriptions.’ I accept the challenge implied in 
the first of these paragraphs, or rather accepted it 
many years before it was given, for already, in 1 84 7, 
in the ‘Proceedings of the Philolog. Soc.’ (vol. iii. 
p. 56) I put forward the theory that such a verb as 
our own ‘ ken ’ or ‘ look ’ as an imperative would 
supply what was wanted. 


Digitized by Google 



QVAERlTVll. 


279 


In the paper to which I refer the problem was 
considered in considerable detail, alike from the 
formal and logical points of view. Thus, as regards 
the mere shape of the words, I showed that pro- 
nouns of the tliird person exhibited an initial gut- 
tural in pretty well all the languages of Europe and 
Asia from the Atlantic to the Pacific, and from the 
Mediterranean to the Arctic Sea. On the other 
hand I produced similar evidence for the presence 
of a final nasal, and so accounted for the form of 
the Sanskrit him, which is set down as the ‘ dh^ltu’ 
of the relative, but by its final letter has been, I find, 
a stumbling-block to Sanskritists. In short, I con- 
sidered that a syllable ken, or something like it, 
appeared to bo the basis of pronominal words of the 
third person, including in that term demonstratives, 
relatives, and interrogatives, which I held to be of one 
stock. On the other hand I regard this basis of pro- 
nouns to be one with our English verb ken ‘see.’ 
But of course I could not rely on our English language 
alone, or even its German congeners. As ken, or if it 
be preferred con, is the simple root whence comes our 
derived verb k{e)n-ow or k{o)n-ow, in precise agreement 
with the verbs bell and bellow, so the root in question 
virtually exists in all those languages which possess 
a representative of know, as Latin with its gnosco, 
Greek with its ytyvaaKw, and Sanskrit with its jnd. 
Nay, the Latin itself has traces of the simpler verb 
gon. I refer first of all to the participles a-gn-itus, 
co-gn-itus, which come from stems Orgon-, co-gon-. 
From agnosc- and cognosc- we must have had 
agnotus cognbtus. Secondly, cdtus ‘shrewd’ seems 
to be a participle from a stem cen, just as in Greek 


Digitized by Google 



280 


QVAERITVR. 


we find words ending in tf>aTot from <f>«v- ‘ kill.’ Over 
and above this I pointed to the suffix ce of Latin 
demonstratives, as hie, istic, illic, sic, nunc, &c. and 
the so-called interjection en ‘ behold,' as exhibiting our 
root ken in two fragmentary varieties, much as a 
particle of totally different origin yet identical form, 
the Homeric ksv, takes in Greek the several corrupted 
forms of Ke or ku and av. Further, as the range I 
claimed for the pronominal base ken extended to the 
Pacific, so I quoted from the Chinese itself a verb ken 
‘ see.’ But I failed to notice the simple verb in San- 
skrit Let me now supply this omission by producing 
the reduplicate verb in mi, chi-te<-mi ‘ I see.’ This 
verb Bopp himself identifies as regards root with the 
Sanskrit verb chit ‘ perceive, know,’ and this again 
with the Zend chin {Y. G. 109 b 2, Anm. p. 239), so 
that the change of ken to ket is no difficulty for Bopp; 
and T confirm this by the parallel case of the Latin 
pronominal form cit-ra, cit-ro, cit-erior, cit~imus. I 
am farther indebted to Bopp for a knowledge of three 
other analogues of my verb, quita or kita of the Philip- 
pines, the New Zealand kitea, and Malagash hita, 
words also signifying ‘ to sec,’ and identified by him- 
self with the Sanskrit iet (§ 87 2). Again, the root in 
its purest form is found in the Keltic family, as in 
Cornish gon, and Old Irish gen ‘ know ’ (W. Stokes’s 
edition of the Middle-Comish poem ‘The Passion,’ 
notes, p. 94). Lastly, the Lithuanian has the par- 
ticle kat ‘ see there.’ Thus the area of the verb is 
as extensive as that of the pronoun. On the side 
of form then there remains nothing to desire ; and 
as to meaning I would ask wliether any idea could 
be in better keeping with pronominal demonstratives 


Digitized by Google 



QVAKiUTViJ, 


281 


tliau that of ‘ see,’ ‘ look.’ The very word ‘ demon- 
strative,' which Prof. Max Muller selects for his defini- 
tion, suggests this interpretation ; and ho himself adds 
that their oflSce is ‘ to point,’ and so determine ‘ loca- 
lity.’ It would be more correct to say that it belongs 
to the finger to point and to the voice only to call 
attention to the finger’s direction by uttering the 
word ‘ look.' It is with this feeling that the French 
has formed its void and voild, and of these the latter 
is often cut down in rapid pronunciation to v’ld, an 
abbreviation which is in keeping with what has been 
seen in our assumed corruptions of ken. I have 
already pointed to Terence’s lucisdt hoc iam and 
Virgil’s nunc dextra ingeminans ictus, nunc ille 
sinistra, as instances where the most graphic mode 
of translation is to treat the two demonstrative pro- 
nouns as practically the imperatives of a verb with 
the signification of ‘ look,’ repeated if preferred, for 
both hie and ille when closely analysed turn out to 
be reduplicated forms. It may be observed too that 
our own is an example of a verb, look, cut down 

to what is little better than a particle. Nay, when 
we ourselves utter the word this or that, we do 
little more than invite the person addressed to direct 
his eye to some object at which we are pointing, so 
that in real power these words are equivalent to an 
imperative ‘ look.’ No doubt the mind is not at once 
reconciled to the identification of a verb with an 
adjective, much less to the declension of a verb as 
though it were an adjective. Yet if the Latin ecce 
‘ behold ’ is a verb, and few will venture to deny it, we 
have a perfect parallel in such phrases as eccum me, 
eccos uideo incedere patrein et magistrum, as used by 


Digitized by Google 



282 


QVAERITVR. 


Plautus. For the full details of my argument I must 
of course refer to the paper itself But whether my 
theory be right or wrong, I trust I have said enough 
to show that Prof Max Midler’s broad denial of the 
possibility of finding a suitable ‘ predicative ’ root is 
untenable. 

On the other hand let us look at the general theory 
of roots, whether ‘ verbal’ or ‘ pronominal,’ as put for- 
ward by the German school. Bopp indeed puts aside 
for the most part the question of the origin of words, 
as not falling within the scope of his work, but Prof 
Max Muller speaks somewhat more definitely on this 
subject Yet his views, I think, will not be found 
satisfactory to others, and seem not altogether satis- 
factory to himself, for, after touching on the topic at 
the beginning of his book, he • practically postpones 
the question to his last chapter, pp. 349 — 399, and 
even then he nearly reaches the end of the chapter 
before he comes to the point It is oidy in page 391 
that he says : ‘ And now I am afraid that I have but 
a few minutes left to explain the last question of all 
in our science — How can sound express thought? 

I find another reason for doubting whether he is a 
firm believer in his own theory. The said chapter 
begins with an admirable extract from a work of 
Dugald Stewart’s which spurns with contempt ‘that 
indolent philosophy which refers to a miracle whatever 
appearances, both in the material arid moral worlds, it 
is unable to explain.’ I say then that when Prof Max 
Muller transcribed these words, he had not yet given 
a thoroughly cordial assent to the view of language 
with which the* chapter ends ; for he himself, in his 
distres.s, practically summons to his aid the deits ex 


Digitized by Google 



QVAERITVR. 


283 


machina, first telling us (p. 392) that ‘man in his 
primitive and perfect state possessed the faculty of 
giving expression to the rational conceptions of his 
mind/ and then adding that ‘ that faculty was an 
instinct, an instinct of the mind as irresistible as any 
other instinct’ Further in a note he says : ‘ The 
faculty peculiar to man in his primitive state, by 
which every impression from without received its 
vocal expression from within, must be accepted as an 
ultimate fact.’ For myself, I can only look upon this 
last passage as a simple admission that he has no 
solution of the problem to ofier, while the preceding 
assumption, that language is the result of instinct, 
seems to savour of that indolent philosophy which the 
Scotch philosopher is quoted to condemn. Again, the 
assertion that language first came into play when man 
was ‘ in his primitive and perfect state,’ seems hardly 
consistent with the tone of the first lecture, in which 
he led his hearers to anticipate a very different con- 
clusion. That lecture begins with a justification of 
the phrase ‘ Science of Language,’ and then refers the 
origin of every one of our sciences to the agency of 
man, as stimulated by his ‘ wants,’ when society was 
yet semi-barbarous or half-savage ; and his argument 
further implies that all the sciences, including of 
course that of language, were things of gradual 
growth, beginning in what was humble and lowly. 
All this is surely at variance with his later theory, 
that ‘the 400 or 500 roots’ which are ‘the constituent 
elements’ of language are ‘ phonetic types produced 
by a power inherent in human nature,’ and ‘ exist, as 
Plato would say, by nature ; though with Plato we 
should add that, when we say nature, we mean by the 


Digitized by Google 



284 


QVAERITVR. 


liand of God.’ One cannot but think that such ex^ 
planations must have been intended for the class of 
people so well described by Prof. Max Muller himself 
(p. 364), those ‘who prefer the unintelligible which 
they can admire to the intelligible which they can 
only understand.’ 

But before I pass from his Lectures, I take the 
opportunity of commenting on two other kindred 
matters. In p. 351, having said that ‘man could 
not by his own power have acquired the faculty of 
speech, which is the distinctive character of mankind, 
unattained and unattainable by the mute creation,’ 
he confirms his proposition by a reference to Wilhelm 
V. Huml)oldt’s writings : ‘ Man is only man through 
language, but to invent language he must already 
have been man.’ This is a taking ai’gument, and one 
that would be thoroughly valid on the assumption 
that language must have been created, so to say, at 
one gush, like a metallic casting. But if we include 
in our view the possibility of a gradual and slow de- 
velopment of the faculty, such as Max MuUer himself 
in his first chapter assigns to the creation of all the 
sciences, including by implication the science of lan- 
guage itself, the whole difficulty is dispelled. On this 
theory the human mind and the faculty of speech 
react each on the other, and thus ‘the foundation- 
stone of what was to be one of the most glorious 
structures of human ingenuity in ages to come may 
have been supplied by the pressing wants of a semi- 
barbarous society’ (‘ Lectxires,’ p. 5). 

But there is another witer, and he not a German, 
who, as agreeing in one of the two phases of the 
Oxford Professor’s book, claims our attention. In the 


Digitized by Google 



QUAEllITVR. 


285 


‘Study of Words’ (p. 16) there stands the sentence: 
‘God gave man language, because he could not be 
man without it.' This seems to imply that language 
was contemporary with man’s creation. May I be 
permitted then to ask how this doctrine is to be 
reconciled with what I suppose will be allowed on all 
hands as a fact, that the primitive language must 
have been wholly wanting in terms for spiritual and 
metaphysical ideas, seeing that the roots of language 
in their first meaning are very generally held to have 
a special reference to the material world. Hence, if 
the said doctrine be well-grounded, at the very time 
that primitive man existed in the most perfect, the 
most spiritual condition, he was yet destitute, it would 
seem, of terms to correspond with all the sublimer 
elements of his mind. In saying that terms for spiritual 
ideas are generally traceable to a material origin, I 
have in view such ca6es as the derivation of anima 
‘soul’ from an ‘to blow,’ of spirit iiom spirare ‘to 
breathe,’ and of ghost as connected with gust, with 
gas, with yeast (‘ Lectures on Language,’ p. 387). But 
while I take these examples from Prof. Max Muller, 
I must demur to his derivation of soul. Gothic saivala, 
from saiv-s ‘the sea,’ and still more to his explana- 
tion that ‘the soul was originally conceived by the 
Teutonic nations as a sea within heaving up and 
down with every breath, and reflecting heaven and 
earth on the mirror of the deep.’ As I have said 
before, I am always alarmed when I find poetry doing 
duty for logic. Still, in reliance on more sober ex- 
amples, I venture to afiflrm again that the late for- 
mation of spiritual language is more consistent with 
the theory of man’s progressive improvement than 


Digitized by Google 



286 


QVAERITVR. 


with the converse theory of his degradation, — in other 
words, more consistent with the first phasis of Prof. 
Max Muller’s lx)ok than with the second. 

The ‘instinctive’ origin of language, as laid down 
in the ‘Lectures,’ might to some minds have sug- 
gested the inference that language ought then to be 
the same for all people in all countries, and that every 
infant at the outset of its little life would have been 
possessed of useful speech ; but a condition of things 
so much to be desired is sadly at variance with fact 
This difficulty however the author of the theory at 
once meets by a little corollary to his theory, that 
‘man loses his instincts as he ceases to want them.’ 
Yet in speaking of his ‘ demonstrative’ roots (p. 272) 
he seems to imply that the instinctive movement still 
retains its force. ‘ The sound to, or sa,’ says he, 
referring to the Sanskrit pronouns, ‘for “this” or 
“there” is ’ (note the present tense) ‘ as involuntary, as 
natural, as independent an expression as any of the 
predicative roots.’ It must be due to some unhappy 
idiosyncrasy, I suppose, that I myself feel not the 
slightest tendency to follow such an impulse, however 
natural, however involuntary it ought to be. If I 
want to say ‘this,’ I say ihis\ if I want to say 
‘there,’ I say there. I certainly do not say either sa 
or ta. 

But, admitting for the nonce the new doctrine of 
pronominal or demonstrative roots, let us consider the 
purposes to which they are applied by Bopp and the 
Oxford Professor. In the instances I am about to 
quote from these two writers I wish special attention 
to be paid to the habitual, almost universal, assumption, 
that if the conditions of outward form be satisfied, it 


Digitized by Google 



QVAEEITVR. 


287 


is unnecessary to enter into any logical proof of the 
appropriateneass of the idea. As the references on 
this head to Prof. M. Muller will be but few, those to 
Bopp numerous, it may be convenient to give prece- 
dence to the disciple over the master. In the index 
to the ‘ Lectures,’ under the word declension, I find 
the proposition that ‘most of the terminations of 
declension’ are ‘demonstrative roots.’ Again, in the 
text (p. 274), we are told that ‘the Latin word luc-s' 
is formed by ‘ the addition of the pronominal element 
«,’ and signifies literally ‘ shining-there ;’ and he goes 
on to say that by adding ‘other pronominal deriva- 
tives’ we get ‘ lucidm, luculentus, lucerna, &c.’ 
What these other pronominal elements are, or how 
they are fitted for the purpose, he deems it unneces- 
sary to tell us. So in p. 221 he says that ‘the short 
i of the Sanskrit locative hridi “ in the heart” is a 
demonstrative root, and in all probability the same 
root which in Latin produced the preposition in’ 
He goes on to deal with the formation of the genitive, 
dative, and accusative, but in a manner so misty to 
my comprehension that I fail to pick up a single idea, 
and can solely refer to his book, pp. 221 — 224. 

