A Comparative Grammar of the Sanscrit, Zend, Greek, Latin, Lithuanian ...

Google

This is a digital copy of a book that was preserved for generations on library shelves before it was carefully scanned by Google as part of a project

to make the world's books discoverable online.

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover.

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the

publisher to a library and finally to you.

Usage guidelines

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing tliis resource, we liave taken steps to
prevent abuse by commercial parties, including placing technical restrictions on automated querying.
We also ask that you:

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for
personal, non-commercial purposes.

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the
use of public domain materials for these purposes and may be able to help.

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for in forming people about this project and helping them find
additional materials through Google Book Search. Please do not remove it.

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner
anywhere in the world. Copyright infringement liabili^ can be quite severe.

About Google Book Search

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web

at |http: //books .google .com/I

J'J'J'J'J'J
J'J'J'J'J'J
J'J'J'J'J'J

■J'J'J'J'J'J
'J'J'J'J'i'J

■J'J'J'J'J'J
■J'J'J'J'J'i

'J'J'J

-J-J
/•J
^.

if.

fJ'J'.
- - - >^;
j'j'j'j'j'.
j'j'j'j'j',
j'j'j'j'j'.

J'j'j'j'^. 1

j'j'j'j'j't

j'j'j'j'j'j ,

i'i'j'j'j'i '

j'j'i'j'j^i

i'j'j'j'i'i
j'j'j'j'j'j
j'j'j'j'j'j

j'j'i'j'j'i
j'j'j'j'j'i
j'j'j'j'j'j
j'j'j'j'j'j
j'j'j'j'j'j
j'j'j'j'j'j
j'j'j'j'j'j
jfj'j'j'j'j

j'j'j'j'j'j
j'j'j'j'j'j
j'j'j'j'j'j

j'j'j'j'j'j

j'li'j'j'j'j

t^ i' ^ j^ j^ j^ j^ J' j^ j^ j^ j^ ^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ ^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^
f i^ j^ j^ ^ j^ j^ j^ j^ J' j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^ j^

jt jt jc jt jc jt jt je jt jt je jt js jt jt je jt jt jt jt je jt jt je jt jt jt jt jt jt jt jt jt js js jt jt j^

i'' J^ J^ J^ Jr j!^ ^

f^ J^ Jf^ J^ J^ J^ J^
^ ^ j£ Jt JfC Jt JC

Jr JF JF Jr J^ JF iT

if^jF jT JF JF JF J^
if^ JF JF JF JF JF JF

^ jc jfjf jt je jt
JF JF Jr JF J^ JF JF

j£ jt jt jF jF jt jt

^ JF JF JF JF JF JF
JF JF jT JF iT JF JF

^ Jt Jt Jt Jt Jt Jt

WORKS SUITED FOR PERSONS
PROCEEDING TO INDIA.

I.

CAPTAIN D. L. RICHARDSON'S

OVERLAND GUIDE TO INDIA,

Outward and Homeward. Second Edition, with numerous Engravings,
and a Map of the Orerland Route, post Svo. cloth, 5s.

II.

ji jt jjt jc jF jt jt

JF JF JF jT JF JF JF
JF J^ iT JF JF JF JF

Ji JC Jt Jt Jt Jt Jt

\jf^\JF JF JF J^ JF JF
]J!^JF JF JF JF JF JF

J^JFJ^JFjtJFJF

jcjFjFjrJF

WHAT TO OBSERVE,

Or the Travellers' Remembrancer, by Col. J. R. Jackson. Second
Edition, in one thick vol. post Svo. 12s.

III.

THE ORIENTAL INTERPRETER

AND

TREASURY OF EAST INDIAN KNOWLEDGE,

By J. H. Stocqueler, being an immediate key to the numerous Oriental
words, terms, and phrases, which perpetually occur in English books,
newspapers, &c. Svo. cloth. 10s. 6d.

IV.

THE HITOPADESA.

The first book, or Mitralabha. The Sanskrit text with an analytical
glossary, Sanskrit and English, shewing the construction, as weU as
the signification of the words, by Professor F. Johnson. 4to. 15si

V.
PROFESSOR WILSON'S INTRODUCTION

TOTHB

GRAMMAR OF THE SANSKRIT LANGUAGE.

Second Edition, containing an additional Chapter on the Grammar
of the Vedas. Svo. cloth. ISs.

VI.

CONCISE GRAMMAR OF
THE HINDUSTANI LANGUAGE,

To which are added selections for reading, by £. K Eastwick, Professor
of Urdii in the East India College, Haileybury.

JAMES MADDEN, 8, LEADENHALL STREET,

J. Wertheimer ft Co., Printers, Circus Place, London Wall.

Jt Jt Jt Jt Jt Jt Jt

jejtjtjtjcjeji
J' J^ J' J^ J^ J^ >

J^ J^ J' J^ J^ J^ /
J' J' J^ J' J^ J^ /

jc jK JF JF JF JF ^

j^ J' j^ j^ J' j^ ^

^^^^^^^

X*^XXXX*I
J^ J* J^ J^ J^ J^ ^

VfXXXXXXi?fJfXXX*^*^XXXXXXXXXi?f*?fi?fXXXXXXXi^X^

vf*Kxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx^

Vi^XXi?fi^*^XXXXJfX*^XXi^Xi?fXXXXXXXXJfXXXX>«tA-t

^-^

'yfyU*^'^^ C 6^ ^

/■/> 2:4.^1^ -'<^<

//vT^

COMPARATIVE GRAMMAR.

COMPARATIVE GRAMMAR

OF THE

SANSCRIT, ZEND,

GREEK, LATIN, LITHUANIAN, GOTHIC, GERMAN,

AND SCLAVONIC LANGUAGES.

BY

PROFESSOR F. BOPP.

PART III.

TBANSLATED FROM THE 8ERHAN

BT

LIEUTENANT EASTWICK, M.R.A.S.

MEMBER OF THE ASIATIC SOCIETIES OF PARIS AND BOMBAY, AND PROFESSOR OF URDU

IN THE EAST-INDIA COLLEGE AT HAILEYBURY.

CONDUCTED THROUGH THE PRESS

BY H. H. WILSON, M.A.F.R.S.

BODEN PROFESSOR OF SANSCRIT IN THE UNIVERSITY OF OXFORD.

LONDON :
JAMES MADDEN,

LEADENHALL STREET.
1850.

LONDON:

WILLIAM WATTS, CROWN COURT, TEMPLE BAR.

COMPARATIVE GRAMMAR.

PART III.

VERBS.

FORMATION OF THE MOODS.

POTENTIAL, OPTATIVE, AND CONJUNCTIVE.

716. In the dialect of the Vedas the Mt mood or con-
junctive is also formed by the insertion of an o, in cases
where, in the corresponding indicative form, an a is wanting,
by the lengthening of which the mood in question might
be formed. Thus, from the aorist abhut^ ** he was,"' comes
the conjunctive bhuvat, ** he may be ;" where, by the aug-
ment being dropped, the meaning of past time is also re-
moved, as is likewise the case in the potential and impera-
tive : from akar, "he made " (for akart, according to §. 94.*),
comes karat, " he may make ;" from chikit-ti, *' he recog-
nises **' (R. kit. CI. 3.), chikMaii, " he may recognise.'^ So in
Old Persian, ahatiy, "he maybe,"' from astiy/*he is" (Be-
hist. IV. 38. &c.), where the Sanscrit ^ s in Old Persian is
retained before t, but before vowels becomes h.

From the aorists also, in the Veda dialect, come con-
junctive moods with the terminations of the present ; hence,
karati, " he may make'" (Rig V. 46. 6.), from akar. The Veda
dialect even forms the conjunctive mood by the simple

* Aorist of the fifth formation, which in the Veda dialect is more ex-
tensively used than in classical Sanscrit.

3q

954 VERBS.

annexation of the personal terminations of the present to
the base of the aorist, thus e.g, vivdchati (vi prep.), *' he may
announce,^^ from vyavdchat (Rig V. CV. 4.).

IMPERATIVE.

717. This mood, which, in classical Sanscrit, is formed
only from the present indicative, is distinguished from the
latter merely by the personal terminations (the first person
of the three numbers excepted : see §. 713.), which have
been already discussed. The dual and plural, with the
exception of the third person plural, have the secondary
terminations ; so that e.g. bharatdnif " let the two carry,"
is distinguished from abharat&m, "the two carried,"" only
by the omission of the augment. In Greek the diflPerence
of the termination twi/ of (pepenav, from rrjv of the imper-
fect eipepertjv, is unorganic, as to)v and rrjv are originally
one, and both rest on the Sanscrit tdm,

718. The second person singular of the Sanscrit first
principal conjugation — i.e. that which corresponds to the
Greek conjugation in o, to the Latin fourth conjugation,
and to the German strong and weak conjugation — is
distinguished from the second principal conjugation, which
corresponds to the Greek /x/, inasmuch as in the active
(parasmaip.) it has lost the personal termination ; so that
e.g. bhar-a, "let him carry'" (Zend, bar-a) terminates
with the class-syllable, to which, in the dual and plural, the
personal terminations are annexed (^TOTT bhar-a-tam^
^ef>-e-Tov, ^TOT bhar-a-ta = (pep-e-Te). The loss of the per-
sonal termination appears of great antiquity ; as in Greek
too, (pep-e is said for (pep-e-di ; and in Latin leg-e, am-d,
mon-i, and aud-h are likewise devoid of the personal sign.

* The e of lege is, in its origin, identical with the « (fix)m o, see
§. 109^ 1.) ofleg-i-te, and rests on the principle, that in Latin, at the end
of a word, e is preferred to i ; hence, e. g. mare from the base mart.

FORMATION OF THE MOODS. 965

719. In German the strong verbs have, in the second
person singular of the imperfect, rejected the class vowel,
and terminate, therefore, with the final letters of the root,*
without, however, in most cases, containing the actual root
itself, as the vowel of the root, according to the analogy
of the present indicative, appears at one time weakened ;
as e.g. in Gothic, bind, from the root bandt "to bind" =
Sanscrit, bandh; at another time with Guna, hence, in
Grothic, biug, "bend,*" from the root 6M5r = Sanscrit, bhit; ;
belt, " bite," from the root bit = Sanscrit, bhid, " to cleave ''
(see p. 105). The Sanscrit also, and Greek, retain, in the
present imperative, the Guna gradations of the present
indicative, or, most generally, that of the special tenses ;
hence, e.g. in Sanscrit, bddha, " know" (from baudh) from
buuVif and in Greek, ff>evye from ^vy. The German weak
verbs retain their class character (see §. 109*. 6.) corre-
sponding to the Sanscrit aya, of the tenth class : the sylla-
ble ya, however, is contracted to i (Gothic ei = {\ as in
general the syllable ya at the end of a word lays aside its
vowel, and changes the y into one. Compare, e.g. the
Gothic tamrei, " tame," from tamya, with the Sanscrit causal
dam-aya ; Latin dom-d ; Greek Sifx-ae. In the second
weak conjugation, let laig-d, "lick," be compared with the
Sanscrit causal Uh-aya, from lih " to lick :" in the con-
traction of a{y)a to A however, laigd approaches nearest
to Latin imperatives like dom-dj as the Gothic d = d (§. 69.).
In the third weak conjugation, compare hah-ai, thah-ai,
sil-ah with the Latin forms of like signification, ftofe-^,

* Thus in Latin die for dice. With regard Uifer it is to be observed,
thaty^ also, in the indicative, is to be joined rather with the Sanscrit bkar
(Jbhri) of the third class than with that of the first. Thns, as fers^fe^^t^
fer-tisy corresponds to bi-bhar-shiy bi-bhar-ti, bt-bhri-tha, so fer answers to
Mbhri-hi (from bibhar-dhi), the personal termination being suppressed, as
in e«=Greek, io-»dt, Sanscrit i-dhi from ad-dhi (for as-dhi).

3q^

956 VERBS.

tao^, sil-i, where the ^ is a contraction of aU and answers
to the Sanscrit ay of aya (see p. UO). In the second per-
son plural tamryi'th (from tam-ya-th) corresponds to the
Sanscrit dam-aya-ta, Latin dom-d-te, Greek Sa/x-ae-re. In
Greek and German the imperative second person plural is
not distinguishable from the present indicative. In San-
scrit, however, the imperative has the termination of the
secondary forms (to) opposed to the tha of the primary ;
thus ^innr damayata^ " tame ye,'' opposed to ^IHT^ dama"
yatha, '* ye tame." In Latin domdte is distinguished from
domdtis, where the latter form answers to the Sanscrit
dual indicative present (?fRirW damayathas, Gothic tamyats),
the former to f[Pm( damayata, " tame ye '' (see §. 444.).
The termination to, of the second and third person of the
so-called future of the imperative, and the Greek termina-
tion TO) of the third person singular, correspond to the
Veda termination tdt, which answers for the second as
well as the third person ;* and in the latter, as has already
been remarked, is most correctly retained in the Oscan
titd (licitud, estiuJL) As in wnr the expression of the per-
son is twice contained, so it is in the Latin second person
plural tdte, for which in Sanscrit ?rnr tdta might be ex-
pected, which, however, does not occur. In the third
person plural nto answers to the Greek irrtov {legu7do =
K€y6vT(»)v), which was before compared with the Sanscrit
middle forms in antdm {(fyepovToiv =bharantdm.)

720. The Sanscrit termination w, plural ^TtT, is derived
from the pronominal base ir ta, by weakening the a to a
vowel of middle weight, while in the present indicative, as

* See §. 470. The edition of the First Book of the Rig V. by Fr. Roeen,
which has appeared since this work was commenced, has confirmed tdt to be
the termination of the second person of the imperative. H. XL VII 1. 15.
occurs H fft n^d l ri pra no yachchhatat^ "give us" and CIV. 5. ^cSmri
charkritdt from the intensive of the root TSf kri^ " to make."

FORMATION OF THE MOODS. 957

generally in the primary forms the extreme weakening to i
takes place. We have, therefore, the forms -td, -tu, -li, as in
the interrogative, in the isolated ease lea, leu, kL In Zend
the u of the imperative termination is occasionally length-
ened ; e. jr. in the frequently-occurring ^coiiAs?^ mraotth " let
him say :" on the other hand. Vend. Sade, p. 142, ;coas7a>^
kharatu, "let him eat,"" )^m^jjjI^ vanhatit, "let him put
on.

721. The Sanscrit middle termination sva (from iva, see
§. 443.) of the second person singular is in Zend corrupted
with a preceding a to anuha (for anhva), where the v is
changed into the vowel u, and has stepped before the h;
the nasal, however, which, according to §. 56*., is placed be-
fore the A, remains, though otherwise j n occurs as a gut-
tural nasal, only in direct combination with h. The com-
bination nhv appears, however, too uncouth to be admitted
in Zend ; and wherever, therefore, it would occur, we find in
its stead »»;j nuh : hence, too, \t^xs^)^»jl^ vivanuhatd=
Sanscrit (\i<|MH^ vivastxdasy "of the Vivasvat" (Vendidad
Sade, p. 40.). Several examples of imperatives in anuha
occur in the eighteenth Fargard of the Vendidad, where,
however, the text corrected by Burnouf (Ya^na, Note A.
p. 17) according to the manuscripts is to be referred to, as
the lithographed copy (pp. 457, 458) has, more than once,
anha faultily for anuha : A*o»;3ui/^^»»jgjai/o a)7i^j9a)9 jqkOas
aiwi vastra ydonhayanuha* " put on the clothes ;' ** auO«
Asw^jui/^^Ajyjj A)^j)As^ frd zasta snay anuha, " wash thy
hands f'^ a5^;3ui/j)am/0 9^9j)ma) au & aismanm ydsanuha,

* This form is based on the causal of the Sanscrit root im ya^ ^^ to
strive.*

1 1 take xs^)^^^AifM may anuha as a passive verb with a middle
signification; thus Vend. Sade, p. 331, twice a)^xia)^^a)JJ9 9^/^S^ ^^
us tanum inayaita^ "let him wash his body " (Anquetil, p. 360, " tl lavera
son carps*) : on the other hand, p. 330, tisa (use ?) tanum snayaita^ with
a conjunctive vowel between the preposition u« (= Sanscrit TiT <«') And

the

958 • VERBS.

" sprectd out wood" (compare Sanscrit m? yam^ in the spe-
cial tenses ^qi^ yachh, with the preposition w d, "to ex-
tend''). So also in the Vend. Sade, p. 39, for Ai^^»pw
hunvanha we ought to read as^)^»j)^ hunvanuha^ accord-
ing to the manuscripts made use of by Burnouf, and for
As^juu j)^ visanha, ** hearken " (Vendidad Sade, p. 123.), per-
haps also A»^;9ui>j>^9 visanuha should be read.

Remark. — ^In the Latin Edition of my Sanscrit Grammar
of the year 1832 (p. 330) I have taken the form Ai^)jjj»)j)^
hunuvanuha, or, as the lithographed manuscript reads,
M^^»p^ hunvanhof as the imperative middle, and trans-
lated /r^lmanm hunvanuha kharetei (according to Anquetil,
" qui me mange en minvoqwird avec ardeur,'') by '* m>e celebra
ad edendum^'' The root hu is^ as is remarked L c, added to
the conjugational character of the first class, besides that of
the fifth class nth for without this unorganic adjunct the
form would be hunushva ( = Sanscrit ^^^ sunushva). It
is certain that the Zend root hu must in Sanscrit be su, and
the opinion which Bumouf ascribes to me (Journal Asia-
tique, 1844, Dec. p. 467), that the Zend hu rests on the
Sanskrit ^ Am, " to oflTer," has been neither expressed by
me at p. 781, nor in my Critical Grammar, p. 330, nor any-
where else. That a Zend w* h never corresponds to the
Sanscrit ^ h has been expressly remarked in §. 57. ; and it is
also remarked in §. 53. that »• h, in an etymological respect,
never corresponds to. the Sanscrit w A, but always to the
pure or dental ^ s. Had I wished to compare, therefore,
/. c. its Sanscrit type with the Zend hu I could only have
referred to one of the roots ^ su, of which one, like the

N9

the following word (see §.518. p. 737). The transitive meaning of the
root md is, on the other hand, usually represented by (pamjjj snadh in

the active ; c. g. Vendidad Sade p. 233, 8. : cam^^jjas^ ^juj^ma)
^g^^A5(oAMijt9A)0« aetao vastrdo framddhay^n **let them wash these
clothes."

FORMATION OF THE MOODS. 959

Zend hu, belongs to the fifth class. On the meaning
** cekbrare,^'' which I have given to the Zend hu (according
to Anquetil *' invoquer avec ardeur'*^) I did not desire to lay
any particular stress ; for my chief object was to settle the
value of the grammatical forms which Anquetil mistook,
and I wished to recognise, in the interrogative form, an
imperative termination based on the Sanscrit a-sva, and in
khareteij the dative of an abstract substantive, while, accord-
ing to AnquetiFs translation (^' qui me manye''^) it might
be taken for a third person present. In both respects I
now find myself supported by the Sanscrit translation of
Neriosengh, which is given (/. c.) by Burnouf, which renders
M^)jjj»pky hunvanuha by nPiMi^X^^af^ parisanskdram kuruj*
and ^PCog^A5^ kharetei by 9l<«f||i| khddandya (" for the
eating,'' or "the food.") The explanation of the appended
commentary is ^I^KIv| ^ «H M ^ dlidrdrtham sanmanaya^ i,e.
** on account of the food honour (me)."t The root ;»» hu
occurs several times in the ninth Ha of the Izeschne, from
which our passage is taken ; and indeed in the third
person of the imperfect huntUa (once hunvata with the
addition of the character of the first class), which Anquetil
everywhere paraphrases by " ayant invoque et sStant
humUier I have translated it (Z.c.) by '* laudabatr and
regret that Burnouf has not given us Neriosengh's trans-

* Burnouf remai'ks, " Nos manuscrits sont tres-confus en cet endroit :
celui de Manakdji a ^l<tiKHI^ sanskarascharu^ mais je ne suis par sur
du ^ ich; le numero II. F. lit. ^[^^g(i^ sahsMrariku avec ^ sch au-
deBsus de la ligne." However, I have no doubt that Burnouf is right in
reading "SR kuru,

f So Burnouf reads for the iniTFC^ sanmaraya of the manuscripts,
which is unmeaning.

\ Burnouf translates " Aowore-woi comme nourriturey*^ in which I cannot
agree with him; for dhdrdrtkam, can only mean ''on account of the
food," not " as food ;" and in khddandya, as the translation of khareteS^ the
relation of cause is apparent.

960 VERBS.

lation of this expression also. Undoubtedly, however,
the circumstance that the verb derived from hu every-
where refers to as^^^as^ haoma, the personified Soma-plant,
speaks in fia,vour of Burnouf s opinion, that the Zend hu
has the same signification as the Sanscrit root ^ su; viz.
" to press out the sap," where it is to be further re-
marked, that in Sanscrit the verb from this root is
especially used in relation to the Soma-plant. I avail
myself of the occasion which has led me to speak of the
ninth Ha of the Izeschne, to correct an error to which I
was led by a false reading of the lithographed manuscript
of the Vendidad Sade. Four times in this Ha the mas-
culine nominative of the interrogative occurs before the
accusative of the pronoun of the second person. The
lithographed manuscript reads once ^^oxTc^ ^^as^ kas^
thwanm (p. 42), once if'^oji^ H^•^^A)^ kasi thwanm (s m^ for
M i, p. 40> by mistake), once ^^qxTuxijias^ kasithwanm (p. 4l),
and once ^^cudAsojj9A5^ kasithawanm (p. 39). Here, there-
fore, two readings support the separation of the two pro-
nouns, and two their combination; and at first I supposed that
the form of writing in which they were separated was the
right one, where, in the ^ or i of kas^ and kdsi, was to be
recognised an appended pronoun, like the Greek demon-
strative I (ovToaly eKetvoal: see §. 157*., and Gram. crit. Add.
ad r. 270). The i, however, I regarded as the sign of the
nominative, and this it really is ; for though the Sanscrit
termination as in Zend regularly becomes 6, but s in the
middle and beginning of a word before vowels A, there
might, however, be an exception in the case of the termi-
nation as occurring before an enclitic, where as might
retain its original form; for in Zend J3 « is not so
much the palatal sibilant as the ^ in Sanscrit is, for the
latter occurs before no .other mutes but palatals only;
while jj occurs before mutes of all organs (see §. 49.), and
before mutes which are not palatals always corresponds to

FORMATION OF THE MOODS. 961

the Sanscrit ^ s, except before p, where this springs from the
Sanscrit t?, as e.g. in jm^m spd = Sanscrit^ sva. As, however,
we learn from the notice of the various readings of the Paris
manuscripts, which have been in the meantime published
by Burnouf (Ya^na, Note R. p. 134), that ^^as^ kcLse, and
the combination of the interrogative with the following
^^<ijS'(^ thwanm, " thee," is the prevailing reading (we find
the words joined seven times, and separated only five times,
and seven times e occurs — ^for i twice, and for i three times)
it admits of scarce any doubt that the vowel which stands
between kai and thwanm is inserted only to assist the
utterance, and that we must regard kasthwanm as the
original form ; so that, as is the case before the enclitic
particle Ara, the sibilant of the nominative has maintained
itself under the protection of the following consonant, and
remained too when a conjunctive vowel was inserted
to aid the pronunciation.* 1 shall not decide whether
this vowel must necessarily be an j e, and could not be
either i or a. Let, however, the quite similar case be
considered, where, between the preposition ^) ui, and the
verb j^Au^eK)'^^ histdmh in the lithographed manuscript at
least, at one time ^ h at another «$ i, at another as a oc-
curs as the vowel of conjimction (see §. 518. p. 737). We
may indeed expect, that in all places where the litho-
graphed manuscript has i or a some one or other of the
manuscripts has e; and undoubtedly this, the shortest of
all the vowels, is best adapted for insertion as a mere
vowel of conjunction, as, too, it is regularly used for this

* Thus, as ought to have been remarked at §. 47., the forms m^^^Js
hitya^ **of the second,* and A5^^^j7(3 thritya^ "of the third," point to a
time when the I of the Sanscrit dvittya, tritiyay was still present, on which
account the y has not commimicated an aspiration to the preceding con.
sonant, as is the case e. g, in meretkyu^ where the combination of the
T- sound with the semi-vowel is primitive.

962 VBRBS,

purpose, to prevent the direct combination of r with a
following consonant (§. 30.), without any other vowel being
used for this object. Here, too, the question might be
started, why no interposed vowel is to be found in the
combinations kasti, " who to thee," and kasnd, " which
man ?"" (for *' who "" generally : see p. 281,) mentioned by
Bumouf L c. (p. cxxxix), while hakhwanm nowhere occurs ?
The reason of this, I doubt not, lies herein, that thwanm,
on account of its double consonant, less easily unites with
a preceding h than ti and nd ; while ^ j) st and y j) hi are
quite favourite and usual combinations. On the other
hand, histdmU though its initial sound is one of weak
consonants, required the interposition of a vowel when
combined with us, because sh is an impossible combina-
tion in Zend. At the beginning of the twenty-first
Fargard of the Vend. (Vendidad Sade, p. 498) we five times
find nemase t^, i.e. " adoration to thee r ( = Sanscrit tfiR^ ^
each time written separately, though the two
words evidently ought to be joined, as the vowel of con-
junction 6, and the retention of the termination as, for
which 6 would otherwise be substituted, sufficiently de-
monstrate. It appears, however, that on account of the
polysyllabicalness of the word, to which in this case the
enclitic t^ is attached, the phonetic combination appears
less intimate, and this may also be the reason why the t
cannot, as in kas0, follow the s without an intermediate
vowel. We may see how much the Zend inclines to use mo-
nosyllabic pronominal forms enclitically, in that it attaches
them even to prepositions, which have become detached
from the verbs to which they belong : hence, ^^^Auoi
A5»»;jju»y;^ frdmanm hunvanuha in the passage cited
above ; so Aj»»;yAJOJJUj^ 9^9*^ dmrmm ydianuha, which

* That Anquetil's translation '^addressex voire priere" is incorrect
requires no proof.

FORMATION OF THE MOODS. 963

Neriosengh translates by ^^ w(f^^, ue. **wish or obtain
me ;" and Bumouf (Joum. As. Dec* p. 465) by '' iwrnque-mou^*
We may also here preliminarily remark that, for the first
time, we have learned, through Rawlinson^s late ingenious
discoveries, that in Old Persian also the pronouns readily
attach themselves as enclitics to the preceding word, and
that if we read without the a (which in old Persian is
sometimes contained in the consonants, and sometimes not),
y, which is regularly added to the i at the end of a word,
as well as to the diphthong aiy the old Persian enclitics
will, in like manner, be all monosyllabic. For this, as
for other reasons, I read auramazddmaiy, ** Auramazda
to me," for Rawlinson's -rnaiya (former reading miya).

722. The first person of the three numbers of the im-
perative follows in Sanscrit and Zend a peculiar principle
of formation, which, as has already been remarked, cor-
responds rather to the conjunctive or LH than to the
other persons of the imperative. An d is prefixed to the

* Anqnetil altogether omits to translate this expression, for which, in
the lithographic manuscripts (p. 89), occnrs by mistake ydsanha. Bur-
nouf thinks he recognises in the root t/ds, the Sanscrit ITT^ y^^) '^ to
demand, ask ;" but a difficulty arises in the aj i for Sanscrit ^ cA, of
which I have elsewhere met with no example. The root 1^3 yachh, as
substitute of H^jR yam, answers better, on account of its final consonant, as
"g chh in Zend is regularly represented by * ; on which account 1 have
above (§. 721.) preferred dydianuha, "spread out,'* to this root. Here,
however, the meaning of the Sanscrit TfR yam, 7|9 yachh^ preposition
^d, does not suit. Perhaps the a (mdhm) ydianuha in question is radi-
cally identical with the frequently-occuring dyiiSy " I praise " (or " in-
voke "%) which leads to a Sanscrit root yas^ which is only retained in
^T^m yasas^ " glory." With regard to the Zend 6 for the Sanscrit a or
a see §. 42. It is probable, however, that in dySii, as also in genitives in
yihi for yahe^ and in present forms in yhni, besides the preceding y the
vowel also of the following syllable has an assimilating influence in the
change of a or a to ^: hence we find, indeed, dySiS^ but not dySianuJia, but
dydianuha*

964 VERBS.

personal terminations, the terminations of the present indica-
tive middle which end in ^ lengthen this diphthong to du
and the verbal theme keeps, in the second principal conju-
gation, the strengthened form, which elsewhere enters only
before the light personal terminations. The first person sin-
gular has ni for its ending, where n is clearly a corruption of
m and is suppressed like the latter in the Sanscrit middle,
while the Zend maintains this decided advantage over the
Sanscrit, that it for the most part retains the personal
character, and presents dni to match the Sanscrit di. This
jjjyjttj dni therefore bears the same relation to the active
jyjoi dni, that, in the Greek present indicative, fiai does to
/Lu. In order to exhibit the principle of formation of the
Sanscrit first person imperative I here present the said per-
son of the three numbers of the two active forms of the
root fg^ dvish, " to hate," compared with the corresponding
forms of the present indicative.

ACTIVE. MIDDLE.

INDIC. IMPERAT. INDIC. IMPERAT.

Sing, dveshmi, dvish-d-ni. dvishS, dvhhdi

Dual dvishvas, dv6sh'd-va. dvishvahi, dv^sh-d-pahdi.
Plur. dvishmas, dvish-d-ma. dvishmahS, dvish-d-mahdi.

So in Zend, Vendidad Sade p. 477, several times jfMSfM^
jan-d-ni ( = Sanscrit han-d-ni) "I will smite, destroy," f

* The lingual /i occurs on account of the euphonic influence of the
preceding lingual sibilant according to §. 94*. of my Sanscrit Grammar.

t In Sanscrit also the first person imperative sometimes occurs in the
sense of the future or present indicative, to express a decided volition of a
positive impending action, e. g. Sunda and Upas. I. 26. Anquetil takes
Jandni as the third person of a preterite^ and renders it (p. 413.) by <' il
/rappa," and once by " seront arUantis" It needs, however, no proof
i^tjanani is r^^y the first person imperative, for 2ioroaster speaks to
Ahriman the words <^»x}.^ as^au^ •^i^^i^^^^^j'^^^^ v^9^ A5^«b^
^c^AU4 duschda anrS mainyo jandni oAma daivd aatSmy &c., ^'Vicious

Ahriman,

FORMATION OF THE MOODS. 965

pp. 132, 479. (JJ/J^»J^/f7g3 kerenaV'd-nS " I should make/'
( = Sanscrit krin-avdni from kamav-d-ni).

723. In verbs of the first principal conjugation and of
the ninth class, as also in roots in d of the second or third
class, the modal d combines with the preceding a or d;
hence €.g. vfTXf^ bhardni, "let me carry;**' Zend jijm?:^ bardnU
middle ^ijm?>a^ bardni (Vendidad Sade, p. 48o). So mjmsm^
visdni, " I will obey ;" * CJ/juuja5;^ yazdnS (see p. 278), " I
should offer ;"" j^yAuyf 7f o) perendnif " I should destroy " (Ven-
didad Sade, p. 335, compare Burnouf, Ya^na, p. 530, ff.) ;
j^yjttf<3j^«bi^AjjC^ yaoschdathdn^, "I should purify ''I.e. p. 480).t

Ahriman I will destroy the Daeva-created people." Upon which (p. 478)
Ahriman says to 2ioroaster,

mami ddma m^r^chanuha ashdum Zaratusira
•* Slay not my people, O pure Zoroaster !

* Vendidad Sade, p. 124. ^yAUj^^vp j^^ ^gC-^ axSm tS vtsdnSy "I
will obey thee," so /. c. are other imperatives in the sense of the future,
as jyjo^^^^g/AJ9 ^juiC^masaa ^^ ^c^xs azSmU gaSthaovaridhaySniy
" I will make thy lands increase (" mcJce fruitful,*' Anquetil p. 271. "^*e
rendrai voire mande fertile et abondante").

t See §. 637. I am now, however, of opinion, in departure firom what
has been remarked at p. 112, that the th of daih is a substitute of dh^ and
I take da as the syllable of reduplication, as in the Sanscrit dadhdmi. The
I^^^^'^^^^J ^idaithyahnf " deponant," mentioned at p. 112, corresponds
to the Sanscrit fVf^lZRT nidadhyus, A)Ct>^^J^jy ni-daithita to the San-
scrit fff^^Vhr nudadhita. (§. 702.) In the genitive of the participle of the
reduplicated preterite yyffl^OASg dathiisho corresponds to the Sanscrit
dadhushas; while in the nominative cau»^as^ dadhvdo (= Sanscrit

t^lV^lH ^£?A-i-van) and in the accusative 9g^^gAM»^^^ dadhvdonh&fn
(= Sanscrit dadh-i-vdhsam\ the alteration of dh to th does not take place,
an alteration which most probably is found only in the weak cases.
Perhaps in Zend th is considered weaker than dh and d^ and this may be
the reason that the interrogative verb, where it appears without a prepo-
sition or other incumbrance of composition, or even with compositional
incumbrance, but without reduplication, also exhibits no th in the examples

with

966 VERBS.

After s^ y comes i for A\ hence e.g. J/«A.Vw^f ^S? varedhayini,
"I will make to grow'' (Vend. Sade, p. 124) ; (^yxi^yM7jai%>*A»^^
firahArayifUi* In the production of this S, however, the t
or i of the termination bears the most important part, for
if the y alone was the efficient cause, it would also influence
the following vowel, if £ or ^ did not occur in the termina-
tion ; this, however, is not the case, hence e.g. aj^^as^j^asC?
varedhaya, "make to grow'' (Vend. S., p. 124); As^^^dij)?
radchaya, " make to give light, kindle" (p. 457) ; As^Aj^^Ajy j^ aj jC^
yasnayata, "sacrificio colUe" (Burnouf, Ya^na, Note A. p.l3.)f
So in the second person plural middle, 9^Qxr(D^^^2ui^
vArayadhwem, "defend yef ^^oj^^yi^^xs^Mi^^dhdraycidhwemy
"preserve ye" (Bumouf, l.c. Note D. p. 38.)

with which I am acquainted ; while, where the reduplicated verh is
burthened by composition, th almost universaUy occurs in the base-syllable,
though dh also is occasionally found, e.g. in yaoshdadhaiti (Bumouf,
Ya^na p. 860.) In cases where the forms with th follow the analogy of
the Sanscrit first class, as e. g. in nidath^m^ *' I have made," (Bumouf 1. c.)
I regard the vowel which follows th not as the class vowel, but, as in the
conjugation of the root 7^ 8thd^ amC^J;) itd^ as the shortening of the radical
vowel (see §. 508.) I also now consider the verbal- theme mddha, " to wash,"
as a compound of the root snd and dhd^ the radical vowel of the latter
being shortened (compare Benfey Wurzel lex., 11. 34.) The perhaps not
numerous forms may appear surprising in which the vowel of the syllable
of reduplication of the Zend root dhd (without a vowel preceding, da) is
long, as in the example mentioned by Bumouf (1. c.) nidhdthaygn, '* they
may lay down." Here either the lengthening of the syllable of repetition
is a compensation for the shortening of the base-syllable, or the genius of
the language takes ddth for a secondary root, without being conscious that
the <f, with its vowel, is in fact a syllable of reduplication, as in Sanscrit
the forms dS-hi^ '^ S^^®/' (from dad-4hi^ Zend daz-di) and dh^-hif " place,"
(from dhad-dhi\ no longer give the impression of reduplicated forms.

* Yendidad Sade, p. 82. 9^^a5 9SCCK3-^A)(p ^^JJ^yy^) ^^
j^ixi^^A5Z\WW*A50© hS urvdnim vahist^m ahum frahdraySni^ " I will
make his soul go to the most excellent place ;" Anquetil, p. 139, ^^jeferai
alter librement son ame aux demeurea celestes."

t YaSnaySmi is a denominative from yaina = Sanscrit yc^na, *' offering."

FORMATION OF THE MOODS. 967

Remark. — ^An explanation — ^and I am now much inclined
to adopt it — might be given of the a of the terminations
Aniy dva, &c., in the first principal conjugation, as follows ;
viz. by recognising in it only the lengthening of the short
a of the class-syllable, while only ni, &c. is regarded as the
personal termination. There is a twofold occasion, how-
ever, for the lengthening of the a of the class-syllable;
first, that in the LSt mood, or conjunctive, to which, ac-
cording to its principle of formation, the first person of the
imperative belongs, the a of the class-syllable is lengthened
(see §. 713.) ; and secondly, that especially before pronominal-
consonants of the first person, in case of their being fol-
lowed by vowels, an a originally short is lengthened ; and
hence forms like ami, avas, ava, &c. nowhere occur, where-
fore ani also is not to be looked for. On the latter prin-
ciple may be explained the d of dvesh'd-ni, bibhar-d-ni,
yunaj-d-nh kinav-d-ni, and karav-d-ni; so that we may assume
that the a, which, according to §. 716. is added in the con-
junctive, is lengthened simply on phonetic grounds. It is
certain that the first person plural of the 1. c. cited, >t^
bhuvat, "let him be," can only be bhuvdmch and this is at
the same time the imperative of the fifth aorist-formation
(see §. 573.). The first person plural of the Old Persian
ahcdy, "let him be," quoted in §. 716., is most probably ahdma,
which would correspond to the Sanscrit imperative ^nmr
as-dma. If this view be correct, then in the ninth class
also the words yu-nd-ni, yurnd-va, &c., must not be divided
into yu-nd-dniy &c., but we must assume that, as here, an d
in the original word precedes the personal termination, no
further a-sound could be added. The ninth class already
meets the requirement for fulness of form in the first per-
son in this way, that the syllable nd is not, as in the weak
forms, weakened to nt The roots dd and dhd, which reject
their d before the heavy terminations, retain the same in
the imperative by reason of their inclination to fulness of

968 VERBS.

form ; thus e. g. da-dd-ma, da-Md-maf not dad-nutt dadh-ma
(compare §. 481.)-

724. Besides the middle termination dn6, which surpasses
the Sanscrit in correct retention of the original form, the
Zend also recognises the abbreviated form di, of which, how-
ever, it makes but unfrequent use. An example is jMi^^
viidi in the fourth Card of the Vispered (Vend. S., p. 65),
where jjos^)^ ^iS^ azem viidi, occurs seven times, which
Anquetil renders by ^'fobeisy With the preceding impe-
rative dstdya, "bring,"* the present indicative accords
best ; so that, in the want of positive examples, we might
believe jjmm^ vildU to be only a more energetic form for
the present indicative vis6. The form jms^jmj^ yazdU
which occurs several times in the twenty-second Fargard
of the Vend., is rendered by Anquetil ** rendez hommager
and the context requires also the second person, for yazdi,
&c., expresses the command of Ormuzd directed to Zoroaster,
to whom he promises, as the reward of the reverence required
of him, that which follows, dathdni, "I will give" (= San-
scrit ^>*ir«f daddni, first person imperative). I see also no
reason to assent to Bumouf in placing (Yajna, p. 495) the
words JAMjAj^ yazdi, &c., in the mouth of Zoroaster ; and
I take yazdi to be the imperative active of the causal form,
and, indeed, as a contraction oi yazaya; whether it be that
this expression really has a causal signification, and means
*' let honour," or that the causal form has here the same
meaning as the primitive form, as in Sanscrit also is not
unfrequently the case. In a phonetic view, the relation of
ya^sdi to yazaya resembles that of .jAuy ndi, " conduct," to
the Sanscrit «pi naya. With regard to yazdi, as well as to
ndi, we must assume that, in compensation for the suppres-

* Literally, "make to come," the causal oiltok, "to stand," with the
preposition a. Anquetil takes the adjoining accusative as a nominative,
and dstdya as the third person.

FORMATION OF THE MOODS. 969

sion of the final a, the a of the preceding syllable is length-
ened, or, which comes to the same thing, the a of the final
syllable is transposed, nearly as in the change of ashavan
** pure,V into ashdum (with m for n) in the vocative. The
form JAW/ ndh ** conduct," occurs six times at the end of
the ninth Ha of the Izeschne in combination with ndiem*
(Vend. S., p. 47). Anquetil (p. 112) renders 9gQ)^»gj ^JMj^Q<MMi»
ndsemndi kehrpem by ** enseignez-moi le moyen d'aneardir son
carps," The literal meaning, however, is "conduct the body
to destruction," (e.g. azdis, "of the snake," =^r^ ahis.)
Here, perhaps, the composition of the imperative with the
accusative kehrpem may have given occasion for the con-
traction of naya to ndi. This, however, does not prevent
the assumption that, without any special occasion, a trans-
position of the a of the syllable ya may also take place,
since the Zend is particularly fond of transposing the a of
the syllables ya and va, and forming them into a diphthong
with the vocalized semi-vowel. I shall return to this sub-
ject in the emendations to §. 42.

726. In respect to Syntax, it deserves notice that the
first person of the imperative in Zend not only, as has been
already shewn by some examples, sometimes supplies the
place of the future indicative, but is also used as the conjunc-
tive, governed by ajwaj;^ yatha, "that." Thus, in a pas-
sage quoted by Burnouf (Ya^na, p. 427) with a difierent
object from the fourth chapter of the Yescht de Gosch,
j/jo<>i)as^^^^ 9^^^ ajwasjCI^ yatha azem bandayini, "that I
bind;" j/m^^as^jjd^ ^g^«>}i^ a5^; uta bastem vddhayinU &c.,
"and (that I) beat those who are bound;" (g^J^)^ a5^;
jyjo^^AsyAso); via bastem vpanayini, "and (that I) conduct
those who are bound." On passages of this kind Burnouf s

* This word is not once written quite correctly in the lithographed
manuscript; the correct reading, ho wever, may be easily found by a com-
parison of the several erroneous ones.

970 VERBS.

opinion may be based, that the forms in dni (or Sni), in
point of sense, belong as well to the imperative as to the
potential, while he denies in toto that the middle form in
dn^ (or in6)f which was first brought to light by Fr. Win-
dischmann (Jenaische AUgemeine Litt. Z. July 1834, p. 138),
belongs, in point of signification, to the imperative, and ex-
plains the forms in di according to their meaning as
genuine imperatives middle of the first person (Ya^na, p. 530,
Note). I cannot assent to this opinion, as e.g. (^/J^-w^
yazdn^, ** ofier," in the passage quoted above (p. 278), has as
imperative a meaning as the first person for the most part
admits of, while visdi (§. 724), according to its signification,
is rather a present indicative, and yazdi (l.c.) is explained
as the second person imperative active of the causal.

726. Among the European sisters of the Sanscrit, the
Gothic alone presents a first person of the imperative,
but only in the plural, where, e.g., visam, '* simus,^'' (Luc.
XV. 23.) corresponds to the Sanscrit vcisdma, " habitemusy''
without, however, being formally distinguished from the
present indicative ; as the Sanscrit terminations mas and
ma in Gothic are represented by mere m, except in the
conjunctive, where ma corresponds to the Sanscrit if ma
of the secondary forms. It has been already remarked
that, according to its formation, the imperative of the
Sclavonic and Lithuanian does not belong to the proper
imperative (see §§. 677. 699.).

I here give a general view of the points of comparison
which have been arrived at for the imperative present.

SANSKRIT. ZEND. GREEK. LATIN. OOTR.

l.p. sg. act. han-d-ni, jan-d-ni,

bhar-d-m, bar-d-ni,^ . . . • . . . .

l.p.sg.mid. karav-di, karav-d-ni, .... . . . .

bar-di, * bar-d-ni,

I Bardni cannot be sapported by quotation, but is clearly deduced from
the middle bardni (§ 728.) and the plural bardma(Y, S. p. 208).

FORMATION OF THE MOODS.

971

SANSCRIT.

1. p. pl. act. bar-d-ma,

2. p. sg. act di-hi,^

bhar^a,
vah-a,
vah-a-tdt,^
2. p. sg. mid. dat-sva,^

bhar-a-stm.

ZEND.

bar-d-ma,
daz-di^

bar-a,
vaz-a,

GREEK.

iar-Ot,

LATIN.

GOTH.

bair-a-m.

veh-e,
veh'i-to,

bair,
vig.

• . • •

bar-ariruha,''

bar-a-ta,

{(pepov, from )
^ep-e-a-o, j

^ep'€-TOV,

^ep-e-re,

ex-e-je.

2. p. du. act. bhar-a-tam,
2. p. pl. act. bhar-a-ta,

bibhn-tay ....

vah-a-ta, vaz-n-ta,

2. p. pl. mid. bhar-a-dhvam, bar-a-dhwem, (jyep-e-ade

3. p. sg. act. vas-a-tu, vanh^a'tu,

vah-a-tdt, vaz-a-tdt,^

3. p. du. act. bhar-a'tdrrit

.... bair-a-ts.
.... bair-i'th.

veh'i'te, vig-i-th.

veh'i-to,

ipep-e-TOiv,

3. p. pl. act. bhar-a-nfut bar-a-ntu?

2 ns-hi from daddhi for dadd-hi from dadd-dhi, See §§ 450. 481.
3^ (A5J dazdi from dad-di^ See % 450., where for dazdhi read j^ (A5 ^
dazdi, as ^^dh occurs only between two vowels. Thus we twice read in

V. S. p. 50, |^9^<A5^ daxdi-mS, "give to me," with m^, "to me,"
enclitic, where we must remember, that in Sanscrit, also, the forms ^ tn^,
" met, wiAiV' «"^d ii' tS, " tui, tibi/' are used only enclitically ; just as in
Old Persian maiy and taiy. We must therefore take the (in V. S. pp. 505,
507, 608) frequently recurring w^^ jjam^ma dathdni ti^ " I will give to
thee," as = dathdnitS, since composites in Zend are frequently separated in
writing. If, however, dathdnitS is to be taken as one word, 1 should then
explain the th as being for dh^ on the same principle as that by which the
root dd, " to lay," in the reduplicated forms, when they appear in compo-
sition, regularly exhibits th for dh in the radical syllable. (See p. 964,
Rem. **.) 4 From ad-dhi for as-dhi, 5 955 Rem.

6 For dadd-sva, (See § 481.) 7 See § 721. » See p. 653,

Note t-

727. In the Veda dialect and Zend occur forms also
which correspond to the imperative of the aorist in Greek,

3 R 2

972 VBBBS.

and, like the latter, have with the augment, which is the
true symbol of past time, also laid aside the past signifi-
cation. To the Greek first aorist corresponds tj^ bhusha,
" be '^ or ** become "' (see Westerg. r. w, pret ^n) euphonic
for bhii-8a=i>tMrov, The v of the termination <rov, if or-
ganic may be deduced from s*, and this from 6, as, e.g., So^

* See § d7. With regard to the transition of final s into v compare also rjv,
^ he was, "with the Dorfc ffs and inv ^ of the Vedas : moreover the
Buff. 6€v = Sanscrit tas, Latin tus (§§ 421. 631.). The form -Bevy as it
approaches closer to the Sanscrit tas and Latin tus than $€ does, must he
regarded as more organic than the latter, which, as Buttmann remarks,
(§116. 4. Rem. 1.), is of frequent occurrence only in certain particles, in
which the original meaning (*' whence") is not so perceptible, and is found
elsewhere but seldom where the metre requires it {dvTp63€ Pind., Ktm-po^e
Calimm., Aifiva^e, iravroOe Theocrit.). Observe, also, the complete rejec-
tion of the V in the ace. of bases ending in a consonant (7rar€pa=Sanscrit
pitaram, Latin pcUrem), as well as, in particular, the abundantly demon-
strated fact, that final letters are the most exposed to weakening and
complete extinction. The weakening of « to n is too, in itself, not more
remarkable than that of s to another liquid, viz. r ; which, in Sanscrit,
so firequently takes place •according to settled laws, and occurs dialecti-
cally also in Greek (see §. 22.), and is found in several kindred lan-
guages in certain parts of Grammar; as, 6.^., in Irish the termination
Tnar of the 1st p. pi. represents the Sanscrit mas, Latin musy Doric ft^f,
which latter, in the common dialect, is corrupted to fiev. The Sanscrit
secondary termination ma^ which also occasionally occurs in the present,
is very probably an abbreviation of mas (see §. 439.), which first appeared
after the separation of dialects; an abbreviation which enters more
extensively into Old Persian, since there the final * after a and d has
become the weakened form of all terminations. Therefore I cannot
agree with Pott (Etym. Forsch. II. 306.)~to whom G. Curtius (Forma-
tion of the Tenses and Moods, p. 27) assents— in deriving only fxes from
mas, but ficv from ma, as if the v were only a later suffix or echo. Why,
it might be asked, have similar enduring resonant letters (not used like
the V €<l>€\icv(mK6v to prevent the hiatus) not been suffixed to distinct
vowel-ending forms, e.g, to the e of the voc. of the 2d decl. (§. 204.), or
to that of the dual (§. 209.)? The Doric termination vra in the
3d p. pi. imper. (XeycJvro), iroiovvrto, aTroTiaavTO)) may be regarded with

at

FORMATION OF THE MOODS. 973

from Sodt. We should therefore have to regard -aadt as
the original form, and from that -aa^f and afterwards -o-oi/,
with the change of a to o, which is preferred before nasals
(see p. 104). In this manner, if the v of tvtt-o-o-v appears
to be the personal termination, and, in &ct, in a place
where the Veda dialect has lost the personal termination
(bhu-sha frojn bhu-sha-dhiX then it must be remarked that,
in Prakrit also, the termination hi, which is a mutilated
form of dhi, is much more extensively used than in
Sanscrit (see Lassen, p. 338. Hofer, p. 185). From <Tadt
a middle termination aaadt may be developed, according
to the principle of rvyl/dcdfa from Tin/raro), rvyjrao-de from
TvxjraTe ; for as all terminations, which in the active be-
gin with T, are preceded in the middle by cr, where t passes
into 6 (see §. 474.), sq it cannot be matter of astonishment,
if, from the to-be-presupposed Tvy\faQi is formed Tv\l/aa-di,
and hence, by rejecting the a$, Tvxjrat, which presents an
accidental agreement with the infinitive active of the aorist.

at least equal justice as an abbreviation of vt<ov ; as, vice verad^ prav may
be looked on as a lengthened form of vra, for the Doric dialect has not
in -all cases preserved the most ancient forms. Pott (1. c.) finds, in a
physiological view, the interchange between s and v difficult to compre-
hend ; as, though both are dentals, yet the difference in their pronuncia-
tion is vast. Still greater, however, is the difference between that of a
mute and the nasal corresponding to its organ ; and yet, in Sanscrit,
final mutes, if they occur before a nasal, pass into the nasal of their
organ (atishthan murdhni, " he stood at the top," for -tm) ; and in Latin
sanmtis stands for wpntts ; in Greek a-cfivos for a-t^vos : while reversedly,
in Lithuanian and Sclavonic, without its being occasioned by the
neighbouring letters, the n of the number nine (Sanscrit navan) has
become d (see §. S17.) ; and in Greek the n of the suffix ITW man,
Latin men, has become r (o-yofiar=sffnnT ndman, nomen). I am also of
opinion that the Veda termination tana, in the 2d p. pi., has arisen from
tata, and therefore is only a reduplication of the common termination to,
and rests, therefore, on the principle of the Latin imperative-ending tSte,
and the Veda tdi of the 2d and Sd pers. singular.

974 VERBS.

as in Latin also, ama-re, ** be loved "' (the last syllable of
which is only a fuller form of the reflexive, which we,
see §. 476., have recognised in aivio-r, &c.) is in sound
identical with the active infinitive. If, however, the
imperative rtnt-aou has arisen from TtnT-aairSt, the abbre-
viation is only one degree greater than, in the indicative,
that of ennr-o-aHTb to ervir-o-a). We return to the Veda
dialect to remark, that to forms like Tvn^iri'Tiaf irrespective
of the personal termination, corresponds the ^^ ni-shortuish
euphonic for s, see §.21.), which is cited by Panini (III. 1. 81.
Schol.) ''let him conduct/' In the second person dual ^J^lf^
bhUshatam (nMJ^flH upahh&shatam, see Westerg., r. )| hhth
prefix 7^ upci), corresponds admirably to (ftvcarovy and in
the third person plural, T^^N^ srd-sha-ntu, " they shall hear"
(Rig. V. 1. 86. 5), in respect of the aoristic suffix, to forms
like Au-cra-vTwi/.

728. In Zend as yet no imperatives have occurred, which,
like the Veda ^ bhmha, &c., would correspond to Greek im-
peratives of the first aorist ; on the other hand, j^j^ dAi-du
"give" (Vendidad Sade, p. 311 twice, pp, 421, 422), corre-
sponds to So- 9, from So-5/, as^am^ dd-ta, "give ye" (Vendi-
dad Sade, p. 224)* to iore, and dA-ta "do ye," "make ye,"
(in comp. as^-uj^c^aj^ ymschdMa, " purify ye," Vendidad *
Sade, p. 367, frequently) to Se-re. I think I discover a mid-
dle imperative aorist in au^^^am^ ddonhd, "give thou"
(Vendidad Sade, p. 222, 1. 1 from the bottom) ; but we re-
quire to understand the passage where this expression oc-
curs by the aid of Neriosengh's Sanscrit translation, as well as
a comparison of manuscripts. It is probable that we ought
to read au^^^^au^ ddonuhd, where the long A would present
no difficulty, as in this passage other originally short as at
the end of a word are found lengthened. In the Veda

• * I write data for data, as in this passage long a stands for short a
everywhere at the end of a worJ.

FORMATION OF THE MOODS. 975

dialect the forms are very numerous which answer to the
Greek imperative of the second aorist ; thus, srudhi, " hear
thou," =kKvdtf* from srinSmi (R,^, CI. 6, irreg.); sag-dhU
" be able,'' from sakndmi (R. sak, CI. 5) ; pur-dhi, "fill thou,''
from fftoS^ piparmi (R. ^ pri, i.e. par, CI. 3). .To ^iPJir abhiit,
"he was" (aorist of the fifth formation, §. 573.), corre-
sponds hhu'tUf " esto.'* Forms like mrfhi mumugdhi " loose
thou" (R.f7iucA, third person, mumdktii), strongly resemble
the Greek like KeKpaydu The Sanscrit form, however,
as appears (see Westerg.) from the indicative form amv^
muhtam, distinctly belongs to the aorist, which in the Veda
dialect also exhibits similar reduplicated forms, combining
the personal terminations direct with the root, which
therefore stand in the same relation to the fifth formation
(see §. 573.), which in the Veda dialect is used also in roots
ending in a consonant, as that in which forms of the seventh
formation (§. 579.) do to those of the sixth (§. 576.). The
^T^^l^ t?dt?nd/ia*t?a, "grow thou" (Rig. Veda, 1.31. 1.), which
has been differently explained above (§. 709. Note), is per-
haps an imperative middle of the seventh aorist formation :
it would then stand for vavridhasva, as from mrigy in the
aorist indicative active, comes amamrigam. The lengthen-
ing of the syllable of reduplication would, according to
§. 580., be more authorised in the aorist referred to than in
the Veda perfect indicative vdvridS (Rig. Veda, 52. 2.), for
vavridM of the common dialect. The circumstance that no

* So long as a pres. of the 2d cl. irSmi does not occur, I am inclined
to regard the forms of the indicative cited by Westergaard, cUravam,
^' I heard "; a^rStf '^ he heard," as aorists of the 5th formation, with Guna
of the short radical Yowel, which appears lengthened in the Greek
Kkv6i; as, in forms like beiKPVfUj the v corresponds to the Sanscrit u
with Guna. Remark, that also in the Veda aorist aJcary "he made,"
akaram, " I made," the broader and here the original, but according to
Indian Grammar the Gunized, form of the root occurs, while the imper.
kridkL " make thou," has the shorter form.

976 VERBS.

indicative occurs corresponding to vdvridhoivaf when re-
garded as an aorist, would not be a sufficient reason for
rejecting this view ; for hitherto no indicatives abhiisham,
anSsham, asrSshamt have been found to correspond to the
aorist imperatives mentioned in §. 727., bhushof bhmhatam,
mishatu, srdshardu. If, however, with Westergaard, we
assume potentials and imperatives of the perfect, we can
then, with him, derive* vdvridhasva from the perfect indica*
tive vdvridhi. But, according to the signification, the
reduplicated imperatives and potentials, which all have a
present meaning, are better derivable from the aorist
(which in its moods lays aside its past signification together
with its augment) than from the perfect, where the re-
duplication expresses past time, and which, therefore,
must remain in the moods likewise ; as, e. g., in Gothic,
haihaityau signifies " I was called," not " I am called.^' If,
however, in the Veda dialect the reduplicated modal forms
spring, in part at least, from the perfect, we must then
assume that they have, through a perversion, surrendered
the past signification, which belonged to them, so that the
Grerman conjunctives of the preterite in this respect stand
on older ground. The explanation of the reduplicated
modal forms from the intensive, attempted in §. 709. Note,
is now far from satisfactory to me ; and I now hesitate
between the derivation of them from the perfect, and their
deduction from the reduplicated aorist. To the latter
might be referred ni • . s^da, " seat thyself '^ (see Westerg.
pp. 177, 179.), as ^l^^p^ an^sam (see §. 682.) presents an ana-
logous indicative. To the avdcham mentioned in the said §.
belongs the imperative sanvdchdvahdi (l p. du. mid. Rig. V.
I. 25. 17.).

729. Traces of an imperative of the auxiliary future
occur in classical Sanscrit. But the few examples hitherto
found all belong to the 2d person pi. of the middle ; viz.
U^nmufH prasavishyadhvam, " shew ye "' (Bhagavad-Gita,

FORMATION OF THE MOODS. 977

3. 10.) ; ^fr^icn^ bhavishyadhvam, " be ye '' (Maha-Bharata,
III. 14394. Ramayana, ed. Sehl. I. 29. 25) ; and ^i^vvpr
v^syadhtiam, ''find ye/' "obtain ye" (Maha-Bhar. I. 1111.).
The conjecture elsewhere expressed, that by sanvakshyata
(in Stenzler's Brahma- Vaivarta-Purani Specimen I. 35.) a
future imper. act. of the 2d p. pi. is established, I must now
retract ; as, by repeated examination of the passage, I find,
by the context, that for ^^N^^ sanvakshyata, which Stenzler
renders ** aUoquiminir we should read sanrcmchata (i. e.
" arcete'"'').

CONDITIONAL.

730. The Sanscrit conditional bears the same relation in
respect of form to the auxiliary future that the imperfect
does to the present, i. e. the augment is prefixed to the root,
and the secondary personal terminations supply the place
of the primary : hence, e.g., SBlt^l^i^ addsyam,** I would give,'^
and also *' I would have given,'' answering to ddsydmi,
" I will give." We may therefore, as in departure from
my former opinion I am now inclined to do, regard the
conditional as a derivative from the auxiliary future; so
that, although the substantive verb is contained in it, there
is no necessity for assuming the existence of an obsolete

* Observe, that in manuscripts written in Bengal, and especially in the ma-
nuscript used by Stenzler, the r is frequently not distinguishable from the t;,
as is remarked 1. c. p. 10. The V y after the 1B| ksh is added by Stenzler
as an emendation. The meaning aUoqtdminij however, does not agree with
the context, whilst arceie principem corresponds to the sense of the prece-
ding SI. ,In SI. 32 of the same Spec, occurs a form worthy of notice in
respect of syntax, viz. the imperative bruta as representative of the con-
junctive governed by yadi : yadi satyam bruta^ '^ if ye speak the truth."
So in the fifth book of the Mah4 Bhar. the second person plural middle of
the im^^TSkiiye pray achchhadhvamgoyerned. by ch^t : nachit prayachchha-
dhvam amitraghatino yudhishthirasyd 'hsam dbhipsntan svakam, *' if ye do
not give the fiend-slaying Yudishthir his required share." In the Rig
Veda (I. 27. 12) we find the first person plural of the imperative, or LH^
after yadi : yadi Saknavdma, " if we can."

978 VERBS,

dsyairij *' I would be," or " I would have been ;" and even
though such a form should have existed, we might still
regard dsyam as a derivative of asydmi, " I will be " (=Lat.
era, eriSf see. §. 650.), which has disappeared from use ; just
as addsyam as a derivative from ddsydmi. The circum-
stance, that in none of the European kindred languages a
mood analogous to the said one in Sanscrit is to be found,
might lead us to the conjecture, that it is of comparatively
late origin, as in Latin the imperfect conjunctive (see
§. 707.), which resembles it most, but has evidently sprung
up on Roman ground. Compare da-^em from dd-sem, for
dd'Sdim with H^ll^i^ OHldsyam,

731. The Sanscrit employs but seldom its conditional,
which, in the earlier period of the language, is commonly
supplied by the potential : a few examples, therefore, may
be given here (manuscript vii. 20.), yadi na pranayidy rdjd
dandan dandyhhv atandritaK i sdl^ matsydn ivd 'paJcshyan
durbaldn bdUwattardH, " If the king did not indefatigably
punish those worthy of punishment, then the stronger
would roast the weak on spits." But here follow four poten-
tials, all standing in the same relation, which are nevertheless
explained by the Scholiast by conditionals ; viz. adydt,
** would eat," by akhddishyat ; avalihydt ** would lick," by
avdUkshyat; sydt, "would be," by abhavishyaf ; and pravartSta,
*' would become," by prdvartishyat In the eighth book of
the Maha Bh. (SI. 1614) we read, vrijinan hi bhavSt kinchid
yadi karnasya pdrthiva l nd ''smdi hy astrdni divydni prddd-
syat bhrigunandanaKy " If any fault attached to Karnas, O
Prince, the son of Bhrigu would not have given him the
heavenly weapons." The conditional occurs as well in the
antecedent as in the relative sentence, and, in fact, the first
time in the sense of the pluperfect conjunctive, /. c. SI. 709,
nachSd arakshishya iman janam bhaydd dvishadbhir Svam

* For arakshishyas on account of the • following,

FORMATION OF THE MOODS. 979

balibhir prapiditam I tathd ''bhavisht/ad dvishatdm pramddanam
'* If thou hadst not freed from danger -this band assailed
by powerful fiends, then they would have been the joy of
their enemies/' Thus, in the Naishadha-Char. 4. 88, api
sa vajram addsyata chit tadd tvadishubhir vyadalishyad asdv
api, "If he (Brahma) had given also the thunder-bolt (to
thee, the God of love, as a mark), so would even this have
been rent in twain (have been split) by thy darts/'

Remark — In Zend I know of no instance of the con-
ditional ; some resemblance to it, however, may be traced
in the form ^^^yM56iAi»AiM fravacsyanmt at the end of the
44th Ha of the Izeschne (V. S. p. 359), which Anquetil
translates *'je parle clairement.^'' I consider this form to be
the first person of the auxiliary future, which, in the
absence of examples, I formerly thought must end in yemi
(see §. 664.). The fact, that the first person of the future is
very frequently replaced by that of the imperative, is per-
haps the reason of the rare occurrence of the former. If,
however, I am right in explaining the {orm fravacsyanm
as the first person of the future, it has lost the i of the ter-
mination ; as in Prakrit, where, except in the form in
himi (see §. 615.), the termination mi of the future auxiliary
has everywhere dropped the i, whereby, however, the
preceding a has been shortened ; hence, e. gr., T!irft#
sumarissany "I will call to mind,'' correspionding to the
Sanscrit smaHshydmu In Zend, through the loss of the final
i an occasion also for the mutation of the d preceding the w
to i has disappeared ; the termination dm, however, must,
according to §. 61., become 9^ awm\ thus, 9^^,>av3^a}»a}7«
fravacsyanm = Sanscrit if^iVHIlOl pravaJcshydmu In the same
Ha, at the end of which occurs the form ^^^^M^j^»xi/^
fravacsyanm, occurs also six times the form fravacsyd (V. S.
p. 356), which Anquetil, in like manner, translates by *'je parle
clairem^nt " or "je vous parle clairement^^ Then follow
the words which Zoroaster (not Ormuzd, as Anquetil

980 VERBS.

supposes) speaks. If, however, fravacsyd is really a first
person, it must still belong to the future only; and it
would then, in this form, as compared with that in annif be
an abbreviation similar to that of the dual case-termina-
tion bya — ^for which, in accordance with the Sanscrit bhydm,
we should expect byanm — ^and to that of the feminine pro-
nominal locative termination a (see §. 202.) for the Sanscrit
dm. The occurrence in fravacsyd of a long a is in agree-
ment with the fact that, in the Ha above mentioned,
particularly at the end of a word, d is found for an origi-
nally short a ; e. jr. in jui^^as^j) sraotd, '* hear ye." If,
however, JM^yi^M»AiM fravacsyd is not the first person of
the future, it can only be taken as the second person of the
future imperative, and must then be regarded as a com-
mand addressed by Ormuzd to Zoroaster.

DERIVATIVE VERBS.

732. The appellation " derivative verbs " strictly belongs
only to denominatives ; for passives, causals, desideratives,
and intensives, stand quite as near the root as the ten
classes of the so-called primitive verbs, excepting the second
class (see §. 109*. 3.), which latter may be regarded as the
base-form of all the rest. The passive, also, is identical
in form with the middle of the fourth class, and the causal
with the tenth class ; while that form of the intensive which
joins the personal terminations direct to the root is dis-
tinguished from the third class only by the strengthening of
the syllable of reduplication, and in that this extends also to
the universal tenses. And here we must observe that the
tenth class also extends a part of its class character to the
universal tenses. We might — as the passive agrees with
the middle of the fourth class, and the causal with the tenth
class — reckon in all twelve classes of verbs ; so that, per-
haps, the intensives would fall under the eleventh class, and
the desideratives under the twelfth. It is, however, certain

PASSIVES. 981

that the verbs called derivative in idea, and as regards
their origin, must be classed under those which express
only the simple verbal notion along with the relations of
person, time, and mood ; and must also be regarded as
later, and originating in the first place from these latter.
For before there could exist a verb signifying, e.g.,
" I cause to hear,'' or " I wish to hear,'" or " I am heard,"
there must have existed one more simple with the mean-
ing '* I hear ;" and though ^TR^RTftl srdvaydmi, susrusMmu
and srth/i, may be derived from the root itself, sru, more
readily than from srindmU " I hear," or its theme srinu (a
contracted form of srunu), still srunu may stand as the base
form from which the so called derivative and secondary
verbs have proceeded, by the suppression of the class-
syllable mi before the characteristic affix of the derivativie
base referred to ; just as the causal bases, when passives
are formed from them, lose their characteristic affix ay
before the passive character ya ; as, e. g., from srdv-aya-ti,
" he causes to hear," comes srdv-t^a-U (for srdv-ay-yati),
"he is made to hear." According to this scheme the
derivative verbs have, in point of fact, only the bare root at
bottom as formative material ; but the sole reason of this is,
that from the primitive verbs, whose offspring they are,
all ingredients are removed which do not belong to the
expression of the radical idea, in order that the derivative
form should not be too unwieldy ; just as certain compa-
ratives and superlatives spring, not from the full base of
the positive, but from it abbreviated by the removal of the
formative suffix (see §. 298. pp. 395, 396.)

733. Let us now consider the formation of derivative
verbs severally, beginning with the passives. These in
Sanscrit, in the special tenses, annex the syllable ir ya to
the root, and join thereto the personal terminations of
the middle. The conjugation agrees exactly with the
middle of the fourth class (see §. 500.), so that in the present.

982

VERBS.

in the example given at p. 696, we have only to annex the
middle terminations (see §. 512.) in the place of the active.
I give below the 3d per. sing, and pi. with the corresponding
persons of the middle (for the class peculiarities of which
refer to §. 109*.) of the roots budh, CI. 1, " to know " (Goth.
ana-bud, " to command ") ; tud, CI. 6, *' to push"" (Lat. tud,
tundo) ; vas, CI. 2, *' to dress oneself '" (Goth, vasya, '' I put
on" = caus. vdsaydmi) ;* bhar (bhri, see §.1.), CI. 3, "to
bear;'' yuj, CI. 7, "to bind'' (Lat.^*ii^, Gr. ^uy); star (stri,
stfi, see p. 680. Note), CI. 5, " to spread," " to deck ;" pri,
Ci. 9, " to gladden,'^ " to love " (Goth, friyd, " I love ").

ROOT.

budh, CI. 1,
tud, CI. 6,
t?a«, CI. 2,
bhar (bhr), CI. 3,
yuj, CI. 7,
star (stri), CI. 5,
prif CI. 9,

3d per. singular.

PASSIVE. MIDDLE.

3d per. plural.

PASSIVE. UIDDLB.

budh-ya-tif

tud-ya-tS,

ixis-ya-t^,

bhri'^ya-ti,

yuj-ya-ti,

star-ya-t^i

pri-ya'4S,

bddh-a-tL

tud-a-tL

vas-t^.

bibhri'ti.

yunk'tS.

strt-nu'tL

• m

budh-ya-nti,

tiidrya-nti,

vas-ya-nt^,

bhri-^a-ntij

yuj-ya-nti,

sfar-ya-ntS,

pri-ya-ntS,

bddh-a-nti.

tud-a-nt^.

va^-ati.

bibhr-ati.

yv/nj-ati.

stri-nv-afS.

pri-na-ti.

' See §. 459. ^ Roots in ar, which in the pure or light forms con-

tract this syllable to ri^ when only a single consonant precedes the radical
vowel, exhibit the (syllable ri before the passive character ya^ which ri
I consider to be a transposition of tr, and the latter a weakening of the
old form ar^ which has remained after a double consonant; hence,
star-ya-U corresponding to bkri-ya-tS. With regard to the protection
which two combined consonants afford to the primitive syllable ar^ com-
pare the circumstance, that the imperative termination At (from dhi)
remains in verbs of the 5th class after two combined consonants, but
cannot be supported by a single consonant; thus, chinu, '' collect,"
opposed to dpnuhi^ " obtain " (see §. 451 .). By this principle I would also
explain the fact that, the Latin root std (=Sanscrit ^qr »thdy ^' to stand ")
has, almost in every case, preserved the original length of the base- vowel
in opposition to d& (= Sanscrit dd). The transposition of ^^^ bhir to

See §. 169*. 6.

PASSIVES. 983

fij bhri^ reminds us of Greek forms like irarpaa-i^ which has been ex-
plained above as a transposed form of narap-a-i : I am also now of opinion
that in Gothic-plural bases like brothru, dauktru - whence come hrdthryus^
"brother;" dauhthryu-s, "daughter" — we must assume a transposition
of ur to ru ; so that the to-be-presupposed bases, brtohur, dauhtur^ corre-
spond, as weakened forms of hrdthar^ dauhtar^ to the Sanscrit genitives
bhratur^ duJiitur^ which are deprived of their case-termination (see §. 191.
Note).

734. It must be observed, that the incumbrance which,
the root receives in the passive by affixing the syllable ya,
occasionally introduces irregular weakenings of the root ;
as, e. g.j the contraction of vach to uch {uch-ya-tii *' didtur '"'),
analogously with some anomalous forms of the active
(uchima, " we spoke," from u-uchima) : so, too, the contrac-
tion of the syllable ra to ri in the root ITS prachh, " to
ask f ' "^^^ prichchhyaU, " itUerrogatur ;' as, ^^l(i4 pri-
chchkdmit " I ask f ' paprichchhima, " we asked," compared
with paprachcha, "I asked;" prashtum, "ask ye." This
principle also explains the fact, that some roots in d change
this vowel in the passive to the lighter i ; hence, e.g., diya
is the passive base of the root dd, "to give" (diyat^,
" datur "). The Zend, on the contrary, as a consequence
of the same principle, shortens the long am d to a5 a, at
least in the examples which occur to me : j^^^-5x»6^A5^^y
nidhayfint^f " deponuntur "* ( == Sanscrit nidhiyant^) ;
Aj^^j^Aj^^AjyjJ snayanuhoy "be washed""^ ( = Scr. sndyasva);

* Vendidad Sade p. 246 : (?A5(en>^7^) As^e^Og^-J V-^/ ^^^^P^
H^^JVg» Jxi^^AJ^i yamnya naro irifsta (irista ?) nidhayHncM, ** in qud
{terra) homines martui deponuntur;" according to Anquetil (p. 326),
*• dan8 les quels on a mis des hommes morts" see Note t.

t With middle meaning, "wash thyself" (jsaSta, "the hands") (see
p. 967, Note **). Bumouf (Ya^na, p. 361, Note) takes the syllable ya
of this form not as the passive character, which according to him (1. c.
p. 369) must be looked for in Zend little more than in Greek and Latin.
It appears to me, however, that we may be very nearly right in regarding

the

964 VERBS.

As^MAs^^yjj snayaStOf ** let him be washed/'' or '' wash
himself '^ (see. p. 957, Note). In support of the view, that
the forms snayaniJia and snayaita may be taken as passives
with a reflexive signification, it may here also be adduced
that in Old Persian a similar phenomenon occurs ; viz. in
^.yyy^.ff.^-Y^'-.^ff.^y^.yyf paHpayanvd* (Beh. IV. 38.),
which Benjfey, in my opinion rightly, renders " guard thy-
self " (Rawlinson by " te expeditum habe% and refers to the
Sanscrit root m pd (with the preposition po^i =prati,) which,
therefore, in agreement with the Zend, has shortened the
long d before the passive character.

735. If, with the Indian Grammarians, we regard the
Sanscrit ^dy^ (irregular (or janyS) " I am bom," as a middle
of the fourth class (see §. 500.), then the corresponding
Zend verb may be explained in the same manner. As, how-
ever, the meaning " to be born " is strictly passive, and

the syllable ya in the form above mentioned as the passive character,
and the whole as a hy-no-means-surprising change of the passive into a
reflexive or middle meaning, while in Greek, Gothic, Latin^ Lithuanian,
and Sclavonic, the reverse is the case. If the form ->^%!^?^,>A5(0J i
nidhaygnti^ " Us deposmt" which Bumonf has mentioned at p. 361, and
which I am unable to quote, he only a difierent reading of the nidhayiintS
mentioned above in the lithographed manuscript, I would also then re-
cognise in it a passive, and draw attention to the fact, that in Sanscrit
also, in the passive, the active terminations not uncommonly take the
place of the middle, so that the passive relation is to he discerned only in
the syllable ya (see Lesser Sanscrit Gram., 2d Edit. §. 44G). If, how-
ever, we take nidhayifnti as active, we must then explain ^'they lay
down" in the sense of " one lays down," and consider nard irista as the
accusative (see p. 247). Constructions of this kind, as far as I know, are
not confirmed by unmistakeable forms, and I therefore prefer explaining
the verb as passive.

* Rawlinson and Benfey read patipayuwd ; I doubt not, however, that
the a inherent in y^*- y must be here read in conjunction with it. The
termination uvd, for huvd (euphonic for hvd\ corresponds to the Sanscrit
imperative termination sva.

PASSIVES. 986

the form of the middle of the fourth Class iid identical with
that of the passive, I prefer to explain in both languages
the forms with passive signification as really passives ; and
I adopt for the Sanscrit a middle jan of the fourth Class, a
kind of deponent with the active meaning " to bring forth,"
of which, however, but few examples occur, as, e.g., Ramay.
ed. Schl. I. 27. 3. iH ^vfNri putran vy-ajdyata, '* she bore a
son" (with the prep. vi). The Zend root yAjj zan, the
passive of which frequently occurs in combination with the
preposition ^y us ( = Sanscrit gir vt), likewise rejects the
final n before the passive character ya : the preceding a,
however, is not lengthened, or the long A, which had been
introduced, is again shortened ; which cannot surprise us,
as from the first the long A at the end of a root is shortened
before the passive ya. Hence, e. g., (^C^j^Jxj^^ajjjj; us-zayiintit
"they are born,"* corresponds exactly to the before-men-
tioned nidhayiinti (§. 734), Of the imperfect we find the
second and third person singular; viz. aj»«3lu^^a5ja5J3;
usazayanha, " thou wast bom," (see §. 466. and §. 518.), and
uszayata, "he was born".t

736. As the middle of Sanscrit verbs of the fourth Class
is identical in form, and, as I believe, in origin also, with
the passive, and therefore f^ mriy^r " nwri(yr,''' Thm^ rnriyai^,
" m&riturr may also stand for the passive, it may here be
remarked, that the corresponding verb in Zend, the con-
junctive of which, mairydith frequently occurs (Vendidad

* Vend S., p. 136, aj^ajj A5>>^ Asi^g?^ i as^as^ M^jyi^M^

AJ^JJAS^^ZjAsy AS^J^^Ji) AsyAJQicf^^J^ tJJ^^-^ i'^^^^^^) dvaSibyo hacha
nffribya dva nara tiizayS inti mithwana Stricha nairyascha, '' duobus ex
haminibus duo homines nascuntur, par^ feminaque masque," Anqaetil
(p. 278), translates " de deux hommes naquirent deux hommes distinguish le
male s*Siant uni d lafemelU"

t Vend. S., p. 39, yat h^ (so I read for w he) puthrS uszayata^ "that
a son was bom to him."

3s

986 TBBBS.

Sade, p. 24*), has replaced the middle termination by the
active, as also in Sanscrit the active termination frequently
takes the place of the middle in acknowledged passives. The
above-mentioned mairyditi is so far older than the corre-
sponding Sanscrit verb, in that it has experienced neither the
transposition of ir to ri mentioned at §. 733. Note 2. (mri-
ycrfA like bhri-yati) nor the weakening of a to i, but mairyditi
" moricUur ^' stands for maryditu in consequence of the assi-
milative power of the y (see §. 41.), and affords us a new
proof of the unoriginality of the Sanscrit ^ ri\ and shews
that in Sanscrit not mrh but mar^ is the true root, whence
comes, in Latin, mor, which presents to us in the to, iu, of
morior, moriuntur, a fine remnant of the Sanscrit passive
character ya ^. Compare iu in mor-ivrrdur with the Sanscrit
ya of mri-ya-ntS, The conjunctive mor-ia-r, mor-id-risy gives
us still more exactly the character of the Sanscrit passive,
only that here the Latin d appears long, inasmuch as it has
absorbed the modal exponent i. The Lithuanian also has,
in the said verb, preserved the passive character, which we
have already (§. 600.) recognised in gemmu from gem^yu,
** I am bom," gim-yau, " I was bom.'' So we have mir^
iau, " I died," while the present mir^ztu, " I am dying,"
belongs to a different conjugational form. In Latin, too,
may be mentioned fio as a remnant of the old passive. I
divide the word thus,/-20, and regard it as an abbreviation
o{ fur-iOf (just as in Old Persian 6-iyd,t "let him be"==
Sanscrit bhuydt), and therefore analogous to the Sanscrit

* The Gothic also present* a remarkably analogous form to the Sanscrit
Jd^Sy '' I am bom," in the isolated form tu-kiyanata^ ^^enatum" (Luc.
viii. 6.), which presupposes in the present us-Juya^ ^^emucor" and there-
fore a simple verb, ki-yay ^' nascor" for kin-y a, as in Sanscrit, y^y^ for
janyi,

t Euphonic for by&y as y unites very often with a preceding consonant
without a preceding i.

PASSIVES. 987

bhi^i*, exclusive of the middle personal termination of the
Sanscrit. Compare, therefore, f-iu-nt, with bhu-ya-nlS, f-ie-t
with bhu-yS-ta, f-ii-mus with bhu-yS-mahu As the Sanscrit
passive is frequently used impersonally in expressions like
WniTH srw/atdm, " let it be heard," instead of ** hear thou,"
>Hl^rilR dsyatdm, "let it be placed,'' i?^ mamr^, '*let it be
dead,*" I will also here further observe, that in Georgic,
whose grammatical relations with Sanscrit I have elsewhere
pointed outt, such modes of expression are very common,
viz. in the verbs or tenses called by Brosset "indirect,"'
whose element of formation, ia or ie, presents an unmis-
takeable resemblance to the passive character; compare,
c. g., SffsoSob m-gon-ia, ** it is thought by me '' ( = Sanscrit
^m FR^ mat/d jnd-ya-U, *' it is known by me ") for " I
think,'' ^a^oggs^gcios shS-mi-qwareb-ia, "it was loved by
me " = " I had loved " (see "The Caucasian members," &c.,
p. 59). But the common Georgic passive also, where it is
retained, corresponds, in its principle of formation, to the
here mentioned ij ya, and most clearly in the third person
plural, e.g., in ^flog^^^j^ioiC she-i-qwarebian, *' amarUur,''^
answering to the active "<)^33^3^flK she-i-qwareben,
** amant,^^ the termination of which, in its abbreviation,
corresponds to our German forms, as Ud)en (from liebent)
1. c. p. 56.

737. Originally the Sanscrit passive character ya may
perhaps have extended over the universal tenses ; and in
roots ending in d or a diphthong I think, even in the pre-

* The passive of bhu 'Ho be," must be looked for as impersonal only in
the dd per. sing., as we also find the nent. of the part. fnt. pass, in con-
structions of this kind; e,g, (Hit. ed. Bonn. pp. 17* 20.), tavd 'nticharSna
may a bhavitavyam, ''mine is it to be thy attendant "=" I must be thy
attendant." The idea "to be" is expressed by the active of bhu^ as
bhavdmi means as well " I become," as " I am."

t '' The Caucasian members of the Indo-European family of languages."

3s^

968 VERBS.

sent state of the language, I recognise a remnant of it,
viz. in the y» which, in the aorist, the two futures, the
precative, and the conditional, precedes the conjunctive
vowel i; e.g., in addyUhi* "I was given," ddyiidhi and
ddyishyS, " I shall be given," d&yisMya, " may I be given,"
addyishyS, " I might be given." I am led to this view
principally by the circumstance, that that form of the in-
tensive which, on account of its passive form and active sig-
nification, I term deponent, retains the passive character in
the said tenses and moods after vowels other than d ; hence,
e.g.^ aMchiyishi, ** I collected," chichiyifdhi, cMchiyishhyi,
'* I will collect," from f^l chi* If the ^ y occurred only
after ^t d, it might be assumed, as was formerly my
opinion, to be a mere euphonic insertion (see smaller
Sanscrit Gram. §. 49*.), as, e. y., in inftnT yd-y-in, " going,"
from yd with the suffix in. The reduplicated preterite of
the passive is in all verbs, like the corresponding tense in
Greek, exactly like that of the middle ; so that, e. gr., ^^;$
dadmi signifies, as middle, " I or he saw," and as passive,
" I or he was seen." Moreover, the reduplicated preterite
or perfect is that one of the universal tenses of the passive,
which, with the exception of the third person singular of
the aorist, is the only one in common use. I cannot re-
collect to have seen in any author other universal tenses, or
other persons than the third singular of the aorist.*!*

* Before the y of the passive character i and u are lengthened, as gene-
rally the y exerts a lengthening power over i and u preceding it, except
when the iy is only a euphonic developement of f or t, as, e.g., in bhiyas,
"/tworw," from fcAi+a*. Observe, with respect to the lengthening in-
fluence of the Sanscrit V, that in Latin also J within a word alone pro-
duces for itself length by position.

t This ends in t, and wants, the personal sign, e.g., ajam, "he was
bom." In this t might be recognised a contraction of the passive cha-
racter If ya : to this view, however, are opposed forms like cMyi, ^' he

was

PASSIVES. 989

738. With respect to the origin of the passive character
If yoy a very satisfactory explanation, I think, is given of
it by Sir G. Haughton, wherein he mentions that in Bengak
and Hindustani the passive relation is expressed by an
auxiliary verb, which signifies "to go": "^(T^jdnA (from
ydndy see §. 79.), in Hindustani, and in y& in Bengali ; in the
latter, e. jr., "^FTT ^Tf^ hard ydi signifies " I am made," as it
were " I go in making." Now in Sanscrit both ^ i and IJT
yd. Class 2, signify "to go"; but of these it appears best
to keep to the latter root, which, in Bengali, also expresses
the passive relation : and I believe that the shortening of
the syllable ^ yd to ij ya is to be ascribed to the root being
burthened by composition, which rendered a diminution of
the weight of the auxiliary verb desirable. The a of the
passive ya is therefore radical, and not, as in the first and
sixth Class, a conjugational affix *. it follows, however, the
analogy of the class syllable a, just as, according to §. 508.,
the root WT sthd, "to stand," after its abbreviation to ^
stha subjects its final a to the analogy of verbs of the first
and sixth Class. Through the middle terminations com-
bined with the appended auxiliary verb, and expressing the
reflexive relation, the auxiliary keeps the meaning " to go
oneself"; and while the Bengali kard^ydi signifies simply
" I go in making,"*^ the Sanscrit composite implies more,
viz. "I go (betake) myself in making." Compare the
Latin constructions like amatum irU "to be gone in love":
remark, also, veneo in opposition to vendo; as also the ex-
pressions of such common occurrence in Sanscrit, like " to

was given," becanse here y is the passive expression : the t, however,
most probably is identical with that of addy-i-shi^ " I was given," addy-
ishnuiy "we were given:" addyi^ therefore, would be an abbreviation of
ciddyiahta,

* In his edition of Mann, B. I. p. 829, and in his Bengdli Grammar,
pp. 68 and 95.

?

990 VERBS.

n ) go in joyi'' "to go in, anger/" for " to be rejoiced," " to
be angered"": we even find grahanan samupdgamat "he
went in seizure,"" for "he was seized,"" in the Ram. (of
Sehl. I. i. 7a).

CAUSALS.