Bopp starts (§ 105) with the doctrine that the class 
of roots he calls pronominal ‘give origin to the 
pronouns, to original prepositions, to conjunctions, 
and particles.’ In § 115 he advances a step farther, 
claiming ‘ the case-endings as, at any rate for the 
most part, of like origin.’ Looking upon the nouns 
of language as the Personae Dramatis of the World 
of Speech, he holds that ‘ the original oflSce of case- 
sufioixes was to express the mutusd relations between 
these “ Personae ” in respect of place ;’ and with this 


Digitized by Google 



288 


QVAERITVR. 


feeling he asks ‘ AVhat class of words could be better 
qualified to fulfil such an office, than those which at 
once express personality and the idea of place, whether 
nearer or more remote, whether on this side or on 
that?’ Accordingly (in § 134, p. 277) the s of the 
nominative is referred to the pronoun sa ‘he, this, 
that,’ fern, sd ; (in § 156, p. 320) the m of accusative 
masc. and fem. to the compound pronouns i-ma ‘this,’ 
o-mu ‘ that ; ’ and the final t which presents itself in 
the neut. nom. and acc. of certain pronouns, as tat 
and kat of the Veda dialect, to the neut pron. ta, Gr. 
TO. Again, in § 158 the suffix d of the instrumental 
case is ‘ as he believes ’ but a lengthened variety of 
tlie pronoun a, and one with the prep, d ‘ to’ (Germ, 
on), a meaning however which one might have 
thought would be more in place in the accusative. 
In § 164, p. 329, the datival ^ <is said probably to 
belong to the demonstrative S, ‘ which S however is 
apparently only an extension of the stem a,’ — that is, 
the very pronoun which has already done duty for 
the instrumental. In § 179, t we are told is the 
characteristic of the ablative, and ‘ no one’ (I quote 
his own words) ‘ who has once acknowledged the in- 
fluence of prepositions on case-endings, can have any 
doubt in referring it to the demonstrative stem ta 
“ this,” which has already in the neut. nom. and acc. 
put on the nature of a case-symbol, and will presently 
be found supporting the character of a personal suffix 
in verbs ;’ so that Bopp seems to think that the fact 
of its employment in two duties is a reason for adding 
a third duty. Most people, I think, would have 
arrived at an opposite conclusion. In § 184, p. 378, 
and § 194, p. 393, the genitival suffix s is held to be 


Digitized by Google 



QVAERITVR. 


28D 


one with that of the nom., and so the same as sa ‘this,’ 
while the longer sulfix si/a of genitives is the Vaidie 
pron. s;/a , — that is, a compound of two pronouns, sa 
‘ this,’ and the relative ya. Lastly, the i of the loca- 
tive he identifies, like Prof. M. Muller, with the 
demonstrative i 

I might be charged with a want of fairness to 
Bopp if I omitted to report an argument by which 
he defends his theory as regards the nominatival s in 
the masc. and fern. In the declension of the simple 
pronorm sa ‘ this,’ he observes that it is only the nom. 
m. and f. that present the s, the neut. nom. and all 
the oblique cases having an initial t, just as in Greek 
we have o, v, with a mere asperate, but afterwards to, 
TOW, Tijt, TOW, &c., so that there is a peculiar fitness in 
the employment of this pronoun for the two forms 
for which he claims it. However, he subsequently 
damages his theory by admitting (§ 345) that origi- 
nally the s may have been carried through all the 
cases and numbers, excepting only the neuters, and 
quotes the Vaidie locative sasmin for tasmin, and the 
old Latin sum, sam, &c., for eum, earn, &c. And even 
this persistence in excluding an s from the neuter is 
at variance with his own statement (ibid.) that the 
Greek o-tjTes, arffitpov, stand for (ro-ejes, oo-rifiepov, 
which 0-0 he himself holds to be of the same stock 
with the Sansk. sa. 

Thus for all the case-endings it is enough with our 
author to find some pronoun signifying ‘ this’ or ‘ that’ 
or ‘ what,’ it matters little to him which, and to de- 
fend himsdf behind the position that case-endings 
are in their nature of a locative character. He fails 
to see that the pronouns in question are but pointers; 

u 


Digitized by Google 



290 


QVAERITVR. 


and define only position, and even then had no defi- 
nite meaning in the outset of things, until aided by 
the pointing fingers. He himself indeed admits 
(§ 371, p. 180,) that the same pronoun originally 
signified ‘ this’ or ‘ that,’ ‘ nearer’ or ‘ farther,’ the 
mind supplying the necessary limitation. But while 
the demonstrative pronouns at most define only the 
‘ here’ or the ‘ there,’ it is the special office of case- 
endings to deal with motion as well as rest, to talk of 
the ‘whence,’ and the ‘whither,’ as well as the ‘where.’ 
Nay, if Bopp’s system were valid, we might freely 
interchange all the case-endings. 

But I have yet two other objections to offer, which 
seem each of them fatal to his doctrine. In the first 
place, the form he assigns to the case-endings is, in 
most instances, a very late and degraded form. For 
example, the locative and dative, which I believe to 
have been of one origin, have assigned to them as 
suffixes nothing but the vowels 6 and i respectively. 
But the Latin in i-bi, ali-bi, utru-bi, exhibits a 6, and 
as the Greek habitually has ^ as the representative of 
a Latin b, there can be little doubt that the Homeric 
ovpapo-(f>i presents the suffix in a more accurate shape 
than the ordinary Sanskrit locative. There is still 
another letter to re-establish in its proper position, a 
final n; and Bopp himself admits that ovpavo<f>w is 
the older form whence ovpavo<fn was derived. The 
Latin nobis, vobis, by their long vowel, also betray 
the loss of an n, and still more accurately defined is 
the suffix in the Old Prussian dat. pi. in man-s (§215, 
p. 424). Nay, I cannot but suspect that the Sanskrit, 
in its masc. loc. tas-min, has also in the last three 
letters a satisfactory equivalent for the <f>tp or bin, for. 


Digitized by Google 



QVAERITVR, 


291 


on grounds independent of the present question (see 
Proceedings, iii. p. 66, note §, and iv, p. 30), I should 
claim tas, rather than ta, for the root-syllable of the 
pronoun, and this view is confirmed by several other 
cases of the pronoun. So too the Umbrian locative 
appears to have had a suffix tmn or mem (§ 200, p. 
400), and the Zend for the dative of the first personal 
pronoun has mai-hyd, the long a of which would have 
a satisfactory explanation in the disappearance of a 
nasal But, to take a more general survey of the 
question, I would object to the fragmentary manner 
in which the school of Bopp pursue the inquiry into 
the form of case-suffixes. Each case must originally 
have had a common form of its own, no matter to 
what declension a noun belonged, no matter what 
its gender; and again, it is easy to see in nearly 
every case that the plural and the so-called dual 
forms (which in fact are but varieties of plurals) 
contain, in addition to the case-suffix of the singular, 
a second suffix denoting plurality, either a nasal 
syllable, as in our ox-en, or a sibilant, as in ouf cows. 
Hence in our search for the full forms of case-suffixes 
we are entitled, and therefore bound, to include all the 
forms belonging to a given case without distinction of 
declension, or gender, or number. 

Then again, on the other side, Bopp appears to be 
unhappy in his dealings with his so-called pronominal 
roots. These also he has robbed, as it seems to me, of 
a final n, which readily interchanged as well with the 
liquid m as with members of its own dental class, 
t and s. Thus for the first syllable of the Latin is-to- 
I find a more satisfactory explanation of. the s than 
Bopp’s own theory (§343) that it results from ‘ a petri- 

u 2 


Digitized by Google 



•292 


QVAEKITVR. 


faction’ of the nominatival s of the simple pronoun 
is. But I go further. In his zeal for pronominal 
roots he seems positively to invent them, jis for 
example ma (§ :IG8), u (§ 1,002), and above all his 
favourite sma (§ 165, &c.), of which he makes a most 
abundant, but I fear most unsatisfactory, use. 

But it is a special office of Bopp’s pronominal roots • 
to supply a corps of prepositions, and accordingly he 
lays himself out for at least an easy solution of the 
problems likely to present themselves. The ideas of 
‘above’ and ‘below,’ of ‘before’ and ‘behind,’ of ‘in’ 
and ‘ out,’ stand in the relation of opposite poles to 
each other. The metaphor is Bopp’s own. Hence 
the demonstrative pronouns are admirably suited to 
act aj# the needful symbols for these ideas, and so, 
what is particularly convenient, as they signify at 
once ‘this’ and ‘that,’ ‘on this side’ and ‘on that 
side,’ from one and the same pronoun we may deduce 
prepositions of directly opposite powers (§ 995). Thus 
from the pronoun a, to take that first as exhibiting the 
most wonderful fertility, with the aid of various suf- 
fixes, whose meaning seems to be a matter of not the 
slightest moment, for he never stops to explain them, 
we have S. a-ti ‘ over,’ S. a-dhas ‘ under ;’ Lith. a-nt 
‘ up,’ Germ, eni, Lith. a-t ‘ to,’ ‘ back ;’ S. Ordhi ‘ over,’ 
‘up’ (§ 997), with Lat. ad ‘to;’ S. Orpi ‘over,’ ‘up’ 
{§ 998), with e7T»; S. a-bhi ‘to’ (§ 999), with 
Lat. amb or am ‘round,’ Germ, bei, and Lat ob; 
S. a-pa ‘from’ (§ 998), with a-iro, Lat o-6, Eng. o-f 
(the hyidiens are Bopp’s) ; and (§ 1,007) from a-pa 
itself, through an intermediate apara-s ‘the other,’ 
cut down to para, we have no less than five S. pre- 
positions, — viz. pra ‘before,’ prati ‘towards,’ pard 


Digitized by Google 



QVAEUITVK. 




‘ back,’ ‘ away,’ puras, and jjan. Of these again pra 
(insepar.) ‘ before’ has for its cognates vpo, Lat. pro, 
Germ. ver. Then y>raii (§ 1,008) is represented by 
irpoTt and irpos; while pard ‘back,’ ‘away’ (§ 1,009), 
gives us irapa; and through a second aphaeresis a 
prep, m ‘ back’ in some other language, which is one 
with the Lat. re ‘ back.’ So much for one extensive 
family, all the progeny of the tiny pronoun a ‘ this’ 
or ‘ that,’ including too at once ano and vapa, at once 
pro and re. 

To the S. pronoun u, if indeed such a pronoun exist, 
are to Ije referred, it seems, S. u-pa ‘ to,’ S. u-t ‘ up;’ as 
also the Gr. v-iro, Lat. suh, and the adj. v-r-repo-^, 
together with Germ, aus, Eng. out. To meet the little 
difficulty about the asperate of vvo and the s of sub, 
Bopp proposes two theories : ‘ The s is either a simple 
phonetic prefix or the remnant of a recently prefixed 
pronoun sa,’ which however, he adds, would be ‘ here 
devoid of meaning.’ 

The S. pronoun ana gives birth to S. anu ‘after,’ 
Old Pruss. and Slav, na ‘ up,’ and ava ‘ up ;’ also to 
S. ni ‘down,’ Germ, nie-der; also to S. ni-s ‘out,’ and 
perhaps to the Slav, i-su ‘ out,’ ‘ which may possibly 
have lost an initial n.’ — The loss is the more to be 
deplored, as we lose at the same time all resemblance 
between i-su and its parent ana. 

Thus Bopp has thoroughly fulfilled the promises he 
held out, as we have from the same sources words 
denoting ‘above’ and ‘below,’ ‘to’ and ‘from,’ ‘back- 
ward’ and ‘forward,’ ‘absence’ and ‘presence,’ ‘up’ 
and ‘ down.’ And then how magical the changes. 

With this wonderful manufacture by the Bopp 
school of prepositions and case-endings from pro- 


Digilized by Google 



294 


QVAERITVE. 


nominnl roots, it may be useful to contrast a few 
specimens which may show the possibility at least of 
deducing prepositions and case-endings from verbs. 
Thus, to commence with a quotation from one of 
Bopp's own followers, we find in the ‘ Lectures’ (p. 
221): ‘The instrumental (in Chinese) is formed by 
the preposition y, which preposition is an old root 
meaning to use.’ So in a little paper of my own 
(Proceedings, vol. vi. p. 120) it is stated on Preraare's 
authority that the syllable commonly used in Chinese 
to denote the genitival relation, tci, -ia at times em- 
ployed as a verb equivalent to the Latin prqficisci. 
Again the Sanskrit inseparable preposition ni, Lith. 
nu ‘ down,’ is to be identified with the Lat. vb. nu-, Gr. 
vev-, ‘ lower,’ ‘ hold down,’ and the Chinese ni ‘descend.’ 
In the French chez, Ital. cam, and in our own through, 
Germ, durch and dur, we possibly Imve prepositions 
formed from substantives, — viz. the Lat. casa ‘house,’ 
and Germ, thiir, Eng. door, Gr. 6vpa. So little is it 
necessary to invent pronominal roots, as the source of 
prepositions. 

On Bopp’s derivation of partichjs from pronominal 
roots I must be brief. That words denoting ‘yes’ 
should be derived from pronouns signifying ‘ this’ can 
surprise no one. Thus we assent at once to such a 
derivation of the Lat. sic and ita and si of the 
French, &c. But Bopp is bolder j he hesitates not to 
deduce the S. na ‘ not’ and Lat. ne ‘ not’ from his 
pronominal stem na ‘this or that ;’ the Greek /iij ‘not’ 
from his stem ma; and the Greek ‘a privativum’ 
from a ‘this’ (§ 372, 1, p. 180). And here again he 
relies on his old doctrine that as such pronouns arc 
qualified to denote alike ‘this’ and ‘that’ (‘dieses’ 


Digitized by Google 



QVAERITVIi. 


295 


und ‘jenes’), in the second of these senses they may 
well represent negation, for what is there is not here. 
It is somewhat unfortunate that the pronoun a has on 
his own showing a marked tendency to express pre- 
sence (§ 366), as a-tra ‘here,’ a-tas ‘from here,’ ordya 
‘to-day.’ Nor is this to be set down as. a late inno- 
vation in the life of Sanskrit, for its position must 
have been already well established before the breaking 
up of the primeval language, seeing that (to use his 
own illustrations) it is found in the old Irish a-nochd 
‘to-night’ of the far west, and in the Ossetic a-hon 
‘ to-day’ of the far east. But be this as it may, the same 
pronominal a, once ^rmly possessed of negative power, 
is deemed by Bopp a fitting symbol for past time. 
‘ 1 hold the augment,’ says he, ‘ the initial a in Orbhav- 
am “ I was” for example, and so corresponding to the 
syllabic augment « of e-Timr-ov, &c., to be in its origin 
identical with the a privativum, and look upon it as 
expressing the negation of present time.’ Nay even 
in such forms as legS-bam (the division is Bopp’s) he 
once thought the long quantity of the middle vowel 
was referable to a suffixed augment, but his confidence 
in this theory was ultimately shaken (§ 527). 

Even among the verbs he is inclined to think that 
his pronouns play a part over and above their use in 
the personal endings. Of the suffixed t in tw-t-w, 
V in haK-v-w and BetK-p-v-fu, av in Xa/tfi-av-ay, he speaks 
with the greatest hesitation, yet still (§§ 494, 496) ‘the 
most probable explanation’ is that they are one and all 
of pronominal origin, their office being ‘ to convert the 
abstract of the verbs in question into a concrete.’ Nay 
even the so-called connecting vowels, as in (fiep-o-fitv, 
tbep-e-re, must be ascribed, he thinks, to a similar origin 


Digitized by Google 



QVAERITVR. 


•29G 

(§ 500), and indeed to our old friend a, for the o and 
e of the Greek verbs just quoted are represented in 
Sanskrit by an a. 

I now leave the pronominal roots with a strong im- 
pression on my mind that Bopp has failed to derive 
from his theory anything that adds to the value of his 
book. Even in his other division of roots I cannot 
divest myself of a fear that he has been wanting in 
eaution. In § 109 b he gives us a list of thirty-two 
root- verbs. In looking over these I find at least 

fourteen which I have little doubt are secondary, that 
is, derivative verbs, and eight others that have been 
shorn of their fair proportions, having lost an initial or 
a final consonant, or both. On the present occasion I 
cannot deal with more than a few of them, but ta 
avoid all suspicion of undue selection, I will take a 
batch that follow one another, those which stand 3d, 
4th, 5th, and 6th in his series. The verb gn^ {or jnd} 
is of course the Lat. gnosc-o, Eng. know, but in these 
verbs all that follows the liquid constitutes a suffix, 
while our obsolete English vb. ken, or rather con, ex- 
hibits the simple verb ; and, as I have already noted, 
the Latin participles a-gn-itus and co-gn-itus are 
deduced from compounds, not of gnosc-o, but of a 
primary verb gon, corresponding to our con. The 
4th in the series, vd ‘blow,’ has sufiered curtailment 
of its final consonant, and is really one with the 1 7th, 
an ‘ blow,’ which has lost its initial consonant, the two 
being truncated forms of a fuller van which appears 
scarcely altered in the Germ, wann-en, and is the 
parent of our winn-ow, wind, and fan, as also of the 
Latin vannus and ventus. This double corruption of 
van to vd and an would be exactly parallel to my 


Digitized by Google 



QVAEUITVR. 


297 


assumption that the Lat. ce and en come from ken. 
The 5th stA, Lat. sta, though very generaU}' set down 
as a root-verb, haa a sufl&x, or rather the remnant of 
a suffix, in the a. The proof of tliis I find in the 
Latin sist-o as compared with gign-o, yiyv-o/iai, fttfip-co, 
wtiTT-o), for as these are admitted to be reduplicated 
forms of ytp, nev, irer, so sist implies a primitive set, or 
something like it. To this primitive I assign the idea 
of ‘ stop,’ a verb which is itself probably of the same 
stock, st-op ; and I quote in support of this translation 
the familiar siste viator or better still s. aquam of 
Virgil, s. lacrimas of Ovid, s. alvom of Pliny : I say 
better, because there is in these phrases no trace of the 
upright position, which eventually attached itself to so 
many of the derivative forms. I may be asked here 
whether I propose to connect the assumed root set 
with the sed of Lat. sed-ere, sld-ere (for seid-ere), &c. 
and our own set, sit. My answer to this is at present 
neither yes nor no ; but on the logical side I see no 
difficulty, as we ourselves have the phrase ‘ to set up,' 
equivalent to the Lat. statuere. Again, if I am asked 
to account for the fact that sta- and its derivatives 
eventually possessed as an important part of their 
meaning that of standing or the upright position, I 
think I see two explanations. First the compound 
a-s<o- in Plautus has the simple notion of ‘standing 
up’ rather than that of ‘standing near,’ so that the 
preposition is an (= ava), as in an-hela-re ‘to send 
up a blast of air,’ Or-scend- ‘ climb up,’ rather than the 
familiar etc? ‘ to or near.’ It should be noted too that 
it is precisely before an initial s that the Greek ava, 
commonly reduced to av, or rather ov, in the Aeolic 
dialect, becomes further reduced to a or o (Ahrens, 


Digitized by Google 



298 


QVAERITVR. 


De Dialectis, 28, 1). The assumption that astare was 
in the end cut down to stare, has its parallel in our 
own truncation of arise to rise, for arise is the 
original form. This theory further explains in a 
thoroughly satisfactory manner the prefixed vowel of 
the Fr. Stat, Slats, Stablir. But independently of this 
argument, if the original notion of stopping be con- 
sidered in connexion with man, and it is of man that 
we commonly speak, the first result of stopping is 
standing. 

The 6th verb i ‘ go,’ though found alike in Greek, 
Latin, and Sanskrit, I believe to be doubly corrupted. 
Already it-er, com-it-ium, in-it-ium, ex-it-ium, comes 
(them, com-it-), pedes (them, ped-it-), claim a final t 
for the root, and the forms so familiar in Plautus, 
per-blt-ere, inter-hit-ere, red-hlt-ere, praeter-bU-ere, 
e-bit-ere (the last in Plant. Stic. 608, according to the 
pahmpsest), exhibit an initial b. I have marked the i 
as long on the uniform authority of Plautus, though 
Forcellini hastily assigns a short i to these words. 
Then as regards the simple verb, Eibbeck has done 
well to follow the guidance of Fleckeisen in exhibiting 
haetere as the reading of Pacuvius in vv. 227 and 255. 
Thus bat, the root of baetere (as edd of caedere), is 
the Latin analogue of jSat* in patv-a, and so only a 
variety of udd ‘ go,’ whence the imperfect tenses udd- 
ere, &c. We have here an explanation of the appa- 
rent anomaly in the corresponding French verb, 
which unites in the same conjugation a stem va 
and a stem i, — these, although wholly different in 
form, being in origin one, — as je vais, tu vas, il va, 
with j’irai, &c. These two verbs sta- and i- may 
indeed be pointed to as containing the best evidenee 


Digitized by Google 



QVAERITVR. 


299 


of the close intimacy between the Greek, Latin, and 
Sanskrit languages ; but it is in Greek and Latin, not 
in the Sanskrit, that we find the truest forms of the 
two roots. 

If it be replied to what I have here urged, that the 
Indian grammarians, when they put forward a so- 
called ‘ dli4tu,’ do not claim for it the honour of being 
an ultimate root, nay, that they apply this term to the 
base of any verb, though it be doubly or even trebly 
a derivative, I still contend that Bopp applies to his 
words the very name ‘roots’ (Wm’zeln), and that his 
whole argument implies that the verbs so called are 
ultimate forms. 

It would not be right to be wholly silent on his 
treatment of matters connected with the conjugation 
of verbs, but I must limit myself to the use he makes 
of the so-called verb substantive, whether as or hhH 
‘ be,’ though I may refer also to similar proceedings 
on the part of Professor M. MuUer. That this verb 
is employed in the processes of conjugation I of course 
do not deny, for I have myself sought to explain many 
forms by means of it. For example, I have contended 
that such phrases as ‘ I am a-dining,’ ‘ I am from 
dining,’ ‘ I am to dine,’ are found in many languages 
besides our own as formula? of presents, imperfect or 
perfect, and of future verbs; but then it is in the 
prepositions a (Ang.-Sax. an), from, and to that I find 
the essential part of the tense-idea. Indeed the very 
fact of the verb ‘to be’ entering into all the three 
phrases is the best proof that it contributes but bttle 
to the notation. But Bopp and his pupil proceed with 
far greater boldness. Thus the latter (‘ Lectures,’ p. 1 74) 
teUs us : ‘ Bam in cantaham was originally an inde- 


Digitized by Google 



300 


QVAERITVK. 


pendent auxiliary verb, the same which exists in the 
Sanskrit hhdvami and in the Ang.-Sax. beom “ I am”.’ 
Again (p. 234) he says: ‘ In the Latin bo of amabo we 
have the old auxiliary bhil “ to become,” and in the 
Greek futures in aw, the old auxiliary as “ to be”.’ (See 
also Bopp, § 52(j and §§ 648, 656.) This is to give 
to the past imperfect and the future of the Latin the 
very same origin, so that the Romans, it would seem, 
thought it no inconvenience to confound the two oppo- 
site ideas of time. Let me note too that the author of 
the ‘Lectures,’ by quoting in the one case the first 
person of the Sanskrit verb and in the other the 
mere base or ‘dhd.tu,’ gives a deceptive plausibility 
to his argument, for one sees some resemblance to 
bam in bhavdmi and some resemblance to bo in bM. 
My own views on the formation of the Latin tenses 
am-ab-a-m and am-ab-o are given elsewhere (Trans. 
Philolog. Soc. 1856, pp. 308, 309). I will here merely 
repeat that I find the symbol of past time, not in ba, 
but solely in the final a of am<tb-a-m, just as I find it 
in the corresponding vowel of the Latin er-a-m, Gr. r)p 
(= eav) or e-Ti0e-a, and S. a-hhav-a-m. I have said 
that the two German Professors explain the a of \e(w 
as the substantive vb. ; but according to Bopp it is 
equally applicable to the aorist eXe^a (§ 542) and to 
the perfect rereXe-a-fiat (§ 569). Nay even the k of 
eBwtca and BeBwxa is deduced from the same source 
(ibid.), a change which will prepare us in some 
measure for a still bolder doctrine, that the strange 
k which appears in the Lithuanian imperative duki 
‘give,’ is also a variety of the s of the substantive 
verb (§ 680). As to the ofiice it performs in this 
place, as in the others, not a word is vouchsafed. 


Digitized by Google 



QVAERITVR. 


301 


As a final specimen of the sort of reasoning which is 
allowed in the explanation of tense-forms, I may point 
to a passage in the ‘Lectures’ (pp. 317, 318). From 
such phrases as ‘ I have loved,’ ‘ amatum habeo,’ it is 
inferred that the notion of ‘ habeo ’ is specially fitted 
to denote the past or perfect, the fact being that the 
essence of this idea lies in the dental suflSxes of 
amort-um and U»yed. And then, as something 
parallel, the writer quotes a Turkish phrase, which 
he tells us is literally ‘ Paying belonging to me,’ but 
practically signifies * I have paid.’ I fear his know- 
ledge of Turkish is not of the soundest, for at any 
rate the Latin phrase ‘ soluendum est mihi ’ and the 
English ‘I have to pay’ sound more like future than 
past tenses. 

I shall conclude my comments on the ‘ Vergleichende 
Grammatik’ with a brief notice of the free use made by 
Bopp of grammatical figures as they are called, and 
these too of the very class which the soberer philo- 
logers of late years have been disposed to reject as 
inadmissible, except in rare cases — I mean the figures 
which imply an extension of words, whether at the 
beginning or end or within the body. Bopp’s 
much-used terms vorschlag, einschiebung, and zusatz, 
strengthened occasionally by the epithet unorgan- 
i^che, stand in the place of our old friends prosthesis 
(or prothesis), epenthesis, and paragoge. To the 
curtailment or compression of words no reasonable 
objection can be made, as it is the general law of 
language that forms should be abbreviated. 

I propose to take the said figures in order. 

Prothesis. The initial vowels of the words avep- 
(V. G. 2d ed. vol. i. p. 650, note), ovopMj- (ibid. 


Digitized by Google 



302 


OVAERITVR. 


1st ed. p. 311, note), o<f>pi/~ (ibid.), ovvx- (ibid.), are 
declared to be inorganic additions. The first of the 
set is further declared to represent the Sanskrit nr or 
nara ; but unhappily for this doctrine the noun avep- 
happens to be the example given by Dionysius of 
Halicarnassus, when he is speaking of Greek words 
that originally had the digamma ; and as this letter 
V) habitually interchanges with an m in many lan- 
guages, there arises a strong suspicion that ^av-ep- has 
its root in the first syllable, and so is identical with 
our own 'man. This is further confirmed on the one 
side by the English corruption of man to one (pro- 
nounced with a digamma) in such forms as one says 
and no one, compared with the German man sagt and 
nie-mand, and on the other by the Greek compounds 
voi~fiavtap and Ava^i-fiavSpot compared with arvy-avap 
and A\e^-avSpo$. 2. As ovop,aT- is always held to be 
one with the Latin nomen, and as this, being a deriva- 
tive from nosco, must originally have had an initial g 
(cf. co-gnomen, a-gnomen), we are driven to an older 
yop-ofMT-, of which yov alone is radical. Indeed Bopp 
himself in his Glossary (s. v.) deduces the Sanskrit 
naman irom. jnd. 3. Oil>pv being compared with the 
Sanskrit hh'rd (gen. bhruv-as) is pronounced guilty of 
having in its first vowel something to which it is not 
entitled. But let us rather compare it with our own 
eye-brow, to which eye contributes no small portion of 
the meaning. Surely then if a reasonable explanation 
can be given of the Greek word, such as shall include 
the idea of ‘ eye,’ we shall have what is more satis- 
factory. Now the most familiar root-syllable for ‘eye’ 
or ‘ seeing’ is in Latin oc {oculus), and in Greek with 
the usual letter-change oir {oirrofiai). But before an 


Digitized by Google 



QVAERITVR. 


303 


asperated letter ott will of coiirse become o^, as ui 
o<f>-6a\fj,ot. I suggest then that o<f>pv- stands for 
or I should myself prefer to say o^pv-, seeing 
that the Greek language habitually drops an initial 
labial when followed by p. Thus we have ptjypvfit 
rather than ^priy-wfu, Eng. break, and pay- rather than 
Fpay-, Eng. berry. 4. The noun I have little 

doubt is to be divided as here marked ; and I say so 
partly on the evidence of the Latin unguis, ung-ula, 
uncus, and the Irish ionga, partly because is a well- 
established Greek suffix, as seen in op-vx~ ‘dig’ {opvaaa), 
the sb. hi-<op-vx~ ‘ a trench,’ and virtually in op-(u)x-o- 
‘ a trench’ (especially for vine-planting), and so closely 
related to the Lat. or-d-on-, which has precisely the 
same for its first and original meaning. Compare too 
for suffix ^oarp-vx,-, Poarp-vx-o-, fiorp-vx-o-, as well as 
Porp-v. Indeed most nouns in u have lost a final 
guttural, as the Latin genvr, metu-, anu-, contrasted 
with genuc-ulum (Eng. knuck-le), metuc-uhsus, anic- 
ula. I might also have included the suffixes vy and 
VK, of irrep-vy-, kuX-vk-, as of the same origin with vx. 
I am myself too further moved by the long-established 
belief in my own breast that words with an initial n 
have generally sufiered decapitation. 

. Epenthesis. This doctrine is called in aid by Bopp 
not imfrequently, but especially when dealing with 
the genitive plural of certain vowel-ending Sanskrit 
nouns (^ 246, 249), which he says ‘insert a euphonic 
n between the ending and the stem.’ Among the 
instances he gives of this ‘inshoving’ are ah)d-n-dm 
‘equorum,’ tri-u^m ‘trium,’ sAnA-n-dm ‘filiorum.’ 
And he notes it as something very remarkable that 
the Zend, the Old German, Old Saxon, and Ang.- 


Digitized by Google 



304 


QVAEKITVR. 