739. The Sanscrit and Zend causal is, in its formative
character, identical with that of the verbs of the tenth Class
(see §. 109\ 6.). In explanation of the affix vrr ay, in the
special tenses ^n aya, the Sanscrit furnishes the roots ^ i,
"to go,"" and ^ i, "to wish,"" "to demand,"" "to pray"":
from both arises, by Guna, before vowels ^htt ay, and in
combination with the character of the first Class, ^vi( ayia.
The meaning "to wish,"" "to demand,"" appears, perhaps,
adapted to represent the secondary notion of the causal
verbs, in which the subject completes the action, not by
the deed, but by the will : thus, e.g., kdraydmi, " I cause to
make,"" would properly mean "I require the making,"" whe-
ther it were intended that " any one made,"" or " any thing
was made."" But if the causal character springs from a root
which originally signifies " to go,'" we must then observe,
that in Sanscrit several verbs of motion signify also
"to make""; e.g„ vidaydmi might properly signify "I
make to know."" '•' ^

740. Although, as has been remarked (p. 109), all Ger-
man weak verbs are based on the Sanscrit tenth Clads, still
that form alone, which has most truly preserved the Sanscrit
aya, viz. that which in Gothic, in the 1st per. sing, pres.,
terminates in ya (Grimm's first weak conjugation), is used in
the formation of causal verbs, or of transitive from intran-
sitive verbs, but not in such a manner that the language,
like the Sanscrit, could form a causal from every primitive
verb, but rather so that it is content with those handed down
from old time. These, in Gothic, agree with the Sanscrit
causals also in this point, that the radical vowel always
appears in the strongest form that the primitive verb has

CAUSA LS. 991

developed . Hence, the weakening of a to i, which the
primitive or strong verbs have frequently experienced in
the present, is not admitted in the causal ; and the vowels
% and t£, which are capable of Guna, are Gunized ; and, in
fact, through the original heavy Guna-vowel a, not as in
the present of the primitive through i (see §. 27.). Generally,
in Gothic, the causal exhibits the vowel of the monosyllabic
forms of the preterite of the primitive, yet without its being
possible to say that it is derived from the latter ; but the
causal and the singular of the preterite of the primitive
stand, with respect to their radical vowel, in a sisterly, not
in a derivative relation. Compare, e. jr., satya^ "I place,"'
(R. sat) with sita, " I sit,"" sa/, " I sate," and with the San-
scrit causal sddaydmU from the root sad, perf. sasdda ; thus,
lagya, ** I lay," from the root lag (liga, ** I lie," lag, " I lay");
nasya ** I make whole,'' " I heal," from the root nas {ga-nisat
" I recover," pret. ga-nas) ; sagqvya, ** I sink, make to sink,"
from the root sagqv (sigqva ** I sink," pret. saggv) ; dragkya,
" I drank," from the root dragk {drigha, *' I drink," pret*
dragk) ; ur-rannya, " I cause to go up," from the root rann
{ur-rinna " I go up," pret. ur-rann). Examples of Gunized
u in the Gothic causal form are the following : ga-drausych
** I make to fall down," " I throw down," from the root drus
(driusa, ** I fall," pret. draus., pi. drusum ; compare Sanscrit
dhvans, " to fall," §. 20.) ; lausya, ** I loosen," from the root
lus {fra-liusa, " I lose," pret ^laus, pi. ^lusum ; compare
Sanscrit lu, " to tear away," " to cut oflF"). So in San-
scrit, e.g., bddhaydmi (6=au), "I make to know," "I
awaken," from the root budh ** to know," " to wake up."
The following are examples of the Gunizing of i to ai : ur^
raisya, " I set up," from the root ris (ur-reisa, " I stand up.

^s

* Those forms only are admitted which have arisen from the contrac-
tion of reduplicated preterites (see §. 606.) : in Sanscrit, however, the 4,
e.g.y ofiddaydmi is heavier than the 4 (=a+t) of Mima.

99!i VERBS.

pret. uT^rais, pi. ur'risum) ; hnawya, " I lower," from the
root hniv Qineiva, ** I bow myself," pret. hnaiv, pi. hnitmm).
So in Sanscrit, e. g., vidaydmi (^ 6=^ at) " I make to know,"
Zend. j^^^^^AAs^ va&dhayimi , from vid, " to know.***
Our new High German causal remains, such as seize,
" place," lege^ " lay," senke, " sink," are, by reason of ab-
breviations of their endings, no longer to be distinguished
from their primitives, and furnish a remarkable proof of a
corruption of form gradually reaching a point where it
becomes imperceptible. Without the fortunate preserva-
tion of Gothic forms like scdya, and other formations of
the Old German dialects, corresponding more or less, it
would have been impossible to trace in the e of setze a re-
lation to the Sanscrit ay&mi 6f sddai/dmi, and hence an
agreement in the principle of formation of the German and
Sanscrit causals. So early as the Old German the causal
character appears much defaced ; e, g., in nerent, '* aJunt "
(vivere faciunt) to be found in Notker, for neriant, Gothic
nasyand ; kgo ^^porw^ for legio, legiu, Gothic lagya ; legent,
** ponunJtr for legiant, Gothic lagyand, 1. c.

741. In Old Sclavonic that conjugation corresponds in
which we, in §. 505., have recognised the Sanscrit tenth Class :
it therefore corresponds also to the Indo-Germanic causal
formation : it also contains the verbs which by their sig-
nification alone rank as causals, and to which, as primi-
tive, corresponds a non-causal or intransitive verb. In
accordance with the Sanscrit-Gothic principle noticed in the
preceding §. these casual verbs exhibit a heavier vowel than
the primitive, or they contain a vowel, while the primitive
has lost its radical vowel. Thus, as in Sanscrit, from the

* It often occurs in combination with the prep, ni; j^^^^as^joas^jjj
nivaidhaySmi^ according to Anquetil, ^^jeprie;" according to Neriosengh,
fvfYff^nnfH nimantrayami^ i. e. '^ I summon" (see Burnouf, Ya<;na, p. 410).
With regard to the foundation of the 6 of the termination hni see p. 963,
Note.

CAUSALS. 993

root mar, "to die" (in its abbreviated form, ^, which
Grammarians regard as the primitive), comes the causal
mdraydmit '* I kill," "I make to die"; so in Sclavonic,
from the radically abbreviated m^& mru^ " I die," comes
a causal, MO^Ii^ moryuy "I cause to die" (Dobr. p. 36l),
which perhaps no longer admits of citation in Old Sclavonic,
but is confirmed by the Russian Mopio moryu. The same
is the case with ba^hth var-i-ti, "to cook" (trans.), com-
pared with B^tmH vr-ye-ti (intrans.), with b&ahth bAd-i-ti,
"to wake," compared with B'bA*niM bhd-ye-th "to awake"
(Sanscrit hodhaydmU " I wake," budhyS, " I awake "). For
the e of the primitive the causal receives the heavier o ;
hence, e.g., hoaoAhth po-losch-i-ti, "to lay," compared
with AE^sATH lesch-a-th "to lie." The a of sad-i-H, "to
plant," properly " to set," corresponds to the Sanscrit d of
add-ayd-mi (Goth, satya, " I set"), while the t ye of CtCTH
syeS'tif " to place oneself" (euphon. for syed-ti, see §. 457.),
has probably first weakened the short a of the root to e,
and then (as is commonly the case in Sclav.) prefixed a y.
Compare the Lithuanian sidmU "I sit," answering to sodinu,
"I plant," with the remark that the Lithuanian o frequently
supplies the place of the long d, as, e.g., in the nom. pi. of
feminine bases in a (aszwos = Sanscrit asvdff, " the mares ").
Here may also be noticed the Irish suidiughalmt "I set,"
"plant" (answering to suidhim, "I sit"), where gh, as
generally happens in the Irish causal verbs, represents the
Sanscrit y (compare p. 110, and Pictet, pp. 148, 149). Of
Sclavonic causals notice also ^acthth rast-i-tU "to increase,"
properly, "to make to grow," (rast-ye-ti, "to grow"),

* Sanscrit vardhaydmi^ Zend varedayimiy " I make to grow," " I in-
crease." llie Sclavonic verb has retained the affix ty whence the radical
d must become *. As, however, the primitive verb had already an «, an
augmentation of the vowel in the causal was impossible. Compare also
the Sanscrit ridh (from ardh\ " to grow," which is probably an abbrevia-
tion of vardh*

994 VEBBS.

BftCHTH vyes-^i-ti "to suspend,"'' (vU-ye-ti, ''to hang ^^), nc^po-
i-ti, "to give to drink'" (na prep., pi-ti, "to drink"), po^ko^
i'th "to quiet/' (po-chi-ti, "to rest"). As the Sclavonic
% ye is the usual representative of the Sanscrit '^ is=ai
(see §. 255. e.), so is the vowel relation between vyes-i-ti, " to
suspend," and the root vis, "to hang," like that of the
Sanscrit vi&^yd-mh " I make to enter," to vUdmi, "I go in.**'
The Sclavonic root vis is also probably identical with the
Sanscrit vis, which, in combination with the prep, f^ ni
in the causal, signifies, among other things, "to adjoin,"
" to annex," and brings us, therefore, very near the sig«
nification of the Sclavonic causal, viz. "to suspend," as
fi:enerally the Sclavonic and Sanscrit roots meet one ano-
Lr in the idea of "approaching" (^nf^ d.U means
" to approach," T^flr^ upavii, " to place oneself"). The
formal relation of (na)poitif " to give to drink," to piti, " to
drink," cannot be correctly measured without taking in
the Sanscrit ; for from a Sclavonic point of view it would
seem as if poUi had arisen from piti by the insertion of
an o, while, in fact, the o of poiti rests on the Sanscrit d
of the root p4 to which corresponds the Greek w of 11(0-61,
itentaKa, and the o of eiroOrjv, as also the Latin 6 of pd-tum,
pd-turus, and the Old Prussian uo of puo-ton, " to drink ** : the
i of piti is based, like the 7 of the Greek m-dt, mv<a, on the
weakening which has already occurred in Sanscrit of pd
to pi, whence the passive pi-yati, " hibUur,"'' the perf. pass,
part pi'ta-s, " drunken," and the gerund pi-tvd, " having
drunk." The Sclavonic causal has, according to the gene-
ral principle, preserved in po the heavier vowel of the
root, and that which stands nearer to the original d. The
relation o{ po-koiti, "to quiet" (po-ko-i-ti, po prep.), to
po-chi-ti, "to rest," is, however, of a different kind. For
if, as I doubt not, Miklosich is right (Radices linguae
Sclav, p. 36) in comparing the Sclavonic root mh chi with the
Sanscrit Si (from ki), " to lie," " to sleep," it must then be

CAUSALS. 995

observed that the said Sanscrit root, as also the kindred
Greek root Keiiicu, assumes an irregular Guna augment, which
extends throughout, and which appears in Greek either in the
form of Kelt or in that of koi {koItyj, Kol'Tog, Kotfxdiji), see §. 4.).
To the latter form corresponds the Sclavonic ko of po-ko-i-tU
where, however, the radical vowel is lost, for the following
i is the expression of the causal relation.

742. The form i, in which, in Old Sclavonic, the causal
character for the most part appears, corresponds exactly
to the form into which, in Gothic, the causal ya contracts
itself before the appended auxiliary verb of the preterite
(see §. 623), and before the sufiBx of the pass, participle ;
therefore, as we have in Gothic, sat-i-da, " I placed,"' sat-i-th^s,
'* placed" (Gen. sat-i'di-s) ; so in Sclavonic, sad-i-ii, ''plan-
tare,^'' sad-i-fy, **plardatr sadni-shU ** plantas,'''' sadni-m, "plan"
tamus,^'' sad-i-te, ** planiatis.^'' In the 1st per. sing, and 3d per.
pi. of the pres. I& yu (from yo-m), lATb yaty (from yanty), cor-
responds to the Gothic ya, yandf Sanscrit ayd-mi, aya-nti,
provided that euphonic laws do not introduce an alteration,
as is the case, e.g., in CAilsA^ saschdu for sadyu. In the im-
perative (see §. 626.) the causal character is lost in the mood
exponent; hence sadi, ** plardes,'*'' "plantef^ {Goth, satyais,
satyat), idLjs^^wbsadyem, '^plardemus,'*'' caa*te sadyete, "plantetis*^
(Goth, satyaima, satyaith), as nesi, ^yeras^ ^yeraV With regard
to the preterite of the Old Sclavonic causal, corresponding
to the Sanscrit aorist see §. 561., where, however, the i of
B&A«% bud'i'ch, "I did wake," corresponds, not to the
Sanscrit i of abodh-i-sham, " I did know," but, as has
already been remarked (§.562.), to the exponent of the
causal relation ; while in Sanscrit the aorist is, with the
exception of the precative active corresponding to the
Greek aorist optative, the sole tense in which the Sanscrit
divests itself of the character aya (in the universal tenses
ay). As, however, all causals assume the reduplicated
form of the aorist (see §. 580.), so the incumbrance of the

996 VERBS.

root by the reduplication, combined with the augment, is
perhaps the reason of the loss of the causal character :
perhaps even the reduplication is held as compensating for
the causal expression, just as, in Latin, sisto, opposed to the
unreduplicated and intransitive st(h or as in gigno = Sanscrit
jajanmi, " I beget,'* opposed to nascor from gnascor.

743. The Lithuanian very seldom uses for the formation of
causals from primitive verbs the forms contrasted in §. 506.
with the Sanscrit ^n? aya. The only examples which
occur to me are zindaUf " I cause to suck," ^rom zindu,
" I suck," and grdu-yu, " I pull down (make to fall in) a
house," from gruw-u, " I fall in like a house." The w of
grUw'U appears to be only a developement from the iJ, as,
in Sanscrit, forms like babhuva, " I was," " he was," from
bhu. If we take gru as the root, the causal form gr&u-yu
corresponds in its vowel increment to Sanscrit causals like
bhdv-ayd-mi, " I make to be," " I bring into existence,"
from hhu, " to be." The usual termination of Lithuanian
causals is inu (pi. ina-me), by which, as in Sanscrit by aya, are
formed denominatives also, as e.g., ilg-imi, "I make long,"
a denominative causal from itga-s, " long." The n of these
forms, in departure from that mentioned above (§.496.),
extends over all tenses and moods, as well as to the parti-
ciples and the infinitive ; for I cannot agree with Mielcke
(p. 98. 10.), in considering it to be a deviation from this
rule, that before s (according to Sanscrit principles) it
passes into the weakened nasal sound, which I express,
like the Sanscrit anusvdra, by n (see §. 10.) ; thus, e.g.,. laup^
sin-su, " I will praise."

744. The Lithuanian formations in inu agree with the
Sanscrit, Zend, German, and Sclavonic causal verbs in this,
that they love a heavy vowel in the root ; so that many have
preserved an original a, while the primitive has corrupted
that vowel to i or e ; whence they appear to us exactly in
the light of the German Ablaut system (see p. 38, Note).

CAUSALS. 997

Thus, as e.g,9 in Gothic, to the intransitive sita, "I sit"
(which is a weakened form from sata), corresponds a pre-
terite saty and a causal satya^ ** I place '' ; so in Lithuanian, to
the neuter verb mirsztu, '* I die,'' answers a causal marinu,
" I cause to die *" (Scr. mdraydmi, Sclav, moryu) ; and to the
gem-mu (from gem-yu), " I am bom,''' represented above
(§. 501.) as passive, corresponds a causal ga-minu, " I beget"
The following are causals, with a answering to the e of
the corresponding intransitive : gadinu, " I ruin," " kill,"
opposed to gendu, nagendu, "I am ruined"; kanJdnu, **I vex,"
opposed to kenchiu, " I sufier." In the Lithuanian causals
also, in place of the organic a, o is found answering to the e of
the intransitive (as in Sclav., §. 742.); for example, in sodinu,
"I plant," answering to sedmi, "I sit." There is much
that is interesting in the vowel relation of pa-klaidinu, " I
mislead," " bring into error," to pa-klystu, " I mislead my-
self" (euphon. for pa-klydr-tu), for the y is, in pronunciation,
identical with i; so pa-klaidinu, in respect to its Guna
form, corresponds very well to the Gothic causals like
hnaivya, "I humble," and Sanscrit, as vidaydmi {^'Daida-
ydmi), "I make to know" (see 109.* 16.). The same is the
case with at-gaiwinuj "I quicken" (properly "I make to
live," compare gywas, " living," Sanscrit jiv, " to live "),
the primitive of which, ** I recover myself," " become fresh
again," " lively," is probably an abbreviation of at-ginyu ;
waidinU'S, ** I shew myself" (see §. 476.), contains a stronger
Guna vowel than weizdmU " I see," and corresponds to the
just-mentioned Sanscrit causal vidaydmi. An example of
the manner in which a Lithuanian causal has, just like its
corresponding intransitive, corrupted an original a to e, is
deginu, '^uro^ answering to the intransitive degu , "ardeo^

* In Sanscrit the fourth Class of the root dah {dahydmi " ardeo ")
represents the intransitive meaning, and the first Class (dahdnU *'uro**)
the transitive. On the latter is based the Irish daghaim « uro"

998 VEBBS.

745. The circumstance that the Lithuanian formation ina
(1st per. sing, inu), like the Sanscrit ayOf forms as well causals
as denominatives, and that the causals so formed, like the
Sanscrit, German, and Sclavonic, prefer a powerful radical
vowel, gives us ground, (in variance from the assertion set
forth at the end of §. 495. which I gladly retract), for seeking
to compare the Lithuanian ina and Sanscrit aya. We might
in the i of ina recognise the weakened form of an original
a, as it appears also in the forms mentioned at §. 506. in
iyu, iya. The n, then, as semi-vowels are easily inter-
changed, must be held to be a corruption of ^ j^ . The i,
however, of ina, inu, as in the forms in iu, plural i-me
{myt-i-me, " we love " §. 506.), might correspond to the San-
scrit y of the derivative aya ; so that, €.g,, the syllable in
of sod'in-ti, " to plant," would be identical with the i of the
Sclavonic sad-i-ti of the same meaning, and with the Gothic
i of sat'i-ta, " I placed," (compare §. 743.). The n of the
Lithuanian form would then be an unorganic affix, like a rind
which has grown upon the vowel termination of the verbal
theme, according to the same principle by which, in Ger-
man, so many nominal bases have received the affix of n;
so that, e.g., to the Sanscrit base vidhavd, "a widow" (at
the same time a nominative, see §. 137.), to the Latin vidua,
and Sclavonic vdova, corresponds a Gothic base viduvdn
(Nom. -vd, §. 140.) ; and to the Sanscrit feminine participial
bases in anti respond Gothic bases in andein (Nom. andei).
If this view be taken, we must then assume that the verbal
theme of sodi (Sanscrit sddaya), extended to sodiu, has taken
up the character of the Sanscrit first conjugational Class, and

* See §.20. As regards the transition of the y into another liquid, re-
mark the relation of the German Leber (labial for guttural, as in Greek
^Trap, see Graff, II. p. 80) to the Sanscrit yakrit (from yakart) and Latin
jecur. With respect to the transition of / to n, observe, e. g,y the relation
of the Doric ^vOov to fjkBov.

CAUSALS. 999

has thus entered into the Lithuanian first conjugation ; thus
sodifira-me, " we plant," as suk-a-^me, " we turn,'' In favour
of the first mode of explanation might be adduced the cir-
cumstance that, together with szhwinu, " I praise," " extol,"
exists a szlmjoiyu,^ which latter is clearly identical with the
Sanscrit irdvaydmu " I make to hear," and Russian CAasAio
Blavlyuy " I laud." Since in Latin, as I think I have
clearly proved, three conjugations — ^the first, second, and
fourth — correspond to the Sanscrit tenth Class, we have
reason to look among these for the Latin causals, as already
(p. 110.) TRoneo has been compared with the Sanscrit mdna-
ydmi and Prakrit mdnimi, " I make to think." The causal
meaning, however, is no longer apparent in the Latin moneo,
as it has not any primitive verb corresponding to it, from
which it might have been derived in a regular way, and
one, as it were, often trodden for similar purposes; for
memini may be regarded as a sister form connected with
it, both in sound and sense, but not as the parent of which
it is the ofispring, Sedo, which corresponds to the Sanscrit
causal sddaydmi and its German-Sclavonic sister forms
(sed-d'S = '^J^^f^ sdd'a(y)a'si), might, according to the
sense, be regarded as the causal of sedeo; but the latter is
in form likewise a causal, and there is a want of other
analogous cases for the formation of causals by the change

* Ruhig doubles the n of laupsinu in both the plural numbers and in
the third person singular of the present and perfect. Mielcke, on the
other hand, makes no remark, p. 98, 10. with regard to the necessity of
such a reduplication, where it does not already occur in the first person
singular of the present. For the rest it may be remarked, that liquids
especially are easily doubled, and that, e.^., in Sanscrit a final n, if pre-
ceded by a short vowel, is doubled in case the word following begins with
a vowel.

t The kindred klausau, ^' I listen," has, like the Greek ieXv<o, preserved
the original guttural, which in azhwiyu^ as in the Sanscrit iru^ has be«n
corrupted to a nbilant.

1000 VERBS.

from the second to the first conjugation. In Latin, there-
fore, the three verbs sido, sedeo, and sedo, can only be re-
garded as three kindred verbs, which, each in its own way,
are referable to the Sanscrit root sad. To the Sanscrit
trdsaydmi, (Prakr. tdsimi), " I make to tremble," *' to fear,''
" I terrify," corresponds ierreo by assimilation for terseo,
from treseo. The fourth conjugation presents sdpio as a
form fairly analogous to the Sanscrit causal sv&payAmi, " I
make to sleep," {svapimi, " I sleep," irregular for svapmi).
Old Northern svepium, " sopimus,'*'' (singular svep). Old High
Grerman in-su^epiu, Russian ycbiOAaio usyplayu . The causal
notion, however, is lost in this sdpio also, as there is no
intransitive sopo of the third conjugation corresponding to
it as a point of departure. The German dialects have, in-
deed, preserved the primitive (Old High German sldfu).
but it has become estranged from the causal by the ex-
change of the semi-vowel v for I (see §. 20.). In Russian,
on the other hand, chak) splyu, " I sleep" (euphonic for spyu),
corresponds, as verb of the Sanscrit fourth Class (see §. 500.),
to the causative u-syplayu (u preposition), the y of which
is based on the Sanscrit u of contracted forms like sushu-
pima, " we slept," ^upta, " having slept ;" with which, also,
may be compared the Greek inr of vnvo^. I here place
opposite to one another the corresponding forms of the
Latin and Old High German languages for comparison
with the Sanscrit svdpaydmi and its potential svdpayi-y-am
(see §. 689.) :

svdp-ayd-mi, s6p-io, in-suep-iu,

svdp-aya-si, sSp-i-s, in-suep-i-s.

svdp-aya-ti, s6p-i-t, in-suep-i-t

svdp-ayd-mcis, sSp-i-miLS, in-suep-ia-m.

svdp-aya-tha, sdp-i-tis^ in-suep-ia-t.

svdp-aya-ntif s6p'iu-nt, in-siLep'ia-nt

* The I is only a euphonic affix required by p ; ayu therefore=sfl^4mi

GAUSALS. 1001

svdp^yS^-am,* sdp^ia-mf in-«tiep-fe.J

svdp-ayi-St s6p^ii'S,'f s6p4d'St in^auep-ii-s.

svdp-ayi4, sSp-ie-t, sdp-iort, in-siAep-ie*

svdp-ayi'ina, adp-ii-mits, adp-id-mtis, in-suep^i-mis.

svdp-ay^ta^ sSp^i-tiSf s6p-id'tis, in-8uep-i&-t

svdp-ay6-y-u8, sdp-ie-nt, udp-ia-nU ifirsuep-iS-n*

746. In the Latin first Conjugation, which has preserved
the two extremes of the Sanscrit causal character aya in
the contraction d, the verbs necdrei pWrdre, lavdre and c/o-
mdre, as well as the above-mentioned seddre, present them-
selves as genuine causals, both in signification and in
origin, though they are no longer perceived to be suel^ by
the genius of the language, since their primitive haa either
been lost or estranged in form. Necare, which, specially
regarded from a Roman point of view, must be taken as
the denominative of nex (nec^s), corresponds to the Sanscrit
nds-ayd-mi ** perire facio,^^ causal of naS^d-mi, CL 4. pereo.
Another form of hi^NiIh ndSaydmi, with softened meaning,
is noceo. In Greek vckvs and veKpog are to be referred to
the Sanscrit root nas, from nak. I believe I am right in
regarding pldro as a corruption of pUvo for the reason
mentioned at §. 20. It would consequently correspond to
the Sanscrit pldvaydmi ; properly " I make to flow," from
the root plu, "to flow," which, in the Latin fluo, has ex-
perienced an irregular phonetic modification ; while in pbdtf
which belongs to the same root, the original tenuis is re-
tained. In lavare (Greek \ovu>) one of the two combined
initial consonants is lost ; in other respects, however, lavo
corresponds still better than pMro to the Sanscrit p^vaydmi,
" to wash," " to sprinkle " (in middle " to ,wash oneself,")
on which also is based the Old High German flewiu^^ '' I

* See §. 689. 1 See §§. 691, 692. X See §. 694.

§ This is, like lavo when compared with its intransitive yfuo, estranged
from the primitive ^tuivu, '4 flow," in that it has kept itself free from the

inoTganic z (see p. 114).

St

1002 VEEBS.

wash." In Carniolan plevA-m, "I water,'' "I dissolve"
(Metelgo, p. 115.), is the regular causal from plav-a-m, "I
swim" (= Sanscrit 9TTf^ plav-d-mi). Clamo properly sig-
nifies (if I am right in explaining its m as a hardened
form of t? (see p. 115.)» '*I make to hear,'' and possesses,
therefore, a concealed aflSnity to chw^ kXvw and is identical
with the Sanscrit stdv-^tyd-mi (s from *), " I make to hear,"
" I speak," with the Zend srdv-ayi-mi of the same meaning,
the Carniolan slav-i-m, " I praise," (sluyem " I hear "), the
Old Sclavonic caoba& slovlyu (from blagoslovlyu, " I bless "),
the Russian slavlyu, ** I praise," and the Lithuanian szlounyu,
id. (see §. 745.).

747. Roots, which in Sanscrit end in d, or in a diphthong
to be changed into d, receive before aya the affix of a p ;
hence, e.g., sthdp-ayA-mU " I make to stand," from sthd; ydp-
ayd-miy "I make to go," "I set in motion," from yd.
As labials in Latin are not unfrequently replaced by gut-
turals*, I believe, with Pott (Etymol. F. p. 195.), that the
Jj&imjcuio should be deduced fronjapio, and be identified
with the above-mentioned ydp-ayd-mi; though properly only
the io of the fourth, and not that of the third Conjugation
(= Sanscrit Tf of the fourth Class), corresponds to the San-
scrit causal character. The agreement of forms like capio,
capiunt, capiavrh &c., and the analogous forms of the fourth
Conjugation, might, however, easily favour a transition of
the latter into the third. The same appears to me to be
the case with /ado, which I compare with the Sanscrit
bhdvaydmh "I make to be," "I bring into existence;"
but in so doing I assume that the e is a hardening of
the radical vf (see §. 19.), as roots in 4 in the Sanscrit
causal never assume a p. The Gothic gives us bau-a, " I

* Compare, e. g., quinque with panchany nevre ; coqtto with pacMmi,
wea-cra, SeTYiaji pechem^ " I roast."
t From it— for aw, before vowels dv^ is the Vriddhi form of w ; see g. 39.

CAUSALS. 1003

build'' (from bau-ai-m), as the kindred form to the Sanscrit
bhdv^yd-mi and Latin /acio: in the second and third per-
sons, therefore, the character ai of bau-ai-a, bau-ai-th,
answers to the Sanscrit ay a of bhAv-aya-si, bhdv-aya-tu
From a German point of view, however, we could as little
perceive the connection between our bauen, " to build," and
biru " I am,"' as recognise in Latin the afSnity of the roots
otfac-io and fu-i. If, however, I am unable to compare
the c of the said form with the Sanscrit causal p^ still I
think I can shew in Latin one more causal in which c takes
the place of a Sanscrit p, viz. doceo, which I take in the
sense of *' I make to know," and regard as akin to disco
(properly " I wish to know '') and the Greek eSativ, $/$<ia-icci>.
If the d of these forms has arisen from g (compare Afffirjnjp
from TfifiYJiTip), then doceo leads to the Sanscrit jndp-aydmU
"I make to know" (jd-nd-mi, "know," (or jnd-nd-mi), and to
the Persian dS-tie-m, " I know ". As an example of the
Latin causal, in which the original p has remained un-
changed, let rapio be taken, supposing it to correspond to the
Sanscrit rdpaydmi, *' I make to give," from the root TT rd,
*' to give," which, in my opinion, is nothing but a weaken-
ing of dd. There also occurs, together with rd, in the
Veda dialect, the form rds, just as, together with rfd, exists
a lengthened form dds. In its origin the root Id, to which
are ascribed the meanings "to give," and "to take," appears
to be identical with rd and dd,

748. To the roots which, in Sanscrit, irregularly annex Ap

* The derivation (elsewhere admitted as possible) from lup (Jumpdmi\
"to rive," "break," ** destroy" (compare Pott. I. 258), to which rumpo
belongs, is less satis&ctory, as a in this explanation must be taken as the
Gona vowel, With the loss of the proper vowel of the root. The Latin,
however, avoids the use of Gnna, and generally retsdns the radical vowel
rather than that of Guna ; e,g. in video, which is based on the Sanscrit
causal vidaydmi, ** I make to know," from the root t^.

3t 2

7

1004 VBRBS.

in the causal, belongs ^ ri, i.e. ar (see §. 1.), "to go/' whence
arp-ayd-mi, "I move," "cast,"^ "send"' (Sardn arpaydmU '*»a-
gittas mitto'"), with which, perhaps, the Greek epemta is con-
nected,^ which, however, as causal, should be epevaeia^ or kpei-
ircccd, or epeiira^o) (see. §§. 19. 109*. 6.). Inasmuch as the theme
kpeiir has lost the true causal character, this verb has acquired
quite the character of a primitive verb, just like /airro), which
Pott has referred, in the same way as the previously men-
tioned Latin jacto^ to the Sanscrit ydp^yd-mi, '* I make to
go.'' UphTia does not belong to k8hip,f *• to throw,*' but, like
the others, to arpaydmi, it is then a transposed form of Ipn-riaf
749. The Sanscrit root m pd, "to receive," "to rule,"
assumes, in the causal, I; hence paldydmi. So, in the Greek
/SaAAo), oreMo), loAXo), the second \ of which appears to have
arisen by assimilation from y, as oAAo; from dAyo;=Grothic
ALYA, Latin aliuSi Sanscrit anya-s (see p. 401). BaWa>,
therefore, is for )8aXya), from fia (see §. 109\ 1.), the radical
vowel being shortened (ejSdXoi^), which, however, in the trans-
position /8\)7 {lie^Kfj'Ka) has preserved its original length;

* PciTT might be taken as a tTansposed form of ccpir, and the c as a
vowel prefix, as, e.g,y in eXaxv-ff=Sanscrit Ictghu-s. Observe, also, that
the n of (roXfl-iy^, which Sonne (Epilegomena to Benfey's Gr. Roots, p. 24),
identifies with the Sanscrit cansal py belongs to a root, which in Sanscrit
ends in ar (rt), viz. to svar {8vri\ to which Pott also (£t F. p. 225) has
referred it : (roXiriyf, therefore, properly =" making to sound." Should,
too, the Lith. amjoUpinu^ *' I whistle," notwithstanding its 9« for a, belong
here, then iremark the shorter form adduced by Ruhig of the Sd per. sing.
sxwilpyat ^'the bird whistles," where pia corresponds to the Sanscrit
forms iapayatiy such as arpayati, ''he makes to go," ''he moves."

t The derivation of kship pre-supposes an abbreviation of piTrra from
Kpiirra ; so that p would have taken the place of the Sanscrit sibilant, as
in Kp^'mvy which Fr. Rosen has compared with the Sanscrit root kahiy " to
rule"; see his Rig VSda Sanhita, Annot. p. xi., where, too, Kpavirv6s is
compared with kahipray "swift" (firom kship^ "to cast"), and the Latin
creptuculum with kshapd^ "night" (better with kshapas).

CAUSALS. 1005

OTeAAo), from (rreKyia {e<TTaXKa), for <TTaKy<a, from era (lar&fxi,
i<rT)7/x/)= Sanscrit sthA, which, in combination with various pre-
positions, obtains the notion of movement*; mWco, from
iaXyta, is to be referred, in a manner different from /aTrro), to
the Sanscrit root m yd, *' to go," to which also belongs
n7/x«, as reduplicated form for yiyrjfii (fiit. ^0'ci)=:inFBnfH yd-
sydmif compare Lithuanian yS-su, " I will ride '"). Perhaps
iceA-Aci) from ic6\-^a)= Sanscrit chdlaydmi, "I move," causal
of the root ^qw chal, "to move oneself ^ perhaps, also,
TraVAci), from TroX-yo), for 7ra$^ci)= Sanscrit pddaydmU causal
of pad, ** to go," to the causal of which may be referred also
the Latin pel-lo as by assimilation from pel-yo. All these
forms, therefore, if our explanation of them be correct, have
lost the initial a of the Sanscrit causal character aya of the
special tenses, and are hereby removed, as it were, from the
Sanscrit tenth Class to the fourth (compare Pott 11. 45.).
As in Greek, verbs in eco, aw (for eyco, ot^co), afw, are the proper
representatives of the Sanscrit causal form or tenth Class ; and
as these extend their character also over the present and im-
perfect ; so here, too, may icaXea) be considered as a concealed
causal, which, like the Latin clamo, properly signifies " to make

to hear,'' and answers to the Sanscrit srdvaydmi (s from k).
Accordingly I take KoAeo) as a transposition of icAa-eo) for

750. The Zend, it appears, has no part in the use of
the p, which, according to §. 747., is, in the causal, to be
added to roots in d ; at least I know of no example where
it is found : on the other hand, we find evidence of the
discontinuance, of the addition of a p in as^^jlm^jjau dkdya,
" make to come,'' " bring " (Vend. S. p. 55. several times)

* Observe, also, tliat together with athd there exists a root sthal^ and
with pa a root pal To sthal belongs onr stelle, "place," Old High Ger-
man steila^ from stelyu; properly, " I make to Btand"=: Sanscrit stSilayami,

1006 VERBS.

= Sanscrit dstMpaya, from ^ITT sthd, " to stand," with the
preposition d, " to approach." In M^y^^^Mi dstdyOf from
dstd-ayOf the a of derivation has coalesced with the radical
vowel ; so in Old Persian ^ . ^j^ . |yy . jJE . j f j^ • f|f • y^*" • ^^H
avdstdyam (from ava-astd'ayaw), " I restored " (Beh. I. 63.
66. 69.), In Prakrit, on the other hand, those roots also
which end in a consonant frequently take, in the causal, the
said labial, in the softened form of 6, where, however, the
root is previously lengthened by the addition of an a ;
€. g.y jivdJMhi, " make to live," jivdMdu, " let him make to
live" (see Delius, Radices Prakrit s. r. jiv). In Sanscrit
also, in the unclassical language of popular tales, forms of
this kind occur ; and indeed jivdpaya, for the just-men-
tioned ^^t^dfr^/ii (Lassen's Anthol. Sanscrit, p. 18), which latter
surpasses the Sanscrit in the preservation of the imperative
termination hi from dhi. In the 1st. per. sing. pres. is
found, L C.9 jivdpaydmi (Prakrit jivdbimi), and in the part,
perf. pass. ^*M>4pi^aA'= Prakrit jivdbidd. Lassen, in men-
tioning these forms, remarks (Institut. linguae Prakrit, pp.
360, 36 1), that causals of this kind still exist in Mahratta ;
and I was surprised at finding myself able to trace the
analogy of these formations even to the Iberian Ian-
guages ; since in Latin, as G. Rosen remarks, the affix ap
(only p after vowels) always gives a transitive meaning to
verbs. Thus gnap, " to unveil," " to make evident," cor-
responds to the Sanscrit Jn^pai/^wi, "I make to know,"
while gna, " to understand," agrees with the Sanscrit root
'^jnd, "to know." In Georgian the said causal affix ap-
pears in the form ab, eb, ob, aw, ew, ow, without, however,
the very numerous class of verbal bases which so terminate
being regarded as causals in meaning, which cannot sur-

See " The Caticasian members of the Indo-European family of lan-

guages

*»

CAUSALS. 1007

prise us, as in Latin also, and German, the form of
the Sanscrit causals, or tenth Class, is so prevalent as to
extend over three Conjugations in Latin, and the three
Classes of the weak Conjugation in the German dialects
(see §. 109*. 6.).

DESIDERATIVES.

751. We now betake ourselves to the examination of
the Sanscrit desideratives, which, as h^-s been already
elsewhere remarked,* are retained also in Greek ; if not in
signification, at least in form, in verbs like /8//8|0(d(ricb>, yiyvcl^'
a-KO), fJujxv^aKU), SiSacKia, SidpacKia, nrpciHrKfa, itvniO'Kiiii vtirpi-
<ric6), vt<l>avaK(a, where the guttural is most probably, as in ecKov
and the Old Latin i^uture esdt, only a euphonic accompani-
ment of the sibilant, which in all Sanscrit desideratives is
appended to the root, either directly, or by means of a
vowel of conjunction, i The roots beginning with a vowel
repeat the entire root, according to the principle of the
seventh aorist formation (§. 585.) ; e.g., dsis-i-shfj^ " to wish
to sit,"' as a weakened form of dsdsish ; arir-ish, " to wish
to go," for ararish, from m ar (^ ri). So, in Greek, apa-
ptaKCi>. Roots which begin with a consonant repeat it or
its euphonic representative, with the radical vowel, where,
however, a long vowel is shortened, and the heaviest
vowel a weakened to i (see §. 6.),+ according to the same
principle by which, in Latin, the a especially is excluded
from syllables of repetition (see §. 583.). On this account
the i prevails in repeated syllables, and the agreement

* Annals of Oriental Literature (London, 1820), p. 65.

t The appended sibilant is originally the dental (jB s), bat, according
to §. 21., subjected to a mutation into sh,

:{: Though roots with ri in their middle receive an i in the repeated
syllables^ still this is based on the original form ar.

1008

VERBS.

with the kindred forms in Greek is thus the more striking.
We find, e.g.f yuyvisdmi, "I wish to contend*' (R. yudh),
hubhushdmh "I wish to adorn*' (K,bhush), but not jaga-
dishdmi, hMt jigtidishdmif *' 1 wish to speak*'; not jqjmdsdmi,
but 1^91411 Pi jyUdsdmi, Mid. jijndsif *' I wish to know," "to
learn," '* to inquire." To P^Q 1^ I Of jijndsdmi corresponds in
form the Greek yiyvciia-KU}, and Latin (gr)no-sco; which latter,
like all similar Latin formations, has lost the reduplication.
To mimndsdmif desiderative of mnd {memorare, nundare,
latidare), corresponds fUfxv^(rK<ji), and the Latin reminiscor.
In the special tenses the Sanscrit places an a by the side
of the desiderative sibilant, which, according to the ana-
logy of the a of the first and sixth Classes, is liable, in the
first person, to production (see §. 434.), and also in Greek
and Latin, in the same way as the said class-vowel is
represented (see §. 109.* 1.). I give, for comparison, the
present and imperfect active of r^^ifufH f/ndsdmi over
against the corresponding forms of Greek and Latin.

PRESENT.

SANSCRIT.

Sing, jijnd-^d-mi,
jijnd-'Sarsi,
jijndrsa-ti,

Du. jijnd-^d-vas,
jijnd'Sa4has,
jijud'sa-tas,

Plur. jijnd'sd'mas,
jijrid-sa'tha,
jijUd-sa-nti,

GREEK.

yiyvci>'<TK€t-£,
♦yiyvcj-cTKei,

yiyvca-fTKe-Tov,
yiyvay-trKe-Tov,

yiyvu)r(rKe'T€,
yiyv&'CKO-VTh

LATIN.

nO'Sco*

no-sci'S,

n(hsci-f.

nO'Sci-mus.

no-sci'tis,

no-scu-nU

* Clearly only a trangposed form of man, "to think," with the radical
vowel lengthened, as, e,g,^ in Greek, ^PKrjKa from jSaX, niimixa from

ITfT,

DESIDERATIVES.

1009

SANSCRIT.

Sing. ajijnd'Sa-m,
ajijnd-sa-s,
ajijnd'Sa-4,

Du. ajijn6rsd-va,
ajijnd-sa'tafn,
ajijnd-sa'tdm,

Plur. ajijud'Sd-mai
ajijudsa-ta,
(ijijnd'Sa-n,

IMPERFECT.

GREEK.

LATIN.

eytyvui-CKO-v,

eyiyvio-aiKe-^,

€yiyv<»hO'K€f

eyiyvco- (TKC'TOv,

eytyvcii'^Ke^vjv,

eyiyvciy^rKO'fiev,

eyiyvci>-<TK€rTef

eytyvci>'<TKO'V,

In the universal tenses Sanscrit desideratives lay aside
only the vowel which is added to the sibilant ; while in
Greek and Latin the whole formation extends only to the
special tenses; and, e.g., yvci'trui springs from the simple
unreduplicated root, and hence stands in no closer analogy
to the Sanscrit jijMs-i-shydmi. That in Latin the future
noscam departs from the Greek arises from this — ^that the
future of the third and fourth conjugations, according to its
origin, is only a mood of the present ; and hence, e.g., noscis
corresponds to the Sanscrit jijhdsis, and Greek yiyviUxrKoi^.

762. It may reasonably be conjectured that the deside-
rative form is no stranger in Zend, but I am unable
to furnish satisfactory examples. Perhaps the forms
j^W)^j^J^^ jijisanuha and j^^Au-^oJ^ji^ jijisditi, in the
Fifteenth Fargard of the Vend. (Vend. S., p. 431, Anq., p. 393),
are to be referred here. The first-mentioned form, which
Anquetil translates ** est vivante,'''' is evidently, like the
AJw^juUJSf ?gQ> piereianuha, " ask,'' which follows it, an impe-
rative middle ; and ^jJMJ^j^^j^diti, which Anquetil
renders " on iapprocherar is, like the j^jjousq/^q) peresditi,
" irderrogetr which follows it, the 3d per. sing, of the con-
junctive active. Perhaps xs^y^jj^^^^ jijisanuha may
correspond to the Sanscrit f^H^RRfr jijndsasvaf ''inform
thyself," and j^jMiju^j^^jijisditi be based on a to-be-pre-

-?

1010 VERBS.

supposed Let-form fiidffWtxflf jijndsdti ? I will not venture
to decide this point, any more than as to the forms which
occur in the same page of the Vend. S., M^)^M^^Ai^j^
mimaresanuha, and j^jJOMod^g^As^j^ mimarecsdUi, which like-
wise have the appearance of desideratives. As regards
the origin Of the desiderative character s, it is probable it
springs, like the s of the auxiliary future and of the aorist
of primitive verbs, from the root as of the verb substantive.
Compare, e.g., didik-shdmh "I wish to shew," with dik-
shydmif " I will shew,"" and adidik-sham, " I wished to shew,"
with the aorist adik-shanif and the imperatives of the aorist
mentioned above (§. 727.) like bhushth nishatu^

INTENSIVES.