Saxon exhibit a similar peculiarity. Surely then he 
ought to have asked himself whether this n may not 
be the substantial part of a genitival suffix. Had he 
done so he would have found, I think, abundant evi- 
dence in his own and other cognate languages. I have 
myself long been satisfied with this explanation of the 
en of the German compounds mond-en-licht, has-en- 
lage, and our own earth-en-ware, Ox-en-ford, Buck- 
en-ham and its equivalent Buck-ing-ham, as well as 
the adjectives wood-en, lin-en, silk en, &c. And then 
again we have in as a genitival suffix in Gaelic, as 
bo-in from hd ‘ cow.’ It is the more remarkable that 
Bopp should have- failed to hit this explanation, when 
he himself interprets (§ 248) the Mm of the Sanskrit 
tS-sdm ‘horum,’ td-Mm ‘harum,’ as containing a double 
suffix, of which s represents the genitival element so 
familiar in the singular. Secondly, in § 97 and again 
in § 727, note, he further teaches that while a final n 
in Greek has often originated in a final s, such inter- 
change is confirmed by the Prakrit On this view 
tS-s-dm and ah)d-n-dm would go well together. 

Again, as an n is ever apt to become silent before 
an s (cf. eis, j(apui9, rv(f>deis, cosol, toties), it would have 
been more prudent perhaps, when dealing with the 
suffix of the dat. pi. in Old Prussian, mans, not to have 
considered the n as inorganic, on the sole ground 
that mas would agree better with the Sanskrit hhyas. 
His illustration too from the Latin ensis and mensis 
beside the Sanskrit asis and mdsis involves a similar 
assumption. 

But we need not hunt up particular instances, 
when we find a wholesale manufacture of epenthetic 
vowels established by A. Kirchoff in the ‘ Zeitschrift’ 


Digitized by Google 



QVAERITVB. 


305 


(i. 37) and K. Walter (ibid. xi. 428). Thus epeptvOot 
and opoffof and the Old German araweiz of like mean- 
ing are convicted of having stolen the vowel which 
follows r on the sole evidence that the Lat. ei'vum 
exhibits no such vowel. HXcAfrpov cannot be entitled to 
the vowel e, because forsooth the S. ari ‘ shine’ proves 
the original root to have been alk. Again, the Greek 
having the two forms opoyvta and opyvia, the former is 
declared to have a vowel that does not belong to it, in 
spite of the evidence of opey-to. Nay, even the long 
vowel of a\-a>-veK- is ‘eingeschoben.’ Walter’s argument 
turns chiefly on the assumption that forms ending in 
rk, Ik, rg, &c. are ultimate roots. Thus, according to 
him, <o\aK~, FwXa*-, avXuK-, a\oK-, all varieties of the 
same word, signifying ‘ furrow,’ come from a root 
valk = FeX*-. Now my own conviction, foimded on 
a long and wide examination, is that such verbs are 
all of them secondary. I do not believe in his sug- 
gestedde rivation of av\aK- from FeX/t- ; but if it were 
true, the Latin uel- (uello-) exhibits the verb in a 
simpler form. But it is enough to place beside each 
other such pairs of words as talk and tale, hark and 
hear, pluck and pull, sparg- and <nrup-, terg- and 
reip-, calc- and heel, stirk and steer, hoik and holl, both 
Scotch verbs signifying ‘ to dig,’ the latter of which is 
one with the Latin col- ‘ dig.’ 

Paragoge. Bopp’s instances of ‘ unorganische Zusatz ' 
are numerous, but I shall be satisfied with quoting the 
Latin genetric- ‘mother’ and iunic- ‘heifer,’ which are 
declared to have a c of this character, inasmuch as the 
Sanskrit janitrt (§ 119) and ydnt (§ 131) have no 
such letter. The Greek vocative ywat by the side of 
yvvMK-os, kc. might have suggested the possibility of 

X 


by Google 



306 


QVAKQITVK. 


» final K being lost ; and, again, the Latin vb, nutri-re^ 
being a denominative from nutric-, has suffered the 
same loss. 

His use of Metathesis however is carried to the 
greatest extreme. Indeed, the term ‘ Umstellung,' 
which is his name for this ‘ figure,’ incessantly presenta 
itself to the eye. I am one of those who believe the 
doctrine implied in these words to be carried to an 
unjustifiable extent by even the more sober of philo- 
logers ; but I will here confine myself to three 
examples selected from Bopp’s book, which I cannot 
but expect all persons will agree with me in condemn- 
ing. In § 308, p. 60, he takes in hand the Gothic 
adj. hanfa (nom. hanf-s) ‘ one-handed,’ and first pro- 
nounces ha to represent the ka of the Sanskrit 4ka 
‘ one.’ This assumed, he holds the residue nfa to 
stand for nifa. By transposition of nifa he then gets 
fani, which would correspond no doubt with all accu- 
racy to the Sanskrit pdni ‘ hand.’ This taken alto- 
gether must be admitted to be a strong proceeding } 
and a German philologer, in discussing a Gothic word, 
would have done well to cast an eye for a moment on 
the other Low German and kindred dialects. Had 
Bopp done BO, he would have found at home that for 
which he travels to the far East, viz. Old Norse, hnevi 
‘fist’ and Lowland Scotch, not to say Yorkshire nieve, 
Nay, Walter Scott (‘Guy Mannering,’ xxiv.) has : * Twa 
land-loupers . , . knevelled me sair aneugh or I could 
gar my whip walk about their lugs ;’ and, to quote 
from a more Southern dialect, Shakspere has : ‘ Give 
me your neif’ (‘ Midsummer Night’s Dream,' iv. 1), 
and : ‘ Sweet knight, I kiss thy neif’ (‘ Henry IV.,’ 
Part II. ii. 4). 


Digitized by Google 



QVAERITVK. 


307 


In voL i. 580, note, attention is drawn to an 
Armenian noun signifying ‘ man,’ of which the crude 
form is said to be aran. Of this the initial vowel is 
first discarded as a mere phonetic prefix, and then by 
‘ Umstellung’ ran is identified with Sanskrit nar or 
nr. Would it not be simpler and quite as justifiable 
to affirm that the Armenian aran was formed from the 
Sanskrit nara by reading it backward ? 

Lastly in his Glossary s. v. nakha ‘ nail’ we have 
the words ; ‘ bib, ionga fortasse litteris transpositis e 
nioga.’ The Greek ov-vx- should have prevented this 
statement 

In terminating my remarks on Bopp’s somewhat 
free and bold use of ‘ grammatical figures,’ I must he 
permitted to throw out the hint that if by any pos- 
sibility the Sanskrit forms just compared with the 
classical have been advanced to a dignity which is 
beyond their due — in other words, if they are, after 
all, the more degraded of the two — then all the diffi- 
culties which have presented themselves disappear. 
From the objectionable figures prothesis, epenthesis, 
and paragoge, we should pass respectively to aphae- 
resis, synaeresis or crasis, and apocope. In plainer 
English, instead of assuming words to grow and 
extend themselves, we should have nothing but abbre- 
viation, a principle which seems to recommend itself 
to the common sense of every one. A man need not 
be much of a phdologer to account for the abbreviation 
of caravan, forecastle, and cabriolet to van, foxel, 
and cah. 

In concluding these remarks, the length of which 
find their only excuse in the importance of the subject, 
I must be permitted to say that I have written in no 

X 2 


Digitized by Google 



308 


QVAERITVK. 


spirit of hostility either to Comparative Grammar or 
to the Sanskrit language. On the contrary, fully 
believing that the science must be benefited when the 
philologer extends his views over many languages, 
especially in the older varieties, but to the exclusion 
of none, I sincerely trust that some of our own classical 
scholars will apply themselves with independence and 
diligence to the study of Sanskrit. My chief object 
in the present paper has been to check that slavish 
sequacity which has long interfered with the advance- 
ment of linguistic science, and I lay down my pen 
with something like a conviction that my readers 
will not so readily give their assent to such propo- 
sitions as the following. Prof. Max. Muller tells us 
(‘Lectures,’ p. 167) that ‘His (Bopp’s) work will form 
for ever the safe and solid foundation of Comparative 
Philology.’ Again (p. 216), ‘Comparative Grammar 
has weU-nigh taught ua all it has to teach.’ And 
another writer, if indeed it be another writer {Satur- 
day Review, Jan. 10, 1862), speaks of Comparative 
Grammar as ‘ a science which has always prided itself 
on the exactness and almost mathematical precision 
of its method.’ 


Digitized by Google 



309 


XIV. 

POSTSCRIPT. 

I HAD hoped that the arguments put forward in this 
paper would have drawn out some reply in print from 
the Sanskritists in Germany and England. Six years 
have now passed since it was printed. But, with one 
exception, they have been silent. That exception is 
Prof. M. Muller, w-ho, in the second series of his 
‘Lectures’ (pp. 13, 14), says: — 

‘ But while we are thus told by some scholars that 
we must look to Polynesia and South Africa if we 
would find the clue to the mysteries of Aryan speech, 
we are warned by others that there is no such thing 
as an Aryan or Indo-European family of languages, 
that Sanskrit has no relationship with Greek, and that 
Comparative Philology, as hitherto treated by Bopp 
and others, is but a dream of continental professors ; ' 
to which he appends as a note : — 

‘ See Mr. John Crawfurd’s essay On the Aryan or 
Indo-Germanic theory, and an article by Professor T. 
Hewitt Key, in the Transactions of the Philological 
Society, “ The Sanskrit Language as the basis of Lin- 
guistic Science ; and the labours of the German School 
in that field — are they not overvalued ? ” ’ 

Now the word ‘ others ’ is a plural, and the note 


Digitized by Google 



310 


POSTSCRIPT. 


naturally suggests the idea that Mr. Crawfurd and 
Mr. Key are included in the term. But Prof. M. 
Muller well knows that I have always accepted the 
Sanskrit language as a member of the Indo-European 
family, the study of which is important for linguistic 
science. Indeed, although my other engagements have 
rendered it impossible for me to acquire any direct 
acquaintance, much less a thorough acquaintance, with 
the language, I have read largely and with care what 
others have written on the subject ; and have not un- 
frequently employed the knowledge so attained in the 
explanation of Latin words' and Latin grammatical 
forms. I therefore here caU upon him to withdraw or 
to justify his assertion. Perhaps he will think this 
the more necessary, when I tell him that a friend, well 
known not more for accurate and refined scholarship 
than for caution and urbanity, on reading the above 
passage from his ‘ Lectures ' made the remark : ‘ I call 
that a suggestio falsi! 

But while this volume is going through the press, 
the ‘ North American Review’ (Oct. 1867, p. 521) 
brings me a paper Yrritten in a very different spirit, 
and claiming the more attention, as report from several 
quarters ascribes it to a distinguished Professor of 
Yale College. 

After reading this article, I rise with the satisfactory 
feeling that my inquiry into the doings of German 
Sanskritists has not been in vain, for the last para- 
graph of the review, so far as concerns my criticisms 
of Bopp, runs : — 


* As in the explanation of the Latin words Umert and Umtrare. 
(See Trans. Philolog. Soo. for 1866.) 


Digitized by Google 



POSTSCRIPT. 


311 


‘ In a considerable portion of the criticisms which 
Professor Key makes upon his (Bopp’s) works, the 
majority of comparative philologists, we believe, of 
the German or any other school, would be free to join, 
yet without abating a jot of the admiration and grati- 
tude which they pay to the founder of their science.' 
Again, in p. 530, he says : — 

‘ In two respects, especially, his (Mr. Key's) objec- 
tions are to be regarded as valuable protests, requiring 
to be well heeded, against modes of etymologizing 
which are too common among Sanskritists : namely, 
the over-ready referral to a Sanskrit root, of doubtful 
authenticity and wide and ill-defined meaning, of 
derivatives in the various Indo-European languages; 
and the over-easy persuasion that the genesis of a 
sufiix is sufficiently explained when it is pronoimced 
“ of pronominal origin.^’ ' 

And he then goes on to say : — 

‘ As regards the former point, we think our author 
entirely justified in casting ridicule upon the facile 
derivation of words meaning “water,” “earth,” “cow,” 
and the like, from alleged Sanskrit roots claimed to 
signify “ go.” ' 

Soon after he adds : — 

* As regards, again, the use of pronominal elements 
in explaining the genesis of grammatical forms, we 
deem Professor Key’s interpellations not less in 
place.’ 

And at the close of the same paragraph there 
occurs' — 

‘ Meanwhile, no one is to be blamed for feeling a 
kind of indignant impatience at seeing this and that 
ending complacently referred to such and such a pro- 


Digitized by Google 



312 


POSTSCRIPT. 


nominal root, as if no further erplanation of it were 
necessary to satisfy any reasonable person.' 

So far then we agree ; but there are points as to 
which he expresses somewhat vaguely a difference of 
opinion. 

It has been however often said that if two opponents 
were brought together face to face, an amicable dis- 
cussion would result in a belief that their differences 
were far less than at first supposed. This is a truth 
which applies, I believe, to the present case ; and 
perhaps a few additional words may lead to the same 
result. 

The reviewer opens his argument (p. 521) with a 
statement that ‘ the change of ground and of point of 
view which philological science has undergone during 
its later history amounts almost to a revolution, and 
naturally provokes the opposition of ancient opinion 
.*\nd of the prejudices engendered by it;' and he in- 
cludes me among the ‘conservative spirits who are 
under such influence.' I think that few readers of the 
present volume will think that he has been happy in 
his theory that I am scared by innovations in linguistic 
matters. 

Again, in p. 535 he observes that while the labours 
of the German school are overvalued by some, they 
are ‘ undervalued by those who, on account of faults 
of detail, reject the whole method, as well as by those 
who, having the acuteness to detect such faults, yet 
lack the sound learning and enlightened judgment 
which should enable them to adopt the method wher- 
ever it is truly valuable. And we fear that our author 
is to be ranked in the latter class.’ 

How he jumps to this conclusion I do not see. In 


Digitized by Google 



POSTSCRIPT. 


313 


the first place, it is scarcely correct to say that the 
objections urged by me are only ‘ faults of detaiL’ 
Many of those objections deal with the backbone of 
Bopp’s work, as, for example, his assumed explanation 
by pronouns of the case-endings of nouns, as to which 
the reviewer fully coincides with me. His space did 
not allow him, he says, to discuss more than a few of 
the difficulties raised by me ; and thus it happens by 
a strange piece of good fortune for me that in every 
one of these he is at one with me, except that he can- 
not altogether go with me in questioning the existence 
of pronominal roots as a separate class ; but he does 
not enter into any particulars as to the ground on 
which this qualified dissent is based. 