75a Besides desideratives, there is in Sanscrit another
class of derivative verbs, which receive a reduplication,
viz. intensives. These require a great emphasis on the
syllable of reduplication, and hence increase the vowels
capable of Guna, even the long ones, by Guna, and lengthen
a to d; e. ^., v^vesmi (or vivi&inii)f plural vivismas, from
vii, *' to enter f ' dAdipmi (or d^dipimi) from dip, " to shine ;"
Idlopmi (or ISluptmi) from lup, "to cut off;'' bdbhushmi (or
bdbhushimi) from bhush, " to adorn ;" sdsakmi {sdsakimi),
from sak, " to be able."*' As in Greek co is a very frequent
representative of long a (see §. 4.), so, as has been else-
were remarked Glossarium, Sanscr. a. 1830, p. II3), Tw^afw
has quite the build of a Sanscrit intensive, only that it is
introduced into the a> conjugation. In 7ra/7ra\Aci>, daiSaMo),

* After the analogy of verbs of the third Class, regard being had to the
weight of the personal terminations (see §. 486.)> To the light termi-
nations, beginning with a consonant, i may be prefixed as conjunctive
vowel, when, however, the Onna of the base syllable is dropped ; hence,
e.y., vSvitUmi,

INTENSIVES. 1011

Trai<l>a<r<r(ji}, fiatfjidCf^, fxaifMo-a-ci), the insertion of an / in the
syllable of repetition supplies the place of the lengthening of
the fundamental vowel ; so in Ttonrvvcn (R. irvv, irvita, from
Ttvefiti, fut. Trvev<Tijii), /xotfjvau), /xot/n^AAo), where the t; of the
root is, in the syllable of repetition, replaced by o, since vt
does not form a convenient diphthong. On this analogy
rests also SolSv^ and KotKv7\?ua.

754. Roots beginning with a vowel, of which only a few
pQSsess an intensive, repeat the whole root twice, in such a
manner that the radical a is lengthened in the second place ;
hence atdl from at,** to go,'' asds fix)m ai " to eat." I be-
lieve I recognise a clear counterpart to these intensive
bases in the Greek dyay, though this forms no verb, but
only some nominal forms, as d7c«)7d^, dyc^yev^. The case of
the ci) for d is just the same as in the above-mentioned
Ttadd^o), On the other hand, in ovivtjfu, oTriTrretyco, aTiraWo),
the base syllable has experienced a weakening of the vowel,
like that which enters into Sanscrit desideratives (§. 751.
ad init), which does not, however, prevent me from referring
these forms, according to their origin, rather to intensives
than to desideratives (compare Pott II. p. 75); so also
d\aXd^(>) and kKeKi^ia exhibit the same weight of vowel in
the base and in the syllable of repetition.

755. Roots, also, which begin with a consonant and end
with a nasal, in case they have a as the base vowel, repeat
the whole root twice in the Sanscrit intensive, but lengthen
the radical vowel neither in the syllable of repetition nor
in that of the base. The nasal, in accordance with a uni-
versal rule of sound, is influenced in the former syllable,
so as to conform itself to the organ of the following con-
sonant ; and in roots which begin with two consonants, only
one enters into the syllable of repetition; hence, e.g., dcm-
dram from dram, " to run ;'' bambhram from bhram, " to
wander about ;" ^fjpr jangam from gam, ** to go." *So in
Greek, itaix^aivu) from tpalvta, the v of which, though not be-

1012 VERBS.

longing to the root, is nevertheless reflected in the syllable
of repetition (see §. 598.). On ITJfi^ jangam is based, I be-
lieve, the Gothic gagga (i.e. ganga, see §. 89. 1.); so that
therefore gam, in the syllable of the root, has lost the ter-
mination am*f and gagg has entirely assumed the character
of a root, which in High German has produced a new re-
duplication (Old High German, giang from gigang, our
gieng, see §. 592.). And in the formation of the word, gang
holds as an independent root ; whence, in Gothic, gah-ts'ff
*' gait'''' (inna-gahtSf fram-gakts). The Lithuanian presents
zengiu '* I step,'^ as analogous form k

756. Some Sanscrit roots also, which do not end in a
nasal in the intensive, introduce a nasal into the syllable
of repetition ; e.g,, chanchai (or chdchal) from chal, " to move
oneself ;" pamphvl from phal, *' to burst," with the weaken-
ing of the a to t£ in the base syllable ; so chanchur from
char, '* to go.**' As liquids are easily interchanged, it may
be assumed that here the nasal of the repeated syllable is
only a changed form of the radical liquid / or r. So in
many Greek reduplicated forms ; as, mfnrTojfu, TtiyLirprifii, yiy-
ypalvoiy yiyy\vfJLO^, yayya\i^ci}, yayypatva^ rovOopv^ci}, rav^
ra\ev(M>, revOprfiw, 'ir€fx<l>pfjSci>v. The following are examples
in which the liquids remain unchanged in the syllable of re-
petition : fjLapfxalp(ji}y fiopfivpui, fiipfiepo^y fjiepfialpci}, jKe/0)Lu;|o/^a),
KapKatpoi, yapyatpu), ^p^opvi^uif itopfftvpa, Ttopffntpta. Com-
pare with these the intensives of those Sanscrit roots in ar
which contract this syllable in the weakened forms to i^
ri : these, in the active of the intensive, repeat the whole
root twice, except when this begins with two consonants, in

* The final a is the class syllable ; dd per. pi. gagg-a-nd.

t Eaphonic for gag-ts, the nasal being rejected. With respect to. the
suffix, compare the Sanscrit ga-ti-s, ^^gait," for garUi-s^ see §. 91.

X In Lithnanian k often stands for the Sanscrit g orj. Compare^ e, g.^
zadasy " speech," with the Sanscrit gady " to speak."

INTENSIVES. 1013

which case only one enters into the syllable of repetition ;
e.g,i dar-dhar-mi, pi. dar-dhri-mas, from dhar, dhri, ** to stop/**
** to carry T' but sdsmarmh according to the universal prin-
ciple, from smar, smrif "to remember/' To dardharmi,
potential dardhriydnif 3d. per. dardhriydJt (from dardharydm,
dardharydt), corresponds the Zend, daredairydt in a pas-
sage of the Vendidad (Vend. S. p. 46a) . ^5?»»g(? . a»(3A}/^

tAJM^^/jM^9 yatha vehrkd chathwarezangrS ' barethry&t hacha
puthrem nischdaredairydt " as the fourfooted wolf tears away
(carries off) the child (the son) of her who bore him (the
mother ?) : according to Anquetil (p. 407), " comme le hup
a quatre pieds erdeve et dSchire Verdant de ceHe qui a porte
(cet e7fanty\ If, however, i»AM^^jj^g^5j«bjy nischdaredairydt
does not come from the Sanscrit root dhar, dhri, it springs
from ;^ dar (^ dri), " to split,'' " tear asunder '' (Gr. Sepia,
Gothic taira) ; whence, in the Veda dialect, the intensive
dardar (see Westerg. R 5 dri), in classical Sanscrit dddar.
The first derivation, however, appears to me for the more
probable : at all events, the form in question is a sure proof
that in Zend also intensives are not wanting.

757. Some Sanscrit roots, which have a nasal as their
last letter but one, take this in the syllable of repetition ;
hence, e. gr., baTiibharymi from bhanj, " to break f ' dandansmi
from dam, " to bite" (Gr. 5a#c) ; chan-i'skandmi from skand,
"to mount'' (Lat. scando); the latter with 2! as vowel of
conjunction between the syllable of reduplication and that
of the base, as also in some other roots of this kind, and
at will, also, in those roots in ar which admit a contraction
to ri, and which nevertheless may assume a short i instead
of a long one; hence, e.g., char-i-karmi, or char-i-karmi,
with chavkarmi, from kar, kri '* to make.""

* With regard to the e inserted in daredairydt, see §. 44.

1014 VERBS.

768, The intensive forms pan-i-pad and pan-i-pai, from
pad, " to go/' and pat, " to fall " (Pan. VII. 4. 84.), appear
obscure. In explanation of these it may be assumed,
that together with n^ pad and in^ pat there have existed
also the forms pand and pant with a nasal, as together
with many other roots which terminate in a simple mute
there exist also those which have prefixed also to their
mute the nasal corresponding to their organ ; as, e.g., panth
with path, "to .go.''* Together with dah, "to burn,''
exists also a root ?fv danh; and hence may be deduced the
intensive form dandah (Pan. VII. 4. 86.), to which the
Gothic tandya, ** I kindle " (with the causal character ya,
see §. 741.), has the same relation, as above (§. 755.) gagga=>
ganga, ** I go," to jangam.'\

759. In Latin, gingrio has the appearance of a Sanscrit
intensive, and is by Pott also referred here, and radically

* Wiih panth are connected the strong cases of pathin, " way," as also
the LaXin pons, pont-is, as '' way over a river," and the Slavonic n^Tb
puty, "way" (see §. 225*.): with, path is connected, amongst other
words, the Greek wdros (see Glossarinm Sanscr. a. 1847, p- 206).

f With regard to the t for d of tandya, see §. 87. The retention of the
second d of the Sanscrit form dandah is to be ascribed to the influence of
the n preceding it (compare §. 90.). Hemark, also, the form sandy a, '' I
send/' in which I think I recognise the causal of the Sanscrit root sad,
"to go," {sddaydmi, "I make to go,") with a nasal inserted. Graff sets
up (IV. p. 685) for the Old High German a root zand (ss for Gothic t, and
t for d, according to §. 87.), which he likewise endeavours to compare
with the Sanscrit dah, but without finding any information as to the n
and t through the intensive form ^ft^^ dandah. On, the primitive root
dah, if not on the causal form ddhay^ is based also the Old High Grerman
dah-t or tdh-t (our Docht^ Dacht), which by more exact retention of the
radical consonants is completely estranged from the intensives (in mean-
ing causals) zand or zant. Initial Mediae remain in German frequently
unaltered, e, g,, in the above-mentioned gagga^ " I go"=jangam ; while
the Gothic root qvam, "to come" {qvima, qvam\ which is based on the
primitive gam, has experienced the regular change of MedisB to Tenues.

INTENSIVES. 1015

compared with gn, i. e gar, gir (whence gir, ** voice '').
The syllable of reduplication exhibits n for r, as in Sanscrit
chanchur, and similar Greek forms (§. 756.)« To girdmi
(also gilAmi)y ** deghdior belong, amongst other words, the
Latin gula and gurgulio, which latter, in its repeated syllable,
replaces the liquid / by r.

760. The passive form of the Sanscrit intensive has
usually an active meaning, and then, by Indian Gram-
marians, is regarded according to its formation, not as
passive, but as a particular form of the intensive, which
I nevertheless call deponent, as in its origin it is evidently
nothing else than passive. This appears more frequently
in classical Sanscrit as the form without ya» yet still sel-
dom enough. I know of no examples besides ^V^^^l^
chanchuryantS, " they convey " (Mah. I. 1910.), from ^ir: char
(see §. 756.), Mihyasi, ** thou lickest,"" from Rh (Bhagavad-G.
11. 30.); didipyamdna, "shining,'' from dip (Nal. 3. 12.
Draup. 2. 1.). In dddhuyamdna (1. c), from dhu or dhu,
the passive form has also a passive signification. Of the
form without ya there occurs the participle present ISlihat,
Mid. mihdna " licking,'' Mah. III. 10394, 12240. The Veda
dialect makes more frequent use of the active form of the
intensive : the following are examples : nAnadati, " they
sound," Rig. V. I. 64. 8. 11.; abhipra-n&numaSf ** we praise,"
from nu (prep, abhi, pra, 1. c. 78. 1.) ; jdhavimi, "I summon,"
with i as vowel of conjunction (see §. 753. note), from hu,
as contracted form of hv^, 1. c. 34. 12. ; d-navindt, " he
moved," " stirred," from mirf, " to move," " to drive "
(prep, d), Rig. V. V.f

* All reduplicated fonns, which combine the personal terminations
direct with the root, suppress the n of the 3d per. pi. (compare §. 459.).
To the root nad corresponds the Welch no^u, "to cry."

t See Westerg., Badices, p. 45, and root nu, to which dnaz^nSt likewise,
according to its form, might belong ; the meaning, however, in the pas-
sage

1016 VERBS.

DENOMINATIVES.

761. Denominatives are not so frequently used in San-
scrit as in the kindred languages of Europe. Their for-
mation is effected either by the addition of the character
of the 10th Class, or by the affix ya^ sya, and asya ; both
which latter ought probably to be divided into s-ya and
as-yOf so that in them the root of the verb substantive
as is contained, either entire or after dropping the vowel
(compare §. 648.). As the Latin verbs of the 1st, 2d, and
4th conjugations are based on the Sanscrit 10th Class
(§. 109 . 6.), forms like laudrA-s , nomin-d-s, bi-min-d-s, co-
lor-As, fludu'd'Sf wsturdrs, domin-d'S, regn-d-s, sorori-d-s'f,
ccen-d'S, planf'd'Sf pisc-d-rist oBi-i-s, calv-i-Sf can -is, miser -
i-ris, feroc-i-s lasciv-i-s, lipp-i-Sy aborC-i-s, fin-i-s, sif-i-s,
correspond to Sanscrit forms such as kumdr'-aya'Si, ** thou
playest,''' from kumdra, " a boy ^ t svkK-ayash " thou

sage cited leads to the root nud : the t^ therefore, of the form in ques-
tion is not a sign of the person^ but radical (euphon. for d), since the per-
sonal character of the 2d and 3d pers. sing, of the imperf., according to
§. d4., cannot combine with roots ending in a consonant; hence, e.g,y
ayunak, ^^ thoa didst bind," and " he bound," for ayunaksh^ ayunakt (see
smaller Sanscrit Grammar, §. 289). With respect to the syllable of redu-
plication, the form a-nav4nBt for dndnSt is remarkable on account of the
insertion of an I, as, according to grammatical rules, such an insertion
occurs only after r and n, see §. 767*, and smaller Sanscrit Grammar,
§§. 500. 501. 508.

* I give the 2d per., as the 1st exhibits the conjugational character less
plainly, and presents the least resemblance to the other persons.

t From wroriiu, not from sorar ; for from the latter would have come
sororo^ not sororio.

^ The Indian Grammarians wrongly exhibit a root kumdr, ^^ to play" —
which, if only for the number of syllables, is suspicious — and thence derive
kumdra, " a boy ;" in which I recognise the prefix ku, which usually
expresses '^ contempt," but here " diminution," and mdray which does not
occur by itself, but is joined with martya, "man," as ^^ mortal." In
general there occur, among the roots exhibited by Indian Grammarians,

many

DENOMINATIVES. 1017

rejoicest," from sukhay " contentment ;'' ydldr-aya-si, " thou
encirclest,"'' from ydktra, " band" (R. yuj " to bind"") ; ksham-
aya-si, **thou supportest,'*'' from kshama, ** patience.*"
From these examples we see that in Sanscrit also the
final vowel of the base word is rejected before the verbal
character; for otherwise, e.g., from ydktra-aya-si would
come ydktrAyasi. That in Latin forms like coen-d-s the
d does not belong to the base noun is seen from this, that
the final vowel of bases of the second declension is rejected
before the verbal derivatives d, i, and i] hence, regn-d-s,
calv-6-s, lasciv'i'8> As to the retention, however, of the
organic t*, viz. that of the fourth declension before d (aestu-
ds, fluctu-d-s)f I would remark, that in Sanscrit also u shews
itself to be a very firm vowel, inasmuch as it maintains
itself before the vowels of nominal derivative sufiixes ; and,
indeed, it moreover receives the Guna increment, while a
and i, i. e. the heaviest and lightest vowel, are dropped ;
hence, e.gr., mdnav-a-s, " man" (as derived from Manu), from
manu ; T^Nr sauch-a-m, " purity," from ^f^ mchit " pure ;"
ddmrath-i-s, "Son of Dasaratha," from dasarathcu Before i,
however, in Latin, the u of the fourth declension disappears
in denominative verbs, as in the above-mentioned ahorf-i-s.
762. As a consequence of what has been said in the
preceding §., I believe that a suppression of the vowel of
the base noun is also to be assumed in Greek denomina-*
tives in aw, eo), o(a, a^co, <fa). I therefore divide, e. g.t ayop-
afo)*, dyop^'do-jxou, fiop(l>'6<jt), ici//(r(r'-da), 7ro\efi*-o(i), woAe/x'-ew,

many denominatives, amongst them also sukh^ " to rejoice," which con-
tains the prefix m (Gr. cv), as certainly as cft^ duUkh, " dolore qffU^e/*
(from duUkha^ "smart,") contains the prefix dus^Gieek 8vs. By the
Indian Grammarians, however, duhkh likewise is considered as a simple

root.

* I have already, in §. 602., pointed out another mode of viewing the
forms a^6> and i^o), but in §. 503. I have given the preference to the

3 u above

1018 VBRBS.

'no\€fX'i^<a, and recognise in the a of a^co the Sanscrit a of
aydrmi, and in the f the corruption of ^ y, as in ^evyvvfu
compared with the Sanscrit ti^ yuj and Latin jungo (see
§• 19.) ; while in forms in aio, €<a, o<a, the semi- vowel is sup-
pressed ; and, moreover, in the two last forms the very
common corruption from a to e, o has taken place (§. 3.).
It admits of scarce any doubt that in forms in t^ta also the
I is only a weakening of a ; for though the weakening of
a to i is not so frequent in Greek as in Latin and Gothic,
still it is by no means unprecedented, and occurs, to
quote a case tolerably similar to the one before us, in
r^o), i^ofjiah compared with the Sanscrit root sac?, '' to place
oneself,'' Gothic SAT. (sUa, "sat'').

763. The lightness of the vowel i may be the reason why
the form in /^o) has become more used than that in a^a>, and
that those bases which experience no abbreviation before the
denominative derivative element by the relinquishment of
their final letter admit scarce any letter but i before f ; hence,
e.g.t 7roJ-/fci), dytav-i^ofJLat, d^oi/T-Zfo), dvip-i^o), ai/uar-Z^a),
d\ofC-/fc«), yvvaiK'i^oi, ^oipaK-Zfo), icuv-Zfo), /xua)7r-/fa>, Kepar-t^ia,
KepfjLaT'i^cii, epfioT't^fa; ejO/Lt-ctfo), 6vo/X'dl^<a, yoi/v'-dfo/xai *,
which, I think, ought not to be divided epp.d'lQui, ovo/xd-l^o}; so
easy is it, from the point of view of the Greek in particular,
to identify the a of epixd^it), ovojLtdfw, dyopdl^(»}, dyopdofxat, and
the like, with the a of the base noun. For then the analogy
of these verbs with lirn-d^ofiat, Aid'-dfo), eiK-d^o) (from the
base ciKOT), ev^i'-dci), Yevei'-do), 7re\eK'-da), vejiea-dci}, and with
the Sanscrit denominatives in aycL, would be unnecessarily de-
stroyed ; for as o and tj, and occasionally t; and /, are dropped

above, and do so now with the greater confidence, as the other members
also of oar family of languages, the denominatives of which I had not
then considered, follow the same principle.

* Not from yow, but from the base yovvar, whence yovvar-os, yovvar-a.

DENOMINATIVES. 1019

before the derivation aco, afo)*, there is nothing more natural
than that a also should give way before the same. But as
bases in a and rj (from a, see §. 4.) produce principally de-
nominatives in aoi, a^<ji>, and those in o principally such as end
in 00), /fo), from this the influence of the final vowel of the
base noun on the choice of the vowel of the derivative may
be inferred ; a and tj favour the retention of the original a,
while o, wliich is itself a corruption of a, readily permits the
a of the derivative to be weakened to o, in which it seems to
re-appear unchanged, but which (if we wish to allow in its
full extent the transmission of apparently autochthonic Greek
forms from the time of the unity of language) presents no
obstacle to our placing on the same footing as regards their
principle of formation, verbs like wo\e/z(o)-oa), ')(pv(T{oy6(»>,
ayKv\{o)-6(ji), and such as ai/xar-oa), dppev'6<i>, ttujo-oco, Karo^pv^
6<jii, da\a(T(T(a)-6(ji>, icv/(7<r(a)oa),and to our recognising such verbs
as dyop{ayao'fxai, ToA/x(a)aa), S/x|r(a)-dc«), vi/c(j;)-aa), as analogous
with Kvv-aci>, y€vei[o)'&(»>, \ox(py&o)f dvTi(o)-aa), i/ejLte<r(0-««> "tg"
Aefc(i»Vdci). The proposition appears to me incontrovertible
that the Greek denominatives in afco, aw, eco, ow, ifw, corre-
spond to the Sanscrit in aya (1st per. ayd-mi, Zend ayS-mi) ;
and that, as in Sanscrit, Zend, and Latin, so also in Greek, the
final vowel of the theme of the base noun is, for the most
part, suppressed before the vowel of the derivative f: where,
however, it is retained, which is only at times the case with
i and V, the vowel of the verbal derivative also remains after
it (prjpi'do'iJLat, o^pu-oco, /%^t;-dco). Forms like Brfpt-o-fxat, infTv-
O'fjLat, fiYjvi'<ji)j [xedv-ia, SaKpv'ta, belong to another class of de-
nominatives, which exists also in Sanscrit, of which hereafter.
764. In German, also, the final vowels of nominal bases

* Examples, in which i and v are retained, are #cXavc]rt-oa), oKpi-a<a,
df/pt-ao/uu, IxdV'dco,

t G. Curtins is of a different opinion {^^Contributions to the Comparison
of Language" pp. 119, 120).

Sv2

1020 VERBS.

are suppressed before the vowel or y (for ay) of the verbal
derivative, which is based on the Sanscrit aya\ hence,
in Gothic audag'-ya, **I account happy,'" from the base
audaga (nom. avdag'-s, see §. 135), " happy ;" gaur-ya, " I
sadden," from gaura, nom. gaur-s, " sad ;" skajf-ya, ** I
make," from skcffti, ** creation," nom. skafC-s ;" mawo-ycij
"I prepare," from manvu, nom. manvu-Sf ** ready ;" maurthr-
ya, ** I murder," from maurthra, nom. maurthr (see §. 153.)
** murder ;"•!• tagr-ya, "I weep," from tagra, nom. tagr-s,
" a tear," (Greek S&Kpv, Sanscrit asrih from dasru). Among
those Gothic denominatives which have retained in the pre-
sent the last syllable of the Sanscrit derivative aya, the
verb ufdrskadv-ya, " I overshadow," stands alone, since this
verb has retained the final vowel of the base skadu (nom.
-us) before the verbal derivative (with euphonic change into
v)t while other bases in u follow the general principle;
hence, thaurs^-yan, " to thirst " (impers. thaursyith mik, I
thirst," literally, " it is a thirst to me,") from thaursu
(nom. -Ms), ** dry ;" dautK-ya, " I slay," from dau-thu-s,
** death ;" X as in Greek, Oavar-oci) from Bavaro. The follow-
ing are derivatives belonging here, and springing from
bases ending in a consonant: namn-ya, "I name," from
naman(noin. namd, see §. 141.) ; and aiug'^-ya, " I shew," from
avgan (nom. aug6)f *' an eye." The former, like the Latin
nomin-Or and Greek forms like aifxaT-ou), a/jLtar-Zfo), preserves
the final consonant of the base, but has, however, admitted
an internal abbreviation, like that of the Sanscrit weakest

* This does not occur in the simple form, but compounded: ga-
fikafC'Sy "creation," "creature;*' ufar-skaft* -Sy "commencement."

t Compare Sanscrit mar-ay ami, " I make to die ;" the Gothic sufiix
^Ara=Sanscrit tra^ of which hereafter.

X Scarcely from dauth{a)-8^ " dead," for the Old High German clearly
comes from t6d (theme t6da), " death," not from tot (nom. masc. t6tir\
"dead."

DENOMINATIVES. 1021

case (ndmn-as, " nominis '') : on the other hand, aug-ya (for
auganrya or atigin-ya) follows the principle already men-
tioned in §. 503., by which Sanscrit denominatives are
governed, such as varm-ayd-mU " I harness," for varman-
ayd-mi, from varmaru Compare, besides the Greek forma-
tions discussed I c, also derivatives from comparatives ; as,
l3e\Ti (ov)-oa), iJL€i{ovy6(ji), eAa<r(r(ov)-oa), ^aKi(oi/)-oa).* In Greek,
also, bases in 2 reject their final consonant, together with
the vowel preceding it, which is the less surprising, as this
class of words has in the declension, too, preserved but
few traces of the <r of the base (see §. 128.). Hence, irKrfp
(e<r)-oa), from irKtjpe^ (see §. 146.); a\7(e(r)-ea), from aAyej;
d(rdei/(ea")-ea), {rom do'deve^ ; Teu%(e(r)-/fa), from reuj^ej; yi^p^
(a(r)-aa), frouiyYjpas (§. 128.).

765. We return to the Gothic, in order to adduce some
denominatives from Grimm's second and third conjugations
of weak verbs. The second conjugation, which exhibits 6 = d
(§. 69.) for the Sanscrit aych and has therefore, like the
Latin, first rejected the ti y of aya, and then contracted
into one long vowel the vowels which, by the loss of the
y, touch one another, yields, e.g., fislc-d-s, '*thou fishest,""
for comparison with the Latin pisc-d-ris. The Gothic base
Jiska (nom. Jisk'^-s, see §. 135.) has abandoned its a, as the
Latin pisci its i, before the vowel of the derivative (see
§. 761.). The Gothic thiudan-6-s, " thou reignest,"** from
the base thivdana (nom. -n -s), " king,"' resembles, in its
principle of formation, the Latin domin-d^s, as the Gothic
first strong declension masculine and neuter and the Latin
second on one side, and the Gothic second weak conjuga-
tion and the Latin first on the other side, are in their
origin fully identical. To Latin denominatives from the
first declension, like ccen-d-s (see §. 761.), correspond Gothic

* On the other hand, TrXcov-afo), not TrXc'-afo).,

1022 VERBS.

verbs of the same class ; as, fairings, ** thou blamest/'
from the base fairind (nom. ^na), " blame.'" To aestu-d-Sj
fludu'd-89 corresponds lusf-d-s, from the base lustv^ ** desire/'
"longing," with the rejection of the w, however, of the
nominal base. Bases in an weaken their a to i, as in the
genitive and dative ; hence, frauyin-d-s, " thou reignest,"
{rom frauyan^ **lord" (nom. frauya, ^gen. frauyin-s), as in
Latin, nomin^d'S, luminrd-s (§. 761.) ; so gudyin-d-s, " thou
administerest the priest's office," from gudyan^ nom. gudycu
" priest." Some bases terminating in a add n before the
formation of a denominative, and likewise weaken the a of
the base to i ; thus, skalkin-d-s, " thou servest," from skalka,
nom. skalJC'6i " servant," gen. skaUci-s (see §. 191.) ; kdrin-S'S,
fiot^evetgf from h&ra, nom. hOr-s, "adulterer;" reikin-d-s,
" thou rulest," from reikya, nom. reiki (see §. 153.), " rich."
That class of weak verbs which has contracted the Sanscrit
aya to ai, and stands on the same footing with the Latin
second conjugation (Grimm's third weak conjugation), pre-
sents, e* g.y arnt-ai'Sy " thou commiseratest," from arma,
nom. arms ; as, in Latin, miser -^-ris from miseru (miser for
miseru-s); ga-hveit^ai-s, "thou stayest," from hveild, nom.
hveUa, "time," "delay."

766. The Sclavonic uses, for the formation of denomina-
tives, that conjugational form which corresponds to the
Sanscrit tenth Class. But, as has been remarked in §. 605.,
not only Dobrowsky's third conjugation belongs to the
Sanscrit verbal class just mentioned, but also the greater
portion of those verbs which, in §. 500., I wrongly classed
all, without exception, under the Sanscrit fourth Class ;
whilst I can now recognise as sister forms of the Sanscrit
fourth Class, of Latin verbs like capio, and Gothic like
vahs-ya, " I grow," only such verbs of Dobrowsky's first
conjugation as combine the formative elements commen-
cing with a consonant ; for example, the ch of the preterite,
the / and v of the participle preterite active, and of the

DENOMINATIVES.

1023

gerund preterite, as also the suffixes th ti and n^b t of the
infinitive and supine, direct with the root, a circumstance
which occurs only with respect to a few roots terminating
in a vowel; e.g., from na, "to drink"' (Sanscrit pi*, Class 4,
middle), comes un\&.pi'yu, "I drink'' (Sanscrit pif-y^), nnicmH
pi-ye-shi, "thou drinkest" (Sanscrit pi-ya-s4), nnj^i* pi-chf
" I drank," unAii pi-l, ''having drunk," nuB'b pi-v (gerund),
nHTH pi-ti, " to drink," sup. hhti* pi4. Those verbs, how-
ever, in I& yu or A\K ayu, which, in the said forms, inter-
pose an a between the root, or the verbal theme, and the
formative element which follows (Paradigm B. of Do-
browsky), I am now of opinion must be compared with
the Sanscrit tenth Class ; so that yu, and more fully ayu, of
the 1st person, corresponds to the Sanscrit ayd-mi. and
the Lithuanian oyu, uyu, iyu (see §. 506.). Compare, e.g.,
p'biAAI& ryd-ayu, " I lament," with the Sanscrit causal
rdd-^y&mi, " I make to weep" (R. rud, " to weep"), and the
Lithuanian ratidroyu*, **I lament"

SANSCRIT.

rSd-ayd-mi,

rdd-aya-si,

rdd-aya-ti.

SINGULAR.
OLD SOLAVONIC.

rydrayu,

rydraye-shi,

ryd-aye-ty.

LITHUANIAN.

raud'OytL.

ravd-oyi.

raudnoya.

DUAL.

rddr-ayd-vas,

rdd-aya-thas,

rdd-aya-tas.

ryd-aye-va,
ryd-aye-ta,
ryd-aye-ta.

raud^ya-vm.

raud-oya-ta.

raud-oya.

* As the Sanscrit ^ is a contraction of a«, so in this respect the
Lithnanian form corresponds still more than the Sclavonic to the Sanscrit
causal. The Sclavonic 'biy corresponds (according to §.225. c.l to the
Sanscrit radical u.

1024 VERBS.

PLURAL.

SANSCRIT. OLD SCLAVONIC LITHUANIAN.

rdd-^yd-mcis, ryd-aye^m, raud-oyoHtne-

rdd-aya-thay rydnaye-te, ravd-oya-te.

rdd-aya-rdi. rydraytUy , raud-oya.

767. Both in Sclavonic and in Lithuanian the y of this
conjugational class is dropped before the formative elements
which begin with a consonant, and then, in Lithuanian, only
the is left, and, in Sclavonic, the more ancient a, which
corresponds to it ; hence, the infinitive in Lithuanian is
raud'O-ti, in Sclavonic ryd-a-th and the future in Lithuanian
raud-o-su. The Sanscrit, on the contrary, preserves the ti y
before formations beginning with a consonant, by the in-
sertion of a vowel of conjunction, viz. i ; hence, rdd-ay-i-
shydmi corresponding to the raud-o-su just mentioned ; and
in the infinitive rdd-ay-i-tum answering to ravd-o-ti, ryd-
a-ti^t sup. piiiAATTb rydnd-t The verbs under Paradigm
B. in Dobrowsky and Kopitar have lost, in the present
and the forms connected therewith, the a of the class
character, and retain only the y (ghgol-yu, " I speak," for
glagol-ayu) before formations beginning with a consonant,
but exhibit the a in other places, in accordance with the
verbs which have ayu in the present ; thus, e.g., rAAroAA;^^'^
glagol-a-ch, " I spoke," glagol-a-ti " to speak," like p'biAA;^'^
ryd-a-ch, piiiAATH ryd-a-ti. The Lithuanian presents no
forms analogous to verbs like glagolryu, since forms like
myl-iut plural myl^i-me, correspond to Dobrowsky^s third
conjugation (e.g., vol-yu, plural vd-i-m, see §. 506.), while
forms like penu, laikau, plural pen-a-me, laik-o-me (see
§. 606.), exhibit the Sanscrit aya in the abbreviated form.

* From rydayo-nty^ see §. 266. g.

t I do not mean by this comparison to assert that the Lithuanian and
Sclavonic InfLiiitive suffix is connected with that of the Sanscrit language.

DENOMINATIVES. 1025

which in ravd-oyuy p^iAAfc^ ryd-ayu, enters, save in the
present indicative and its derivatives, only before suffixes
beginning with a consonant.

768. The Lithuanian and Sclavonic nominal bases, like
those of the kindred languages already mentioned, when
they terminate with a vowel, which is generally the case,
reject this before the verbal derivative ; hence, in Li-
thuanian baif-oyu, "I appear white," baU'-inu, **I make
white," * from baUa, nom. -tar-s, " white ;" duwan-^oyu " I
bestow," from duwana tern. " gift ;" czysC-iyu, ** I purify,"
from czysta, nom. -ta-s, " pure ;" "j* gatato-oyu and gataiv-iyu,
" I make ready," from gatawa-s, " ready ;" ddC-iyu, " I
divide,'' from dali-Sy " portion ;" apyole-m, " I deride," from
apyoka-s "jest ;" didd'-inu, ** I enlarge," from diddles; brang^^
inu, " I render dear," from brangu-s. The following are
examples of denominatives in Old Sclavonic : a^aai&
dyeV-ayuy " I make," a^aa^^-b dyet-Onch, " I made," from
A*AO dyeh, " work ;" noAOB*Tb podob'^-ye-ty, " it is fitting,"
infin. noAOBATH podob^'Orih irompodobdy ** use ;" 3NA]iienaI&
J^namena-yu, ** I denote," from ^namen ^namen^ nom. Cnamya
(see §. 264.), " mark *" (Kopitar Glagol. p. 73.) ; rAAroAl&
glagoT-yu, " I speak," infin. glagoT-a-ti, from glagolo, nom.
glagoly ** word." In forms in »l& uyu, infin. ov-a-ti, the »
u appears to me, in departure from what has been re-
marked at §. 255. h. as a contraction of an or ou (§. 255. f.),
and the v of ov-a-ti as the euphonic alteration of the final
element of the diphthong » ti = ov. The corresponding
form in Lithuanian is auyu, the first u of which, before
vowels, likewise changes into its equivalent semi-vowel ;
hence, e, jr., naszT-duyUi " I live in widowhood," from naszle

* Denominatives in inu have all a causal signification, compare §. 744.

t With the formations in tyu compare the Greek, in t^(fi>=iyfi>, see
§. 762 ; iyu and oyu have the same relation to one another as i^m and a(a>
have to one another in Greek.

1026 VERBS.

" widow/' pret naszt-auxm, fdt. naszt-au-su. So in Old
Sclavonic; BAOB{rl& vdov-u-yu, pret. BAOBOBA^'b vdov-ov-
ach, infin. baobobAth vdov^ov^a-th from baobA vdovOf
" widow **" s Sanscrit vidhavd. hm eh&I& imen-u-yUf *'I name/'
infin. HMEHOBATH imen-'OV'arii, from the base hmen imen.
Other examples of this kind occur in Dobrowsky, p. 372.
We may regard the u, av, of these forms as a lengthening
of the theme of the base noun, and divide, therefore, as
follows : vdavu-yUf vdovov-ortif imenu-yu, imenov-a-ti, where
we must recall what has been observed at §. 263. regarding
the unorganic introduction of Sclavonic bases into the de-
clension in i»i y. In denominatives in %l^ y^yu, as, eg.,
BorAT:bl& bogaC-yeyu, " I am or become rich,'' infin.
BorATtTH bogaJC-ye-tii from the base bogato, nom. bogat,
A ye corresponds to the Sanscrit a of aydmi, which will not
appear surprising when we consider the peculiarity of the
Sclavonic in constantly prefixing to vowels a y. The
following are examples of denominatives from Dobrowsky's
third conjugation (see §. 505.): 2kEHl&(A schen^-yd-sya "I
marry," infin. ^Renhthca schen'i-ti'Sya, from ^ena schena,'
'* woman ;" roTOBAl& gotov-lyd (euphonic for ryd), '* I pre-
pare," infin. roTOBWTii gotov*<-tu from totobo gotovo, nom.
m. roTOB* gotov " ready ;" vfhA^ zyeF-yH, " I heal," infin.
ig'bAHTH zyeV-i'ti, from q*AO zyelo, nom. q'bAi> zyel,
" healthy."

769. I have already, in §. 502., compared the Greek
denominatives in cro-co, as ai)iiaa--(ra) from alfiar-yta (see
§. 501.), with those in Sanscrit formed with i? ya. While,
however, in Sanscrit, the final vowel of the base noun, if
short, is lengthened, the same in Greek, according to the
analogy of §. 762., is dropped; hence, e.g., dyye?\!\(a from
dyy€\(pyyci>, TrotK/AXo) from 7ro/iciA(o)-ya), a/icaAAco from al-
Ka\(o)-yci), fiaKa(T(r<a from fiaKaK^oyyu}, fietXta-aci) from /xei-
AiX(o)-ya). Bases in p, po, and v, transfer the y, vocalized to
i, to the preceding syllable, instead of assimilating it to

DENOMINATIVES. 1027

the preceding consonant ; hence, reKfiai-p-o-fJiat from reKfiap^
yo'fiatt from reKfiap; icadaijo-o) from Kadap(oyyci), from Ka-
Oapo ; fieyaip-io from ixeyap-yca, not from fjteya-g, but from the
base of the oblique cases fieyaKo, the \ being exchanged for p
(see §. 20.) ; /xeAa/vo) from fieKav-ya}, from the base fieKav ; iroi"
fxaivta, TreTraiVo), reKTaivfji, a^joafVo), ev^paivoif from notfiav^yci},
&c., from the bases irotfiev, irenov, tcktov, d<f>pov, ev^pov, with
the retention, however, of the original o, instead of the unor^
ganic vowels e, o (see §. 3.). In denominatives from substan-
tive bases in /xar, as ovofiaivoD, Kt//iaiW, airepfialva}, <n;/xa/i/a>,
yetfiatvci}, the v probably springs from the original form of the
suffix fiar, as this is a corruption of fxav, and answers to the
Sanscrit man, and Latin men, mm,* It appears, however, to
me impossible to determine with certainty as to the case of
the preponderating number of denominatives in atvca, whose
base nouns terminate neither in v, nor in a letter which can
have proceeded from v. I cannot, however, believe that the
Greek language has produced such formations independently,
and that, therefore, they are entirely unconnected with the
kind of forms handed down from the period of the unity of
language. Perhaps the bases in v, and those w^ich termi-
nate in a consonant which is a corruption of v, have only
supplied the type for the formations in atv(ti ; and verbs like
oAea/vo), diCTa/i/a), yKvKaivca, depjjLatva), epiSaivc^, Krjpaivci>, have
followed the beaten path, in the same way as, in German,
many bases have pressed into the so-called weak declen-
sion, in that they have extended the original limits of the
base by the addition of n, or the syllable an>. Perhaps,
too, aii/o), in a portion of that class of verbs which have
this termination, viz. those which have sprung from other
verbs, is some way connected with the Sanscrit formation
aya, with which we have before compared Lithuanian

* See §. 497., and compare G. Curtius De nominum Gnecorumjbrtna''
tionef p. 40.

1028 VERBS.

causals and denominatives in inu (see §. 745.). If the v in
those denominatives which have not proceeded from bases
in V, or /xot for /xav, is a corruption of the y (compare §. 745.),
then the ai preceding might be regarded as representing
the d (compare §. 753.), which, in most Sanscrit denomina-
tive bases in ^ ya, precedes the semi-vowel ; for though
this d belongs to the nominal base, and is in general a
lengthened form of short a {chird-yatU " he delays,^' from
chiroy "long"^), still the same, in course of time, might
come to be regarded as a portion of the derivative, and
be suppressed before its Greek representative ai, as in' the
formations in aco, aC<>>» &c. Those verbs in a/i/o) which ap-
pear to spring from more simple verbs, might, in their
principle of formation, be contrasted in a different manner
with the Sanscrit; as, e.y., avalvca (auo)), Spaivfa {ipdai),
Kpaialvca (KpaSdto), %a\a/va) (^aXaa>), stand in the same rela-
tion to the corresponding short forms, as, in the Veda dialect,
charanydmh "I go/'* does to chardmi. The broader
forms come from the noun of action ^inc^ charana, " the
going '' (euphonic for -if -na, on account of the r preced-
ing). Some Sanscrit verbs, however, of this kind do not
exactly correspond to the noun of action, from which they
spring, but exhibit a weakening or contraction of the
vowel, or the pure radical vowel instead of the gunised
one of the base word, seemingly on account of the incum-
brance caused by the verbal derivative ; thus, bhuranydmi,
" I receive " (Rig. V. 50. 6. bhuranyantam arm), from bhet-
rana, " the bearing," " receiving " (R bhar, bhri) ; tura-
nydmU " I hasten " (Rig. V. 121. 1. turanyari) from tvarana,
•* the hastening '' (R. tvar) ; churanydmU " I steal " (see
Westerg. Radices p. 337.), from ch&rana, "the stealing"'
(R. chur). As, according to rule, a noun of action in ana

* It occurs in combination with the preposition ut, ^' out," in the
Yajur-Veda, see Westergaard Had. p. 337.

DENOMINATIVES. 1029

may be formed from every root, and on this, too, are
based all the German and Ossetian infinitives^, it cannot
surprise us that, in Greek, a few denominatives of this
kind remain, whose base nouns have been lost; and thus,
e. g.f avaiv(a, from avavya), would come from a lost nominal
base avavot or avavrj. Mapaiv(»>, which has no short verb
corresponding to it, reminds us of the Sanscrit noun of
action mard-na-m, ** the dying,'^ from mar, mri, " to die,''
causal mdraydmi. Let attention be given to the Greek
feminine abstracts in ovrj, which correspond to the Sanscrit
in and, or a»d.f Verbs in avw may, in part, owe^ their
origin to obsolete nominal bases in avo.

770. How necessary it is, in the explanation of denomi-
natives, to look back to an earlier state of language, and
at the same time to examine the kindred dialects, is shewn
by an interesting class of Gothic denominatives, in which
the n likewise plays a part, though it is no way connected
with that of Greek verbs in ouvca, in whatever way these
latter may be explained. I rather recognise, as already
stated in my " Conjugational System/' (pp. 115, 116), a con-
nection in Gothic verbs like ga-fuUna, " impleorr us-gutna,
*^ effundmr distaarna, '* disrumpor,'''' and^bundna, "solvor,^^
ga-haUna, " sanorr frorqvistnOy " perdor,^^ ga-vaJcna, " excUor,*^
uS'biJena^ ** aperwrr dauthna, ** morior,^'' with the Sanscrit
passive participles in na; as, bhug-ruh *'bent," to which
the Greek verbals in vo-j correspond (oruy-i/or, (refi'vog &c.),
and from which the Gothic passive participles have some-
what diverged, in that they do not append the suffix na
direct to the root, but retain the class syllable ; thus, Mug^
a-n^ays, " bent," answering to uiif^ bhyg-na^s ; while the
verbs just mentioned point to a period of the language.

* E.g.^ Gothic hindan, Osset. ^a^Ain, "tobind'ssSanscrit handhana^
"the binding."
t Examples are: ySbchand, ^^precatio;" arhand, ^^ honoris test^kaUo.**

1030 VERBS.

when the suffix was still, as in Sanscrit and Greek, added
direct to the root; so that, e.g., ga^skaidna, **1 separate
myself" (l. Cor.vii. II. yabagaskaidnai, eav x^P^^^S)i answers
better than skaidru^ns, ** separated/' to the Sanscrit f^
chhinr-nas (euphonic for chhid-nas), " cleft.'' Compare, also,
and-bundrna, ** I am loosed (set fipee)," with bund'a-n(a)s,
*' bound ;" bi-auJe-na, ** I am enlarged," with bi-auk'a'n(a)-s,
" enlarged ;" fralus-na, " I am dissolved, destroyed, lost,"
with lusa-niays, " loosened " (Sanscrit lu-na-s '* cut off,"
"torn off"); galuk-na, "I am closed," with ga'luk'a-n(ays,
" closed ;" and-Mt-na, '* I am unloosed," with Ut-a'n{a)'S,
"tranquil;" qf-lif-na, "I am left remaining," "I remain
over " (TrepiXeiirofiai), with the to-be-presupposed Ub-a^niays,
" left remaining " (laiMs, " remnant "), for lif-a'-n(ay$9 as
the law for the transposition of sounds (§. 87.) would lead
us to expect, in answer to the Greek \enra)*» from the lost
verb feifta, laif, libum (Old High German, bi-libu, " I re-
main," bileib, '* I remained," bi-lihumSs, " we remained ") ;
tifar-haf-na, **I raise myself above " (yrrep-atpofJLai)^ with ufar-
haf-ya-niays, " raised over," " elevated ;" dis-taur-na, " dis^
Tumpar,^'* with dis'-taur-a-niaysf ** diruptm ;" gor-thaurs-na, " I
dry up" (j^rjpaivofjLou)f with ga-tlmuTS'a'n{a)rSt *' e^fjpafJLfxevo^,^
from the non-existing verb ga-thairsct, ga-thars, gaihaursuTru
Dis-hnaup-nay *' dirumpor,'"' from the root hrmp (hniupa,
hnaup, hnupum, hnupans)t is so far irregular as it has the
radical vowel gunised, whilst otherwise denominatives in
noy like the passive participle with the same termination,
attach themselves to one of the lighter forms of the verbal
theme. Us-geis-na, also, *' perceUor,'''' " stupeo^ from the to-
be-presupposed geuoy gais, gisum (Grimm. II. p. 46.), is con-

* In departure from what has been remarked at p. 441, I now agree
with Benfey (Greek Wurzellexicon II. p. 11) in taking the Sanscrit root
rick (from rifc), " to separate/' " to leave," as the root akin to the Latin
Uc (Unquo)f Greek Xitt, and Gothic lif, Ub.

DENOMINATIVES. 1031

trary to the common analogy, and should be us-gisna.
But dis-shriUnat ^'findoTt'' and tundna, *' uror,'*'' the base
verbs of which are likewise lost (skreita, skrait, skritum,
tinda, tand, tundvm), exhibit the regular vowel.

771. After that na in Gothic, as in the above-mentioned
instances, had once raised itself to be the exponent of the
passive relation, it might also extend itself to the adjective
bases, and thus denominatives in na and ya (for ya also ai,
see §. 109.* 6.), as passives (or verbs neuter) and transitive
active verbs, stand mutually answering to each other.
The final vowel of nominal bases are dropped as well before
na as before ya ( = Sanscrit ayth see §. 674.); hence, e,g,,
from the base fuUa (nom. masc. fulT-s), ** full,'" fuW-na,
" impleiyrr faW-ya, " impko f from mikila, '* great " (nom.
mikiC'SJy milciC-na, ** majgnificort'' mikiT-ya^ " magnifico '*''
(compare fxeydKtXocl) ; from veiha ipdJi-s), " holy," veM-na,
" sanctijicor,'*'* veiK^a (veiK-ais) ** sandifico f from ga-ndha
{ganjoK-sj^ "enough," ga-ndK-nay '' expleor,^'' gan6K-ya, " ex-
pleoj'' from managa{manag''-s), "much," iaanou^-na^ "alfundo''''
("I am made much"); manag'^'yay **augeo\''' from gabiga
(gabig'^'s), " rich," gahig'^-na, " locupletattts sum^ gabig'-ya
" hcuphtoy It cannot surprise us that the base words of
denominatives in wx cannot be all cited from the lingual
sources which have been preserved to our time, nor that
some were already obsolete in the time of Ulfila, but sur-
vive only in the denominatives, of which they were the
parents. Thus, e. gr., an adjective base dr6ha (drdhs),
" troubled " (Anglos, drof), does not occur ; whence comes
drdV-ya, " I trouble," " excite," " shake," and drdb'-na. " I
am troubled." Inseparable prepositions precede the de-
nominatives, as they do the primitive verbal themes, though
the base word be simple ; as, e. gr., from blinda (blind'-s),
** blind," comes ga-bUnd'-na, ** I am blinded," and ga-blindr-
ya, " I blind," " dazzle ;" from dumha (dumb'^'-s), " dumb,"
af'dtmV-na, " I become dumb," ** grow speechless " (Mark

1032 VERBS.

iv. 39. c^umbn me^iiuMro). It is possible, that from the
simple adjective bases at first simple denominatives pro-
ceeded, and from these, which no longer exist, or cannot be
cited, compound denominatives ; thus, from drnnhn came,
at first, dumbnot and thence afdumbna ; as, in Latin, from
mutur-s, mutescOf and thence obrmdesco.

772. To return to the Sanscrit, we must remark that
denominatives formed with i| ya partly express a wish ; as,
e. g., pati-ydmi, " I wish for a spouse,^' from paii ; putri-
ydmif " I wish a son, or for a son, or children,^' from putra.
These forms lead us to the Greek desiderative denominatives
in lao), which, however, in departure from the Sanscrit,
reject the final vowel of the base noun, while the latter
lengthen it, but in doing so weaken d to i; thus, patri-^
ydmi for pvird-ydmi* And Greek forms like Oavar-iiw,
oTpaTJ/Y'-iao), KAav<r*-iaci), are properly based on the causal
form of the just-mentioned Sanscrit denominatives in ya;
thus, davar-idoiy 0ai/aT-iao-/L(6i/= Sanscrit forms like putri-
yayd-mif pviri-yayd-mas, while putri-yd-mi, pidrt-yd-TnaSi
would lead us to expect Greek forms like davar'tia, davar-
io^fjLCv, or, according to §. 502., davaa-aa, davaa-aofxev. It de-
serves, however, notice, that, in Sanscrit, denominatives in
ya occasionally adopt the causal form without a causal
signification; thus we find, without a causal meaning, 'j*
the gerund asiiyayUvd, which belongs to the causal form,
but is used as coming from the denominative asur-ydmi,

I curse," "execrate" (intrans. "I am wrath," from asu

life").

«

* Bat we find in the Veda dialect cuvor-ydmi^ ^^ equas cupio," from aiva,
«a horse" (S. V. II. 1. 1. II. 2.).

t Nal. 14. 17. : krodhdd asuyayitvd tarn, " ird easecrando eum," On
the other hand, dhiimdyaydtni, the causal of dkumd-ydmiy ^^fumo," has
also a causal meaning: dhumdpayan diiati, ^^ causing the regions of the
world to smoke."

DENOMINATIVES. 1033

773. With the causal form of denominatives in i| ya
may be compared also the Latin in ig&. The i would then
be the final vowel of the base noun, either in an unaltered
form, as in miti-gd'S, levi-gd-s, navi-gd-s ; or the weak-
ening of a heavier vowel (see §. 6.), as in fumi-gd-s (for

fumu^gd'S, or fumo-gd-s), remi-gd-s, dari-gd-s, casti'gd-'S (but
puT-gd'S with i suppressed) ; or the unorganic extension
of a base ending in a consonant, as in liti-gd'S opposed to
jur-gd-8. The g must be taken as the hardening of y, which
indeed occurs, perhaps, nowhere else in Latin, but is not
uncommon in the kindred languages (see pp. 110. and 993.),
and with which is connected the fact, that in Greek C often
stands as the hardened form of an original y (see §. 19.).
The d of the forms in question, as generally of those in
the first conjugation (except where it is radical), must be
the contraction of the Sanscrit a{y)a ; and thus fumi-gd-s
would be, as it were, the Latinization of the Sanscrit
dhumd'ya(^)a'Si, "thou makest to smoke '^•f'. If, however,
we agree with the common opinion, which, however, is op-
posed by Duntzer, ("Doctrine of the Latin Formation of
Words " p. 140,) in recognising in the verbs in igo com-
posites with ago, we must then divide thus, mif-igOffurn-'igo,
&c., and assume a weakening of the radical a of ago to i,
and a transfer of igo from the third conjugation to the
first, both of which things occur in facere, which, at the
end of compounds, becomes jicare.

774. Bases which, in Sanscrit, end in n, reject that letter
as well in desideratives as also in other denominatives in
ya. Other consonants, also, are occasionally dropped before
the denominative suffix ^ ya ; hence, mthd-t/^, ** I become
great'" (Mid.), from vrihat, in the strong cases vrihant, pro-

* I retract the conjecture expressed at §. 109**. 1,
t See p. 379 and §. 772. note **.

3x

1034 VEBBS.

perly a participle present from varh, vrih, "to grow/^ Thus
tripd-$fif rdlid-yi, from the participles tripant, tripat, rdhant,
rdhat (see Westergaard Rad. pp. 337, 339). We might con-
sequently expect from the participle of the auxiliary future
forms like dd-syd-yi for dds-yat-yS, or ddsyant-yi ; and it
follows that "we may regard the Greek desideratives in (re/ci)
as denominatives, i, e. derive them from the participle, and
not from the indicative future. The e, for instance, of
7raf>a*$a)-a'6/a) must then be looked upon as the thinning of
the o of the sufl^ ovr, and Trapa-ScM-e-iia must therefore be
derived from 7ra/oa5aKro(vT)-ia) ; just as above, §. 503., deK-
a^6/jLevo£ from deKovr. But if Greek desideratives in (reica
spring from a future participle, then Latin desideratives
in turio, as cccnaturiOf rmpturio, parturio, esurio (from es^turio,
see §. 101.), may be placed by their side as analogous
forms in which the i appears to correspond to the San-
scrit suffix i| ya, though usually the i of the Latin fourth
conjugation corresponds to the Sanscrit ayoj while the sim-
ple ya is represented by the i of the third conjugation.
As, however, the i of the third conjugation is occasionally
altered to the i of the fourth •!•, it cannot surprise us that
some denominatives of the Latin fourth conjugation should,
in their origin, belong, not to the Sanscrit formation aya,
but to ya ; and so equ-io, equ-is, both as regards its base
word and its derivation, might be compared with the Ve-
dian asvdydmi, '*equos cupio,'*'' mentioned above (§.772. Note ).
775. Denominatives with a desiderative meaning are

* The short u of verbs in turio occasions me no difficulty In deducing
them from the participle in turu-s* The incumbrance of the verbal deri-
vation appears to have occasioned the shortening of the vowel, as in deno-
minatives like colorOy hahorOy compared with color^ colo-ris, honar^ honor-is.

t See §. 500., and Struve On the Latin Declension and Conjugation^
p. 200 (from^(2to, in Plant., ybrfCny icoiagrcLdiorf aggrediri; frompariOf
in £nn., parire ; from morior^ mofimur).

DENOMINATIVES. 1035

also formed in Sanscrit by the suffixes sya and asya ; e. g.p
vrishasydmif " to long for the bull f ' asva-sydmi, " to long
for the stallion'^ {equio); madhv-dsy&mh "to wish for
honey/** We have already noticed the agreement of these
forms with that of the auxiliary future, as also, as respects
the sibilant, with the desideratives which spring from
verbal roots. From Latin may be adduced imitatives in
9S0, as has already been done by Diintzer ("Doctrine of
the Latin formation of words'' p. 135). Whence, e. gr.,
patri'Sso would stand by assimilation for pairi-^o (compare
the Prakrit futures, §. 656.), with i as the extension of the
base noun, as in pairi-hua. The i of attici^saot grtBci-sso, is the
weakening of the final vowel of the base noun. The first
conjugation, however, does not admit of comparison with
Sanscrit desideratives like asva-aya-ti, which leads us to
expect the Latin third conjugation, as in derivatives from
verbs like cape-ssot incipi-sso, lace-sso, peti-sso, which admit
of comparison with Sanscrit verbal desideratives in sa — ^in
so far as their s really stands for sy — or also with the aux-
iliary future. The e or i of Latin forms is, however,
most probably the class vowel of the third conjugation,
though usually this does not extend beyond the special
tenses. Incesso, from cedo, is probably an abbreviation of
incedesso ; and arcesso, if it comes from cedo, of arcedesso.

776. Outwardly a similarity presents itself between the
Sanscrit nominal desideratives in sya or dsya, and the
Latin inchoatives in aaco and esco : these, however, as re-
spects their principle of formation, are scarcely transmitted
from the time of the unity of language, but most probably
first originated on Roman ground, by the annexation, as
it appears to me, of the verb substantive with the meaning
" to become " to nominal bases, which, when they terminate
in a vowel, drop this before the vowel of the auxiliary
verb (compare §. 522.). Thus, as poa-sum from pot-sum for

poti-aum, pot-eram for poti-eram ; so, e.g., puelT-aacot ir-

3x2

1036 VBRBS.

CL8C&r, puer^asco (from the base pueru,^ro)f tener-asco, and
tener^scOf acet-asco, gef-asco (from gem), herV^escOf exaqa-
escOi phmC'-esco, flamrn^sco, amar^escOy aur-esco, clar'^esco*
vetust^-^scOi dvlc-escotjuven-^sco, celebr-esco, corn^esco. Whe-
ther we ought to divide long*-iscOt vetusf-isco, or hngi-scoj
vetusti-scOf may remain undecided. In the former case the
i of the auxiliary verb might be compared with that of the
Greek imperative la-dt : in the latter i is the weakening
of the final vowel of the adjective base, as in compounds
like bngi-pes and derivatives like hngi-tvdo. Bases ending
in a consonant experience no abbreviation , thus, arhor^
esco, carbon-esco, lapidr-escot matr^esco, nod^esco, dit-escOf but
oput-esco from opuient-escOf which reminds us of the Sanscrit
denominatives from abbreviated participial bases in rd
mentioned above (§. 774.). The verb substantive, which I
think I recognise in these formations, answers to the ob-
solete future esco (escits superescit, obescit), which, however,
in composition, has occasionally retained the original a ; as
in Old Prussian, also, in its simple state, cLs-maij as-sai, as-t,
corresponds to the Lithuanian es-mi, essi, es^ti. How
close the notions of futurity and of becoming, as of future
existence, approach one another needs no mention. With
respect to the guttural which has attached itself to the
root of the verb substantive, asco, esco and the isolated
future esciti resemble the Greek imperfect e<rKov, which,
with the rejection of the radical vowel, enters also in-
to combinations with attributive verbs (Sii/et5e-(r#ce, icotXee-
(TKovt k\a<ra'<rK€).* The Latin esco, also, when added to

* I hare no hentation in ascribing the vowel which precedes the a- to
the temporal base of the simple verb ; for the o of cKoKeov is, in its origin,
identical with c, and stands in place of the e of cKokccs, e/coXf €, only on
account of the nasal which follows: the c of the 3d person of the 1st
aorist is identical with the a of the other persons, which is everywhere
retdned where an ending follows it.

DENOMINATIVES. 1037

verbal bases, relinquishes its initial vowel ; for the a (d), e
(^), and % (i) of forms like laba-sco, amd-sco, consuda-sco,
genera-SCO, palle^scot viresco, rube-sco, senti-sco, obdormi'Scot
are clearly the characters of the first, second, and fourth
conjugations ; on which account we here divide difierently
than above, in py£r-asco, clar-esco, dulc-esco &c. . In com-
pounds with bases of the third conjugation the i of gemi-
SCO, tremi'ScOf must be regarded as by nature short, as it
is identical with the i of gem-i-s, trem-i-s (see §. 109*.
1.), which leads us back to the Sanscrit a. The i of pro-
fici-scoT, concupi-scor, is identical with that oifaci-s, projici-s,
cupi-s ; nanci'Scor presupposes a simple nanco, nanci-s ;
frage-sco exhibits e for the ♦ otfrangi-s (compare §. 6.), and
has lightened itself by the rejection of the nasal of the
root. To Latin forms like laba-sco, ama-sco, palk-sco, cor-
respond, in their principle of formation, Greek forms like
yrfpa-cKCt), ^j8a-<TK<o, IKd^o'KoiJiai, oLKSYj-aKt*) ; where, however, it
is not asserted that the Latin ^ of the second conjugation
is connected with the Greek tj of forms like ire^iKTi-KOL,
<l>iKYj(T(jii, though both lead us back to the Sanscrit aya ; but
of this the Latin contains the two first letters in the con-
traction of ai to 6 (see §. 109*. 6.), while the Greek ri of
fpiKfjiTijii and ee, eo of <f>iKe€Te, i^iKeo^ev, contain the first and
third letter of the Sanscrit aya, either separate (in ee, eo),
or united in ?;. The i of forms like evpi'(rK(»>, aTepl-(rK(a,
a\i'<TKOfJLat, anl3\t-<TK(»>, is scarcely a vowel of conjunction, but,
in my opinion, only a weakening of a heavier vowel ; thus,
evpitJKifi, <TTepi'(TK(»>, for evprjaKO), a-Teprj-a-Kit) ; a/i^Az-cnco), aKt-
(TKOfiai for a/xjSA.co-o'ica), dAa)-<rico)Ltai ; to which, among other
things, the futures evptj-tru), oKcu'dopiai, &c., point. We must
remark the weakening of o to t in ovi-vrjfju for ovovfjfxi,
oTTiTrretJa) for oTroTrrei/o)*; and, moreover, the forms aXdrj'O'KO)

* See §. 754., and compare dTTomrj and oTrwTrco), which forms, by the
lengthening of the radical vowel in the second syllable of the root, which

is.

1038 VBRBS.

and dKdi'O'KOi} which exist together. I am now inclined, in
departure from what was remarked at §. 751., to assume that
the Greek reduplicated forms in o-kw, in spite of their
striking resemblance to Sanscrit verbal desideratives like
jijndftdmi (compare yiyvuKTKia), are nevertheless not histori-
cally connected with them, but, as comparatively younger
formations, have arisen from the junction of the verb sub-
stantive in a form analogous to the iniperfect ea-Kov and
Latin future escit, but deprived of the radical vowel, to
roots repeated according to the principle of the Sanscrit
third class (see §. 109*. 3.). Thus, yiyvwrKo^ fUfxvi^(TK(a,
presuppose simple verbs like yiyvc^fju, fxifjLvvjfUp according to
the analogy of d/Sco/xf, ridtjfu, l3il3tjfAt, or such as 7i7voa>,
fUfivita. And eyvc^v and yvtixTo bear the same relation to
the probably existent yiyvafu that eSwv and Scixria do to
SlSodfu. If, however, the Greek reduplicated forms in erica)
must, with regard to their principle of formation, be
looked on as distinct from Sanscrit verbs like j^Msdmi,
the same must hold as regards Latin forms like no-sco, di-
sco (perhaps from dida-sco), pchscor, na-scar {gnor-scor by
transposition from ganscor), which correspond to Greek
unreduplicated forms like fia-a-KO), dv-fj-cTKoa,

777. In Sanscrit, denominatives may also be formed by
annexing simply an a to the theme of nominal bases in
the special tenses, which o, like that of the first and sixth
classes of primitive verbs (§. 109*. 1.), is suppressed in the
universal tenses. A final a of nominal bases is dropped ;
hence, e, g., I6hif^a-ti, ** he is red,'' from Whita, I am un-
able to quote from authors instances of such denomi-
natives : there occur, however, among the roots exhibited
by Indian Grammarians of the first or sixth class, several
in which I think I recognise denominatives from bases in

is twice repeated in its full form, correspond admirably to the Sanscrit
intensives there mentioned.

DENOMINATIVES. 1039

a; thus, among others, bhdm, "to be angry,"' bhdm-a-tS,
** he is angry," which I derive from bhdm-a, ** anger :" this
latter, however, which also signifies ** light," " splendour,"
clearly comes from the root bhd, " to shine." As the
Latin i of the third conjugation corresponds to the Sanscrit
a of the first and sixth class, so metur-i-t, trUm-i-t, statu-i-t,
minu'i-t, correspond to the Sanscrit denominatives here
mentioned. In Greek correspond denominatives, which
in the special tenses add o and e to the nominal base ;
thus, e.g., fjLYfvi'O-fxev, ^tivl-e-re, iripi-o-iiah fiYfTi-o-fiai, SaKpv-
O'fieVf fxedv-o-fxevj Idv-o^fJiev, d%\tJ-o-/xei/, ^aciXev-o-fxeVf ^Spoe-
^ev-O'fjiev. What, however, are we to say of that rather
numerous class of denominatives in evo), which are not
founded on any nominal base in ev; e.g., Kop'-evo-fxat, "I am
a maiden ;" woAxt -ev-w, " I am a citizen ;" dOK-ev-a, ** I con-
tend," properly, " am in strife f laTjo'-etJ-w, ** I am a phy-
sician;" KpaTi(rT'€v-(ji>, **I am the best;'' KoAocK-ev-o), "I am a
flatterer, flattering;"' 5ot>\'-et5-c«), "I am a servant;" dhfjff*-
eu'U), " I am true" ? If the verb substantive, which in most
of these formations is more or less evidently present in
spirit, be also contained therein bodily, we must then have
recourse to the root ipv (see p. 115), which therefore, in
these compounds, has preserved the original notion, while
in its simple state the causal meaning of bringing into
existence, " making to be," prevails. The e of -euo) would
therefore be the Guna vowel, corresponding to the a of the
Sanscrit bhav-d-mi, " I am," " I become ;" and, with respect
to the dropping of the radical labial ei/o), would stand on
the same footing with uU vi, of Latin forms like pot-ui,
,mon'Uh ama-^if avdi-vi, (see §. 556.). In Gothic the verbs

* The Ossetian also has, in its simple state, lost the labial of the auxi-
liary verb under discussion, and gives, e.g., wa-d^ "he must be," wonth,
** they must be," corresponding to the Sanscrit bhavatu^ bkavantu : see
" The Caucasian Members of the Indo-European Famify of Languctges^"

pp. 43

1040 VERBS.

in na (as fuUrui, " impleor ")» mentioned above (§. 770.), be-
long to the class of denominatives here mentioned. These
verbs in na come from participial- bases with the same termi-
nation, which, like the Sanscrit bases in a (rdhif-a-ti), reject
their final vowel before that of the class; thus, fuUn-i-th,
" impletur,''' from fuUna-i-ihf (or fuUna-a-th (see §. 67.), plural
fvUn-a-nd, as in Sanscrit rdhif-a-ii, rdhif-a-rdu But this
kind of formation holds, in Grothic, only for the present
and its derivatives, while in the preterite an 6 takes the
place of a or i ; so that, e. g., fuUn^da, " I was filled,^' in
its principle of formation agrees with Latin forms like
regn-d-vi, the base noun also of which, regnu (" kingdom
as ruled"), with respect to its derivative suffix, is con-
nected with the to-be-presupposed Gothic haiSefuUna (San-
scrit jwrna, "filled").

FORMATION OF WORDS.

778. With regard to the formation of verbs there re-
mains nothing to be added to what has been already said
regarding the structure of roots and the classes of verbal
bases (§. 109'.) which proceed thence, and subsequently re-
specting the formation of derivative verbs. The primitive
pronouns, and the appellations of numerals, do not follow
the ordinary rules for the formation of words (see §. 105.),
and, with their derivatives, are discussed in the paragraphs
allotted to them. We shall now discuss simply the for-
mation of substantives and adjectives ; and, first, those
which stand in close connection with the verb, and, both
in the organization and in the application of language, play
a very important part : we allude to the participles and
the infinitive. It might be said that we ought to treat of

pp. 43 and 82, Rem. 48. In Persian the present of the verb substantiye
may be combined with any substantive, adjective, as well as with the
personal pronoun ; e.^., piram^ '' sen&e sum ;" manam^ ^^ ego sum**

FORMATION OF WORDS. 1041

the formation of nouns before treating of their inflection,
because words must be formed before they are inflected.
But for practical considerations it appeared more useful,
at first, only to lay down the principle of the formation of
words generally, as is done in §§. 110. 111., and to defer
the more full investigation of the subject to this place.
At all events, the theory of the formation of tenses must
precede that of the participles, as the latter, for the most
part, irrespective of their nominal suffixes, rest on a prin-
ciple of formation similar to that of the corresponding
tenses of the indicative, and bear a sisterly, if not a filial
relation to them. It will, however, be clearly seen from
the following paragraphs how requisite an acquaintance
with the forms of cases, and with the distinction of genders,
is to the understanding of the theory of the formation of
words.

779. The participle present active forms a point of ob-
servation as regards the representation of the original
unity of the Indo-European languages; and it is here
worthy of notice, that several of the still living tongues of
our quarter of the world have, in some cases, preserved
the original formative suffix in a more perfect form than
the Sanscrit in its most ancient sources. The full form
of the suffix is rd ; the Sanscrit, however, exhibits the n
only in a few cases, which in all places, where a division
of the theme into stronger and weaker forms occurs, has"^
retained the original and full form of the base (see §. 129.) ;
hence, e.g., bharan, bharantam = 4>^p(av, ^G^ovraj Jerjenjem,
diial bharantdu, Veda bharantd (nom. ace. voc.) = ^epoi/re,
plural bharantas (nom. yoc,) =z <f)epov7egy ferentes ; but in the
accusative we find bharatast by the loss of the n in the latter
part of the word, opposed to (jyepovr-agt and so in all the other
cases of the three numbers the n is dropped in Sanscrit ;
and in the genitive singular bharatas stands, from this
loss, in an inferior position when compared with the Greek

1042 FORMATION OF WORDS.

^^povToj, Latin/er«t/w, Gothic 6ajran-dtn-» (see p. 138), and our
German strong participial genitives, as stehendes, gehendes.
The Lithuanian also has till the present time retained the
nasal of the participle present through all the cases of the
three numbers in both genders: it extends the theme,
however, in the oblique cases, by the addition of ia; and,
according to a universal law of soimd, changes the t before
t, when this is followed by any vowel but e, into the sound
isch, which Ruhig writes cA, Mielcke cz; hence, e-g., degam,
"the burning" ( = Sanscrit dahan), according to the ana-
logy of Zend forms like barans, Latin likeferens, iEolic as
Ttdiv^, accusative degantin (for degantien, from -ian), geni-
tive deganchio.

780. The Old Prussian, difiering^ from the Lithuanian,
extends the participial base in the oblique cases by the
simple addition of i, and so far agrees entirely with the
Latin, which, e.g., forms simply /erens from the haseferent,
which has not exceeded its original limits, but which, in
all the other cases, follows the analogy of bases in i.
FerentUa and fererdi-um belong as decidedly to the i de-
clension as faciii-a, fadli-um. We are therefore right in
dividing ferente-m just as facUe-m (from /oci/i-w), though
from a base, ferent, the accusative could be in no case other
than ferentem =Zend barent'em. The participles present
masculine which remain to us in Old Prussian are, dilants,
"the workfer,^^ " working f** "j* sidans, **sedensf^ emprikisins,
** prwsens ;'''' dative emprHd^senti^smuy according to the pro-

* v^rhf^o^^ third filp^j ^^ ^ftP"^*^ o^'"g t^ the jn^cM^b"*"^*' of the

sjUable of reduplication, have lost the nasal in the strong cases also;
hence, e,g.^ dadatam compared with hibovra^ dadatas with bibovres (com-
pare §. 459.).

t According to the mode in which the two following examples are
written we should expect dUans; but as respects the retention of the
T-sound, dUanta corresponds to Gothic forms like bairands.

FORMATION OF WORDS. 1043

nominal declension (see §. 170.); niaubiUinti'S, "of the under
age," ** not speaking" (ir/ardia) ;* ripinti-n, '* sequentem f'f
emprild waitiainti-ns (ace. pL), " contradicerdes ; ^ wargu-
seggienti-ns, " maleficos,^'' The following are adverbial da-
tives, giwarUei, *' living," and stanintei (also stanirdi) " stand-
ing," from the bases giwanti (Sanscrit jivant), staninti (see
Nesselmann, pp. 52 and 76).

781. Before the feminine character i, the Sanscrit, ac-
cording to the difference of conjugation of the respective
verbs, either retains the nasal of the participial suffix or
rejects it, and in such a manner as that verbs of the first
principal conjugation regularly retain it, and but rarely
reject it, while conversely those of the second ordinarily
reject it, and only occasionally retain it ; while the Gbthic
and Lithuanian have constantly preserved it Compare,
€,g„ with the Sanscrit vasanti, "the inhabiting" (also va-
satiy Nal. 13. 66.), from vas. Class 1, the Gothic visandei
(Them, visandein, see §§.120. 142.), "the abiding or being;"
and with the Sanscrit dahanti, "the burning," the Lithua-
nian degardi (gen. deganchiSs, see p. 174, Note *). In Greek,
depairovTts is in form a solitary participle present feminine
with <5 = Sanscrit i, according to the analogy of the femi- ^
nine bases in rpiB^^tri, Latin tri^c, mentioned in §. 119.
The root ^tns as. Class 2, of the verb substantive, forms in
Sanscrit scdi, "the being," never sardi; the Lithuanian
esardi therefore surpasses the Sanscrit both in the reten-
tion of the radical vowel and in that of the n of the suffix.

* BiUiy ^'I speak." The inseparable preposition au, combined with
the negation m, corresponds to the Sanscrit ava.

t Also ripintinton^ in the last syllable of which I think I recognise an
appended pronoun or article= Sanscrit torn, Lithuanian tan^ Greek roi/.
As regards the o for a, compare the accusative of the participle perfect
passive ddt<hn, ^' datum" =SaDacTit dattam^ from daddtam, irregularly
for ddtam.

1044 FORMATION OF WORDS.

In the masculine nominative, also, the Lithuanian esans
has two points of superiority to the Sanscrit sam the re-
tention of the radical vowel, and of the nominative sign :
V the latter is shared also by . the Latin sens, of j^ce-

sens, ab-sens, to which the abovementioned X§' '''80.) Old
Prussian sins, of emprild'sins, admirably corresponds. The
Greek, for the most part, with its iv, contrasts disadvan-
tageously with the Lithuanian esans ; for while the latter
has, together with the case sign, preserved the complete root,
we miss in cjv both the entire root and the expression of
the nominative relation. The epic and Ionic form e<av,
however, leads us to conjecture a formerly existing eo-cov,
9.ud the suppression of the o- in this position is not surpris-
ing according to §. 128. It is, however, not less marvellous
that a form which, in Greek, has been corrupted for thou-
sands of years, quite up to remote antiquity, and which
has been tolerably accurately retained by the Latin only
under the protection of the prepositions pros and ab*,
should have remained quite perfect in the Lithuanian up
to the present day.

782. The Indian Grammarians assume at, in the strong
cases ant, as the sufEx of the participle present. I cannot,
however, attribute to the suffix the a of forms like bharant,
any more than the o of the Greek ^e/oovr : the vowel be-
longs in both languages to the glass syllable; i.e, the o
of 0€|O-o-vT is identical with that of tf^ep-o-ixev, ^e/o-o-vTi, and
with the e of ^ep-e-re, e^ep-e-f, &c. That the Greek par-
ticipial suffix is simply vr, not ovr, is clear from the conjuga-
tion in lu, where vr attaches to the final vowel of the root or
of the verbal theme (SiJo-io*, ride-vr, loTa-vT, Jenc-viz-vr) : the
Sanscrit, however, in accordance with a peculiarity, which,
in my opinion, first arose after the separation of languages

* On the other hand, in potens, just as in the simple ens, the sibilant
is lost.

rOEMATION OF WOKDS. 1045

in cases, where the nt or t of the suffix would be added to
a letter other than a or d» prefixes to the suffix an a
(compare §. 437, Remark, and §. 458.), or extends the ver-
bal theme by the addition of an a ; hence, e, g,, strinvant,
" strewing *" (for strinunt), answers to the Greek laase
(TTopvvvT, The € of Latin participles of the third conjuga-
tion, €. g.f of veh-e^nst veh-e-ntem ( = Sanscrit vah-a-ru vah-a-
ntam, Zend vaz-a-ns, vaz^a-niemX is in origin identical
with the class vowel i (from a, see §. 109*. 1.) of veh-i-s,
veh-i'ti &c. (see §. 507.), and is based on the circumstance
that before two consonants the Latin language prefers / to
i (see §. 6.). In the fourth conjugation, ie, e.g., in and-
i-ens, represents the Gothic ya and Sanscrit aya of forms
like sat-ya-nds, ** placing "' = Sanscrit sM-aya-ru "making
to sit" (compare §. 505.). It does not require mention,
that in verbs of the first and second conjugation the a and
Cf as in am-a-ns^ mon-e-ns, belong to the conjugational syl-
lable ; the a, however, of da-ns, sta-ns, fa-nsy and Jla-ns, to
the root : and as little does it require notice, that in Ger-
man and Lithuanian the vowel which precedes the n of
the participle present is identical with that of the class
syllable. Compare, in Gothic, bair-a-nds, ** the carrying,""
Dahs-ya-nds (Zend ucs-ya-ns), " the growing "' (see §. 109*. 2.),
sai-ya-nds, " the placing,"** " making to sit,", saIb-6-nds,
"the anointing,,' with bair-a-m (Sanscrit bhar-d-masX
" we carry,'" vahs-ya-m, " we grow,"" sat-ya-nh " we place ""
(Sanscrit edd-^yA-mas), saJb-d-m, "we anoint;'" and in
Lithuanian, wez-a-ns, " the conveying,"' with wez-a-me, " we
convey ;" myl-i-ns, " the loving,"" with myl-i-me, " we love.""
With regard to the non-correspondence of the Lithuanian
es-a-ns, " being,"" to es-wi, " I am,"" es-me, " we are,"" we
must observe, that here an auxiliary vowel is necessary in
the participle, which in the Sanscrit s-or-n (accusative s-a-
ntam) occurs in the same form, while the Latin ^sens places
in its stead an e, and the Old Prussian -sins an i.

1046 FORMATION OF WORDS.

783. In Old Sclavonic, the so-called gerundives corre-
spond to the participles of the kindred languages, and that
of the present to the participle present active here under
discussion, In the nominative singular masculine, where, e.g.y
BE^'Bi veQ), " vehemr answers to the Sanscrit vahan, Zend
fkizanit Lithuanian tve£am, and Gothic vigandsf we should
scarce observe the analogy of the Sclavonic form to those
of the kindred languages, as, according to a universal law
of sound, all final consonants in Sclavonic are suppressed ,
but in the dual, BE^&qiA velCumhcha'^f corresponds to the
Vedian vaharUd and Zend vazanta; and in the plural,
BE^&^E (ve^nshche) answers to the Sanscrit vahant^as,
and Greek expvr-eg (see p. 618, Note 3.) ; where it is to be
observed, that ^s shch more frequently occurs as the
euphonic alteration of t (Dobrowsky, p. 39, Kopitar, p. 53),
just as d, under similar circumstances, becomes ikA scM :
a sibilant, therefore, is prefixed to the 7*-sound, and, be-
sides, the original t is changed into ch as in Lithuanian
likewise the latter is used before i, with a vowel following.

* See §. 255. 1. I now think that the monosyllabic words also mast
be subjected to the nniyeiBiiL law, as I no longer recognise in the forms
HAC naa and BACb vas of the genitive and locative plural of the two
first persons the Sanscrit secondary forms nas and vas^ but I refer the
CB » of the genitive to the Sanscrit pronominal genitive termination sdm,
and that of the locative to the Sanscrit locative termination «u. The fact
that the s of these terminations is elsewhere changed into ^ ch (see
§§. 255. m. 279. and p. 355, Note 6.), and that in Sanscrit the genitive
termination adm occurs only in pronouns of the third person plural, con-
ceals the causal nature of the ending of the forms nAC1> na-Sy BACl*
va-s ; but in Old Prussian also the ending T^fm sdmy in the form much
nearer to the Sanscrit wn, has made its way into the pronouns of the first
and second person; hence here are found nou-Mn^ ^fioi^, iau-son, vfi&v,
after the analogy of stei-s<my rfi>y=sSanscrit te^ham^ answering to the Scla-
vonic HACl> na-s and BAZ'b va-s.

t As to &=u», see the Remark at the end of the preceding §.

FORMATION OF WORDS. 1047

Compare, therefore, in this respect, the dual BE^^iqiA
ve^umhcha with the Lithuanian wezanchiu. It is probable
^ that in Sclavonic also, as well as in Lithuanian, a y, or the
syllable yo, has, in the oblique casesj mingled with the t of
the participial suffix, and under the influence of the t/
the preceding t has become ip shch. So in Dobrowsky's
third conjugation, in which, in the first person present, a
y is found before the termination vn, forms occur like
mS^ip&^ munshckun, " turbo,'*'' euphonic for muntyun, infinitive
munt-i-ii*. In the feminine singular the gerundive spoken
of is BE^il^ipu t?efMnsAc/i«= Lithuanian weiardi, "the con-
veying '' (genitive weianchi6s\ Sanscrit vahardt

Remark 1. Dobrowsky, to whose grammar I was circumscribed in
treating (§. 155.) of the Old Sclayonic alphabet, makes neither an ortho-
graphical nor a phonetic distinction between & and oy, or Sf, and nerer
uses the first-mentioned letter, as he everywhere writes K> for \S%.* It is
now, however, generally snpposed, and I think with good reason, that the
vowels & (with y, li^) and A (with y, ia) contain a nasal, as was first disco-
vered by Vostokov, but still held by Kopitar (Glagolita, p. 52) to be doubtful.
It is, however, certain that the vowels &, l&» A, y\, in the Old Sclavonic
Grammar, as Kopitar has informed us, occur scarce anywhere but where
the Polish has vowels with a nasal ; and comparison with the ancient
allied languages leads us to expect a nasal, for which reason I have before
assumed a corruption of on (from an) to ^ (see §. 155. ?.). On the other
hand^ however, oy, or sf, and the u contained in K) {yu\ wherever these
letters occur in Old Sclavonic in their proper place, in forms which admit
of comparison usually, according to etymology, represent the Sanscrit
^ 6 (for a+tt), or its resolved form av; hence, c.^., oyCTA iista (neuter
. plural), " movL^'=^shtha" "lip" (Theme) ; Cf oyTH 9rvrti, "to hear "=
'rdtum (irrespective of the infinitive suffix); BOyAHTH bud-i-ti^ "to
wake "=&d<2a^f^m / moyH «Ai2i, left "s^atT^a. So in the termination
of the genitiye locative dual, where, e.p., OBOIO ^^amborum», in ambcbus"
answers to the Sanscrit lAhayoSy and Zend ub6y6 (see §. 273.). Now let
us examine the cases in which nasalized vowels, the nasal of which I now

* Miklosich compares the Sanscrit root month, " to shake f and & uh
therefore stands for the Sanscrit an. See the note to the preceding §.

1048 FORMATION OF WORDS*

express, as in LithuaniaQ, by n (see §. 10. )» in grammatical terminations
or sufifixes, correspond to a Sanscrit n or m with a preceding vowel (a or
d). There appear, therefore, if I have not overlooked any thing, the
following : —

1. Accusative singular of feminine bases in a ; e.g»^ BAOB& vdovuh^
" viduam "=vidhavdTn,*

2. Accusative singular of pronouns of the first and second person: ma
mmi^ TA ton=Sanscrit mdm, tvdm; like the reflexixe CA sah.

3. Accusative plural of masculine pronominal bases of the third person
in ya, and therefore also of definite adjectives compounded with the
base ya. Compare v\ yah^ '^ eos" with the corresponding Sanscrit
ydh^ ^^qtios" and Old Prussian accusatives like scJia-ns^schi-ns^
" ?ios" wira-nsy " vivos" Gothic vaira^ns (see §. 236.).

4. First person singular present, where St Mn=3Sanscrit ami; e.g.^
veCun=vahdmi; ^\Sk ayuh=aydmi, e.g,^ rydayuh=r6daydmi (see
§. 766.).

6. Third person plural of the present, where StTh u/2/y=Sanscrit anti;
e.g.^ BE3&Tb veCuhtyzszvahanti; and in Dobrowsky's third conju-
gation (see Kopitar, p. 61), lATb yanfy = Sanscrit ayanti,

6. The above-mentioned gerundive or participle present.

The nasal vowel in the genitive singular and nominative accusative plural
of feminine bases in ya, e.^., in BO AfA volyahy '^ voluntatis^ and ^' volun-
tates (nom. ace), appears surprising. If we consider, however, that in
the three cases spoken of the Sanscrit grammar exhibits a final «, which
is also contained in the Lithuanian and Lettish, which approximate closely
to the Sclavonic languages, as also in Gothic in all the words which cor-

* Compare §. 266. The Polish also, in the corresponding forms, has
a written nasal vowel, though now, at the end of a word, the nasals,
though written, are no longer pronounced ; just as in the instrumental,
where I regard the Sclavonic vdO'Voy-uh^^anscrii vidhavay-d as join-
ing to the old instrumental termination the new also^ with a corrup-
tion of the my (Dobr. gives only m) to the now probably very weak
nasal sound n. Remark, that in the plural instrumental, the feminines,
especially rather than the masculines and neuters, have the termina-
tion mi (see p. 349) ) for which, in Lithuanian, both in masculine and
feminine, stands mi«, only that the masculines in a have contracted amis
to ais.

FORMATION OF WORDS. 1049

respond to the Sanscrit feminine bases in d *, we are led to infer the
nasalization of a final «, as in the Prakrit instrumental termination
hih = Sanscrit bhis (see §. 220.). The y especially appears to have
protected the nasalized vowels which follow it, as we may conclude
from No. 3. and the gemndiyes mentioned below (Remark 2.). A
place where the Old Sclavonic has a nasal vowel at the end of a word,
while the Sanscrit has a simple vowel, occurs in the nominative and accu-
sative singular of neuter bases in n; in uma imah^ ^^nomen" (from the
base imen from iman), answering to the Sanscrit ndma, from ndman.
Here, however^ the nasal of the Sclavonic nominative and accusative can-
not surprise us, as it belongs to the base word, and the Latin also has firmly
preserved the n of the base in the nominative and accusative singular
neuter. Thus^ as in Latin, nomen, semen^ opposed to homo, sermo, &c., so
HMA imahf Cl^MA syemah^ opposed to RAM^bl kampy ^' stone," from
kamen.

Remark 2. The verb substantive gives Ci»i «y= Sanscrit mriy Lithua-
nian sehsj and in the feminine tSapn stmshchi^a'^Cffisatli (for ^an^e), genti.
After the y in the nominative masculine the nasal and the old a re-
mains; hence BHIA biyahy ^^cadens^ feminine BHl&^ll biyuhshchi. In
Dobrowsky's third conjugation the h/\ extends also to the other forms
withgi; hence BOAIA vofyan, ^^volens/' boaiaijtE volyanshche^ "w-
lentes/' BOAiAipH volyahsh^i, i6e\ov<ra. As regards the use of the
gerund, it is limited to those constructions in which the participle present
stands as predicate, and in German the uninflected form of the partici-
ple is used ; hence (Luc. xxiv. 13.) BikCTA ha&ijiA byesta idunshcha,
" they (two) were going," is the translation of the Greek rja-av nopcvofuvoi^
only with this point of difference, in which the Greek is inferior, that the
Sclavonic has the dual of the verb as well as that of the participle.
Where the participle stands as epithet or substantively, the Sclavonic
uses the definite form of the participle (see §. 234.), and in this the
participle is fully declined; thus, 1. c, KSfirjp aTrexova-av is rendered
BbCb OT'bCTOlAqi&l& vysyotstoyanshchunyuh,

784. The same suffix that forms the present participle

* So, in Lettish, akka-s is both the genitive singular and the no-
minative and accusative plural of akka^ '^ spring of water" (compare
Latin o^tio, Gothic ahva^ '^ stream," genitive singular and nominative,
accusative plural ah-vd-s; Lithuanian up2}e, '' stream;" Sanscrit ap^
"water").