Nor is it enough to say that he was confined by 
want of space, for he deals with several minor matters 
which affect myself, but have little bearing on the 
main question. Thus it is a very unimportant matter 
whether I am right or not as to spelling ‘asperate’ 
with an e, which he attributes to ‘ a w him or a false 
theory.’ Perhaps he may change his view when he 
calls to mind the grammatical terms spiritus asper 
and spiritus lenis, or the Greek adjectives Ba<n>9 and 
y/rt\ot, applied in the same senses. Again, his lin- 
guistic peace of mind is sadly disturbed when he 
finds that T have the courage to assume that Finnish 
has a close relation as regards pronouns and gram- 
matical forms with the Indo-European family. Not 
long ago it was deemed an over-daring act to claim a 
connexion for the Keltic with this family; and the 
time win come when even Germans will be startled 
on coming across a European race whose inflections 
for the dual and plural of the verb run : — 


Digitized by Google 



314 


POSTSCRIPT. 


Dual : molsoimen, molsoiten, molsoikan. 

PI. molsoime, molsoite, molsoin { 
forms which bear a strange resemblance to what is 
seen in Greek grammars. Still greater perhaps their 
surprise when they come in pronominal declensions 
to : — 

mo ‘ of me,’ to ‘ of thee,' so ‘ of him.’ 

And again, when they find that as the Persian uses 
the letters m, t, s, as suffixes to substantives with 
the meaning of ‘ mine,’ ‘ thine,’ ‘ his,’ so precisely as 
suffixes of the same form the language I speak of 
presents them with, — 

parne ‘ son,’ parnam ‘ my son.’ 

nipe ‘ knife,’ nipat ‘ thy knife.’ 

aija ‘grandfather,’ aijabs ‘his grandfather.^ 
And the matter will perhaps be clenched, when they 
see before them mocum, tocum, socum, identical in 
meaning, stiU more than in form, with the Latin 
mecum, tecum, seciim} 

The language in question is the Lapp, one so nearly 
akin to Finn that the admission of one as in any 
way cognate with the Indo-European will insure the 
admission of both. The one can only enter the pri- 
tdleged gate arm in arm with the other. Thus it is 
no such absurdity as the reviewer supposes, to di-aw 
arguments 'from the Finnish, so far at least as- 
concerns the pronouns. 

But although I find Icen and cu as the two leading 
forms of the Finnish relative, and so appeal to them 
in support of my theory as to the origin of the third- 
person pronouns, that theory remains intact, even if 

* See Mr. Wedgwood’s paper, Trans. Phil. Soo. 1656, p. 1. 


Digitized by Google 



fOSTSCIUPT. 


316 


such additional evidence be put aside. For the 
Sanskrit testifies in my favour, and the other members 
of the Indo-European family as well. To this side of 
my argument the reviewer has nothing to oppose. 

I had forgotten to state how the reviewer treats that 
portion of my paper which is directed against Prof. 
M. Muller’s views. A few words wiU suffice on this 
head. After demurring to my implied assumption 
that ‘ accusations made to lie against these two (Bopp 
and Max Muller) will attach to the whole cause they 
represent’ (p. 528), he soon turns to my dealings with 
the latter (p. 529), saying : — 

‘ As regards our author’s other antagonist, Professor 
Max Muller, it is only in England that modem philo- 
logy is looked upon as so identified with his name, that 
a blot on the one must be presumed to sully the other.’ 

And then after general compliments to this writer 
he concludes thus : — 

‘ A notable example of his characteristic weaknesses 
is offered in his theory of phonetic types instinctively 
produced as the beginnings of human speech ; a theory 
which forms one of the counts of Professor Key’s in- 
dictment, and which we should not think of defending 
in a single point from the latter’s hostile criticism. 
Rarely is a great subject more trivially and insuffi- 
ciently treated than is that of the origin of language 
by Muller in the last lecture of his first series.’ 

Let me conclude then, in the absence of all other 
replies to my inquiries as put in the paper entitled 
* Qvaeritvr,’ with a statement that I am thoroughly 
ready to subscribe to the articles of linguistic faith in 
the Bopp school, so far as the reviewer states them (in 
pp. 549, 650), with this one qualification, that all or 


Digitized by Google 



316 


POSTSCRIPT. 


nearly all these truths had been discovered before 
Bopp entered the field of philology. 

‘ Bopp and his school have shown, beyond the reach 
of cavil, that the branches of Indo-European speech 
have sprung from a single stock ; that they are not 
independent growths, upon which certain common ele- 
ments have been ingrafted. They all count with the 
same numerals, call their individual speakers by the 
same pronouns, address parents and relatives* by the 
same titles, decline their nouns upon the same system, 
compare their adjectives alike, conjugate their verbs 
alike, form their derivatives by the same stiffixes.’ 

On the other hand I hold that Sanskrit does not at 
present deserve the high rank assigned to it in lin- 
guistic science, and this partly because no one has yet 
attained to a knowledge of the language at all com- 
parable in accuracy to that which the students of the 
two classical languages have reached ; secondly, be- 
cause as yet the Vedic language, which alone can 
pretend to a rivalry in antiquity with Greek, has so 
far been but little studied; and thirdly, because 
Sanskrit literature has no basis for linguistic study 
comparable in clearness of ideas to the Iliad and 
Odyssey. 

Farther, I believe that in pureness of grammatical 
forms both Greek and Latin have often a marked 
superiority over Sanskrit. But the true course for 
the philologer is to study aU these languages so far 
as the limited opportunities of each permit, and one 
at least of them thoroughly. 

^ I have myself doue something to complete this theory by the 
identification of the Greek 6i/yar(-(p-) with the Latin 
(Trans, for 1866.) 


Digitized by Google 



ADDENDA, CORRIGENDA, etc. 

Pcvge 6, line 24. For -ava-KoxKv- read ava-KOKKv-. 

Page 6, line 28. For -pay^wSe- read -pay^^e-. 

Page 13, line 1. Add after ‘Secondly,’ ‘the notion 
of “back” enters into adim- “take back or revoke,” 
as used in Dig. xxxviii. 4, 1, §§ 3 and 4 : “ Assignare 
autem quis potest (libertum) quibuscunque uerbis uel 
testamento uel codicillis uel uiuus. Adimere autem 
assignationem etiam nuda uoluntate potest.” ' 

Page 47, line 27. For ‘uninteresting,’ read ‘in- 
teresting.’ 

Page 49, line 4. Insert before ‘also’ the word 
‘ compare.’ 

Page 64, line 31. Add: ‘So Sallust (Jug. 18, 5) 
speaks of certain immigrants in Africa, who arriving 
by sea “alueos nauium inuorsos pro tuguriis habuere;” 
and in this way accounts for the shape of the Numi- 
dian huts : “ Adhuc aedificia Numidarum agrestium, 
quae mapalia illi uocant, oblonga incuruis lateribus 
tecta quasi nauium catinae sunt.” The same meaning 
of the adjective is seen in “ duratur nasus incuruus, 
cogimtur ungues adunci, fit bubo Pamphile,” Apul. 
Met. 3, 213.’ 

Page 66, line 28. Add: ‘So too in the French 
anivrisme, “ an aneurism.” ’ 


Digitized by Google 



318 ADDENDA, COERIGENDA, ETC. 

Page 82, line 1. For erhitten, read erhittem. 
Page 89, line 19. For ‘Garnet,’ read ‘Garnett’ 
Page 107, line 31. For verchieben, read ver~ 
schieben. 

Page 113, line 23. For irapaamjvo-, read irapaaicijvo-. 
Page 115, line 4. Add aa an additional paragraph : 
‘ So far I have dealt with the Latin per in the for- 
mation of compound verbs and compound adjectives ; 
but even in the ordinary use of the word in connexion 
with substantives the sense of “over” is placed be- 
yond doubt; and this more particularly in the text 
of Livy, as first (i. 26, 13): "Is (the father of the 
surviving Horatius) transmisso per uiam tigillo ueli^t 
sub iugum misit iuuenem secondly (xxvii. 32, 35) : 
“Ibi equus pilo traiectus quum prolapsum per caput 
regem effudisset thirdly (xliv. 19, 9): “Antiochua 
. . ponte per Nilum facto transgressus . . obsidione 
Alexandream terrebat;" fourthly (x. 19, 21): “per 
uallum, per fossas irruperunt;” fifthly (xxvi. 6, 2) :, 
“ elephantos trausgredientes in ipso uallo conficiunt. 
Quorum corporibus quum oppleta fossa esset, uelut 
aggere aut ponte iniecto transitum hostibus dedit ; ibi 
per stragem iacentium elephantorum atrox edita 
caedes.” So Madvig in his text ; but in his preface, 
to voL iL, part 2, he writes : “ Kecipienda fuerat 
Ussingii coniectura: ‘ibi super stragem iacentium.' 
Caedes hominum per stragem elephantorum edita 
nihil est.” I think I may now assume that there is 
no need for the conjecture of Madvig’s collaborateur. 
But the same use of per is seen in Catullus (20, 9) : 
“ Quondam municip^m meum dd tuo uolo pdnte Ire 
praecipitem In lutum p^r caputque ped^sque, &c. as 
also in Caesar (B. G. iii. 26, 5) : “ Hostes desperatis 


Digitized by Google 



ADDENDA, CORRIGENDA, ETC. 319 

rebus se per munitiones deicere intenderimt.” Perhaps 
too I ought to add from Plautus (Poen. 6, 12, Gep- 
pert’s ed.) : “ Ita repleuero atritate atr(at)ior multo ut 
siet Quam Aegyptini qui cortinam ludis per circum 
ferunt.” In the passages so far quoted the notion of 
“over" seems alone admissible; but there are many 
others in which the same translation is at least as 
satisfactory as that by through for example in 
Qu6 Castalia per struices s4xeas lapsu dccidit,” (Liv. 
And. 36 Ribb.) ; “ Dubil fauentem per fretum intro- 
cdrrimus,” (Naev. 59 R.) ; “Perque agros passim dis- 
pergit corpus,” (inc. fab. 168 R.) ; “Ponti per freta 
Colchos delatus,” (inc. fab. 182 R.) ; “ Rapiunt 
per undas currus suspenses, ” (ib. 196); “Ardua per 
loca agrestia trepidante gradu nititur,” (Pac. 272 R.) ; 
“Nunc per terras uagus extorris,” (Att. 333 R.) ; 
“Multa siti prostrata uiam per,” (Lucr. 6, 1262); 
“Transtra per et remos, &c.” (Verg. A. 5, 663); 
“Unetds saluere (they ran) per utres,” (Verg. G. 2, 
384). Still, as the Latin language had three prepositions 
of the same origin, per, tram, and super, it was to be 
expected that the meanings would for the most part 
be distributed between them; so that the notion of 
“over” might with many writers be limited to super. 
Lastly, to the verbs compounded with per in the sense 
of “ over ” add percurr-, as used by Terence.(Haut. iv. 
4, 11) in “ Curriculo percurre,’’ run over, run across 
(to Charinus’s).’ 

Page 119, line 10. Add as an additional para- 
graph : — 

‘ The instances to which Diez refers are “ Hygin, de 
condicionibus agrorum,” p. 118, 1. 6: pos legem 
datam ; and M. lun. Nissus, p. 294, 1. 6 : pos te 


Digitized by Google 



320 ADDENDA, COBRIOENDA, ETC. 

relinquas orientem ; to which may be added from the 
same class of writers, “ Casae litterarum p. 329, L 12 : 
Casa . . . pos si (= post se) Jinem habet, opposed to 
finis ante se habentem of line 3. So Munro in his 
Lucretius (4, 1186) writes poscaenia, and (4, 1252) 
in a foot-note prefers pos sunt to post sunt, the MSS. 
having possunt ; but his reference to a solitary pos- 
quam as the reading of the sole MS. in Liv. (xlii. 
10, 5) seems to have less weight, as postquam is a 
word of such frequent occurrence in Livy. Other 
instances are to be seen in Ribbeck’s prol. to Vergil, 
p. 442, and Schuchardt’s “ Vokalismusdes Vulgar- 
lateins, 1, 122.” Few words then are better estab- 
lished in the Latin vocabulary than the form pos/ 
Page 164, line 15. Add: ‘ mittere scriptam solet 
(for so MSS.), Pseud, iv. 2, 46 and to note 2 add : 
‘/ore (Pers. ii. 3, 6), fore (As. 214, 57).’ 

Page 165, line 2. Add; * uiuere (Glor. iv. 6, 60), 
perdere (Cure. iv. 2, 18), adducere (A& ii. 4, 32), 
noscere (Rud. ii. 3, 59);’ and Z. 10, add: ‘ reddere 
(Amph. i. 1, 52).’ 

Page 167, line 14. Add: ‘and the German par- 
ticle hin of like meaning.’ 

Page 210, line 33. After Liv. xxvi. 9 add: ‘pos- 
tremis genu nixis, Liv. xliv. 9, 6.’ 

Page 211, line \L For inuenis, read iuuenis. 

Page 211, line 26. Add ; ‘ clift, as in Netherclift 
by the side of cliff, and perhaps graft by the side of 
graff! 

Page 212, line 25. For ‘imperfect particles,’ read 
‘ imperfect participles.’ 

Page 213, line 1. After 7, 4, add ; ‘ So too the 
Danish preposition is i, not in.’ 


Digitized by Google 



ADDENDA, CORRIGENDA, ETC. 321 

Page 215, line 2 . Add: ‘In the Dirae, v, 27, 
uirectis is the reading of two MSS. of the ninth ccn- 
tiuy, B and Y of Ribbeck ; and ought I think to 
have been ' admitted by the editor into the text So 
in the Rosetum, 1. 13, 'the same MSS. have frutectis, 
which is also found in the Cod.iHarl. 2534.’ 

Page 217, line 11 . Add : ‘ So also KpvfiBa> for. icpv<f>a. 
Nor can the numerous adverbs in Bov and Btjv in my 
opinion be opposed to this view, for in these also a 7 
seems to have disappeared from before the S, just as 
a c has from such Latin adverbs as cateruat-im. 
rBoinros too I should regard as older than Bowos’ 

Page 236, line 13. Add: ‘Of course this theory 
assumes that such adverbs as hreuiter ended at first 
in er. A change of this kind would be parallel to 
what has occurred in pater, mulier, &c. It is with 
some confidence too that I venture to assert that I 
have met with many cases in which old writers give a 
long e to the adverbs in er, although at the present 
moment I can only point to the Rudens (ii. 3, 65) : 
“Vt lepide, ut liber41iter, ut hon^ste atque hau 
granite;” the Epidicus (iii. 4, 49), “Redr, peccatum 
lirgiter. Immo haec east;” and the Eunuch (iL 1 , 
24), “Ficie honesta; mfrum ni ego me tfirpiter hodie 
hie dabo.’” 

Page 247, line 18. For ‘Hast.’ read ‘Haut.’ 

Page 262, note 1 . Add : ‘ Because dmi is to be 
claimed for the sufi^ in hhar-dmi, to take this as an 
example, on the ground that the “ dhatu” of the pro- 
noun of the first person is asmat, and the plural 
forms of this pronoun, asmd-kam and astnd-bhis, 
point to the same result. But this admitted, a<eo- 
called causal verb, as vSd-ay-dmi (ich mache wissen) 

Y 


Digitized by Google 



322 


ADDENDA, CORHIGENDA, ETC. 


must have ay for its suffix, a form which corresponds 
with all accuracy to ag or ac, the suffix of Latin 
verba, and to of the Greek Tapaaaa (rap-ax-), for 
a guttural between vowels would readily slip into 
a y. The usual doctrine of Sanskritista, that the d 
of bhar-dmi is a mere connecting vowel, offends by 
its very extravagance For such a purpose a short 
vowel alooe can be admissible 


Digitized by Google 



INDEX 


Thb English Alphabet is here adopted, so that fur Greek words 
Xi fi are considered as representing our ph, ch, th; while words 
with an initial asperate fall under h, and those commencing with ^ 
under hr (not rh). Latin words are commonly given in the crude 
form with an appended hyphen, in other cases are marked l; while 
A. -a denotes Anglo-Saxon, boh. Bohemian, bret. Breton, d&h, 
Danish, t. French, o. German, oa. Gaelic, oo. Gothic, L Italian, 
IR. Irish, LiTH. Lithuanian, n. Norse, s. Sanskrit, so. Scotch dia- 
lect of English, sp. Spanish, BW. Swedish, w. Welsh, o is in some 
cases prefixed to denote Old. 


A, 2UL 

accliui-, 12. 

addorraisc-, 12. 

a = nro, ^ 8L 

accresc-, 12. 

adesnri-, 12. 

a-, A.-8. 3^ 3^ 8L 

accumb-, ^ 22. 

adh-, IB. 21. 

a-, OA. fiS. 

accumula-, 12. 

adhinni-, 12. 

-a- of L. vbs. 22fl. 

accurr-, 14. 

adi-, 14. 

a cinToricoF, 1 28. 

achten, o. 21L 

adim-, 13, £2. 

a, priv. 127, 294. 

acknow, 24. 

adimple-, 12. 

ah, L. ^ 222. 

acknowleg .24. 

a-dining, 22. 

abaft, 119. 

acquiesc-, 12. 

adiuua-, 1^ 14. 

abasit, so. 240. 

actutum, L. 112. 

admiuiculo-, 14. 

abend, o. 218. 

ac ueluti, 157. 

admirn-, 12. 

abici-, IIQ. 

ad-, ^ 1^ 6^ ^ 

admisce-, 13, 52. 

abiet-, 

292. 

ado t OA. 52. 

aboard, 8L 

ad-, w. 1^ 3^ UL 

adole-, 13. 

abolesc-, 12. 

adaequa-, 12 

adolesc-, 12. 

abeorbe-, UB. 

adaestoa-, 12. 

adoperi-, 12. 

acced-, 12, L4. 

adaperi-, 13. 

aduesperasc-, 13. 

accend-, 12. 

adaresc-, I^ 52. 

seghvader, a.-b. 82. 

acci-, 14. 

adbib-, 52. 

Y 2 

set, A.-S. ^ 31, 52. 


Digitized by Google 


324 

aedituo-, 21fi. 
atiKtf; 127. 

af, ao. ^ 1112. 
afal, w. 147. 
afana, ao. 62. 
afar, oo. 121. 
afara, ao. 121. 
affle-, 12. 
afflig-, LLQ. 
afoot, 84. 

aft, 1^ 119. 
aftana, oo. 120. 
after, UT, 112. 
aftra, oo. 119. 

ag, OA. 28. 
agger-, 12. 
aggredi-, 14. 

-agh, KELTIC saff. 2.20. 
agita-, Z2. 
agnito-, 279, 296. 
agnoac-, n, 13. 
a-going, 213. 
agreeti-, 214. 
agul OA. 22. 

M at timea short, 16.2. 
aime-t-il, p. 205. 
ai])]>au, 00 . 179, 181. 
al-, 7^ 122. 
oXSaiv-, 217. 
alder, o. 180. 
alder, 217. 
alderhrst, 217. 
aldiza, oo. 1 .2.2, 
ali-, IIL 
alio-, 170. 
aliqui-, 172. 
uXXtjXo-, 172. 

Allen, 142. 
aller, f. 219. 
alleua-, 10. 
alliga-, 12. 


allinm, l. 228. 
alloqn-, 12, 
alt, o. 72. 
alt-ero-, 212. 
alter-uter-, 168. 182. 
aXA, 222. 
altiro, o. o. 123. 
aXoyo-, 128. 
aloft, 121. 
along, 33. 
aXwra-, 30.2. 

am, L. ^ 292. 
ama-, fiL 
am&lo-, 136. 146. 
amatum iri, 261. 
ambed-, 12. 
a/iflporo-, 127. 
ambula-, 19, 223. 
ambur-, 12, 

Ami, B. snif. 321. 
ammone-, 12, 13, 
amputa-, 12. 

an, OA. prefix, 132. 
an, o. 2^ 33. 

an, L. 170, 183. 
an, B. 296. 
ay, priv. 127. 
ai a, ^ ^ ^ 8, ^ 73, 
293. 

ana-, oa. 132, 146. 

ayaeXwTO-, 146. 

avaKitfiai, 1^ 23. 
anaknmbj-, oo. 23. 
anaout, brbt. 22. 
ayappoipi€-, 216. 
aya(7Ktva(-, ^ 22. 
ayarr/K; 6 , 
ancient, 213. 
andare, l 219. 
ander, o. 171. 
aySijpo-, 24. 


Andrews’s Lexicon, 1 2. 
avifKK, 217. 

OKeFiSKo-, 146. 

aKeiXc-, 22. 

avip-, 139, 301, 32L 
anfangen, o. 32. 
anbang, o. 22. 
anbela-, 2. 
anima-, 283. 
ayifo-, 53 

ankouna, bbet. 21,27. 
annar, h. 171. 
anne, l. 169, 183. 
anno-, ^ 260. 
annulo-, 260. 
anochd, ie. 296. 

avo(ci(-, 22. 

anqnir-, 3 
anser-, 222. 
antara-, s. 170, 177. 
ay&-ia-, 220. 
antidea, l. 125. 
antras, lith. 177. 
any, 171. 
anya-, b. 179. 
aupo-, 127. 
apama, b. 121. 
apara, b. 1 21. 
aperi-, 110. 
apium, L. 228 
am, 62, 292. 
apple, 147. 
appon-, 110. 
apprehend-, 1^ 22. 
aqua-, 258. 
arbusto-, 214. 
arde-, 13 m, 199, 219. 
arduo-, 13 n^ ^ 219. 
arise, 298. 
aroma, 258. 
arti-, 213. 


Digitized by Google 


INDEX. 


335 


apOpo-f 221 . 

ascend-, IL 
asi, s. 304-. 
asperate, 313. 
aspira-, 11. 
assudesc-, IL 
assurg-, LL 
asta-, 11, 297. 
aoTcport), 121. 
Bstreindre, r. 219. 
astru-, IL 
asv&nftm, s. 303. 
at, ^ 60. 

-at-, Ik. suffix, 229. 
areveff-, 129. 
aOararo-, 1 27. 
atque, 149, Ifil. 
atque adeo, 153. 
atque utinam, 154. 
atting-, IL 
auf-, 0. 103. 
aufdecken, o. 26> 
auildsen, a. 26. 
Aufrecht, 124, 208. 
augment, 295. 
aul, PBOv. 137. 
auXox-, 305. 
ainrj'O-, 127. 
aut, Ik 170, 180. 

av^ay-, 222 . 
aveindre, p, 218. 
avol, PROT. 137. 
Avon, 138. 
awake, 2L 
axt, o. 216. 

-ay-, a suff. 262, 
321. 

Ayenbite of Inwyt, 
103, 104, 115. 

Bag, 183. 


flaKK; 294. 

-bam, Ik suff., 299. 
fiav, 298. 

Banergea, 251. 
bay, 183. 
bd, 216. 
behind, 119. 
bei, o. 292. 
beinicht, a. 212. 
bellicuB, Ik 229. 
(itXTioy-, 212 . 
beneath, 68. 
fiiydio-, 129, 220. 
jitvTurro-, 212. 

Bergk, 208. 
berry, 216. 
bet, 115, m 
bey, o. 16. 
biga-, 192. 

Bindseil, 206, 271. 
bis, Ik 62. 
bis, 0 . 69. 
bit-, 298. 
bib, BOH. 275. 
Blindeisen, 12. 

-bo, Ik suff., 300. 
bodkin, 190. 
bo-in, GA. 304. 

Bopp, 119, 120, 122, 
125, 146, 172, 178, 
184, 204, 241, 257, 
260, 261, 276. 289, 
291, 293. 
bore, 190. 

Bosworth, ^ ^ 34, 
105. 

bottom, 228. 
bov-, 257. 
bow, 183. 
braggart, 213. 
Brahmanas. 250. 


break, 96. 
brid, 274. 
broach, 96. 

I urden, 219. 
burn, 224. 
buttock, 228. 

Cad-, 186. 
caduceo-, 219. 
caed-, 186. 
caelesti-, 214. 
caelitus, l. 246. 
calamitat-, 187. 
calf, 232. 

Campe, 91. 
cand-, 218. 
carectum, i* 208. 
carp-, 269. 

case-endings, 287,289, 
cassi-, 193. 
casto-, 189. 
castra, ik 18.5. 
castra-, 187, 189. 
castro-, 188. 
cateruatim, 321. 
c&tus, L. 279. 
causals, 261. 

-ce, Ik 161, 292. 
cendre, p. 218. 
cimec-, 232. 

CIS, Ik 242. 
citra, Ik 280. 
chambre, p. 223. 
chez, p. 294. 
chi, IT. 243. 
chi-ket-mi, s. 280. 
chin, 197. 
chin, ZEND, 280. 
chit, B. 280. 
chordae uocalea, 272. 
X»,222. 


Digitized by Google 


INDEX. 


32 G 

claustro-, 185. 

Clift, 320. • 
co([nito-, 279. 
eognomeiituiu, L. 212. 
cognosc-, 152. 
col-, OIL 
conibler, f. 223. 
commigo, bp. IM. 
coraparativoa, 34,338. 
concilia-, 10. 
connecting vowel, 295. 
contigo, BP. 154. 
contra, L. 120, 212. 
cord-, 219. 

Corssen, 166,209,211. 
corulo-, 217. 
coude, F. 217. 
coward, 213. 
eraindre, F. 218. 
crastino-, 213. 
crawl, 259. 
creep, 259. 
crow, 256. 

cs, 222. 

ct, 20L 
cubil 156. 
cum, L. 222. 
cunde 1 156. 
curr-, 195, 269. 
eurru-, 194. 

Curtius, 117, 119, 

124, 262. 
oustod-, 215. 
cut 1 L. 156. 
cuter t L. 156. 
Czermak, 272. 

D of Latin neuters, 
242. 

dagger, 213, 226. 

1 59. 


SapBay-, 221 . 

Davis, 148. 
de, L. 55. 
declension, 281. 

StSoiKa, 158. 
defenstric-, 215. 
degen, o. 225. 

Stkut, 123. 

8«8-, 158. 

StiyOr, 158. 
d^ismes, f. 205. 

5i)/xi/rcp-, 21L 
demonstrative roots 1 
278.- 286. 

Devon, 138. 
dextoro-, 214. 
dhdtu, - 299. 

&X0a, 222. 

Dies, 1^ ^ 2^ 
217. 

discrib-, 55. 
distrent, f. 216. 
diutius, L. 208. 
dixismus 1 L. 205. 
do, GA. 58. 
do, tR. 54. 
doce-, 264. 
domestico-, 214. 
Donaldson, 176 ; vii., 
xii. 

dorso-, 85. 
dos, F. 85. 
dosso, L S6. 
du, GO. preHx, 51. 
dumectum, l. 208. 
durch, G. 294. 
dyn, MANX, 69. 

. dys, MANX, 52. 

E of Lat; in6n. 163, 
164. 


-6 of Lat. adv. 235. 
earth, 257. 

Ebel, 112. 
ecce, li. 281. 
eccum, li. 281. 

‘X-. 114- 

ed, A.-B. 24, ^ 111. 
eft, 66, 

eigentlich, g. 212. 
fixaO-, 174. 
einhver, n. 176. 
einig, G. 114. 
einst, G. 216. 

222 . 

either, 168, 182. 
iKtiyo-, 161. 

-ola-, L. suffix, 229. 
cXa^o-, 267. 
i)\.cKTpoy, 305. 
eXcvScpo-, 221. 

£\f(a, 300. 
eller, dan. bw. 179. 
elope, 34. 
tfiaOoy, 220 . 
emmet, 7^ 227. 
empfangen, o. 36. 
umpfehlen, o. 36.: 
emprise, 41. 
en = inde, 167. 
en-, G. 24. 
tytprtpo-, 63. 
tyipBiy, 63. 
tytpo-, 63. 
Engonasin, l. 211. 
tyi, 62. 

ent, G. 24, 63, 612. 
enterprise, 46. 
entertain, 46. 
entfallen, a. 48. 
entgegen, o. 212. 
€yBey, 220. 


Digitized by Google 


INDEX. 


327 


entlassen, o. ^ 
entnebmen, a. 42. 
entrebaiser, f. 40. 
entteprendre, F. 40. 
entreprise, f. 40. 
entretenir, F. 40. 
entsagen, o.^ ^ 62. 
ontsinnen, o. 2S. 
entatehen, o. 22. 
entweder, 0. 182, 212. 
entzwei, o. 212. 

-eo- of Lat. adj. 230. 
epentbesis, 301, 303. 
€iri, 121, 124, 222. 
eppicb, o. 228. 
or-, o. ZS. 

.or of comp. 120, 124. 
-er of Lat. adv. 236, 
321. 

eram, L. 300. 
tptiiivdo-, 306. 

.eri-, of Lat. adj. 230. 
erkennen, O. Z2i 
erlosoD, o. 80. 

-emo-, of Lat. adj. 229. 
.ero-, of Lat. adj. 230. 
erscbliesaen, o. 80. 
eremneii, o. 72. 
c(t9Xo~, 221. 
et, L. 149, 166, 107. 
crafft/y, 129. 

6tre, F. 216. 

£-TwaT.«V, 296; 
etwas, o. 176. 

00 . 

evil, 138. 
extero-, 214. 

F&0I-, 264. 
facinos-, 226. 
fall, 180. 


fall-, IfiL 
falte, o. 212. 
far, 106. 

farbroder, dan. 18L 
farther, 100. 
fechten, o. 211. 
fell, 180. 
fer-, 7. 
fertili-, 213. 
feasel, o. 226. 
fetter, 226. 
light, o. 211. 

Ogures, 30L 
filia., 20^ 310. 

Ond, BO. 3L 
Anger, 213. 

Finn, 2^ 2^ 314. 
Fleckoisen, 209. 

Aect-, 207. 
fod-, 120. 
fodica-, 70. 
foUi-, 123. 
for-, 108. 

Forcellini, 1^ 1^ 61, 

210 . 

forcep-, 191. 
fordo, 108. 
forfeo-, 191. 
forget, 108. 
forpec-, 191. 
forswear, 108. 
forti-, 269. 
foxel, 307. 
irago-, 216. 
frang-, 230. 
fratello, L 234. 