3y

1060 PORMATION OF WORDS.

is added in Sanscrit and Zend to the theme of the auxi-
liary future; just as in Greek and Lithuanian, where
5co-<rco-v, Sui-arov-ra, du-se-ns, du-se-rdin, correspond to the
Sanscrit dd-sya-riy dd-sya-ntam. In the feminine the Li-
thuanian du-se-nti, " the (woman) about to give," an-
swers admirably to the Sanscrit dd-sya-nti ; deg-se-fis,
" the (man) about to burn," accusative deg-se-ntin, answers
to the Sanscrit dhak-shya-Uf dhak'Shya-ntam ; and in the
feminine, deg-se-nti to dhak-shya-nti. The Lithuanian root
bu, "to be," gives bu-se-ns, "futurus,^'' bu-se-nth ^'futura^*''
as analogous to the Zend bu-sya-ns, bu-syai-rdi. Some-
what further oflF lies the Sanscrit bhav^i-$hi/a-n, bhav-i-shya-
nti, on account of the Guna of the radical vowel, the in-
sertion of the vowel of conjunction, and the suppression
of the nominative sign in the masculine. As regards the
€ of Lithuanian future participles like du-se-ns, bu-se-ns,
I see in it, not a corruption of the i of indicative forms
like du'Si-me, ** dabimus '' (see §. 652.), but a corruption of
the a of Sanscrit bases like dA-sya-rd : it is therefore iden-
tical with the of the Greek Sw-cro-vr; and the Lettish
also gives an o for this Lithuanian e, as to the a, also, of
the present participle it opposes an o, while for the i of
the future indicative it has, in like manner, i; e.g,
buhschotsi "/u<wrw5"= Lithuanian busens ; buhschcii, **futura'^^
=^bm€nii; as essois, " being '' = esa»», feminine €ssoti =
esard.f

* See §§. 21. and 104.

t The fdtnre participle in Lettish occurs only in paraphrasing the con-
junctive^ and the present participle also has the feminine form in ti only
in this kind of phrase, hut elsewhere scha^ which, in my opinion, comes
from schia, and this 0'om schi ; so that under the influence of the t, with
a Towel following it, the t is changed into schy as in Lithuanian into c^
(genitive e«a9ioAia«= Lettish essodhas). Refer to what has heen said he-
fore (§. 783.) regarding the origin of the ip ahch in the Sclavonic gerund.
The coincidence of the Lettish feminine termination scha with the Greek

era,

FORMATION OF WORDS. 1051

785. The aorist tenses in San scrit have left us no par-

— -— — -- — ■ ^— \j

tieiples; and the Greek language, by forms like J^o-as", ,y\
hiTTCdVf <l>vy6v, TUTTcov, maintains ajsugeriority py^r the^San-
serit. As, however, the first aorist in Greek contains the
verb substantive (see §* 542.), we may compare <raf,
(Tavra, eravref, &c., with the Sanscrit san, santam, sardas.
The forms which appear in composition maintain a similar
superiority over the simple «v, oi/toj, with respect to the
more true preservation of the ancient form, to that which
the Latin sevis of prtEsens, absem, does over the simple ens.
In respect to the accent, and the pure radical vowel, Greek
participles of the second aorist like \/wc5i/, (jyvycl^v, opposed
to Aenrcov, ^et^Ycov, answer to Sanscrit participles of the
sixth class like tvddn, "the pushing,'^ accusative tuddrdam.
As in the Veda dialect many verbs occur in conjugational
classes other than those which they follow in the common
dialect, I still hesitate to concur with Benfey in consider-
ing participles like vridhdnt, " increasing," dhrishdnt, " dar-
ing," in the weak cases vridhdi, dhrishdt, as aorist partici-
ples, though in no other case have the roots in question
been shewn to belong to the sixth class. If, however,
they are really aorist participles, then dhrishamAna-s (Rig.
V. I. 52. 5. ; probably to be accented dhrishdmdna), also a
middle aorist participle of the sixth formation, though in the
common dialect, having no middle voice, belongs to this for-
mation in the indicative. The root pd, " to drink," whence
pivdmi (Ved. pibdmi from pipdmi), in the Veda dialect follows
also the second class, as is clear from pdihd, " ye drink "
(Ved. thd for tha, Rig. V. I. 86. 1.) ; whence I cannot concur
with Benfey in ascribing the participle pSktam, " bibentem,"
to the aorist, and just as little can I allot to it the imperative

cra^ in forms like iwrovcra, ru^ovcro, is also remarkable. This a-a was
probably preceded by a form a-ia (compare r/)(as=Sanscrit tri, §. 119.), so
that the o- was produced from r by the influence of the ( following.

3y2

1052 FORMATION OF WORDS.

pdht, " 6fAe," which likewise belongs to the present of the
second class. With respect to the accentuation of the par-
ticiple present active, I must draw notice to the fact that
the Greek conjugation in jxi agrees with the corresponding
Sanscrit conjugation in this (the reduplicated verbs ex-
cepted), that it accents the second syllable of the par-
ticiple in question, and that therefore, in this respect,
OTopvv^f OTopvvvTa, stand in the same relation to ipepcav,
^epovra, as, in Sanscrit, strinvdn, strinvdntam, to bhdran,
bhdrantam. The Sanscrit, however, diflFers from the Greek
in allowing, in the weakest cases (see §. 130.), the accent
to sink down to the case syllable ; hence in the geni-
tive singular and accusative plural stri-nva-tds opposed
to <rrop-vv'VTog, arop^vv-vra^. The Sanscrit differs from
the Greek also in this, that in the accentuation of
the participle present (the theory of the weakest cases
excluded) it is governed by that of the corresponding
tense ; thus, bmh-a-n, tfld^d-nj shuchyan^ chor-dya-rif accord-
ing to hddh'd-mi, tud~A-mi, Such-yd-mi, chSr-dyd-mu In
the second conjugation (see §. 493.) the participle present
is governed with respect to its accent by the heavy ter-
minations, especially by that of the third person plural,
and, in irregular verbs, participates also in the abbrevia-
tions, which the root experiences before heavy termina-
tions : hence from vdimi, " I will,**^ comes not vdsantf but
usdnt, "willing," according to the analogy of usmds,
ushthd, usdntu The third class has, as well in the entire
singular (with few exceptions) as in the third person plural
and in the participle present, the accent on the syllable
of reduplication ; hence ddddmu " I give," dddati, " they
give'' (see §. 459.), dddcA, "the giving" (see §. 779. Note),
the latter opposed to the Greek SiSov^, Ttdelg, while ddddmi,
dddhdmU agree with S/Sw/u/, rldrjfju.

Remark. The principle of Sanscrit accentuation appears to me to be
this, that the £eirther the accent is thrown back, the graver and more

FORMATION OF WORDS.

1053

powerful the accent ; and I believe I may assert the same principle in
Greek also ; only that here, out of regard for the harmony and euphony
of the word, the accent in polysyllabic words cannot overstep the limit of
the third syllable, while the Sanscrit places the accent on the first syllable,
without reference to the extent of the word, and contrasts bdrdmahS
with the Greek <l>€p6fi€da. A very striking proof of the dignity and
energy of the accentuation of initial parts of words, and, at the same time,
a very remarkable point of agreement between Sanscrit and Greek ac-
centuation, is afforded by the circumstance, that both languages, in the
declension of monosyllabic words in the strong cases (see §. 129.), which,
with respect to their accentuation, are, as it were, pointed out by the
genius of the language as the most important, lay the accent on the base,
but in the weak cases allow it to fall on the case termination. Here,
however, the accusative plural, though in respect to sound it belongs to
the weak cases, yet passes, as regards accent, in most monosyllabic
words in Sanscrit, as in Greek, for a strong case * ; which cannot surprise
us, as this case in the singular and dual belongs, in each respect, to the
strong cases. Compare the declension of voeft, fem., '^speech," ^Woice,"
with the Greek ott (from Fott for Fo#c, Latin, voc).

N.V. «Vr
Ace. SiTra

SINGULAR.
SANSCRIT. GREEK.

N. V. vik
Ace. vaeham
Instr. vdchd
Dat. vdchi
Gen. Abl. vachds
Loc. va4ski

Gen. on-ds
D.

OW-l

PLURAL.
SANSCRIT.

N. V. vichas
Ace. vdchas
Instr. vaghhU
D.Abl. vdghhyds
Gen. vdcham
Loc. vdksM

OREEK.

N.V. ««■€£
Aoc. (iirag

Gen. oirS>p
Dat. ^(

SANSCRIT.

N. A. V. vdchdu
J. D. A. vagbhyim
Gen. Loc. vdchds.

DUAL.

V^d. vachd

GREEK.

N. A. V. 3jr€
D. G. ojroiv

1 consider as a consequence of the emphasis, which lies in the accentua-
tion of the beginning of a word, the circumstance that active verbs, to

* See the exceptions in Bdhtlingk, ^^ A first attempt as to the Accent in
Sanscrit*' (St. Petersburg, 1845), §. 14.

1064 FORMATION OF WORDS.

which the middle yerbs also belong, in Sanscrit principally accent the
first syllable, so that, therefore, the energy of the action is represented
by the energy of the accentuation ; and I perceive an agreement of the
Greek accentuation with the Sanscrit in this, that Greek verbs throw
back the accent as £Bff as possible. In dissyllabic and trisyllabic forms,
therefore, the two languages usually agree most fully in their accentua-
tion of verbs. Compare et/u with imiy dida>fu with ddddmi^ riBrifu with
dddhdnd^ <t>€pofiev vrith bhdramas^ e^pov with dbharam. In forms of
more than three syllables the Greek approaches tlie Sanscrit as closely as,
without a violation of the fundamental law of its system of accentua-
tion, is possible ; hence the already-mentioned <l>€p6fjL€Ba compared with
bhdrdmahS (from -madhi^ see §. 472.), and also €<f>€p6fi€Ba compared with
dbhardmahi, A quite similar agreement, together with a similar con-
trast, appears between the Greek and Sanscrit accentuation in cases in
which the Greek, in accordance with the Sanscrit principle, throws back
the acceut of the base word in the vocative.* This evidently happens, in
both languages, in order to give emphasis to the name of the person
called, and to bring it prominently forward by the voice ; and in the voca-
tive, in the three numbers of all words, the Sanscrit (where this case is
specially accented) always accents the first syllable, however long the
word be, and wherever the accent may fall in the other eases. To the
nominatives pita, mdtd, dtthitd (ace. pitdram, mdtdram^ duhitdram), cor-
respond the vocatives j9t^ar, mdtar^ duhitar, with which the corresponding
Greek vocatives irmp, firfrcp, Bvyarep — as compared with irarfip, irarepa,
/*4"?P (^or firjrrip), p.rfT€pa^ Bvydrrfp (for Bvyarrjp)^ Ovyarepci, — stand in sur-
prising agreement ; and this is the more remarkable, as the words denoting
affinily in our family of languages belong also, in another respect, to those
expressions which have preserved the ancient stamp with astonishing
fidelity. While, however, the Sanscrit also exhibits vocatives like vish-
vamitra, the Greek, owing to accentual limits prescribed to it, can only
shew such as ^Aydp^pvov^ which, however, does not prevent us from re-
cognising, even in forms of this kind, the agreement of the Greek and
Sanscrit vocative theory ; and just as little, in my opinion, could forms
like (I)€p6p€da compared with bhdrdmahS cause us to overlook the affinity
of Greek and Sanscrit verbal accentuation. The principal part of the
Sanscrit first conjugation (see §. 493.) is formed by the first class, which
comprehends almost one half of the whole number of roots, and to which,

* Compare Benfey in the "ITo/fe Journal of General Literature," May
1845, p, 907.

FORMATION OF WOEDS. 1056

with few exceptions, all the German strong verbs belong (see §. 109^. 1.) :
these in the special tenses throughout accent the first syllable. The sixth
clasS; which is properly only an offshoot of the first, and contains, as it
were, the diseased members of that class (about 140 roots), has, with the
Guna, put off also the accenting of the radical vowel, and accents instead
the class vowel, only that the augment, as well in the imperfect as in the
aorist in all dasses of verbs, has the accent; hence, tuddmi^ ^^tundo,"
tuddsi, ^^tiindis" opposed to bSdhdmi, ^^scio" bodhdsi^ ^^scis," The passive
accents its characteristic ^a, and therefore the second syllable instead of
the first, undoubtedly because in it the energy of self-exertion is lost :
this is evident from the fact, that verbs of the fourth class, though their
middle is literatim the same as the passive, nevertheless accent the first
syllable ; hence, sHchi/atS^ ^^purijicat" opposed to Suchydti, ^^purifi-
catur" It is also of some importance for the support of my view of the
meaning of Sanscrit accentuation, that when the passive is used as re-
flexive, the accent may be thrown back on the radical syllable, though
only in roots terminating in a vowel, or which drop their final consonant.
Desideratives and intensives, excepting the deponent of the latter, as is
natural from the energy inherent in them, hold fast to the general prin-
ciple of throwing back the accent as far as possible ; hence pipdsdmi^
" I wish to drink ;" bSbhidmh " I cleave " (intens.)* As to the fact, how-
ever, that verbs of the tenth class, though they Gunise the radical syllable,
stiU throw the accent on the second (cA^^^amt, ''I steal," not Mra-
ydmt\ we may suppose that these verbs feel themselves to be compounds,
and in a measure determinatives; and as such, in accordance with the pre-
vfdling principle of compounds, accent the last member of the compound,^
but the first syllable of it in order to comply with the fundamental rule
of verbal accentuation. The same syllable, in my opinion, is accented in
denominatives formed by ya for the same reason {putrij/dH). 1 consider
it as another consequence of the composition that the auxiliary fdture
accents not the first syllable of the whole compound, but the auxiliary
verb, whether it begins with the second or the third syllable of the whole
expression ; while the Greek, through all tenses, retains the fundamental
principle of verbal accentuation ; hence, baa-o, daxroiicv^ compared with
ddsyami^ dasyamaSy and forms like tamshydmi (^^ extendam*\ tanishyamas.
So in Sanscrit the auxiliary verb, which is added in the potential (optative)
and precative (aorist of the potential=optative), viz. the syllable yd^
draws the accent upon itself; heacQ^ dadyd^t^ ^^ det** (dtdoti;), precative

* See Aufrecht ^^De Accentu compositarum Sanncritioorum" p. 5.

1056 FORMATION OF WORDS.

dhfai (dotiy), bhuydnuif ^simus,"* On the other himd^ in cases where
the modal element coalesces with the preceding class vowel into a diph-
thong, the accent remains on the same syllable as is accented in the indica-
tive ; thus, bhdrhi bhdrSt^ bhdriina=s<l>€pots, <^epoi, <f>€poififv : on the other
hand, tudSa, tudet^ &c., according to the analogy of tuddsi, tuddti. The
analogy of the sixth class is followed by the potentials of the aorist of the
sixth formation peculiar to the Veda dialect; hence, sak^ma, ^^possimtu,**
In the six classes of verbs belonging to the Sanscrit second conjuga-
tion (see §. 4d3.), as also in the perfect of all verbs, the heavy personal
terminations exercise a similar influence on the attraction of the accent
to that manifested in Greek in all classes of words by the length of
the final syllable, only that the heavy personal terminations in Sanscrit
not only attract the accent, but appropriate it, and, if dissyllabic, to
their first syllable. In this way imi (seT/u), ddddmi (=dtda>fu), jdkd-*
mt, '^abandon," are in the plural imds, dadmds (for daddmdsy middle
dajdmdhi,^; jahimds. In the fifih, seventh, eighth, and ninth class, as also
in the perfect, the Guna syllable, or the heavier class affix or insertion,
exercises an influence in throwing back the accent ; hence, chindmu '' I col*
lect" (plural chinumds); yundjmi, ''I bind" (plural ^un/m^); tan6mi,**\
extend" (plural tanumds) ; yundmiy '^ I bind" (plural yunimds) ; tutdda,
^^ I did thrust " (plural tutudimd), instead of the forms chinomiy yUnctjmi^
&c., which, according to the fundamental principle of verbal accentuation,
would be looked for. The heavy suffix of the participle present (n^, ant)y
the a of which, just like that of the third person plural, is viewed, with
respect to the accentuation, as an essential portion of the termination, or of
the suffix, follows, in the just-mentioned verbal classes, the analogy of the
heavy personal terminations, especially, that of the third person plural ;
but in the weak cases (with the exception of verbs of the third class)
allows the accent to fall down to the case termination ; and the feminine
$, in case the suffix loses its n, follows the analogy of the weakest cases.
The same principle is followed by the participle present of the sixth class.

* Sama Ved. II. 6. 2. 16. 2. Bemark the dropping of the * of the com-
mon dialect (hhdydsma\ as in Zend, see §. 701.

t Beduplicated roots accent only thos^ heavy terminations which begin
with a consonant, and accord to those commencing with a vowel no in>
fluence in casting back the accent. The vowel a, which precedes n in the
third person plural, holds as regards the accentuation as belonging to the
personal termination. Hence ydntiy "they go," compared with iU;
but dddaM, " they give" (see §• 469.) not daddti^ like ddd&ti, "he gives."

FORMATION OF WORDS. 1057

I annex the nominative, accnsative, and genitive singQlar masculine (the
neuter also of the genitive)^ and the feminine nominative in i: dvishdn^
dvishdntamy dvisheUds^ dvishati; dddat, dddatam, dddatas, dddati ; ^fim-
jdrij yunjdntam^ yunfatds, yunfati ; cMnvdn, ehinvdntam, cMnvaids^
chinvait ; tanvdn^ tanvdntam, tanvatds, tanvatij yundn, yundntam, ytt-
natds yunad; tunddn^ tunddntam^ tundatds^ tunddnti, — ^As in Gieek, par-
ticiples present active of the conjugation in /xi, in agreement with the pre-
vailing principle in the corresponding Sanscrit conjugation, accent the
vowel which precedes the v, instead of the first of the hase-word, and
trropyvs^ aropyvvrcL, aTopvvvT€f aropvvvTcs^ stand for comparison with the
Sanscrit strinvdn, strinvdntam, strinvdntd (in the YSda dialect) strinvdntaSy
it might he conjectured that originally the heavy personal terminations,
as they exercise (see §. 480.), as in Sanscrit, a shortening influence on the
preceding syllable, have also, in like manner, attracted to themselves the
accent. Then the Doric forms 8iB6vti^ riBtpri, iaravri^ btixvvvri, might
be regarded as remnants of an older system of accentuation. In the op-
posite case, we must look upon Sanscrit forms like strirMmda^ compared
with the Greek ardpwfiev^ as the consequence of an influence upon the
accentuation exercised by the heavy personal terminations, and first ac-
corded to them by the genius of the language after the separation of Ian-
guages. I have no doubt that forms like strinomi (from 8tarn6mi=
arSpvvfii)^ yund^mi^ through the influence of the weight of the second
syllable, first, after the separation of languages, transferred the accent from
the first to the second syllable. This takes place also in some verbs of
the third class, which we find, therefore, in this respect, as it were, in
the period of transition from the original system of accentuation to that
more recent, in which, in the second principal conjugation, the weight,
of the second syllable has made its influence on the accentuation efiectual.
However, in the Veda dialect, in those roots also which adnut the accen-
tuation of the radical syllable, the accenting of the syllable of reduplica-
tion seems prindpaDy to prevail. Benfey {Glossary to the Sdma-Vida^
p. 139.) cites from hhoTy W, Class 3, the forms bfbharshiy '^fers" Mbhrati,
^^ferenti" bibhratiy ^^ferentes" (as Veda pi. fem. for bibhratyas), opposed
to bibkdrti, '"-fertr^

« We must not infer fi^m bibhdrti, and similar forms, that or is really
the Guna of ri: it is natural, however^ that in parts of grammar where
vowels^ capable of Guna receive it, that those verbs which admit of
weakening should preserve the fiill form of the root, as vas, ''to will,"
becomes contracted to us only in places which do not allow of Guna ;

hence,

1058 FORMATION OF WORDS.

A strong proof of ihe emphasis of the aooentnation of the beginning of
words (in Sanscrit always of the first syllable) is afiPoided in Sanscrit and
Greek by the suffixing of the degrees of comparison, fivhsr tydhs (in the
weak cases tyas), lov, n ishtha^ laro, which, where they are added,
always require the accent to be thrown back as &r as possible. Thus,
in Sanscrit, fipom svddu, ** sweet '=i7dv, comes the comparative svddiydhsy
nominative masc. wadiydn^ and the superlative svadishtha-s. To the latter
corresponds the Greek ^biaro-Sy and to the nominative and accusative
neuter of the comparative svadiyas the Greek fjbiov ; while ffSiavy ^diovosy
for well-known reasons, do not exhibit an agreement of accentuation with
svddtiydfiy 8V&d%ya8CL8. The Greek degrees of comparison in repo^ raro,
follow essentially the same principle, t. e. they throw the accent as fer
back as possible, by which, however, only the syllable preceding the
suffix is reached, so that the accent is often necessarily transferred from
the beginning to the middle of a word, as in fic^iArtposy pc^&rarosy
compared with ^epcuos. In Sanscrit, on the other hand, the d^;ree suf-
fixes, corresponding to the Qreek repo^ raroy exercise no influence at all
on the accent; and the positive base retains the accent on the base
in whatever part of the word soever the same may occur ; thus the

hence, usm&Sy " we will/* opposed to vdhniy " I wiU " (Comp. Vocalismus,
p. 168). When Benfey, who, in the ''^HaMe Journal of General Literature "
(May 1845, p. 944) contrasts the Greek Spwfxi with the Sanscrit rinSmi,
remarks, that in Greek ri is Gumsed, because it is accented, and that u is for
the same reason Gunised in Sanscrit, I cannot assent to him in either point.
In the first place, I recognise in forms like SpwfUy <TT6pwfu (the latter=s
8trin6mi)y no Guna, but only the discontinuance of the abbreviation of ar
to n, which was admitted in Sanscrit, just as in rpiros compared with the
Sanscrit trUiyM (Latin tertius, transposed from tretiusy for triiius), the
abbreviation of the syllable ri has ceased. In the second place, I cannot
admit that forms like rindmiy atrirpmiy have, for this reason, Gunised the
second syllable because it is accented ; for if the accent occasioned the
Guna, we should also expect for bibharshi and vivakti (in the Veda dialect),
bebharshiy vevaktiy and for desideratives like pipdsdmiy pipdsdmi. To
me, therefore, the principle set forth above, viz. that the accenting of the
first fifyllable belongs to the verb, but that heavy syllables have often de-
stroyed the original accentuation, and appropriated the accent to them-
selves, appears far more natural. The Greek replaces the Guna of rinSmi,
9tfin6mi, by the lengthening of the vowel {aT6ppvfu opposed to <rr6pvvyuev\
but nevertheless preserves the original accentuation.

9>

FORMATION OF WORDS. 1059

comparatire fmd snperlatiye of mahdt (in the strong cases mahdnt) are in
the nominative masculine mahdttarasy mahdttamcLs ; and the superlative
of frridum^ *' liberal," " giving freely '* (in the Veda dialect), vrishantama-Sy
genitive vfishantamasya (Rig> V. 1. 10. 10.). The reason that tara and
tama, in Sanscrit, exercise no influence on the accentuation lies, in my
opinion, in this, that these sufi&xes are rather enclitic in their nature,
and have not grown up so inwardly united with the principal word,
as the other more rare sufiB^es of comparison; as appears, also, from
the circumstance that the feminine accusative tardm^ tamdm^ may
be added to verbs adverbially also; e,g. vddatitamam^ '^he speaks
very much."

A consequence of the emphasis which lies in accenting the beginning
of a word is this, that abstract substantives, which frequently are merely
intensifications of adjectives, a£Pect, in Sanscrit and in Greek, this kind of
accent Thus the suffix a«, in Sanscrit, is used especially in forming ab-
stracts, and requires an accent on the first syllable of the word ; as in ydkuni,
"glory," compared with yasds, "glorious" (the latter only in the Veda
dialect, see Benfey's Glossary), whence the comparative yakdstara-s^
superlative yasdstama-s ; thus, dpas^ nominative "activity," "work,
^^oflPering" (Latin <>pu8\ compared with apds masculine ^^the active,'
" the warrior,* " the sacrificer." As to Sanscrit neutral bases in as cor-
respond the Greek in os, ey, e((r)-off (see §. 128.), Benfey draws our notice,
as regards the paroxytone accent of the abstracts spoken of, and the oxy-
tone accent of the adjectives, to the relation of the Greek dyos to dyrjt.
It may also be observed, that Greek bases in os, er, when they form pos-
sessive compounds in combination with preceding words, usually throw
the accent on the suffix, while other compounds of this kind accent
the first member of the compound, or, at least, throw back the accent
as far as possible ; thus evpvo-^ei^r, /ieyoXocrdci^r, fieyaBapoTiSy dvo-fcXe^r,
tvKkefjf, compared with forms like fxcyddvfws, ficydbapoSf fieycikddapos, fi€'
yoXddofoff, aloXofiopfjyoSf aloKdireirKost alokoxairris.

786. The suffix of the participle of the reduplicated pre-
terite or perfect (see §. 588.) is, in Sanscrit, in the para-
smaipadam or active (see §. 426.), according to the diffe-
rence of case, vdns, vat, and ushy and in all these forms,
according to the analogy of the heavy terminations of the
indicative (see p. 1057), has the accent. Indian Gram-
marians, however, consider yds as the true form of the
suffix, tTiough it does not appear in this form in a single

1060 FOKMATION OF WORDS.

case, but the strong eases spring from vdns , the middle
from vdt, and the weakest from ush (euphonic for us).
From Ush comes also the feminine theme lushi, to which
the Lithuanian usi is an admirable counterpart ; hence
degusi, "the having burned "= Sanscrit dihashi, for dada-
hushi (see §. 605.). The oblique cases of the Lithuanian
feminine participle spring, for the most part, from an ex-
tended base usia ; hence the genitive singular defftisid^,
as rankd-s, from rankd, " hand/'* Compare herewith the
Greek via of rerv^uia, which has been already elsewhere
compared with the Sanscrit tutupiishi.'\

787. With the weakest form of the Sanscrit participial
suffix above mentioned are connected also, in Lithuanian,
the oblique cases of the masculine, but with the same un-
organic affix of ia, which, too, the participle present has
retained; thus, genitive degusio (as vnUco from mlka-s)
corresponding to the Sanscrit dShiish-as, dative deg-iisia-mt,
accusative deg-usi-n for deg-usia-n. The nominative degens is
based on the Sanscrit strong theme dth-i-vAns {i as conjunc-
tive vowel) ; but the s of the Lithuanian form scarcely be-
longs to the base, but is the sign of case, and extends, as in

* The Yocatiye siiignlar, which in general <1ific1aiTnfl long yowels (see
§. 205.), shortens the long d; hence, win compared with the nominatiye
vduy since annsv^ra (n) after the s is dropped (see §. 9.) becomes n. I am
not inclined with Bohtlingk (Decl. p. 10) to represent vahs as the original
form of the suffix ; for i£^ as we ought to be, we are guided by the strong
cases, which in gener^^ where different modifications of the theme occur,
have preserved the origmal form, we must then take vahs to be the ancient
form, and allow that the yocative, as is its wont, has shortened the vowel,
which perhaps is only a consequence of the emphasizing the beginning of
the word in the vocative by accenting it. Bohtlingk also, in his zeal for
the vocative, represents tyans as the theme of the comparative suffix iyaiMy
tyaa (see §. 298.), the long d of which, in Latin, takes the form of 6 in all
the oblique cases.

t ^'OntJie Influence of Pronouns in the Formation of Words," p. 4.

X According to the analogy of the adjective declension, see § 281.

FORMATION OF WORDS. 1061

the participle present, to the vocative also ; while the San-
scrit, as it cannot bear two consonants at the end of a word
(see §. 94.), in both cases abandons both the nominative
sign and the final consonant of the base ; thus, nominative
d&h-i-v&ni vocative (Uh-i-van, corresponding to the Lithua-
nian deg-ens. The Zend, on the contrary, has retained

^ In the Old Prussian Catechism there occur two perfect participles
in wuns very deserving of notice, viz. klanMwuns^ " having cursed," and
murrawunsy " having murmured," which stand nearer to the Sanscrit
vans than any other European kindred form. The u of fvuns, as also that
of the common form uns (after consonants also otm, and sometimes ans)^ is
evidently, like the e of the Lithuanian ens, a weakening of a, originally
^y as in widdevm^ '' widow "= Sanscrit vidhavd, Latin vidtia, and some
similar feminine nominatives. The u of the plural -usisy accusative usins,
and of the accusative singular iisin, is, on the other hand, organic, and
identical with the Sanscrit u of the hase of the weakest case and of the
feminine, as also with that of the corresponding forms in Lithuanian.
Nesselmann {''^The Language of the Old Prussians^* p. 64) represents the
participles in uns (pns, ans, wuns) as indeclinahle, and takes usis as an
independent form with declinable terminations. I, however, consider
vounSy uns, ons, ans, as the singular nominative masculine, with s as the sign
of case, as in Lithuanian ens. This participle seldom requires declension,
as it is principally used for a periphrasis of the perfect indicative, and thus
occurs in the nominative relation ; e.g.^ asmdi murrawuns bhe kkmt^-
wuns^ ^^I have murmured and cursed" (literally, '' 1 am the person
having murmured and cursed"). The nominative singular usually takes
the place of the plural, as also in Lithuanian the present and perfect par-
ticiples have lost the termination of the plural nominative, and in this
case only have rejected the s of the nomiilative singular : hence, from
sukensj " having turned," comes the plural suken. Where, however, in
Old Prussian, the plural relation of the participle perfect is really ex-
pressed, it ends in usis^ probably from a lengthened base in usi (compare
§. 780.) ; so that i-s of the Lithuanian plural termination corresponds to
the y-s of bases in t (airy-«, " sheep," from the base avn). The examples
occurring in the Old Prussian Catechism may be found in Nesselmann,
p. 31, n. 84. : madUti, tyt imrstai iaus immusis; laukyti, tyt ivirstai ious
aupaJIusiSy ''ask, and ye shall receive (be having received) ; seek, and ye
shall find (be having found)." The future, which is wanting in Old Prua-

sian.

1062 FORMATION OF WORDS.

the nominative sign in its participles ; as, {au»(2,^ chidh'
vdo, "having made," vid-^^, "knowing" (elS^i), which it
has also done in the participle present, a point in which
it is superior to the Sanscrit, and agrees with the Lithua-
nian, Latin, and Gothic ; for from i^H vdn is formed in
Zend, not fui(? vdo, but y^9 vanri. It is clear, however,
that the o of vdo does not represent the s of the theme of
the strong cases, as the suffix vard also, in the nominative,
forms vdo (compare Bumouf Ya9na, Note R. p. 128). In
the accusative, 9g»»9£AM»^^^ dadhvdonhem corresponds to
the Sanscrit dadh-i-vdnsam ; in the weakest cases, and be-
fore the feminine character i, the Zend suffix is contracted,
like the Sanscrit suffix, to ush* ; hence, in the genitive
^jy;(3A5j dathushd (Vend. S. p. 3. for dadhmhS, see p. 965.

sian, is always periphrastically expressed by the auxiliary verb signifying
" to be," with the participle perfect ; hence, p. 12, n. 15., pergubons wyrst^
^' he is come" (is the person having come). The oblique cases of the per-
feet participle, from being little required, seldom occur, and spring like-
wise from the theme increased by t, while the Lithuanian adds ia to the
base. The only instances that occur are, au-law-uai-ns^ "the slain" {mor-
tuos, for which, also, aulausins and aulauvmssens), and ainan-gimm'tisi-n^
" to those bom in" (the place), the latter with passive signification, which,
except in the root gem^ gim^ does not occur in this participle. If we
should not admit a nominative plural in tms^ the above-mentioned forms
might then be taken as singular nominatives, with a plural signification.
The circumstance, however, that the real and frequently-occurring singular
nominative always terminates in ns, and that, too, the participle present
leaves the old base (in nt) in the nominative singular unlengthened, and
in the other cases lengthened only by t, is much opposed to this view. —
The single feminine form of this participle which occurs cteserves mention ;
viz. the nominative singular aidatisi^ ^'mortua" for cadauusi^ as above
aulaU'Sins together with aulautisins. The final S corresponds, therefore,
to the Sanscrit t and Lithuanian t of feminine forms in usM, im.

* The lithographed Codex of the Yendidad Sad^ has, almost in aU
places, jj^ s for t^sh: I, however, agree with Bumouf in reading )mj
sh as probably the sole correct reading.

FORMATION OF WORDS. 1063

Note*) ; in the dative (\jt^25^ tidusM, " to the knowing "'
(1. c. p. 214.)=ftr5^ vidfdsM (eiSdri) ; in the genitive plural
^'^tpy^'^'P^ irtrithushanm, "of the dead'" (1. c. p. lOl); in
the genitive singular feminine ^^^tp^99f^^jaghmushydo
(1. e. p. 91. twice, and 304. twice) * = Sanscrit jagmtt^hyds,
from gam, " to go ''^ ; in the accusative feminine ^4tp)^j^
t?i'^AMsfeif/n= Sanscrit vidusMm,iroiavidi "to know" (1. c. p. 469).
788. With the contracted form ^anc ush of the suffix here
spoken of is connected a word which appears in Gothic
as a solitary remnant of an obsolete participial gender,
and corresponds in a remarkable manner with Sanscrit
forms like dik&sh (theme of the weakest cases) from dah ;
I mean, Mrusyds, ** the parents,"* occuring only in the
nominative plural masculine, and which, I have no doubt,
properly signifies " the having given birth to ;" and, with
respect to its radical vowel, corresponds to the polysyllabic
forms of the preterite of baira (bar, plural birum, conjuga-
tional singular Mr-yan, plural bir-ei-ma (see §. 605.). The
theme is b&rusya, which corresponds in its unorganic affix
ya to the above-mentioned (§. 787.) Lithuanian ia ; e. g.
of deg-usia, dative deg^tisia-m. The nominative singular,
according to §. 135. would be bh'-useis, and the accusative
b&ru8u the latter like the Lithuanian degusi-n.

* With regard to the long u of jaghmushyao^ let it be noticed that the
sibilant is here followed by a semi-vowel, since, as it appears, a lengthen-
ing of the tt, which is, in Sanscrit, always short, occurs especially before
two consonants; hence, also, Vendidad Sad^, p. 615, ^9^C^*^^^9;^^
jaghmuh&m6 (with J9 i for jio »), a superlat^^e formed from the weakest
theme ; and p. 525, dadiUdUdSy an interesting form ; whence it is dear
that in Zend also the middle cases (see §. 130.) of this participle spring
from the weakest theme. There occurs, however, a long u mpipyuthim^
without the occasion of two following consonants, as also in its negative
apipyusMm (Vend. S. p. 429), from pi^ " to drink," virith a causal mean-
ing ('^ the having sucked "). Perhaps the circumstance that two conso-
nants precede has its influence.

1064 FORMATION OF WORDS.

789. To the form vdt, whence come in Sanscrit the
middle cases of the perfect participle*, belongs the Greek
OT, which has preserved the ancient accent (see §. 786.) ;
but after losing the digamma, which is generally lost in
the middle of words, in case it does not assimilate with
a precedmg consonant (see riaaape^, §. 312.), as, for instance,
also in the suJQ^ 6i/t = Sanscrit fxzrd (of the strong cases):
thus, the same relation that a/L(ir6\d(f)6ia' has to San-
scrit forms like dhdna-^vant (" endowed with riches," see §.
20.), t€tv(I>'{F)6t has to tutupvdt, to which, as nominative,
accusative, and vocative, in Greek, rervipos corresponds (see
§. 152.). To the plural locative tidup-vdt-su corresponds
the Greek dative Terv<fe'6(Ty<rt, Mention has already been
made of the feminine form in via, as abbreviation of va-ia,
and of the affinity, as regards formation, of TeTVfpvia with
the Sanscrit tutujyiishi (see §. 786.). The Latin, perhaps,
in secAri-s presents a remnant of these feminine participles
in nshi (euphonic for ust), and the proper translation^
therefore, is, perhaps, " the cutting '^ (instead of ** the hav-
ing cut ''), the u being lengthened, and the sibilant being
changed between two vowels into r.f As several parti-
cipial suffixes are often used also in the formation of de-
rivative words, there is, therefore, ground for comparing
the suffix dsu in words like lapid-dsus, Ivmin-dsu^, fructur-
dsus, forrn-dsus, pisc-dsus, with the Sanscrit vdns of the
strong cases, to which it has nearly the same relation that
the comparative suffix idr has to $]^T^ iydns (see §. 298.),

* See §. 130., where it must be also noticed that the nominative, acca-
sative, and yocatiye singular of neuters in the threefold theme gradation
always are connected with the middle form.

t See §. 22. In the Veda dialect there are abstract substantiyes in
uM^ with the accent on the radical syllable (see p. 1059) ; as, tdpushiy
" ire " (properly, " the burning "), from tap^ " to bum ;" tdruski^ " strife,"
from tar {tri irt, " to overstep."

FORMATION OF WORDS. 1065

only that the original sibilant is retained, though the t) is
lost, just as in 8opio=8vapimi ; soro, sor6rem=svasdr9
svasdram; sol=svar, "heaven'' (from sur, and this from
svar, *' to shine ''), Zend hvare, " the sun." With respect
to the prolongation of the suffix by a vowel affix, compare
the relation of the suffix turn to t&r, Sanscrit tdr (see
§. 647.).

790. In Old Sclavonic the gerundive preterite corresponds
to the participle here spoken of, as is most clearly apparent
in the feminine singular form, in which, in verbal bases
ending in a vowel, siimii vshi corresponds to the Sanscrit-
Zend tishi, and Lithuanian usL Compare BT>iB'i»mii by-vshi,
** having been '' (feminine) with the Sanscrit ^snrwt babhuv-
nshi, and Lithuanian buw-usi. In the nominative plural
masculine (used also for the feminine), BT>mE ishe — ^with e
as the termination of case = Sanscrit as, Greek ej — answers
to the Sanscrit vdnsas, and therefore BitiBme by^she to
babhu-vdtisaa ; on the other hand, in the singular the sibi-
lant is lost in the nominative masculine ; thus, B'bib'b by-v
corresponding to the Sanscrit babhtl-van and Lithuanian
buw-ens (see §. 787.), where it must be observed that gene-
rally the Sclavonic has lost the original final consonant, so
that the s also of the Lithuanian buw-ens belongs not to
the suffix, but to the case sign. After consonants the v
of the gerundive suffix is suppressed ; hence, e.g., hecb nek,
"having carried" (for nes-v), plural HECBmE nesshe (for
HECBi>mE ne»vshe), feminine singular HECi>mH nesshi (for
nes-vshi).

Remark. In the Sclavonic that tense of the indicative is wanting
whence the past participle or gerundive has proceeded: on the other
hand, I am now of opinion that the Lithuanian perfect (also aorist), which
I formerly compared with the Sanscrit first augmented preterite (Greek
imperfect), must be compared with the Sanscrit reduplicated preterite,
Greek perfect and Gothic preterite of the strong conjugation. I assume,
therefore, that in buwau^ '*' I was," or ^^ I have been," instead of the

3z

1066 FORMATION OF WORDS.

augment^ the syllable of reduplication is dropped, as in Gothic preterites
like baug, " I bent," bugum^ " we bent "=Sanscrit bubhoja^ bubhufima;
and I compare it with the Sanscrit babhiivay to which, with regard to its
medial u, it corresponds better than to the imperfect dbhavam. Buwau
does indeed closely resemble also the Sanscrit aorist dbhvvam^ but in the
third person buw^ answers better to babkuv-a than to dbhut ; and in both
the plural numbers the forms given abore (p. 762) answer better to
bcMuv-i'Vd (from -a-va) babhuv-a-thUs (from -thas\ babhuv-i-md (from
-arma\ babhuv-a'(ta\ than to abhutam^ dbhu-ma^ dbhu-ta. The conjec-
ture that the Lithuanian perfect belongs to the universal tenses, and not
to the imperfect, is also confirmed by the consideration that the imper-
fect in Sanscrit and Greek always takes part in the base of the present.
I.e. in the class peculiarities, while the Lithuanian preterite, which is
called perfect, does not; hence the perfect oigdu-nuy "I am acquainted
with," which corresponds to Greek verbs like dax-vo), Latin like ster-no
(see p. 718), is not gau-nau, but gavHiu (future gdu-su). In the perfect,
too, t or st of the present base is dropped, which formerly, when we
sought to compare this tense with the Sanscrit-Greek imperfect, ap-
peared a difficulty (see §. 498.). As to the circumstance that the y or t
(see p. 722) compared with the Sanscrit fourth class is retained in the
perfect, and that from liepyu^ " I order," comes the perfect Uepyau
(future liepiu) ; from traukiu^ '' I draw," the perfect traukiau (future
trauk'8u\ this may be explained from the near resemblance in form of
the fourth class to the tenth, in which the retention of the t/ or t in the
universal tenses is regular. In general the perfect loves a y, and often
adds one in verbs which do not exhibit one either in the present or in
any other tense ; as from dumi (for d^idnd\ or dudu^ ^^ I give," comes
daw-yau (future du-w) ; from demi (for dedmi\ ^' I lay," de-yau (future
de-m=idhdrtydmi^ ^-o-w) * ; from ctmi, or einuy " I go," eyau (future
et-«ii =3 Sanscrit ^-shydnu). In every case the form of the participle may
be safely inferred from that of the perfect indicative ; but when the y of
the first person singular indicative disappears in the other persons, it is
lost in the participle also ; thus, from daw-yau^ second person dauhei,
participle dato-enSy feminine dato-usi; but from deyaii, second person d^ei,
participle d^-enSy feminine dey-usi; from eyau^ ^^ivi" second person ^0»,
participle ey-ens, feminine ^-uH. It is beyond doubt, therefore, that as

* If the Lithuanian perfect belonged to the Sanscrit-Greek imperfect,
then the perfect of dudu and dedu would most probably be dudau^ dedau
ss Sanscrit ddaddm, adadhdm^ Greek ihibia^v, iriBjfv,

FORMATION OF WORDS. 1067

the participle is based on the Sanscrit in vdns, feminine ush% so the pre-
terite indicative, which is most intimately connected with it, must also
be connected with the Sanscrit reduplicated preterite and its European
kindred forms. The Old Prussian simple preterite also, which in signi-
fication usually appears as aorist, appears to me to be a sister form of the
Sanscrit reduplicated preterite, with the loss of the reduplication : hence,
dai, *'^he gave," for (;2a^=: Sanscrit daddu^ for dadd. The present dasU
from dad-ty is, on the other hand, like the Lithuanian d&sti, a redupli-
cated form (see p. 661 ). The ts which often terminates the third person
singular preterite; as in daitSy ''he gave," a form used together with
dai; immats^ "he took," with imma; hiUais, "he spoke,* with biUa:
this ts X regard as an appended pronoun, and abbreviated for tas (compare
Lithuanian tag, "that," and the Sanscrit base to, "he," "this,* "that").
Let it be observed, that in general bases in a for the most part suppress
this vowel before the nominative sign s; hence, deiwa, " God "= Lithua-
nian dieu>a-8y Sanscrit dSva-s (see Nesselmann, p. 49). That^ the ts
spoken of is not characteristic of the preterite is clear from this, that it
also occurs sometimes in the present; for example, in astitSy '^he is,"t
QXid po-quoitStSy "he desires." The former occurs twice, and once in the
sense of the conjunctive : Nesselmann, p. 23, n. 51, nostan kai tans spcarts
astits, *' on which he may have power." Here, therefore, the idea " he"
is three times expressed, once by tatiSy then by the ancient personal ter-
mination tiy of the meaning of which the language is no longer conscious,
and lastly by the appended ts. This ts, however, can scarcely be
admitted in reference to feminines : there are no neuter substantives in
Old Prussian ; and in one place, where astits appears to mean " he is," it
refers to the masculine undsy "water" (Nesselmann, p. 17): adder sen
stesmu wirdan Deitoas astits aind Crwtisnd, " but with the word of God
is a baptism." Here, therefore, the appended pronoun, as the subject of
the proposition, is correctly in its place.

791. The middle and passive participles in Sanscrit, in

* At frequently stands in Old Prussian for a ; as in the nominative
singular feminine, where both a and ai correspond to the Sanscrit d, see
Nesselmann, p. 48 ; and compare quaiy " which V* with the Sanscrit kd^
Lithuanian A», and Latin quae; so stai (also «/^), "this," "the "= Li-
thuaman to.

t Compare Sanscrit asti, Lithuanian esti, the % of which in Old Prus-
rian is contained only in this compound (simply ast)

3 z 2

l()68 FORMATION OF WORDS.

SO far as they attach themselves to any tense of the indi-
cative, have the suffix mdna or dna. I consider the latter
to be an abbreviation of the former, as it is represented in
Greek, just like mdnaf by /xevo : nor is it probable that
the Sanscrit should have originally appropriated to the
participle present of the middle voice two suffixes which
resemble one another so closely as mdna and dna; and
which, in use, are so distributed, that the former belongs
exclusively to the first principal conjugation — only with
the exception, that the tenth class, probably on account of
its greater fulness of form, admits also dna — while the
latter is fixed in the second conjugation ; and, moreover, in
the perfect, to which, as it appears to me, on account of
its incumbrance with the syllable of reduplication, the
shorter form is more agreeable, where we must remark,
that in the present participle active also the reduplica-
tion has an influence on the weakening of the participial
suffix (see §. 779. Note). The auxiliary future has every-
where preserved the complete suffix mdna; hence, dd-syd-
md-na-s, both middle and passive = 5a)-(rd-/ievof . With this
agrees the Lithuanian du-se-ma-s (feminine -ma), " qui da-
biturr since in Lithuanian the said participial suffix has
been abbreviated to ma, which nevertheless does not cause
us to overlook its connection with the Sanscrit mdna and
Greek fxevo. In the participle present dud-a-ma-s, " qui
1 daiur^'* corresponds to the Greek BtSo-fMcvog, and Sanscrit

dddh-^na-s (for dadh-md-nas, and this for dadd-mdnons) :
the latter, however, is middle only, and the passive parti-
ciple is ?fhWR^ di-yd-mdna-s. The Old Prussian, which
approaches the Lithuanian very closely, has, in one of the
two examples of the said participle which remain to us in
the translation of Luther's Catechism, preserved the origi-

* Several roots in d (among them da) weaken this vowel before the
passive character ya to f.

FORMATION OF WORDS. 1069

nal form of the suffix with astonishing fidelity, it may be
said, in its perfect Sanscrit form, unless, perhaps, the a of
the first syllable be short. The example I mean is, po-klaus-
i-mana-Sf ** heard/' or rather " being heard,'' aKovofxevo^ : *
in form, however, inroKKvoixevog would be the correspond-
ing word, as klatis or klus is the Prussian form of the
Greek root k\v (Sanscrit irUf from (kru), and po corre-
sponds to the Greek vtto, Sanscrit {ipa. Besides poklaitsi--
maruis, the Prussian Catechism presents one more form,
which, with respect to its suffix, evidently belongs, in like
manner, to the participle passive present ; viz. em-m-vHnne,
" agreeable," properly " becoming accepted," as the parti-
ciple perfect passive also signifies both "accepted" and
acceptable." f

«

. * The participle present passive suits the passage where the expression
occurs better than the perfect participle (Nesselmann, p. 16), staundas
madias a^t steismu tdwan en dangon enimmewingi bhe poklaustmanas,
^' such prayer is acceptable to and becoming heard (=is heard) by the
Lord in heaven."

t Nesselmann (p. 104) takes enimtmine to be a typographical error,
though he gives no reason for this opinion. The termination mne does
not appear to. me doubtful : the internal vowel is omitted, as in the Latin
al-u-mnu8f Vert-u-mnus (§. 478.), and as in the Zend forms har-OrmnSm^
vaz-a-mngm^ of which hereafter. So in Old Prussian, from kermen-e^
^^ body/' comes the accusative kermnem (also kermenen and kermenan).
This kermens for kermenas is, according to its formation, probably, in
like manner, a passive participle ; so that, properly, its meaning is equi-
valent to '^created," '^made" (Sanscrit karomi^ ^^1 make," compare Latin
creo, crecUura). Pott refers the Latin corpus, and Zend Ueref-s (accusa-
tive k^hrpSm^ to the root kbripy kalp ; which, however, is itself connected
with har {)cn), as Pott also assumes (see my Sanscrit Glossary^ a, 1847,
p. 84). As regards the final e of enimumne, it is either an adverbial or a
neuter termination. The passage wherein the expression occurs requires
properly the nominative singular neuter (Nesselmann, p. 24, u. 56^ sta ast
lahban bhe dygi emmumne prtki Deiwan nausesmu pogdlbenikojty *'*' this is
good and acceptable before God our Saviour"), as hbhan also is really a

neuter,

1070 FOBMATION OF WORDS.

792. With respect to accent in Sanscrit, the participles,
middle and passive, in mdna, dna, follow the same principle
as the active participles (see p. 1057), i.e. they are governed
by the accent of the corresponding tense in the indicative ;
so that the suffix receives the accent only in cases in
which the indicative has it on the personal termination,
which happens in the heavy terminations of the present
of the second principal conjugation (with the exception of
the third class, see p. 1056) and of the perfect of all verbs.
The Greek corresponds, in forms like rervfi-jjievo^ (opposed
to TxmTofjievog), to the accentuation of the Sanscrit cognate
forms, only that the latter have the accent on the final
syllable of the suffix, so that ^u^t^dnd^ corresponds to the
Greek Tcrv/i-ixevo^.

neuter, according to the analogy of Sanscrit nenters in am (see §. 152.).
If, however, emmumne is a neuter, in that case the e stands, as frequently
happens in Old Prussian, for a, and the case-sign is suppressed, as in the
pronominal neuters, «to, ^' this," Aa, " what " (accusative ka and Aran),
and in Lithuanian neuters, as gSra, ^^bonum" (§. 135.). If, however,
there is a typographical error in this word, which is an isolated one of
its kind, we might perhaps conjecture enimumnem=mnan. As regards
the vowel ti, it is prohahly like the Latin u of ahu-mnus, Vert-u-mnus —
for which we might have expected al'i-m{{)nti8y Fert-i-m(J)nu8 — the cor-
ruption of an original a, and corresponds to the Sanscrit a of the first and
sixth class (§.109^.1.).

*. At the time when the Sanscrit suffix ana had not yet lost its fn, it
will probably have had, like the Greek -fuvos of rervfi-fievos, the accent
on the first syllable; for that the circumstance of the suffix beginning
with a consonant or a vowel may have an influence on the accentuation is
clear from this, that the verbs of the third class in the present indicative
have the accent only on those heavy terminations which begin with
a consonant, while in cases where the heavy termination begins with
a vowel, the syllable of repetition is accented (see p. 1088): hence,
Ubhri'Vahi^ "we two cany" (Mid.), but second person Hbhr-dthi^ third
person bibhrdtSf so also in the participle present middle Mbr-dfuiy not
bibhr-dnd : it is highly probable, however, that bibhr%-mdn& would be said
if the m of the suffix were retained.

FOBMATION OF WOEDS. 1071

793. In Old Sclavonic the participial suffix in question
has experienced the same abbreviation as in Lithuanian :
it is in the nominative masculine mi> m, feminine ma ma,
neuter mo mo, and, as in Lithuanian, has only a passive*
signification, but occurs only in the present. Compare
BE^oM'b ve^-o-m', "the being conveyed,"*' feminine be^oma
vel^-o-ma, neuter bedomo ve^-o-mo, with the Lithuanian
wez-a-mors, feminine -m^ the Sanscrit vdh-a-mdna-s, -d,
-a^m, the Greek e^-o-ZLtevo-s", -rj, -o-v, and the Latin veh-i-mini
(see §. 478.). In the German languages this participle,
as such, has disappeared, but the Gothic lauh-m>6ni, ''the
lightning," properly, ** that which lights,"*" from the femi-
nine base Iauh-m6ny6,\ is a substantive remnant of the
participle present middle, and, therefore, the y is an unor-
ganic affix, otherwise m4n6 would correspond admirably to
the Sanscrit feminine suffix m^dnd, as 6 is the most common
representative of the d, which is wanting in Gothic (see
§. 69.). The nominative form -mdm, of mdnyd, is to be
explained according to §. 120.+

794. The Zend has either shortened or rejected the
middle a of the Sanscrit suffix mdna, and weakened the
preceding class vowel a usually to f e. The form m^na
(mna) becomes, as it were, the step of transition to the
Greek /xevo, and Latin minu §. 478), and is identical with

* It needs, perhaps, no remark, that the vowel which precedes the n
in all the langnages here compared belongs to the class syllable, and is
therefore not to be referred to the participial snifix (see §. 507.).

t Sanscrit roch-a-mdnoi, " the shining," firom the root nmh (from ruk),
which is only used in the middle, according to the first class (see §. 109^.).
The Latin luceo is based on the cansal form r6€h6yam% (see p. 110).

X It may also be assumed that the Gothic m&ny6^ moni, is based on a
to-be-presupposed Sanscrit form mdM, as bases in a, especially in sub-
stantives, form their feminines frequently in ^ ; as, ditA^ ^^ a goddess,"
from dSva^ ^^ a god." This i most, in Gothic, according to §. 120., take
the form oiy6 or etn, nominative £, ei.

1072 FORMATION OF WORDS.

the Old Prussian mana, of the (§. 791.) above-mentioned po-
khLus-i-marui'S ; while the form mna, which has lost its in-
ternal vowel, finds an accidental countertype in the Latin
mntt, of al'U^mnus, J^eri-u-mnus, and the Old Prussian mne^
of en-im-u-mne (§. 791.). In Zend, also, this suffix, as in
Greek, has, beginning even with the present, both a middle
(or purely active) and passive signification, while the San-
scrit in the passive prefixes the character ya to the parti-
cipial suffix. Thus we find in the Vendidad Sade, p. 203,
baremanem, ** being carried"' (=s<l>€p6fi€vov), and vazemnenh
" being conveyed," as adverbial accusatives in reference to
the nominative plural mazdayasna. At times the final
vowel, also, of the suffix mana is suppressed, together with
the middle vowel ; so that thus only mn is left, to which
are affixed the case terminations. Thus, in nydiemn-dy
" celebranteSf^'' yizimnd, " venerantes,^^ which indeed, according
to their termination, might also be singular nominatives of
bases in a, but in the passage where they occur clearly
shew themselves to be plurals of bases in n.f We might,

y^j)A5^/« ^jAUJJAty yat aiti yoi maxdayasna pddha ayantSm vd
tach^tem vd har^manSm vd vazSmnSm vd iachi aipya naidumfrajakmn^

" If those, who being worshippers of Ormnzd, going on foot, or or

carried, or riding approach a corpse." Anquetil (p. 312) translates :

^^ Si un MazdHesnan allant d pied^ ou en bateau^ pwte (dans une voiture),
ou eleve de quelque fagon que ce soit (aperfoif) un mart. In a similar pas-
sage (1. c. p. 279) occurs har&mn^m^ and likewise vajehnnem,

t Vendidad Sade, p. 482 : Naro ahhen ctshavano havSyazasta nydianno
yi%imn6 Ahuramaxdahm ; '^ Viri sint puri, Icevam manum hahentes (ItBvd
manu tenentes)^ celebrantes^ venerantes Ahuramazdam.** Anquetil trans-
lates (p. 41 G) : Qu*%l n'y ait que Vhomme pure qui coupe k Barsom ; et que,
le tenant de la main gauche, il/as&e isseschn^ d Ormuad. 1 consider mydi-
emno as an abbreviation of niydsy and refer, on this hand, to the root yds,
p. 963, Note.

FORMATION OF WORDS. 1073

therefore, also distribute the forms barhnvihn and vazhnnem
into haremrtrem and 'oazemn-em, as bases which end in a
consonant have, in the accusative, em as their termination.
That, however, in general in Zend the suffix spoken of has
not lost its plural a, is shewn by forms like vazemna
(Vend. S. p. 52l), which, as nominative plural, can belong
only to a base in a (§. 231. Note) ; thus, csayamana (1. c.
p 543.) = Sanscrit kshdyamdnds, from kshi, "to rule," csa-
yamndo plural feminine (1. c. p. 550) ; frdy(a)zemnananm,
genitive plural = Sanscrit prayajamdndndm, from n^ yaj,
"to honour," "to sacrifice.'^ An example of a form indna
(for mdna) in the second principal conjugation is ui-dna
(1. c. p. 543), as nominative plural for the Sanscrit usdnas,
from vas9 " to wish," with an irregular contraction of the
syllable va to u. The following are examples of parti-
ciples of the future passive : as/jo^as^^^^ zanhyamana or
-mna, " about to be born " (Vend. S. pp. 28 and 103)*, and
Mj^M^^}AJM^) tbzddkhyamna, " being about to be raised
up " = Sanscrit uddJidsyamdna (Vend. S. p. 89, see §. 669.).
795. In close connection with the participial suffix mdna
stands the Sanscrit suffix man, the original form of which
appears to be mdn, which has remained in the strong cases.
The words formed with it have, like the kindred partici-
ples, either an active or a passive signification : some are
abstract substantives, like the Greek formations in [jlovyj
{fpKeyyiOvfi, Xapftov;/, ireicfiovrj, itToja-fiov^t Tnjfxovrj, ffteKTiiovrj),
which, in form, are essentially identical with the partici-
pial feminines in yievYi, as e and o are originally one (§. 3.) ;

* See §.668., where, however, we should read A5^,>^^»a< xanhya^ for
A)^^^ juU< zavihya ; and the remark at the end of the §. on the incor-
rectness of the way in which the word is written mnst be cancelled, and
the h of the participial forms referred to be really regarded as an euphonic
alteration of the n of the root fxse zan.

1074 FORMATION OF WORDS.

— and with regard to the accentuation of the last syllable
of the suffix, they agree with the Sanscrit dnd, and (for
mdnd, mdnd), of the second conjugation (see §. 792.).* But
few masculines in man remain to us in Sanscrit* and these,
too, are, for the most part, but rarely used. The follow-
ing are examples : i'dsh-man, ** fire,"" as " that which
dries f ' ush-mant " the hot time of year," as " the burn-
ing ;" vimarif " weaver's loom," as ** weaving or apparatus
of weaving ;" siman, ** border," as ** binding," from ftr si,
" to bind," with the i lengthened ; pip-man, " sin," as
"that which is sinned" {peccatum), from a lost root
Some masculines in man have a vowel of conjunction i ;
as, har-i-m&n, " time," as '* carrying away," *' destroying ;
sar-i-mdn, "the wind," as "moving itself," "blowing;
" dhar-i-mdn,"" " form," as " borne," " sustained " (thus the
Latin /orma, from the root /er); starH-mAn, "bed," as
"spread out" (compare stramen). Thus, also, the two
abstracts jdn-i-man, " birth," and mdr-i-man, " death,"
which are likewise masculine, but are distinguished from
the other forms in m^n by accenting their first syllable ;
fdnni-man, mdr-i-man — like Bushman, &c. — opposed to
harimdn, sarimdn, starimdn, dharimdn, bharimAn.^

* Compare <l>\€yfiovri with Sanscrit middle participles like yunjdnd,
^' the binding," from yunjmdna,

t See Bohtlingk, " The Unddi AjgHves," p. 58. Wilson renders bhari-
ntdn by "nourishing," "cherishing;" Bohtlingk by "maintenance." I
think, however, I may venture to deduce from the accentuation that it is
not an abstract substantive ; for otherwise, like m&riman, " death," and
jdniman, " birth," it would have the accent on the radical syllable (see
p. 1091). The expression ^VA kutunibuy by which, in the Unl^ Book
of Kdumtu^^ bharimdn is explained, according to Wilson also, signifies, not
" nourishing," " cherishing " (though to the root kutumb, an instance of
which has not yet been met with in books, the meaning "supported"
{dhritydm) is ascribed), but, amongst other things, "family;" and I con-
jecture that bharimdn signifies " family," in the sense of ** that which is

maintained

FORMATION OF WORDS. 1075

796. In Sanscrit the masoiilrnp bases in man are much .
more numerous than the neuter : they all have the accent

on the last syllable, and express partly a passive, partly an
active relation, or are abstracts. The following are ex-
amples : dhaman, " a house," as " that which is made or
built,'" from dhd, *' to place " (vi-dhd " to make ") ; vdrtman,
"way,'" as "that which is gone upon,"" from vart, vrtt,
" to go ;" vismaTi, ** a house,'' as " that which is entered,"
from vis, "to enter;" sddman, "a house," from sad, "to
go," and " to sit ;" kdrman, " deed," '* factum ;" vdrman,
" harness," as " that which covers ;" rdman, " hair " (abbre-
viated from rdhman), as "growing;" ddman, "band," as
"binding;" sthdman, "strength," as "having conti-
nuance," from std, "to stand ;" ^'dnman, "birth," from ^an,
"to bear;" priman, "love," from pri, "to love." The
Zend furnishes the neuter bases JM^^3^^ ddman, " people,"
as "created" ( = Sanscrit VHT^ dhdmxin, "house ;") yAt^jjxiAs^
maSsman, " urina " (quod mingitur, Sanscrit mih, " mingere ;")
and yAs^i^As^ chashman, " an eye," as " telling," " announ-
cing." The last is radically connected with the Sanscrit
chcdcshus, from chalesh, "to say."

797. Adjective bases in man are rare in Sanscrit : one
example is, ^l^ sdrmxin, masculine, feminine, neuter,
" happy " (as neuter substantive, " happiness,") the con-

maintained or supported," as the wife, hMryd, implies ^' she who is to be
supported,*' and the husband bhartdr^ hhartH, " he who supports." Wil-
son and Bohtlingk also regard ^| Ol^I «! iarim&n as an abstract substantive,
and the latter renders it (1. c. p. 149) " to bring forth," " to bear." The
explanatory Sanscrit expression (prasava) is, however, ambiguous: I
have, in my Glossary, assigned to it the meanings partus^ partura, and
proles, progenies, suholes ; and here, where sarimdn is explained by it, I
would adhere to the last signification, on account of the oxytone accen-
tuation of the just-mentioned expression.

* Without any root corresponding in idea. Compare the Greek dcio,
b€a'fiar, from tea-fiap, of which hereafter.

1076 FORMATION OF WORDS.

nection of which with its apparent root (^ iW, "n ffi,
" to break,") is, as regards meaning, by no means clear.
In Greek, adjective bases in fiov correspond, both as to ac-
cent and as to the non-distinction of the feminine base
from that of the masculine neuter; as, fivtjfiov, tA^jhoi/,
hfjcfjiov, iSfJLov, <f>paSfiov, enioTfjfxov, To the paroxytone mas-
culine substantive bases mentioned in §. 795., like s^hman,
"fire," as "drying," correspond in Greek such as wi/eC/xov
(" lung," as ** breathing "), yvS>fjLov, Saifxov (" god," " god-
dess," properly "shining,"* arrjfiov. With the there-
mentioned tri-syllable oxytone masculine bases like
harimaru "time," as "taking away," compare lajSefiov,
^yefxov. Here, too, belong — ^as e, like o, is a corruption of a
— some bases in fiev ; viz. irotfjiev (" herdsman," as " causing
to feed," compare pmco and the Sanscrit root pd, " to sup-
port," *' to nourish "), dvTfi€v,-f TUfxev, itvdfiev (the two latter
from roots now obscured). The suffix /xwv, fmv-o^, of
KevdfJLCiiv, $rifxii>v, yet/JLciiv, Ae<f(cov (from Ket^yLonv), has pre-
served, through all the cases, the long vowel, which, in the
corresponding Sanscrit suffix, is retained only in the strong
cases : so, too, the corresponding Latin mdn of the bases
sermdn, termdn {= terminus, see §. 478.) Um&n, and pidm^rut —

* It belongs to the Sanscrit root divy " to shine ;" whence dSva^ " a god ;"
div, "heaven; divasa^ "day," &c. (See Benfey, Gr. M, L» II. p. 207.)

t With respect to the T-sound in avrfirju and (rra3fjLa>Vy and which is
often added to the root before the suffix fxo, remark a similar circumstance
in Sanscrit, where, before the suffixes van, vara, and the gerundial suffix
ya, a euphonic t is always added to roots which end with a short vowel ;
as from Ji comeBJitvan and jitvara, " conquering ;" jitya (with preposi-
tions preceding), "after the conquest."

X Compare Pott, Etym. Inq. II. 694. and I. 270., where tS-mo, as well
as tig-num, is compared with the Sanscrit tdk9h, ''^frangere,findere,fahri-
cari ;" whence, also, takshan, "a carpenter ;" and our Deichsel, " a chip-
axe " (Old High German dihsila, and Anglo-Saxon dhixJ), and the Old
High German dehsa and dehsala, feminine, "axe** (Graflf, V..126.), as

" cleaving."

FORMATION OF WORDS. 1077

It is also highly probable that to the Sanscrit formations
in man belongs the Latin ho-min, for ho-mdn (in the old
language he^mo, he-mdnis). I take the A, as has been
already remarked elsewhere (" Berlin Annual Beg. of Lit
Crity Nov. 1830. p. 791 ; compare Pott, ** Etymological In-
quiries,"" I. p. 217 ; and Benfey, *' Gr. B. L"" II. p. 105), to
be the representative of the / oifuh &c., and therefore
fio a.s =fo, in fcHre^ fo-^em. Let reference be made to the
Prakrit Admi and havdmi, ** I am,'" for the Sanscrit bhav&mU
and the dative termination hi, of mihi, compared with the
Sanscrit hyam, from bhyam (see §. 215. and §. 23. at the
end). Man, therefore, according to the Latin expression,
is simply "the being,"" as in Sanscrit ^ana, "the bom " (root
fan, " to produce,"' " to bear "). There is also in Sanscrit
an appellation of man, from ^ bhu, " to be," viz. bhuvana
(see Wilson) ; and two appellations of the earth, viz. bhu
(the simple root) and bhmni (compare Latin humus). I am,
however, not aware that bhavat, "being,"" also signifies
'* man,"" as Benfey 1. c. asserts. The resemblance of the
Grothic base gurman, "man,"" Old High German go-m^m,
ho-^mon (nominative guma, gomo, komo), on which is based
our gam, of Brautigam, " bridegroom "" (Old High German
brik-gomon, properly Bravt-Mann) to the Latin ho-min,
he-mdn, is surprising : the relationship, however, I am now
of opinion, is confined to the suffix, and the German ex-
pression in reference to its root belongs to the above-men-
tioned Sanscrit jana (compare Graff, IV. p. 198), with the
retention of the old medial (see §. 92.), and with the loss
of the n, as in the radically, and, by suffix, related M-mmu
"germ"" (see §.799. Note ), and in the Latin gi^minus (see*
§. 478. at the end). Properly, therefore, gib-man, go-mon.

'^ cleaving." With the active signification among Latin formations in
mSn only remsana ptdmSn, "lung," as "breathing," by transposition from
plumSn (lonio rrXeufimv).

1078 FORMATION OF WORDS.

signify "the bom." The circumstance that we have
already the Sanscrit root jan contained in Gothic in the
forms kin (keinat kain, kinum, whence our Kind, " child ''),
kun (kunif ** sex ") and qvin (gveins, ** lawful wife," as " she
who bears," compare yvvrj), need not prevent us from ad-
mitting a form which has preserved the original medial.
I would recall to mind the fact that both the Gothic qvam,
" to come " (qvima, gvam), and gagga, " I go," are derived
from the Sanscrit root gam, " to go " (see §. 755.). But to
return to the Latin suffix m6n — ^from it arise the forms
mdnia, mdniu, by the addition of ia or iu, as, t&rici,
from t&r (viddricLt from victor), with this diffisrence, that the
primitives in mdn of derivatives like qtter-i-fn&nia, al-i-
mdnia, al-i-mdrdum, cer-i-mSnia (root cer := Sanscrit kar, kri,
"to make"), have disappeared. From adjective and sub-
stantive bases also spring, by this double suffix, abstracts
like acri-mdnia, tBgri-mdnia, casti-m&nia, miseri-mSnium, trisH-
mdnium, testi-mdnium, matri-m&nium. I consider the i of
forms like casti-mdnia, (Bgri-mdma, to be a weakening of
the final vowel of the base-noun (see " F'octdismus,'*'' pp.
132, 162, and 223), and the i of nUdri-m&nium to be an ex-
tension of the base, which, in the generality of cases, is
added to all bases ending in a consonant I therefore now
regard the 6 in the nominative plural as a contraction of
ai, and as = the Sanscrit ay (from ai), of a^^os: ovis, for
example, therefore, has the same relation to the Sanscrit
avay-as that mon-i-s has to mAnraya-si, Prakrit mAnr4si
(see p. 119); and thus pedis, amanii-8, come from the ex-
tended bases pedi, amanti. Remark that bases in u also,
in the nominative plural, have simple s for their termina-
tion, and that here the lengthening of the u represents the
Sanscrit and Gothic Guna ; e. g., frudu-s, as in Sanscrit
sunav-as, and in Gothic sunyti-s, *' son," from sunu, sunu
(see §. 230.). Compare, also, what has been said before (§.
780.) regarding the Old Prussian present participle.

FORMATION OF WORDS. 1079

798. In Greek there are some bases in fiiv which pre-
serve the long vowel in all cases, and resemble the San-
scrit strong cases with mdn, to which, with respect to their
7, they bear the same relation that, in Sanscrit, the plural
kri^ni-rndsf " we buy," has to the singular kri-ni-mi (see §.
485.). Compare the accusative singular ptfy/xiv-a, and the
nominative plural prjyfMive^, with analogous Sanscrit forms
like iiushmdn-am, hitshmdn-cts ; while in the genitive singular,
which belongs to the weak cases, the Sanscrit mshman-as
(with short a) stands in disadvantageous contrast with the
Greek pYjyfuv-og, The suffix /xZi/o, feminine fuvrjs is con-
nected with the Sanscrit participial suffix mdna, and, with
reference to the retention of the long vowel, stands nearer
the latter, than the usual fxevo. Here belong icaft7vo-f,
" oven,**" as " burning,'' " glowing,'' from koc/o), Kdco, with
the radical vowel shortened ; va-fiivrj, ** strife," for which no
root occurs in Greek, but which Pott (II. p. 594) rightly
traces to the Sanscrit yudh, " to strive " (whence yudhma-s,
" strife," which would lead us to expect, in Greek, ucr/ioj) ;
KVK\afjLivo£t KVK\afuvov, properly " rounded."

799. To the Sanscrit masculine substantive bases in im
man, mentioned in §. 795., correspond the just-mentioned
masculine bases ahmaru " spirit," as " thinking " (ahya, " I
think ") ; hliuman, " ear," as " hearing " (Sanscrit root Sru,
from kru9 "to hear,' Greek k\v)] bldman, "a flower," as
"blowing" (Old High German bluot, "floret;'" bluont,
"florerd^)', milhmany "a cloud" (probably like the Sanscrit
mJighcu originally " mingens,'*'' see §.140.); skeiman, "a lamp,"
as "shining," ''lighting" (Sanscrit kan, *' to light")*; and

* I have no scruple in deducing skeiman from the root skiriy '^ to shine/'
'' to light " {skeina^ akain^ 8kinum\ with the suppression of the final con-
sonant of the root, as nm is a combination unsuited to the German ; hence,
also, in Old High German, ki-mcn^ chi-mon (nominative -mo)^ "germ,"

from

1080 FORMATION OF WORDS.

with passive signification, mal-man, ** sand," as *' triturated,'*
also neuter (nominative masculine mahnxu neuter malmd,
see §§. 140. 141.) ; and hiuh-manf " heap,'" as " heaped up,"
from the root, lost as regards the verb, huh (euphonic hauh,
see §. 82.), to which also belongs hauhSf " high " (Grimm,
II. p. 50). The Old High German places over against the
Gothic-Sanscrit man the form mon (nominative mo), and in
this form corresponds to the Greek fiov. The following
are examples : wtihs-^t-wMi, and wahsmon, " vegetables,"
" fruit,"' as " growing," or " having grown ;" gliz-e-mon,
** lustre ;" ka-smag-mon, " taste ;" with passive signification ;
sd-moTif " seed," as " sown" (Latin se'meTi).^ As in Sanscrit
the suffix m^n also forms abstract substantive or adjec-
tive bases, as prath-i-m^n, "breadth," from prithA, "broad"
(from prathu, compare Greek irXaTv) ; krishn-i-mAn, " black-
ness," from krishnd, " black ;" t we may also here mention
the Old High German rdta-mon (also rdto-mon, rdte-monj^

:a »»

from the roots kin, chin (chin-i-tf ^puUulaty" ar-hm-i-ty -Mni-ty ^^gignity
^^germinat" see GrafiF, IV. 450.)= Sanscrit ifw Jan, "to produce," "to
bear" (Latin gen, Greek yev), whence Jan-wian neuter, axid Jdn-i-man
masculine, " birth," which agrees with Mmon in root and suffix. Ger-meny
for gen-meny corresponds in Latin. With respect to the rejection of the
final consonant of the root before the m of the suffix, compare the (§. 796.)
above-mentioned Sanscrit ro-?nan, "hair of the body," as "growing,* for
rSh-man ; and Latin forms ]ilie/ulmeny for fulg-men ; lu-men, for htc-men ;
as well as gS-minus (see §. 478. conclusion), which is prohably, in root and
suffix, connected with M-mon, To lu-men corresponds, in root and suffix,
the Anglo-Saxon iSo-man (nominative leoma), " h'ght," for Uoh-many com-
pare Gothic lauh-m&niy "lightning" (§. 793.).

t The kindred Sanscrit root vakah, " to grow," would, in the middle,
form vdJcf^amdria as participle present.

* This has been already explained in the above sense in my Heview of
Grimm's German Grammar {^^ Berlin Ann, Reg. of Lit, Criticism" Feb.
1827, p. 757; "rocaZimw*," p. 131).

X The final vowel of the base word is rejected before the vowel of con-
junction t.

^t

n

FORMATION OF WOBDS. 1081

" redness," from the adjective base rdta, as a very remark-
able analogous form. The Latin uses for this object the
suffix mdniutf or feminine mdnia (see §. 797. conclusion), ex-
tended from mdn.

800. In Lithuanian the suffix spoken of appears in the
form men, nominative mu; and thus, from a Lithuanian
point of view, the obscure piemen, nominative piemu, " shep-
herd's boy,'' corresponds to the Greek itoiyievy Trotfitjv (see
§. 797.) ; and akmen, — mu, " stone," to the Sanscrit, also ob-
scure, dsman, — md. From a Lithuanian point of view, the
bases aug-meriy zel-men, "sprout," "shoot," as "growing
(augu and zelu/*l grow"); yos-men, "apron-string," "girdle
iyds-mi, "I have a girdle on;" ap-si-yds-mi, "I gird myself");
sto-men, "stature" (stawyu, "I stand," compare Sanscrit sthd-
man, " strength, from sthd, "to stand"), are quite intelligible.
Semenys, " linseed," properly only '* seed " (seyu, *' I sow,
future s/e-su), is a nominative plural, as akmeny-Sy " stones,
from the extended base akmeni, and leads us to expect a
singular semu ; and therefore corresponds to the Old High
German base sd-mon (§. 799.), and to the Latin se-m^en.
The Old Sclavonic presents a few masculine bases in men,
which, in the nominative, contrast mi»i my with the. Lithua-
nian mu and Sanscrit mA (see §. 260. at the end, and
p. 348), but prefer, however, the form meny, from the pro-
longed base m,eni (Dobrowsky, pp. 287 and 289, under ENb
eny). From a Sclavonic point of view, however, only plu-
men (nominative plamyt or phmeny, " flame," as ** burning,"

ni

in

* The suffix men forms the entire plnral, with the exception of the
genitive {akmen-H, '' laj)idum"=iSaJiscrit asman-dm\ from the extended
meni. In some cases of the singular the suffix is extended by the addi-
tion of ia; thus, in the genitive, hkmenio (like wilko^ §. 169.), together
with the oiganic akmen-s; instrumental hkmeniu (like toWeu)^ together
with akmeni-mi; accusative dhmem-n; locative dkmeniye^ according to
the analogy oiamye^ from the base azoz, '^a sheep."

4a

1082 FORMATION OF WORDS.

is etymologically intelligible (oaAhSiTHCa planuntisani
" comburi ;" oaahth pal-i-tU ** urere,'*'' &c. ; see Miklos.
p. 62) ; RAmeh kamen^ " stone "" (nominative kamy, or kameny)
answers to the Lithuanian akmen, akmu, and Sanscrit ds-
mails dimd.

801. To the Sanscrit neuter bases in num (nominative
ma, see §. 139.), mentioned at §. 796., correspond the Latin
in min (men in the cases having no termination beyond the
base), the Greek in /nar, for fiav (see §. 497), and the Gothic
and Sclavonic in maUf meh men. The Latin and Greek
formations which come under this class have, like their
Sanscrit sister forms, either a passive signification, which,
indeed, is generally the case ; as praefamenf stram^nf simen,
agmen, segmen^ germen^ vpayfiar, iroirffiaT, p9jfiaTf aKova-fxaT,
ypafifxaT, yXvfifiar, Softar, ^pfjifiar ; or an active significa-
tion, as flumen^ Iwnen, (from lucmen), fabnen (from fulgmen\
tegmeuy teg-i-ment'^ teg-Vrm^en, reg-i-men (" helm,'* as " guid-

* Germen^ from genmen^ is founded on the frequent interchange of
liquids (§. 20.).

t The t of tcg-i-mm^ reg-i-men^ is identical with the class-vowel of the
third conjugation, and leads us, therefore, to the Sanscrit a of the first
and sixth class, which in Latin has been weakened to t or u {veh-i-mus,
veh-u-nty see §. 607.) : this is clear from the long t of the fourth conjugation
{mol-i-mm, ftilc4-men, as mol-i-mini^ fulc-t-mini), and the d of the first
{certdmen^ levdmen, &c.). Forms like agmen, Jragmen^ tegmen, on the
contrary, belong to that period of Sanscrit which combines the suffix man,
without reference to the conjugation of the verb, almost invariably direct
with the root. In the Latin second conjugation we should expect S
before the said suffix, and the mentu derived from it : for it, howeyer, we
find, where the suffix is not combined direct with the root, according to
the analogy of the third conjugation, » or m; hence, sed-i-men^ doc-u-men,
doc-u-mentum, mon-i-mentumy mon-u-mentum. In general, the Latin i of
the second conjugation does not keep its place so firmly as the two other
representatives of the Sanscrit tenth class (see p. 110) ; hence, also, doc-m,
doC'tum, opposed to am-d-m, am-d-tum, aud-i-vi^ aud-Utum.

FORMATION OF WORDS. 1083

ing "), Secfiar, pvfxaTt wi/ev/xar, dijfxaT, ^povrtiyLar, eifiar, eaft;-
jJLar ; or are abstracts, as solamen, certamertf levamen, tentamen,
regimen, moUmen, jSAT/ftar, ISorffiar, l3ptr)(r}fiaT, ietyLar, yapiJiar.
At the end of compounds, the original v of the suffix ^ar,
which is corrupted from /xav, either remains in its original
form, or is entirely suppressed : in both cases, however,
the a is corrupted to o (nominative masculine and feminine
ficov) ; probably because the heavy sounds t and a are
found, through the incumbrance of composition, less ap-
propriate than the lighter v and o ; hence, TroAuirpayixov,
dirpayfxov, dvaifxov, and dvatfxot uKVfxov and aKVfio, avtovtz/xo,
awiavvflo. The form voavvfjivo is interesting, because here
we find intact the old n of the Sanscrit nimaru Latin
nSmen, &c., which, in o-vo/xar, has become t, but elsewhere,
in the compounds of this word, is suppressed : along with
its retention, however, we find the base prolonged by o, and
the vowel of the suffix suppressed (vwvu/xi/o, from vonvvfiavo,
or vcavvfiovo) ; in the latter respect compare the weakest
cases of the Sanscrit ndman, the genitive ndmn-ds, dat. ndmn-Sf
and the Grothic plural namn-a* 'AiraXafjivo points to a
lost substantive TroAa/xar, from vahafiav (of which, also,
TrdKafxvdto^ is a proof),which apparently has been disused for
iraKififl, I would also rather regard Kpri^efivo, " head-band,""

* In §. 285. namdna is given incorrectly, though this form would be
the regular one (compare hairtSna), and would correspond well to the
Sanscrit namdn-i (from namdn-a^ see §. 234.). The form namna^ on the
other hand^ answers to the Sanscrit weakest cases^ while the nominative,
accusative, and vocative plural of Sanscrit neuters always belong to the
strong (see smaller ^^ Sanscrit Grammar" §. 177. Note). It appears,
however, that in Grothic it is necessary, for the protection of the full form
6na, that it be preceded by a vowel long in itself or by position, or by
more than one syllable ; hence augomi, ausona, bamilona, vbtUna, but not
namdna^ and probably, also, not vatdna^ from vatan^ "water," as the
dative is vatnam^ not vataijiym ; compare Grimm, I. p. 609, Gabel. and
Lobe, p. 67.

4a2

1084 FOEMATION OF WORDS.

v^ith respect to its concluding element, as a form analogous
to '(ovvjJLvo (and, therefore, as a derivative from Sefiar, from
$6/Liav), than as a participle for Sefievo : on the other hand, I
look upon hSvfwo, which Passow takes to be analogous to
v^vvfivo-^ and dira\afivo^t as a participle (properly, there-
fore, ** doubled ") from a reduplicated verbal base Siiv, which
has sprung from Svo, and from which a present indicative
SiSvfu might have been expected; thus, SiSvfiva^, like
Ji5o/xei/o-r, only with the suppression of the middle vowel of
the suffix, as in the Latin cd-u-mnu, and in the above-men-
tioned (§. 791.) en-ifn'U'fnne. Compare, also, the participial
substantive bases in fivo, feminine /iva, as, /SeKcfivo, fieStfJLvo,
fiepifiva, which have been already discussed by Pott (E, L
II. p. 594.) under this view, and which have no corresponding
verb, any more than the above-mentioned StSvfjLvo, though
^e\€fjLvo, just like j86A.o;, is visibly connected with ^d7\X<a.

802. The Old Sclavonic neuter bases in meh men have
in the cases, which in Sanscrit and Gothic drop the final
n, retained the original a with a resonant nasal ; hence,
HMA iman, ** names "*" (see §. 783. Rem. 1. conclusion), from
the base emen=s Sanscrit nd-man. Here belong, also, the
bases c:bMEH sye-men, " seed,"' as " sown " (sye-ya-ti, *' to
sow "") = Latin semeuy Old High German sdmon masculine
(see §. 793. Note 3), nHCMEH pis-men^ " letter of the alpha-
bet,"" as " written " (pw-a-fi, " to write "') ; ^hameh J^na-

* I cannot refrain from drawing attention here to the strong agreement
between the Sclavonic root pis and the Old Persian j9t«^, with the prepo-
sition rd : ni-pishy " to write down," " to describe," properly, " to hew in."

RawHnson (Beh. IV. 47. 48.) translates ^^••^^^.^•^•^yfy.^^yty.
nipishtamhy ^^ scriptum ;'* and, IV. 71., ^^'Tf -T^^'^'^f •^•(f^*^).
>']j\ • nh/api8ka(i/a)m by " inscripH" I think, however, that we must,
with the ^ Pf read also the a contained in it ; thus, myapaiahayam : for
whether this form be taken as a causal — thus, '' I have caused to describe "
—or as a verb of the tenth class, in both cases Gima is indispensable.

The

FORMATION OF WORDS. 1085

men, " a sign,"' as " making to know " (^na-4;i " to
know '*'), and a few words from obscure roots (Dobrowsky,
p. 288). The Gothic famishes besides na-man, ** names "
(nominative accusative namd, see §. 141.), which, in the
other German languages, has become masculine, only atdd-
maUf "age,^' if this word really be, as Gabel. and Lobe
suppose, a neuter, which cannot be discerned from the
but once occurring dative aldSmin (Luke i. 36). As the
neuter abstract of an adjective it would correspond to the
above-mentioned (§. 799. conclusion) Sanscrit neuter bases

like krishfiA'm&n, "blackness," from krishn^, "black;"

• • •

while the there-mentioned rdta-mon, " redness,*" like namon,
** names" (nominative namo), has perhaps first become
neuter as it was gradually corrupted. The 6 of the
Gothic aJdd-man I take to be the lengthening of the a of
the base alda (see §. 69.), " old," which, indeed, does not
occur, but may be inferred from the cognate dialects (see
Graff, I. 192). K, however, aM6-man is derived, not from
an adjective, but from a verb, we must suppose a lost de-
nominative a/c/d-m, ** I grow old " (see §. 765.) ; and aldd^
mon would then correspond to Latin formations like certA-
men (§. 801.). We can hardly imagine any similarity of
formation between the above and the Old High German
compounds aU-^uomt aU-tvom (see Grimm, II. 151.).

803. From the suffix men, min^ an extended form m£ntu
has proceeded in Latin {argu-m^entiirm, mon-u-m^rdu-m, incre*
mentvr-my co-gruHfnenttt-m^ sed-i-merdu-m &c.), in which I do
not agree with Pott {E. L II. 594.) in recognising the affix
of a participial suffix tu (tus, ta, turn), but one that is simply
phonetic ; just as, in Gothic, the base hun-da (nominative
hunds) stands over against the Sanscrit ^n of the weakest

The causal fonn of the SaDscrit pish^ Class 7, " to beat down," " to bruise/
whence the meaning "to engrave," "to hew in," is easily dedacible
appears to me the most probable.