Friem Barnet, 205. 
fresher, 213. 
frodoza, GO. 123. 
frutectum, h. 208,214, 
ft, 21L 


fundo-, 228, note, 
furca-, 190. 
lurche, o. 2L 
Furlanotto, 18. 

Gaffbb, 181. 
yaca, 276. 
yaXoKr-, 207.' 
gambol, 233. 
gammer, 181. 
yaft^Xai, 197. 
ganache, F. 197. 
gander, 218. 

Ganges, 208: 
gane, o. 222. 

Garnett, 89. 
gav, 8. 267. 
gd, 210. 
ySovjro-, 32L 
ge-, o. 2. 
geindre, F. 218. 
gelacbter, O. 21 1. 
GelUus, L44. 
gena-, 197. 
yet'ciui', 197. 
genetric-, 306. 
ytyv, 197. 
genuino-, 197. 
ghost, 286. 
gimblet, 7^ 227. 
gingiua-, 197. 
gmft, B. 257. 
yvado-, 197. 
gnoBc-, 161, 296. 
go, 8. vbs. signifying, 
200. 

Goldstiicker, 254. 
gon, CORN. 280. 
Gothic, 24. 
gownd, 207. 

Graff, 200. 


Digitized by Google 


328 


INDEX. 


ypaf; 23L 
Grimm, ^ 2^ 2^ 28, 
8^ 3^ 4^ 43, 44, 
^ 6^ 61, 6^ 74, 
78. 79, 84, 120, 
175. 179, 181, 

204. 
gs, 222. 

guarder, f. 216. 

Guest, 244. 
gums, 107. 
yvytuK-, 305. 
gwaetb, w. 181. 
gwell, w. 181. 
gyn, MANX, 51L 
gys, MANX, 52. 

Haooard, 219. 
al/to-u, 214 . 
haldit, 0 . IL 240. 
Haldorsou, ^ 141. 
hammer, 213. 
haiifa, 00 . 306. 
hanu, & 197. 
hare, 257. 
hau, L. IZ. 

Haupt, 154, 155. 
hazel, 217. 
heart, 213. 
ifiScfio-, 216. 

Hebrew roots, 277. 
imipiy, 152. 
hen = hence, 167. 
hence, 167. 

Hendrick, 218. 
hennen, o. enq. 167. 
liennes, o. bno. 167. 
herz, a. 223. 
hipo; 166, 175, 116. 
hethen, o. eno. 167, 
220, 246. 


heus, L. 161- 
hiatus, 125. 206. 
hie. L. 172. 2AL 
hiemps, L 223. 
hin, o. 320. 
hin-c, L. 167. 
hind, 112. 
him, DOB8ET, 195. 
hita, MALAOABH, 280. 
holl, BO. 96. 

Holmboe, Gfi. 

Oftako-, 136. 
homon-, 139, 
opa-, 215. 

ii^a, 260. 

ftaflio-, 216. 

^X‘-> 

Pay, 

hridi, b. 287. 

po^0<y-f 222. 

po,pc-, 216. 
ht, 211. 

hu-cusque, L. 156. 
humble, F. 223. 
Humboldt, 1^ 284. 
humo-, 139. 
ibraro-, 42. 

95. 
sVep, 43. 
vtpttmjiu, 4S. 
iliro, 4^ 45, 293. 
UTToStX-, 45. 
v^eoXa/iflay-, 4.5- 
ihroXittr-, 28. 

U 1 /. 0 S-, 43. 
burl, 194. 
hurry, 195. 

I, B. 228. 
i, DAN. 321. 

-i-, L. sutBx, 229. 


iecur-, 14. 
ignominia-, 145. 
ignora-, 22. 
ignofic-, 17, ^ 27. 
iKTtpo-, 207. 
iia, 8. 257. 

-ili-, L. suffix, 229. 
-ilia- of L. vbs. 229. 
illo-, 156, 165, 170, 
172. 247. 
illinc, L. 245. 
imbori-, 16. 
imbu-, 18, 223. 
imo-, 63. 
impell-, 19. 
impotent!-, 145. 
in, ‘ down,’ 63. 
inara-, 18. 
inardesc-, 16. 
incalesc-, 18. 
incandesc-, 16. 
incend-, 1 0. 
incid-, 1^ 52. 
incipi-, 18. 
incita-, 18. 
inclina-, 64. 
incoha-, IS. 
inconcilia-, 16. 
incresc-, 16. 
incuruo-, 64, 317. 
indaga-, 16. 
inde, ‘down,’ 218 
indige-, 217. 
inousceme, l. 231 
infami-, 146. 
infecto-, ^ 
infer!, L. 65. 
inlind-, 18. 
infitias-, n. 18. 
infla-, IjS. 
inflamma-. 16 


Digitized by Google 


INDEX. 


329 


inflect-, Ifl. 
informa-, 18. 
informi-, T46. 
infra, l. 65. 
infring-, 19, 48 n., 52. 
ingemina-, Ifl. 
ingrandesc-, 15. 
inhibe-, 16. 
inborre-, Ifl.- 
innutri-, 16. 
insepulto-, 66. 
in'sequ-, ^ 146. 
inaolenti-, 16. 
instanra-, Ifl.. 
instinct, 28.3. 
institu-, 16. 
insimula-, LL 
insurg-, 15. 
insula-, flfl. 
intabesc-, Ifl. 
intolleg-, 29. 47. 
intemperie-, 146. 
intepesc-, Ifi. 
inter, l. 48 53. 

interaresc-, ^ 62. 
interbib-, 48, 62. 
intercid-, 48. 
intercid-, 49, 62. 
interclud-, 49. 
interda-, 5L 
interdic-, 47. 
interfic-, 49. 
interfod-, 61. 
interfrigesc-, 48. 
interfring-, ^ 62. 
interfug-, 61. 
interfulge-, 61. 
interi-, 49. 
interim-, ^ 62. 
interiung-, 47. 
interluca-, 61. 


interluce-, 61., 
intermina-, 47. 
intermisce-, 47, . 162. 
intermitt-, 48. 
intennor-, 48. 
intemeca-, 48. 
intemosc-, 49. 
interpella-, 49. 
interpola-, 49. 
interpung-, 49. 
interqniesc-, 47. 
interroga-, 60. 
interrump-, 48. 
interscindr, 49. 
intersepi-,.49. 
interspira-, 61. 
interstingu-, 48. 
intertenere, L 46. 
interter- 1 l. 48. 
intertrigon-;, 48. 
interturba-, 47. 
interrert-, 49. 
interris-, 61. 
intumeec-, 16. 
intus, L. 246. 
inuestiga-, 16. 
inuido-, 146. 
inuocato-, 56. 

-io- of Lat. adj. 230. 
ioco-, 14- 
ion-, Im suffi 229. 
ionga, IR. 307. 
irai, p. 298. 
isto-, 291. 

-ita- of Lat. vbs. 76, 
229. 

itara, & 177. 
iterum, L. 176, 177. 
itidem, l. 167. 

-itie-, L. suif. 229. 
-ium, L. n. suff. 229. 


iung-, 218. 
iunic-, 306. 

-iuo- of Lat adj. 
230. 

iusticior-, 236. 
iusum, L. 86. 
iuua-, 14. 

Jamibson, 72. 
jeder, o. 182. 
Jefierson, fifi. 
jnft, 8. 296. 

Johnson, 233. 
jungoro, o. o. 123. 

E, suffix, 306. 
trou, 149. 
k&rava, 8. 266. 

KopSta, 219. 
kat, HTH. 280. 

•ret', 161. 

ken, base of pronouns, 
244. 279. 286. 
-kin, suffix, 226. 
knevel, 306. 
know, 246. 
krk, BOH. 276. 
krt, BOH. 276. 

Kpv/l&t, 321. 
ksbip-ra-, s. 146. 

KTa-, 207. 

KTtiy-, 207. 

Kuhn’s Zeitschrift, 
112. 113, 2fiQ. 
Kubner’s Gr. ^ 4. 

L, initial, lost, 14. 
Lacbmann, 164. 
lact-, 207. 
laeuo-, IflL 
lamb, 223. 

Xafijiar-, 220 . 


Digitized by Google 


330 


INDEX. 


Lapp, 24S, 314. 
largiter, L. 32L 
lauch, 0 . 228. 
laugh, 231. 
laughter, 211. 
lauru-, 159. 

Id, 21fi. 
lead, 2m 
left, 12L 
leg-e-bam, L. 295. 
leng, 180. 

212 . 
lepos-, 257. 
lesser, 147. 

-let, suffix, 226. 

Xcfw, 300. 
liberaliter, L. 32L 
libero-, 221. 
Libystino-, 213. 
lictor-, 194. 
liddeU & Scott, ^ 
54, 1^ IIL 
light, 211. 

Ligustico-, 213. 
limen-, 194. 
linen, 246. 

-ling, suffix, 220. 
lira-, 2L 
litera-, 103. 
little, 1^ 22L 
lo, 2SL 

-lock, suffix, 226. 
lointain, f. 21 2, ■ 
lording, 227. 
lovest, 216. 
loveth, 220. 

Ir, 8. vowel, 274. 

It, 212. 

xe, 220 . 

luc-8, L. 287. 
lumbo-, 1^ 223. 


Iz, 223. 

Madvio, 228, 31B. 
mogis, L. 122. 
niagistero-, 214. 
maior-, 146. 
m&la-, 202. 
fxaXaotr-, 199. 

Malden, 273. 
male, l. 129. 130. 
malle, L. 180. 
malleo-, 1£9. 
malmen, a. 199. 
mSlo-, 133. 
m&lo-, 147. 
mand-, 198. 
manda-, 217. 
manduca-, 198. 
manner, 202. 
manoir, F. 202. 
mant, w. lOB. 
Mantras, 250. 
Marriott, xii. 
fiaaa-, 202. 
fuuTa-, 202. 

Massmann, 78, 83. 
fta<r6aXiS-, 221 . 

Matthiae, 2. 
maxilla-, 202. 
mb, 223. 
fu), 294. 
fUfSia-, 174. 
Meissner, 7^ SL 
Hci(ov-, 146. 
ftefipKoiKa, 223. 
men, Umb. suffix, 291. 
raenton, f. 198, 
mentum, L. 185, 196. 
Merkel, 173. 
fifOntlx-Ppia, 223. 
messer, o. 213, 226. 


metathesis, 306. ’ 
metuouloeo-, 303. 
mickle, 138, 227. 
piyta, 217. 
might, 211. 
ministero-, 214. 
minor-, 122. 
miace-, 222. 
fuay-, 222 . 
mix, 222. 
mo, 180. 

/toyif, 202. 

mol-, 199. 
mola-, 108. 
molesto-, 214. 

^oXtc, 202. 
fioXvfiSo-, 216. 
Mommsen, 117, 144, 
166, 208. 260. 
mondenlicht, a. 304. 
raora-, 202. 
morast, o. 119, 2m 
morde-, 110. 
mordre, f. 212. 
Morris, 104, 147. 
mortario-, 199. 
morti, 213. 
mos-, 202. 
motion, 273. 
mould, 199. 
mourn, 224. 
mouth, 198. 
mp, 223. 
mnd, 8. 198. 
mulca-, 199. 
mund, 0 . 198. 
murder, 219. 
murec-, 191. 

198. 

Muller, C. 0., 12. 
Miiller, Lucian, 160. 


Digitized by Google 


INDEX. 


331 


MuUer, Max, 251, 
253, 256. 267. 268, 
278. 281, 299, 308, 
309, 310, 316. 
fivtrraK-, ^ 98. 

N, paragogie, 162. 

V, interchanged with 
fl, 174, 

V = « of u, 163. 
n of s. gen. 206. 
n of ENO. gen. 205. 
n, silent, -304. 
ua, BL&y. 57. 
naitre, f. 216. 
namentlich, o. 212. 
navel, fifi. 
nd, 216. 
n8, L. 170, ISa. 
nS, L. 294. 
near, 124. 
neath, 68. 

Ftaro-, 43 Py 66, 67. 
nect-, 207. 
ned, DAN. 68. 
vtFo-, 66. 
neif, 306. 
neither, 168. 
nello, L 63. 
vto-, 72. 

fcifttXt}, 66 . 

nequin-, 162. 
yipOt, 66 . 
vtv, 68. 
newt, 66. 
ng, 224. 

ni, OHiN. 68. 

ni, a. 67. 

nid, 0. NO. 68. 
nidr, o. no. 68. 
Niebuhr, 140. 


nieder, o. ^ 293. 
niemand, o. 218. 
night, 21L 
ult-, 210, 32a 
niz,- ROBS. 67. 
noble, L. 290. 
noct-, 207. 
nombre, bp. 223. 
non, L., -omitted, 169. 
nonce, 178, 227. 
nor, omitted, 168. 
North American Ee- 
view, 310 «< teq. 
nose-, 1 6.3. 
nouiesimo-, 67. 

UOUO-, 6a 

nt, 212. 

> e, 220. 

nu-, 6a 

nu, LiTH. 67. 

FUX'I ^ 
fwx'O) 297. 
nug, LiTH. 6a 
yvicT-, 69, 207. 
'nunner, 176. 
nutri-, 306. 

Ob, l. 6a 
ober, 0 . 4a 
obi-, HO: 
ube, L. 200. 
obsolesc-, 200, 220. 
obsordesc-, 201, 219. 
obet, o. 147. 
obter-, 110- 
occid-, MO- 
oculo-, 161. 
oeno-, 176. 
of, ^ 222. 

of, N. ua 

off, 62. 


ofguoing, 115. 
ofmickill, h, 1 15. 
of-serve, 104. 
of-take, 104.- 
often, 121. 
oftyened, 1 1 5. ■ 
oiKi/rop-, 236. 
oft<ftaXo-, 6a 
on, 64. ■ 
on = of, 6L 
on-, A.-B, 24. 30. 31. 

^ 6a 

oncUfjan, a.-b. 3a 
onenavan, a.-b. 2a 
ongitan, a.-b. 52. 
oyoptar-, 226, .302. 
onsecan, a.-b. 52. 
opox', 66, 303, .307. 
opv/tar-, 227. 
operi-, 1 10. 
opfel, a. 147. -. 
o<l>pv-, 302. 
opinion-,'226. 
oirura, 122. 
otriaBtv, 221. 
otrupa-, 147. 
oppet-, 110.' 
opple-, 147. 
opprim-, 1 10. 
or, 1^ 18L 
-or-, L. Buff. 238. 
ord, DAN. 142. 
ordi-, 219. 
ordon-, 204. 
opty-, 94. 
opi^Bc-, 222 . 

Orelli, 1.30. 
opyvia-, 306. 
ori-, 94. 

origin of language, 
282. 


Digitized by Google 


INDEX. 


332 

orm, DAN. 142. 
opOo-, 220. 
opOpo-, 220. 

08-, 213. 

OB8-, 213. 

OS, A.-8. ^ 3L 
other, 174. 
bSrum, N. 175, 
otium, L. 208. 
ovK, omitted, 169. 
ovXa, 199. 
ovOap, 222. 
over, 44. 

Ovid, 113. 

Palabt, q. 119, 216. 
Paley, 173. 
voWoc-, 231. 
palo-, 203. 
palumbe-, 93. 
pand*, 218. 

P&aini, 25.3. 
rraofiai, 207. 
vapa, 293. 
paragoge, 301, 305. 
vapanaXviTT-, 113. 
rapayOt-, 113. 
irnpoirayo-, 114. 
irapainjSa-, 113. 
irapa<ftr)pi, 113. 
vapaa€iov, 1 14. 203. 
wapadtppo-, 114. 
naprikiK-, 113. 
itapopa-, 113. 

Parry, 63, xiii. 
parti-, 213. 
passives, 261. 
pater, L. 86. 
paulo-, 202. 
pect-, 207. 


peior-, 148. 
peUec-, 1^ ^ 
pend-, 218. 
rt ydto-, 129. 
per with adj. 109. 
per with sb. 318. 
per with vb. 108,'319. 
percell-, 109. 
percurr-, 319. 
perd-, 262. 
peremni-, 6L 
perendie, L. 111. 
perfuga-, 112. 
perfund-, 109. 
perg-, 96. 
perinde, L. 96. 
perlin-, 109. 
permagno-, 109. 
persuastric-, 215. 
peruert-, 109. 
peruide-, 109. 
perung-, 109. 
petrifaction, 291. 
pf, 223. 

pfad, &c. o. 222. 223. 
<f>tpop€y, 296. 

<l>iy, suffix, 290. 

‘ phonetic types,’ 283. 

<^ 0 , 221 . 

0Oeyy-, &C. 221. 
pignos-, 226. 
pitpit, 080. 96. 
plang-, 230. 
plaustro-, 185, 189. 
plect-, 207. 
plorahundo-, 233. 
plumbo-, 216. 
plural suffix, 291. 
podec-, 191. 
PoggendoriT, 272. 
poll- 


pollice-, 76. 
polling-, 75. 
porno-, 122, 147. 
pon-, 118. 
popina-, 96, 
porca-, 90. 
porro, L. IL 
vopBp/t-, 220. 
pCe, L. 118, 123, 319. 
poshac, L. 118. 
posUla, L. 119, 123. 
posmeri^iano-, 118. 
posquam, L. 118. 
post, ii. 117, 214. 
poste, L. 126. 
postempus, l. 117. 
poeterganeo-, 118. 
postidea, L. 126. 
irorcpo-, 242. 
iro$(v, 245. 
Pott,9.119. 140, 148, 
267. 

prehend-, 217. 
prentili, a. 204. 
irpty, 122. 

Priscian, 145. 
pristine-, 122, 213. 
pro, L. ^ 92. 
wpo, 293. 
procell-, 93. 
procul, L. 151. 
prod, L. Uj 93. 
profano-, 94. 
profundo-, 94. 
prOm-, 93. 
prono-, 92 
pronominal roots, 278 
291 

irpof , 122, 293. 
rpoam, 122 
TTooaOf, 1 67 


Digitized by Google 


1X4K1!.A. 


333 


^LUHtHOSlSf 301 . 

WpOTCpO-, 77 . 

provection, 177, 227. 
pt, 213. 

wToXtfto-, &c. 213. 

Quadriga-, 193. 
quaerita-, 15. 
Quarterly Journal of 
Education, 147. 
quattuor, l. 187. 
que, L. 1^ 1S2. 

queato, L 243. 
quien, sp. 243. 
quinque, 187. 

R, 8. Towel, 274. 
rapi-, 264. 

Eaak, 2^ 31^ 93,115. 

121, 125, 126, Ififi. 
rastro-, 192. 

rd, 219- 

re, L. 74. ^ 293. 

recenti, 89. 
recidiuo-, ^ 99. 
reciproco-, 7^ 90. 
recita-, 88. 
recuba-, 88. 
recubitu-, 98. 
recumb-, 99. 
recupera-, 75. 
tecnruo-, 82. j 

red, L. lA. 

refell-, 9L 
r8g-, 95. 
reg-, 96- 
remane-, 38. 
rep-, 269. 
repando-, 82. 
repon-, 89. 


reside-, 88. 
reeoudre, p. 217. 
retorqne-, 98. 
retro, u 74. 
Rbeiniscbes Museum, 
164. 

Ribbeck, 298. 
ricochet, 99. 
ridge, 20. 
rig, 7^ 90. 
rise, 96. 

Ritschl, 97, 117.118, 
123, 126, 106. 

-ro-, n. sufHz, 229. 
roga-, compounds of, 
60. 

roll, ^ 195. 
rpetro-, 185. 
rouse, 95. 
rt, 213. 
pO, 220. 
rilck, o. 20. 
rz, 223. 

Sa, 8. 289. 
saeculo-, 260. 
saft, G. 121. 
sale, r. 202. 
sali-, 269. 

salictnm, 208, 214. 
salt, 212. 
salz, G. 9.9.H 
Sanders, 140. 
Sanskrit asperates, 
220 . 

sap, 121. 
sasmin, 8. 289. 
sat, L. 115. 

Saturday ReTiew.308. 
sc, 222. 
scala-, 198. 


Scaliger, 151. 
-schaft, G. sufF. 211. 
schlange, o. 259. 
Schweizer, 209. 
sci-, 233. 

Bcribendo-, 218. 
scribenti-, l. 212. 
seouB, h. 122, 209. 
segel, G. 203. 
sepeli-, 110. 
septentrion-, 196. 
serp-, 259. 
serpenti-, 259. 
sesqni-, 209. 
arfrts, 289. 
setius, L. 209. 
sg, 222. 
si, L. ‘so,’ 156. 
si-out, u 156. 
siluestri-, 214. 
Simplicius, 148. 
sinistero-, 214. 
sist-, 297. 
sister, 21.5. 
sitte, G. 202. 
situ-, 201: 
mcapSfio-, 221. 

<rir«r-, 234. 
tTKippo-, 222. 
sluggard, 213. 
slusb, 201. 
smure, so. 181. 
sou, 201. 
sole-, 202. 
soleil, F. 234. 
solemni-, ^ 260. 
sorbe^, 216. 
sordi-, 200. 
sorec-, 191. 

<TO<l>poavyri, 226. 

soror-, 216. 


Digitized by Google 


3:u 


INDEX. 


aorti-, 213, 21ilL 
BOUilltiT, F. 201. 

Boul, 285. 
sparrow, 2M. 

8|)ec-, 234. 

<r0oXX-, 187. 
spirit, 285. 
spiritual words, 285. 
sr, 8. 259. 
srp-, a 2.59. 
st, 213. 

St. Edmund, 180. 

St. Kenelm, 182. 
sta^, 23^ 291.: 
stamp, 223. 

Stewart, Dugald, 282. 
stern, o. 224. 

afl, 22L 
aOiyia-, 22L 

stingu-, 223. 

Stokes, W., HI. 
aTO/laT; 213, 
strag-, 239. 

‘ Study of Words,’ 
284. 

stump, 223. 
sub, L. 38, ill 44. 
submerg-, 4fL 
submitt-, 45. 
subsist-, 48. 
subter, l. 193. 

Bubtus, L. 213. 
suffod-, 45. 
sullage, 29L 
sully, 201. 
summo-, 42 n. 
sumpsi, L. 223. 
super, L. 43. 
superlatives, 238. 
supernl!, L. Ill, 248. 
supparo-, 114. 


T of Eng. neuters, 242. 

tuillour, 238 

talo-, 202. 

tura, a 196. 

r*, 142. 

rt, 15S. 

tegumentum, L. 212. 
227. 

rctvw, 129. 
tela-, 203. 

TtfityKT-, 226. 
temece, L. 310. 
temple-, 223. 
tempos-, 228 
tend-, 211. 

Tt/iifJiyOo-, 220. 
terebra-, 231. 
terion-, 196. 
terrestri-, 214. 

TtpOpo-, 221. 
riTiXiiTfiai, 300. 
that, 179. 
that oon, 178. 

6uT€poy, 177. 
the tone, 227. 
thence, 179. 
thet, old form of ‘the,’ 
178. 

third person pro- 
nouns, 179, 242. 
thorough, 12. 

Thorpe, 28 
dopvjju-, 2.31 . 
through, TT, 274. 
duyartf)-, 202 . 

thumb, 22L 
Tiy-, 157. 
tither, so. 177. 
tithing, 227. 
to-, o. ENO. prefix, 54. 
102 


to, 51. 

tobreak, 192 
toculion-, 213. 
toll-, I. 
toner, 175. 
louslric-, 21.3. 
tunstrino-, 215. 
topincb, 102. 
tordre, P. 219. 
tospend, 192 
tother, 177. 
tranquilh)-. 111. 
trans, 111, 
i'rench, 284. 
triis, F. 111. 
tribunicio-, 234. 

rpi^Oa, 222 . 

tristicior-, 236. 
true-, 2.3.5. 
tuft, 21L 
rvuTo/uffOo, 22E 
TvtrTio, 295. 
turf, 232. 
turn, 224. 
turpi-, 200. 
turpi t8r, L. 32L 
tyrant, 213. 
tz, 228 

-U, L. suff. 229. 
uad-, ^ 228. 
uae, li. 143. 
uallo-, 186. 
uanno-, 144. 
nano-, 148 
uber, L. 222. 
iiber, o. 48 
ue-, L. 143, 148 
uecord-, 148. 
uegrandi-, 143. 
uebement-, 148 


Oigiti?ed by Google 


INDEX. 


335 


Ueion-, 144. 
uel, L. 160. 
uellica-, 75. 
uelo-, 114, 2Q3, 
uelum palati, 206. 
uend-, 262. 
uSni-, 261. 
uento-, 296. 
uepallido-, 143. 
uerber-, 216. 
uenni-, 2fi9. 
uereu-, 91. 
uesano-, 143. 
uespertino-, 213. 
uesti-, 216. 
uide-, ^ 216. 
airectum, l. . 208, 
321. 

-ula- of L. Tbs. 229. 
uld, DAN. 142, 
ullo-, im 

uls, L. 242. 
ultra, L. 170, 212. 

ult, dan. 142. 
um, D. suff. 228. 

um, o. 61. 
umbilico-, 
umbra-, 66. 

un, ^ ^7j 53. 
nnart, a. 139. 
unbild, o. 139. 
undergitan, a.-b. 62. 
undermine, 45. 
undenecau, a.-s. 62. 
understand, 3L 
understanding, Tbs, of, 

22 . 

undertake, 36. 
undertaking, Tbs. of, 
22 . 

nndienst, o. 139. 


unding, o. 139. 

-undo-, L. su£F. 229. 
ungestaltet, o. 145. 
ungethier, o. 139. 
ungewitter, o. 1.39. 
ungezogeu, o. 139. 
ungott, o. 139. 
ungroes, o. 140. 
ungui-, 66, 
un-ion-, 226. 
unkrant, o. 139. 
unkub, 0 . 140. 
unloose, 6. 
unlust, 142. 
unmensch, o. 139. 
unrest, 142. 
unter, a. ^ 3^ 63. 
unterarbelten, o. 38. 
unterbauen, o. 3L 
unterbleiben, o. 38. 
unterbrechen, o. 4& 
unterdriicken, o. 38. 
untereiteni) o. 38. 
unterfangen, o. 37. 
untergehen, o. 42. 
untergraben, o. ^ 46. 
unterhalten, a. 36, 46. 
unterhobleu, o. 38. 
unteijooben, a 38. 
unterkeilen, o. 32. 
nnterkiLssen, o. 40. 
unterlassen, o. 3^ 48, 
62. 

untermauem, o. 37. 
untermengen, o. 38. 
unterminen, o. 38. 
untermischen, o. 38. 
untemehmen.g.36.46 . 
unterrichten, o. 37. 
untersagen, o. ^ 42, 
62. 


unterschneiden, o. 
unterscbreiben, o; 
unterschwaren, o. 
untersiegeln, o. 38. 
unterspulen, o. 38. 
untersteben, o. 37. 
unterstutzen, o. 3L 
unteisucben, a. 32. 
unterwascben, o. 38. 
unterweisen, o. 32. 
unterwerlen, o. 38. 
unterwiihlen, o. 3L 
unterzeicben, o.' 38. 
untendehen, o. 36. 
untbat, o. 140. 
untbier, G. 140. 
untiefe, q. 140. 
untrust^ 142. 
untyme, 142. 

-uo- of L. ad j. 230. 
uoca- = uaca-, 143, 
208. 

uociuo-, ,208. 
uolnes-, 226. 
uolsella-, 192. 
Upanisbads, 250. 
upper, 44. 
nppltiea, bw. 26. 
upptacka, sw. 26. 
uptake, BO. 28. 

-umo- of L. adj. 229. 
urt, DAN. 142. 
UTTOcale of Bopp, 271. 
uslukan, oo. 86 
uatula-, 214. 

-ut- of L. sb. 229. 
utero-, 179, 242. 
utrum, L. 169- 
uxor-, 215. 

Va, f. 228. 


Digitized by Google 




INDEX. 


33H 

vi, B. 296. 
van, 307. 

Varro, 15, 

ver, Q. 101, 106, 233. 
verachten.0.112,211. 
veralten, a. 106. 
verbluten, o. 107. 
verbriicken, o. 106. 
verfecbten, o. 112. 
verboren, a. 114. 
veijiihren, o. 106. 
verkehren, o. 106. 
vemehmen, a. 106. 
vereehen, o. 106. 
vertheidigen, o. 112. 
voilk, P. 281. 
von, o. 6L 
vowel-aasimilation, 
273, 224. 

vowels, iniinite, - 273. 

Wagner, G., 153.166. 
Wagner, W., 164. 


wan-, o. ENG. 142. 
wanhope, 142. 
wantrust, 142. 
ward, 216. 
waur, 146. 
wax, vb. 222. 
Westminster Review, 
261* 

Weymouth, 1^ 206. 
what, 179. 
whence, 179, 246. 
where = whether, 
181. 

where = whither, 
181. 

whether, 169, 242. 
whirl, 1^ IfiS. 
whom, 244. 
widar, o. o. 36, 119. 
Willis, 271, 213* 
Wilson, H* Hy 263, 
266, 268. 
wind, 296. 


winnow, 296. 
wit, 2L 
with, 35. 
withstand, 120. 
womb, 223. 
wot, 21. 

-WP-, W. suffi. 238. 

Ps!i^p(hf 222 . 
f«y-, 222* 

(vyo-, 222. 

Ya, b. Buff. 262. 
-yans, & suff. comp. 

122 , 

year, 260. 
yellow, 236. 
yonder, 119. 
yman, a. -8. 196. 
yun-ty'-mi, s. 219. 

Zend, 274. 
zer, 102. 

Zu, G. 67, 223 


THE END. 


Digitized by Google 


This preservation photocopy was made at 
BookLab, Inc., in compliance with copyright 
law. The paper is Weyerhaeuser Cougar 
Opaque Natural, which exceeds ANSI 
Standard Z39.48-1984. 1991 


Digitized by Google 




Digitized by Google 



Digitized by Coogle ]