1086 FORMATION OF WORDS.

cases, and Greek kw {Kv<av, kwo^), or as, in Latin, the San-
scrit roots tan, "to extend," and han (from dhan), "to
smite," " to slay " (Greek 6av), has become extended to
tend, fend (j^s^dh, d, see §. 29d.X and» i^ Sanscrit itself, kan
and chand (from kand), " to shine," are originally one. A
mute is readily attracted to the side of a nasal, and the
former as easily annexes a vowel ; and thus, for the Latin
extended sufiix mentu, without reference to gender, we find
a parallel in the Old High German munda (from manda),
nominative mund, but only in the solitary base hliurmunda,
nominative hliur-mund (abbreviated liu-mund, our Leumund,
" renown "), ** fame,'' as " that which is heard," as in Gothic
hliu-man, " ear," as " hearing " (compare Grimm, II. p. 243).
The Greek base eAfuvd, " worm," as " winding itself," has
added to the suffix fxiv, mentioned above (§. 798.), only
a 0, but in this respect stands as isolated as, in Old High
German, the just mentioned hliu-mundou The form ehfityy
(ehfityyes) exhibits, instead of the T-sound, a guttural, and
thus reminds us of the relation of our yung, " young "
(Gothic Vugg^Sf theme ytigga = yunga), to the Sanscrit yuvan,
in the weakest cases yun (genitive yun-as), and Latin ^ave-
nis, jumor. Thus the Old High German suffix unga (our
ung) of abstract substantives, as in ar-find-unga, "inven-
tion," warn-unga, "warning," may be identical with the
Sanscrit feminine form of the suffix ana (and) ; so that the
first a has become weakened to u, as in the polysyllabic
forms of the preterite, as bunti, " thou didst bind," com-
pared with the monosyllabic bant, " I bound," " be bound."
In the same way our root sang^ "to sing," (Old High
German singu, "sang," second person sungi), may be com-
pared with the Sanscrit root svan, " to sound " (compare
Graff, VI. p. 247).

804. I think I discover the origin of the medio-passive
participial suffix mdna, and of the cognate nominal suffix
man, in the combination of two demonstrative bases ma

FORMATION OF WORDS. 1087

and na (see §§. 368. 369.) ; the vowel, therefore, being
lengthened in mdnaf and in the strong cases of man, and
the final vowel in the last-mentioned form being sup-
pressed. We must here observe that na readily combines
with other pronominal bases, and then always takes the
last place ; hence ^R«f ana, J^ Sna, in Greek ice?vo^, and in
Old Prussian ta-ns, for ta-na-s, "he,"'* opposed to the Li-
thuanian simple ta-s, "the."*' If the medial relation be
really expressed formally in the suffix mdna, fievo, in that,
case the final element must express the nominative rela-
tion, or that relation which, from time to time, belongs to
the position of the participle ; and the unchangeable md, ijlc,
the dative or accusative (siU, se) ; so that, therefore, 7i na,
vo, denote the person acting, and m md, fte, the person
acted upon, which, however, in the middle, are one and the
same. The suffixes' of participles, as in general those of
adjectives and substantives, represent the personal termi-
nations of verbs, i.e. those of the third person ; and I thus
consider the t of the participle present and future active
as identical with the termination of the third person, and,
like the latter, a derivative from the pronominal base ta,
the vowel of which, in the participial suffix, is dropped.
The n of the active participial suffix probably serves only
for the phonetic intensification and more emphatic desig-
nation of the agent ; while, in the third person plural, plu-
rality is symbolically denoted by the same nasalization
(see §. 536.) : hence the coincidence of bhdrdnt, ij>epovT, f event
Grothic bairand, ** bearing," with bhdranti, ^epovrt, ferwnt,
bairandf "they bear."

805. We recognise the simple pronominal base wia in the
Sanscrit suffix if mxi, which in adjectives or substantives
denotes the person or thing which completes the action

Feminine tarma^ with the favourite repetition of the llqaid.

1088 FORMATION OF WORDS.

expressed by the root, or on whom that action is accom-
plished. Abstracts, also, are formed by this suffix, which,
however, is seldom adopted in that state of the language
which has descended to us ; while the corresponding suffixes
of the Lithuanian and Greek (fno, /xo) are of very frequent
use. The following are examples in Sanscrit : ruJcmd-m,
"gold," as "glittering" (ruchf from rtJc, "to shine"); yttg-
md-nh " pair," as " bound together;" tigmd, adjective (-md-St
md, md-mX "sharp" ("sharpened"), "hot" (root tij, from
tig, "to sharpen"), substantive neuter (figmd-m) "heat;"
hhimd, "fearful" ("feared," root hht, "to fear"); dhrnnd-s,
"smoke," as "being moved" (root dhu, "to move");
yudh-md'S, "combatant," "contest," "arrow" (yudh, "to
fight"); gharmd'S, "heat," apparently as "moistening," by
sweat (root ghar, ghri, "to sprinkle"); ishmd-s, "tone
(root ishf "to wish"); idhmd-St "wood," as " being burned
(root idh, "to burn"). To the latter corresponds the Zend
a)9j)mas aSsma (nominative md). Remark the agreement
of the above-mentioned Sanscrit words in the accentuation
of the suffix with Greek formations like oroA/xo-f, TroA/xo-f ,
Kopfxo-^, dSvpfjLO-s, KoixfJLo-^t TptfXfjLorg, (pTioy/xo-^, dyfJLO'^t pVfJLO-g,
Xi//xa-j, KKav-d-fio-Sf /xv#cj;-6-/Lto-r. In Sanscrit, also, there are
a few words formed with ma, which, like iroTfio^, o?jlco-j,*
av€fjL(y-^, oKfJLo-s, and some others of obscure origin in Greek
(Buttmann, 11. p. 315), have the accent on the radical sylla-
ble. Here belong, for example, bhdma'S, "the sun," as
" giving light," sushma-m, " fever," as " drying." To the
masculine nominatives in mas correspond numerous Li-
thuanian abstracts in i-mas, or, with m doubled, i-mma-Sf'f

1^

*i

* ol is the Guna form of the root t, "to go" (compare §. 609). Thus,
in Sanscrit, vdrtman^ " way," from vart^ vrity " to go."

t With regard to the doubling of the m, compare the doubling of
liquids so common in Old Prussian. I belieye I have discovered it to be
a fixed law in Lithuanian, that the doubling of the m in the said suffix is

only

V

tt

1^

FORMATION OF WORDS. 1089

the i of which, as in Sanscrit forms like jdn-i-man, " birth """
(see §. 795.), is only a vowel of conjunction. The following
are examples: gimm-i'mma'S, ** birth;" ey^i-mma^s, "going
{ei-mi, " I go f ' ey-au, ** I went ") ; pa-gadinnri-ma-Sy " ruin
(pa-gadirm, "I mar'"). In this manner, in Lithuanian,
abstract substantives are formed from adjective bases also,
in which formation a final a of the adjective base is weak-
ened to Uf while bases in u have their vowel unchanged.
The following are examples : gudu-miru^s, " avarice," from
gudu-s, "avaricious ;" gra-zu-mmd-s, " beauty," from graHi'S,
"beautiful;" darku-mnas, "ugliness," from darku^s,
" ugly ;" drasu-mna-s, " boldness," from drasu-s, " bold "
(compare Greek Opacrv^, dapcrvs, Sanscrit dharsh, dhrish,
" to dare "); rietu-mna^s, " hardness," from rieta-s, " hard ;"
auksztu-mma-s " height " from avJeszta-8 " high ;" Hgu-mma^s
"length,'' from ilgi-s (for ilgia-s, see §. 135.), "long."*

806. The Latin has but a few words in mu-s, and those
of obscure origin and etymology, to offer in comparison
with the Indo-Lithuanian in mas and Greek in /xo-; ; as,
an-i-nms, which, like the Greek av-e-fto-y, has originated
from the Sanscrit root an, " to breathe," " to blow " (see
109^ 2.); furmus=dvfx6^, Sanscrit dhu-mds, "smoke" (root
dhu, dvf see §. 293.) ; perhaps pd-mur-m, " apple," as " nou-
rishing," or " being tasted " (Sanscrit pd, " to support,"
and " to drink," compare pa-bulum, pa-scOf pdrvi, pd-tiis, pd-

only then permitted or required when, ezclnsiye of prefixes in combina-
tion with the verb, the verbal base is monosyllabic. If, however, it be
polysyllabic, the m is not doubled; hence, indeed, gimm-i'mma-s, ''birth,"
and also uz-gimm-i-mma-8^ idem. ; su-gruw-i-mma-s, '' circumstance "
\ (gruum^ " I occur ") ; but not gratidSn-i-mma-s, " warning," but grauden-

i-mas {graudenuy '' I admonish").

* Bases in ta, nominative m, drop their t before the u of their abstracts
which has arisen from a ; hence middu-mmas^ '' greatness, from middis^
"great."

1090 FORMATION OF WORDS.

turcL) ; and the adjectives for-mus (compare fermo, fer^men-
tum)i ftr-mus (compare /or-^i*, /ero), cJrmus, In the Ger-
man languages, also, the formations of this class are, for
the most part, no longer conscious of their origin : they
occur in Grimm, II. p. 145, where, however, the bases in
ma and those in mi, which have both lost their final vowel
in the nominative singular, are not distinguished. I re-
gard the sufiix miy which exists also in Sanscrit and in
Greek,* as merely a weakened form of ma, as in the Greek
pronominal base yd (accusative /xiv) = Sanscrit ma (see §.
368.). The Gothic bag-ms, "tree" (theme bag-mxt), pro-
bably means originally " the growing " (Sanscrit barh, brih,
"to grow"): the adjective base ar-ma, nominative arm^,
is perhaps an abbreviation of ard-ma, and a shoot from the
Sanscrit root ard, " to vex,'" with which I would compare,
also, the Sanscrit dr-ma (nominative masculine drma-s,
neuter drma-m) " a malady of the eyes :" bar^mi (nominative
barms), " lap," springs evidently from the root bar (baira,
bar) "to carry." In Old High German davrm, dou-m
(theme -ma, or -mi ?) " vapour," corresponds to the San-
scrit dhur-md'Sy " smoke ;" trau^m, theme traw-m^a (Old
Saxon drd-m, drd-ma), leads us to the Sanscrit root dr& *'to
sleep ;" sau-m (theme savrma), ** seam," to f^ siv, " to sew "
(Old High German siwu, "^wo"); hel-m, "helm," as "co-
vering," springs from the root hal, " to conceal " {hiluy hal,
hulumh),

807. The feminine form of the suffix, viz. md, does not
occur in Sanscrit in substantives ; but the Greek in fxtj, as
yvdfiYff fJLvrjfirj, (TTtyfxrj, ypafifirj, correspond to it ; as do the
Latin, like flamma, from flagma, fdma, spuma^ strtima^ gluma

* ^•^. ^PtsHU daUmUs^ masculine, Indra's " thunderbolt,* from dal^
" to cleave;" ^Thh bhu-mi 8^ "earth," feminine, from bhu, "to be,'' "to
become;'' Bvva-fju-s, (pij-fxi-Sy Oe-fu-s (Ion. genitiye 0€fii-os).

FORMATION OF WORDS. 1091

for gbdrnia; and the Lithuanian in ma, mg;* as wazmh,
"riding;" tudma, "grief" (fuzuhs, "I grieve"); sluzma,
"service" (sluziut "I serve ");t Q^^smS* "song" (jpzedmi,
"I sing"); baime, "fear" (bii/au, "I fear" Sanscrit root
bhi, " to fear," bhimd-s, " fearful," and nominative pre-
terite, whence the patronymic bhdima'S, feminine bhdimi);
drausme,X "prohibition." To this class probably belong,
also, the Lithuanian and Sclavonic abstracts in ba, be, ba
ba ; so that the medial stands in place of the organic
nasal, as in dewhfii, AEBATb devanty, " nine " (see §. 783.) ;
and as in Greek )8poTOf, )8/oa5i5j = Sanscrit mritd-Sf mridu-s.
Thus, in Lithuanian we find the forms tuzbh, "grief,"
sluzha, ** service," side by side with ttizmh, sluzma, which
have the same meaning. Garbe, " honour," " fame " (gir-
riu, " I praise "), corresponds in its root to the Sanscrit
gar, gri (in the Veda-dialect, " to praise "). Abstracts in
be from adjective bases, whose final vowel has been weak-
ened to y (=si), are numerous ; as, silpny-be, " weakness,"

* 3fe from mia (see p. 174, Note *).

t Thus drutu-mci^ " strength," together with drutu-ma-B^ from the ad-
jective base driita^ ''strong.'

X For draud-me (draudziu, " I forbid")? according to the analogy of
the infinitive draits-ti^ in which the change of the d before t into s is re-
gular (see §. 457.)- ^^ ei-s-me, " going" (ei-mi, " I go"), the * is euphonic,
as in Greek forms like be-a-'fir], 8e-(r-fMJff. A euphonic 8 of this kind some-
times precedes the masculine suffix also, but, I imagine, only after gut-
turab, and then the insertion of the vowel of conjunction f, mentioned at
§.805., does not take place; hence, dzaitg-s-mas, "joy" (dxaugio-a, "I
rejoice'); werk-s-mas^ "weeping;" rc/f-»-wa«," clamour." Hence it ap-
pears that, in Lithuanian, krni or gsm is a more favourite combination
than gm^ km. Compare, in this respect, the insertions of consonants
mentioned in §§. 95. 96., from which, however, is to be excepted the s of
the Old High German tarst, " thou venturest," torstay " I ventured," as
here the 8 belongs rather to the root (Sanscrit dharak, dlirUh, " to dare"),
see Sanscrit Glo88ary^ a. 1847, p* 186.

1092 FORMATION OF WOEDS.

from sittmons, ** weak ;'* byaury-^, ** ugliness/" from Infauru^Sf
" ugly.*"" The following are examples of Russian abstracts
in ha : MOAb6a mdyba, " begging '' (moaio mdytif ** I beg") ;
CAy;d[&6a sluschba, ** service "" (cAy Ay sluschu, " I serve ") ;
cmpaA6a itraschbOf ''watching"" (cmepery deregu, **\
watch "") ; aAqfia dllbay " hunger "" (aA^A alcu, " I am hun-
gry ""). Perhaps, as we have seen in Gk>thic m take the
place of 6 in the dative plural (see §. 215.), so we may
assume the converse mutation of m to 6 ; and, in fact, in
the formations in u-Jyni (theme u-bnya neuter, u-hnyd femi-
nine, see Grimm, II. p. 184), occasionally u-fnt If we re-
trace the 6, which is evidently the more genuine form, to
tn, then vii-u-mni {vU-u-bnU " knowledge,"" would resemble
Latin formations like ai-vrmrms (see §. 478. conclusion); and
in my opinion the Gothic like the Latin u is only a class
vowel, and therefore a weakened form of a, or, in GTimm"s
weak form of the second conjugation, of 6 ; and therefore
vund'U'fhh feminine, "wound,"" is for vundrd-fnif from vuiidnd,
" I wound."" It deserves notice, that, together witt/ratsf-
u-hnit feminine, *' attempt,"" there occurs also the {ormfraist'
d'brd (genitive plural fraist-^bayd, Luke iv. 13.), evidently
from a weak verb ^atsf^ (compare the Old Northern yVmfa,
" tentarer see Graff, III. 830.)> which cannot be cited ; for
the strong verb fraisa gives no authority to the f, and
would make us expect only frais-^urinu In fast-u-bni,
** fasting,"" the u represents the a sound of the diphthong
ai of the third weak conjugation, where we must observe
that the i element of this diphthong is dropped also before
personal terminations beginning with nasals ; thus, as fast-
a-m, ** we fast,"" fast-a-nd, ** they fast,"" for fast-ai-my fast-
ai-ndi so fast-u-bni, from fast-u-mni (or fcLst-ai-mnL

808. In order to exhaust the presumptive cognates of
the Sanscrit participial suffix mdnat the Latin suffix midu
must also be here mentioned, the I of which, perhaps, like
that of a^m9 = Sanscrit anya-s, "the other,"' rests on the

FOBMATION OF WOEDS. 1093

favourite interchange of the liquids (see §. 20.). We divide,
therefore, /a-wiwftis, properly "the making'' (for /ac-wirfus);
or if, as Ag. Denary conjectures, it belongs to the Sanscrit
root bhaj " to honour," " to serve '' (compare Gothic and-bah-
ts, " servant," " he who serves ;" sd-mvlus (for stig-mulus\
" sting," as " sticking *' (compare, according to Vossius,
o-T/fo), ariyfjia, &c.). Compare the Irish suffix mhuil, in fas-
a-mhuiU "growing" (fasaim, "I grow")=t?<iArsA-a-md»a-s.*
If, however, the a oi fasa-mhuil is not a class vowel, as in
fas-a-maTy " we grow " = Sanscrit vdhsh-d-mas, but to be in-
cluded in the suffix (to be divided, therefore, /a5-amAw«7),
in that case the last portion of the word properly means
"like," and is most probably an abbreviation of the adjec-
tive samhuil,'\ which occurs uncompounded. Words like
fear-amhuily " manlike," can scarcely be explained otherwise
than as compounds of fear and amhuil The Latin suffix
mulu might, however, be also connected with the Sanscrit
mara; whence, admara ond jasmara, "voracious," from ad^
jas, "to eat," srimara (Wilson), according to some authori-
ties, " a young deer," from sar, sri, " to go," This suffix,
however, as v and m are easily interchanged, is originally
one with the more usual vara ; whence nasvara, " transi-
tory," from naS, " to be ruined ;" bhdsvara, " shining," from
bhds, " to shine ;" sthdvara, " standing," " immoveable," from
dhd, "to stand."

809. Before we pass on to the consideration of those
participles which do not, like those already discussed, be-
long to any tense of the indicative, and make no distinc-
tion between active, passive, and middle, we must mention
one other participle peculiar to Latin, viz. the participle
future passive in ndu. I have already, in my Conjuga-

* It being taken for granted that vaksh Is used in the middle. F for
Sanscrit v is, in the Irish dialect of the Celtic, very usual,
t Compare the Sanscrit sama^ ''like/' Latin nmiHs.

1094 FORMATION OF WORDS.

tional System (§. 109*. i.), considered this, with regard to its
form, as a modification of the participle present active,
and think I must continue to support this view, though it
may be objected that, in this manner, the passive and fu-
ture signification of the said participle will have no foun-
dation as respects form. But words seldom express in
form those relations, to denote which they are destined by
the use of language ; and grammatical forms often change
their original meaning, as, in Persian, the forms in tdr or
ddr (faref'tdr, "deceptor,'' dd-ddr "dator,''*), which are
based on the Sanscrit nouns of agency in Mr, Greek in
Tfjp, and Latin in tor, tdr-is, are used, contrary to their
original intention, with a passive meaning; also, gurif-tdr,
" capfus, captivus, prceda T^ res-tdr, ** liberatus ;" kush-tdr,
"occisus'^ gyf'tdr, '' sermo'"* (see Vuller's Inst. L, Pers.
p. 166) ; while conversely the participles in tah or dafe, which
are based on the Sanscrit passive participles in ta, have
generally an active signification, and retain their original
passive meaning almost only when in combination with the
auxiliary verb shvdan (** to be "); hence burdah, *' qui tulib "
= Sanscrit britd-s (from bharta-s), ^^htus'^ but bvrdah mzska-
vam, "ferorr properly " lotus fioJ*'' The Latin ferendus ap-
proaches very closely the Persian present participle barin--
dak, "bearing;'^ and, like the latter, has weakened the
original tenuis (o{fereni) to a medial, and extended the base
by the addition of a vowel, both which changes take place
also in Prakrit and Pali (see p. SOl)*!*. This opinion that

* The choice of dot tin the sufiix depends on the preceding letter.
Compare §. 91. conclusion.

t The Sanscrit also has a few words which, in their origin, are evi-
dently present participles, hut have added to the nt also an a, or have
preserved the a of the hase ta (see §. 804.). They accent the suffix ;
hence, bhdsantd^s, "sun," as "lighting," opposed to bhdsant (see §. 785.);
rdhantd'Sy "a certain tree," as "growing," opposed to rohant; gada-

yantd «,

11

11

FORMATION OF WORDS. 1095

the future passive participles have proceeded from the
active present participles is confirmed by the circumstance,
that the class peculiarities, which do not extend over the
present and imperfect, and the forms which spring from
the present, are preserved in the form in ndu ; e, g. the n of
sterno (see §. 496.), the t of pedo, pkdo, the reduplication of
gigrto (gen-uif gen-i~tum) ] the gerunds also, which are in
form identical with the future passive participle, point to
an original active and present signification of the participial
form ; docendi, ** of teaching,"' docendo, ** by teaching,
speak for the signification ** teaching,"' which " docendus
must originally have had ; for such abstract substantives,
especially those which, like the Latin gerunds, express only
the exercise of an action, spring naturally from active
present participles ; as abundardia from abundant^ providentia
from provident, and not from passive participles. Partici-
ples in turu, when they form abstracts, or rather raise their
feminine form to an abstract, abandon their future mean-
ing, and then pass as present participles or nouns of agency ;
thus, ruptura, " tearing,"' as the personification of " to tear,'"
properly " the person who tears ;"" junctura, ** joining ;""
mistura, " jningling ;"" genitura, " producing ;"" " having.""
It must be noticed that in Gothic, also, from adjectives
spring feminine forms which are used as abstracts, as
mikiki, '* greatness '" (theme mikikiTi), from the adjective
base mikila, to which it bears the same relation that, in
Sanscrit, mndari, "pukhra''' does to the masculine neuter
base ?r«5x: sundara (see §. 120.) ; so, among others, also
managei, ** a multitude,"" from manag{a)s, " many ;"" siukei.

ydnta-Sy " cloud," as " making to flow," opposed to gadaydnt^ from gad^
'^ to flow," in the causal. So in Latin unguentum^ if it be not an extended
form of ^^unguen" (compare §. 803.), and perhaps argentum^ "silver," as
" shining " (Sanscrit rqfa-td-m), apparently from rd;^ " to shine," with
the vowel shortened.

1096 FORMATION OF WORDS.

** sickness,'' from «tiA(a)-» " sick/' (see Grimm, I. p. 608).
In Greek, too, there are a few adjectives, the feminines of
which represent abstracts ; in such a manner, however, as
that the latter is distinguished from the feminine adjective
by throwing back the accent, in agreement with what has
been before remarked on similar phsenomena in Sanscrit ;
hence, depfirj, " heat," kolkjj, " wickedness," opposed to depfx^,
Kaicfj ; as above, ydias, *' fame," opposed to yaSds, '* &mous "
(see §. 785. Remark) ; jdniman, " birth," mdrimant ** death,"
opposed to words like sarimdn, " wind," as *' blowing " (§.
547.). But to return to the Latin participles in ndu, secundus,
" the following one," has correctly retained the original
design of the suffix ; and the conjecture, therefore, that it
is a contraction of sequ^mndus is unnecessary : yet, in my
opinion, words in bundus in so far belong to this class, as
most probably the verb substantive is contained in them
in the same way as we have recognised it in the imper-
fects and futures in bam, bo (see §§. 526. 663.). When, how-
ever, Voss derives the forms bundu from the imperfect,
as, errabundus from errabam, vagabundiis from voffobar,
gemebundus from gemebam, he appears to be in error, as this
derivation is not supported by the sense; as gemdmndus
signifies, not " qui-gemebat,'*^ but " gemens.'"' I allow, there-
fore, between gemfbam and gemebundus only a sisterly re-
lation, and take bundu-s rather as the participle present of
the root fu, with the extension of the suffix rd to ndti, as
in the future passive participle under discussion. In Per-
sian the participle present of the root bu, "to be," would
probably be bavandah (for bu-andah, compare bavam, "I
may be ") ; and in Sanscrit from bhu really comes bhdvant,
" being " (base of the strong cases), to which the Latin
bundu, exclusive of the suffix u, has nearly the same rela-

* Regarding b for/, see §§. 18. 526.

FORMATION OF WORDS. 1097

tion as bam (ama-bam) has to d-bhavam* The first u of
bundu I take to be not the radical vowel of ^i*, but the cor-
ruption of an original a, as in the third person plural (veh-
M-nf= Sanscrit vdh-a-nti). As a proof that the forms in
bundur'S are, in their origin, participles, may be adduced
also the circumstance that they occasionally govern the
accusative; thus, in Livy, vitabundus castrci, mirabundus
vanam spedem. But should these forms originally belong
to a tense other than the present, we might recognise in
them obsolete future participles, and assume that the use
of the participle in turus has caused them to be less freely
employed, given room for their being dispensed with, and
changed their signification. An especial corroboration of
this view is to be found in the fact that the majority of
forms in bundus belong to the first conjugation, and that
in old Latinity futures in 60 occur also in the third and
fourth conjugation, a form which may originally have be-
longed to all classes of verbs ; as, as has been shewn, forms
like legam and audiam are nothing but present tenses of
the subjunctive mood, and used as a compensation for the
lost futures (see §. 692.). We should consequently regard
lascivibundus and sitibundus as analogous forms of old futures
like sctbo, dormtbo, only with the vowel shortened, as before
the sufl^ bundurs, with the exception of the d of the first
conjugation, only short vowels are found, and, therefore,
we have gemebundus, frenySmndus, opposed to dicibo, and
pudibundm opposed to ptJuMbit.

810. Let us now betake ourselves to the consideration
of those participles which, without any formal designation
of any temporal or lineal relation, have retained their desti-
nation in this respect merely by the use of language.
These are in Sanscrit the future participle in tdr, tri, the
perfect passive participle in ta or na, and the future passive
participle in ya, tavya, and aniya. The first-mentioned

participle, which is, at the same time, a noun of agency, has

4 B

1098 FORMATION OF WORDS.

been already discussed in §§• 646> 647 ; somewhat, however,
remains still to be observed on the subject. And first
must be noticed the coincidence in accent which exists be-
tween the Sanscrit and Greek, since the formations in m^
Mr, like the Greek in ti/jo, regularly accent the suffix ; thus,
dMdTf nominative dAtd (see §• 144.) dator and daiiirm, as in
Greek Sor^p; janitdrf nominative janitd ** genUor^ and
"5Pcnifttrti«"=7ei/€T)Jp. On the other hand, the suffix rop^
which in origin and signification is identical with Trip^ and
the long vowel of whose nominative Ta>/9, is to be regarded
only as a compensation for the want of the case-sign, has
lost simultaneously its organic length and its accent : it
admits, too, of scarce any doubt, that, in Sanscrit, the
weight of the suffix tdr is the cause of its being accented,
according to the same principle by which, in the second
principal conjugation, the heavy personal terminations
assume the accent (see §. 786. Remark). The Greek forma-
tions in TJ/^c, which in §. 145. have likewise been compared
with the Sanscrit in tdr, have, in part, remained true to
the old accentuation, since in forms of more than two
syllables a vowel long in itself by position, with <r gene-
rally, and occasionally also with k, p, v, and \ preceding
the suffix, serves like a dam to the accent which be-
longs to the suffix, and prevents it from receding farther
back ; hence, indeed, Mortis opposed to icrnip^ ddta; but /xax?;-
rrjs, iroiTynj^, O/Xwt^j, SiKO/crri^t i&KQVTi(rn]s, ^aoraicnys', tfwp-
/LUKTiyj, XvjjLavT^g, evdvvTrjg, TroiKtKrfjg, Kadaprtj^, opposed to
forms like yafieTYj^, yevsTf]^, itaviaK&TYi^. The e of forms like
YCv-e-T^/s", yev-e-TYipf TravSaic-e-Tiyy, is most probably a corrup-
tion of I ; for it corresponds to the f , which often occurs, in
Latin, and still oftener in Sanscrit, between the root and
the suffix ; e. g. yev-e-rrfp and yev-i'Tvjs correspond to the
Sanscrit ^an-z-Mr and Latin gen-i-tor.

811. In the weak eases the Sanscrit suffix tdr suppresses
its vowel, and the accent then falls on the case terminations

FORMATION OF WORDS. 1099

beginning with a vowel ; while before consonants the r be-
comes tij and the accent abides on the suffix ; hence ddtr-i,
" to the giver/' as in Greek irarp^o^,] trarp-iy for iraTep-og,
trarep-i, but ddtri-bhycts, " to the givers," The analogy of the
weak cases is followed also by the feminine of the noun agent,
inasmuch as before the feminine suffix i, which usually re-
ceives the accent, the vowel of the principal suffix is sup-
pressed ; hence ddtri, " the female giver." The Greek and
Latin, which possess over the Sanscrit the superiority of
retaining the vowel of the masculine suffix (rrip. Top, t&r)
through all the cases, follow notwithstanding the ana-
logy of the Sanscrit in suppressing, in the feminine
forms rpti, rpta, tri-c (see §. 119.), the vowel of the prin-
cipal suffix, and the Greek rpiS agrees with the Sanscrit
tri also in the retention of the accent, which the form
rpta (perhaps on account of its increase of syllables)
has abandoned ; thus, 7ig(rrpiS, oKerplSf avT^irplSf arffjLavTpiS,
\aXrjTplS, opxTtforrpiS, (rTeya<rrpi% as in Sanscrit ddtri. The
base yaarpi deserves especial notice, which, though also
masculine, is properly nothing but the feminine of ya<TTep,
nominative ya(rrrjp*, in which I think I recognise the San-
scrit root jas, "to eat," whence might be expected a noun
of agency jast&r, feminine jastri ; thus yaan^p, properly
"the male eater," and yaarpi-^ (properly "the female
eater") has indeed experienced a transposition of the accent,
but has kept clear from the inorganic affix of a 5. The
feminine bases in ti5 seem to me, where they appear as
nouns of agency, to be abbreviations of rpiS : they corre-
spond, as respects the loss of the p, to their masculines in
'^v{py^> but have throughout displaced the accent, even
where the masculine has retained it in its original site ;

* In shortening the vowel of the suffix, as also in declension, yaorcp
follows the analogy of the words denoting affinity, see §. 818.

4}B2

UCO FORMATION OF WORDS.

thus, not only iKert-^ compared with UeTri'Sf but also evperi-g
opposed to evpcTij'S.

812. The words denoting affinity in irc tdr, tri, are evi-
dently, in their origin, nouns of agency (see **Vocdlisnmsr
p. 182); for pit&r, weakened from patdr, and this again from
pdtdr, means properly ** nourisher,'* or " ruler," from the
root jpd; and rndtdrt " mother," I regard as " she that brings
forth;" while I dissent from the Indian Grammarians who
derive it from mdn, "to honour," and prefer deducing it from
the root md, " to measure," which, with the preposition nisy
"out of" (nir^md), signifies "to make," "to produce," and
even without a preposition is capable of this interpretation.*
Duhitdr, "daughter," signifies properly "suckling," from
duhy " to milk ;" ndptdr, " grandchild," is in its final ele-
ment essentially identical with pUdr, " father " (this, how-
ever, is perhaps opposed to my former opinion, see p. 387,
Notet), here not in the sense of " father," but to be taken
in its primitive meaning, while we regard the compound
not as a possessive but as a determinative ; so that naptdr^
in opposition to pitdr, as " ruler," or " family chief," would
signify the "not ruler," or "subject," and thus it might
mean any member of a family but the father ; as also in
the Veda dialect, napdi, which has preserved the original

^ I now find a strong confirmation of this opinion, which is elsewhere
expressed (" Fbca/twiiw," p. 182) in the Veda dialect in the First Book of
the Rig. Veda (Hymn 61. 7.)? which has been edited in the interim by
Fr. Rosen, where the genitive mdtur occurs as mascnline, with the
meaning " creatoris." The Old Persian furnishes the noun of agency
framdtdr (Jra preposition), which is connected in root and suffix with
9ndtar, the accusative of which, framdtdram^ occurs repeatedly in the
inscriptions with which we are acquainted, and is rendered by Lassen,
" imperatwremJ' I have no doubt that the above-mentioned Vedian mdtur
has an accusative mdidram (not mdtdram\ and that, therefore, the theme
is properly mdtdr, not mdtdr, as the d is shortened only in words denot-
ing affinity.

FORMATION OF WORDS. IIOI

length of the root jpd, signifies in the passages cited by Fr.
Rosen (on the Rig. V. I. 22. 6.) " son,'^ though in form it
corresponds to the Latin base nepdtf as also its feminine
naptiy " daughter," to the Latin base nepti , Old High Ger-
man nifti (nominative accusative nifl), Bhrd-tar, " brother,""
has clearly lost a consonant before the sufiix, for there is
no root bhrd. If, as the Indian Grammarians assume, the
root is bhrd^, "to shine," we must then observe that the
rdf, which is probably related to it, and from which Pott
deduces bhrdj (for abhi-rdj), signifies besides " to shine," also
*' to rule," and, therefore, ** the brother " may be so desig-
nated as ** ruler " in the family, which, according to Indian
manners, the eldest brother after his father's death really
is (see ^^VocaUsmusr p. 182). But bhrd, in bhraidr, may
also have sprung from the root bhar, bhri, " to carry," " to
support," by the transposition and lengthening of the radi-
cal vowel, just as in Greek from I3a\ : ISXrj-a-u), /Se^hrj-Ka,
)8\7-/xa, &c., from 7reT = Sanscrit pat, "to fall," "to fly"
(Tr/wTO) from TTiTrerco) : ttto) and tttj; (tttSktis, TtrS^iia, irrijaig),
and in Sanscrit from man, " to think," mnd, " to mention.

s^

« 'I

This feminine form leads to the conjecture that the masculine napdt
in the weakest cases (see §. 130.) rejects its a ; that, therefore, the geni-
tive would be napt-as, for napdt-as, since feminine bases in i generally
follow the analogy of the weakest cases ; as, r^.i^ <^ a queen," follows
that of rafnSy to the king/' rajn-as, *^ of the king," &c. Before termina-
tions beginning with a consonant, where napt would be impossible, I
should expect napdt; thus, napad-bhyas^ "to" and "from the sons," If
such forms were confirmed, I still could not assent to Benfey's (Glossary
to the S^ma Veda, p. 106) conjecture, that d in napdt, as also the 6 of
forms like datSr-U^ &c., is a lengthening that originally belongs only to
the strong cases, which, in Latin (nepot), has entered into all ciEises. It is
more natural to suppose the theme of the Sanscrit strong cases to be the
original one, and therefore, also, in the classical languages, for the most
part, carried through all the cases, as is the case in the example before us
with the suffix t^, njp, contrasted with the Sanscrit strong tdr (shortened
in the vocative to tar) and with the participle present in nt.

1102 FORMATION OF WORDS.

which is regarded by the Indian Gb*aniniarians as a dis-
tinct root. If, as now appears to me more probable, this
is the derivation of bhrdridr, viz. from bhar, in that case the
" brother "' is properly *' the supporter/' as the stay of the
mother, sisters, and younger brothers after the father's
death.* So the husband, also, in relation to the wife, who
is termed bhdryd Q* the female to be supported, to be
cherished '"), is " the supporter,"' and as such is called bliar-
tdr^ nominative bhartd; a word, the creation of which still
lies within the clear recollection of the language, and
which, therefore, in departure from its supposed cognate
bhrMar, follows the ordinary declension. The appellation
of ''sister," in Sanscrit svdsdr, has still preserved the long
vowel in the strong cases, but has, on that account, like
the Latin sordr from sost&r, lost a t, which has remained
in the German and Sclavonic languages (Gothic svistar,
English " sister," Old Sclavonic sestra), and in the Lithua-
nian sesser (nominative sessti, genitive sesser-s, see §. 144.),
has assimilated itself to the preceding s. Svd's(f)dr is
properly "the wife belonging" (regarding the pronoun
sva, see §. 341.), and is, in its final element, akin to stri,
"woman," which Pott is undoubtedly right in deducing
from the root suy m *' to bear a child" (E. I. I. p. 126) ; so
that, like fe-mina (see §. 478. conclusion), it originally sig-
nifies " the parturient," and is a regular feminine noun of
agency up to the loss of the radical vowel.

813. The shortening of d to a, which most words de-
noting affinity have experienced in Sanscrit and Zend in
the strong cases, appears to have existed so early as the
time of the unity of language, as it is scarcely fortuitous
that pUdram, pitar-du (Veda -rd), pitdras, stand in the same

* So in a passage of Sdvitrt (p. 16 of my translation of " Ths Deluge' ') :
'^ When the husband (of the mother) is dead, that son is culpable who is
not the protector of his mother."

FORMATION OF WORDS. 1103

relation to ddtdram, ddtdrdu (-rd), ddtdras, as, in Greek,
Ttacrepaj Ttarepe, Trarepe^, to Sor^pay Sorrjpe, ioTtjpeg, par-
ticularly as the Latin makes a distinction between the
declension of words like patera patris, and such as dator,
datdr-4s.

814. In the Veda dialect, formations in tdr, tri, occur
also in the sense of the participle present or future go-
verning the accusative ; and in this case the accent inva-
riably is thrown back from the suffix to the radical syl-
lable ; hence ddtdr, " giving,'" opposed to ddtdr, " giver ;""
pdtdr, "drinking,"' opposed to pdtdr, "drinker"' (Latin p6-
tdr-) ; hdntdr, " smiting," " slaying," opposed to hantdr,
"smiter," "slayer;" dstdr, "casting," opposed to astdr,
" caster." These participles serve principally to represent
the present indicative ; so that, as in the participial future
of the classic Sanscrit, the verb substantive is either to be
supplied or is formally expressed. The former is the case
if the participle refers to the third person ; the latter if
the first or second person is the subject. The forms of
this kind which occur in the Sama Veda are all in the
masculine singular nominative : and it is matter for future
investigation, whether the feminine also occurs in con-
structions of this kind, or whether, as in the participial
future of the classic Sanscrit, the nominative masculine
represents the other genders. I annex a few examples
from Benfey's edition of the hymns of the Sama Veda :
Hdntd yd vritrdn sdnitStd (-td vi6) vajan datd maghAni, "who
(Indra) striking (cleaving) is the cloud» and distributing is

* That in Zend^ also, the form m tdr occurs in the sense of a participle
present, and governing the accusative, is proved by a passage in the be-
ginning of the 1st Farg. of the Yendidad (V. S. p. 498), where 9?C^C5^ali

bactem is governed by ^7oa^ ddtkrd, " to the giving " (genitive in the
sense of dative, as is frequently the case in Sanscrit) : riemai etS ddthro
hacHeniy ^^ worship to thee the giver of happiness (riches)."

1104 FORMATION OF WORDS.

food, giving is riches ""=:** who strikes/' &c. (I. 4. 1. 5. 4.);
yd ddrityd iaiamdndya mnvafi ddtd jaritrd (euphonic for
.... tri) ukthydm, " who is giving that which is com-
mendable to the praise singer, who slays with care, and
expresses the juice of the (Soma)'' (II. 1. 1. 14. 2.) ; tvdshtd
nd ddtvyan vdchaK paijdnyd brdhniancLspdtih\ " Parjanyas
Brahm is creating for us godlike speech"* (1.4. 1. 1.7.);
dstd ''si sdtravi vatVidm, ** thou art hurling death at the foe"
(II. 9. 1. 13. a). I take pdtd as a future participle in the
following passage: pdtd vritrahd midm d ghd gamatf ** po-
turns yritri occisor sdnuB succum adeaV (II. 8. 2. 1. S.).*!*
As regards the cause of the retrogression of the accent in
these expressions, I have no doubt that the aim which the
language has in view is most emphatically to express, by
the accentuation, the energy of the action, which, in the
case where the form in tdr as a participle governs the ac-
cusative, appears in its full force ; and I am of this opi-
nion, as, as has already been remarked (see §. 785. Remark,
at the beginning), the accenting the initial syllable of a
word in Sanscrit is the most emphatic.

* Tvdshtdr is paroxyton also as a noun of agency.

t According to Benfey's translation^ " let the Vritra-slayer drink the
juice," &c., pdtd woiild=:pdtd sydt, " Mbens ait." I doubt, however, that
these participles can, without an auxiliary verb, represent the potential or
imperative ; for the indicative only of the verb substantive is, in Sanscrit,
very frequently omitted, as being by the sense itself understood. The en-
clitic ghd (for gha), which stands in the text in the common dialect ha,
which, as well as hd^ occurs in the Vedas, and attaches itself to pronouns
especially (see F. Windisclunann's Sankara, p. 73 ; and Benfey's Glossary
to the S^ma VMa, p. 206), gives me occasion to remark, that I now, in de-
parture from my former explanation (§. 175.), regard the Gothic Ar, and
our ch in mi-ky thu-k, si-k, mi-chy di^h, si-ch, as well as the Old High
German h in unsi-h, Vf^s, iwi-h, vfms, as a particle which has grown up
with the base, and as identical with the Sanscrit ha, gha, and Greek yc
(Dor. ^oL yd\ and therefore dich as=:Sanscrit tvdh-ha, Greek creyc? ^
in a phonetic point of view, tcA, Gothic ik=aham, ryw.

FORMATION OF WORDS. 1105

815. As to the origin of the suffix tdr, it may be re-
garded as springing from the verbal root tar ( W tfi)* This
root properly signifies " to overstep," " to transgress/' but
also "to accomplish," "to fulfil;" e.g. prcAijnAm^ "a pro-
mise." And it must be observed that several verbs of mo-
tion express also " to transact," " to do ;" as, char signifies
(l) " to go," (2) " to pass through," (3) " to do," " to prac-
tise," " to arrange." Thus, ddtdr, " dator, dans, daturas,''''
may be taken as ** the accomplisher," the " exerciser of
giving," or, also, if we keep to the primitive signification
of the root, as, " the man who passes through the action
of giving ;" as, pdraga, properly " going to the farther
shore," is used in the sense of " perusing." The verbal
roots, therefore, in combination with the suffix tdr, are to
be taken as abstract substantives, which cannot surprise
us, as some of them appear as such without any annexa-
tion of a formative suffix ; as, bhi, " fear," from bhz, " to
fear f hri, " shame," from hrt, " to fear ;" yudh, " strife,"
from yudh, " to strive." It may be requisite here to ob-
serve, that in Latin several formative suffixes beginning
with c can be traced back to the Sanscrit root Icar, hri

m

(with which creo is connected). Thus, for example, cri for
ceri — ^nominative masculine cer, feminine cri-s — and cru in
volucer, " flying," properly *' fulfilling the action of flying ;"
ludicer, hdicrw-s, "sport," "pleasure," "causing enjoyment;"
involu-cre, " that which envelopes or serves thereto ;" lava-
crw-m, " that which makes to bathe," " to bathe ;" amhvla-
cru-m, " that which makes to walk out, gives occasion
thereto," hence " promenade ;" sepul-cru-m, " that which
makes to inter," " a grave ;" lu'Cru-m, " that which causes
to pay," " gain ;" faUcru-m, for jvlc^cru-m, " that which
makes to support," " a support." As r and / are easily

* Compare Benfey, " Greek Etymology^' II. p. 267.

1106 FORMATION OF WORDS.

interchanged, I have no hesitation in referring to this class
also the suffix cuhh and comparing it with the Sanscrit
kara, ** making ;" * thus, ridicu-lu-s, properly " making to
laugh f ' picHmlu-m, " that which makes to atone ^ specta-
cula-m, " that which makes to see,'' " gives to see ;" vehi-
ddu^m, " that which makes to ride ;" pd-culu-m, " that which
makes to drink ;" mira-culu-m, " that which makes to won-
der ;" ba-culU'S, " that which makes to go " {^i^rniiy e^ti-v).
816. From tdr springs, in Sanscrit, by the affix of an a,
and with the suppression of its own vowel, as in the weak
cases, and before the feminine character iy the neuter suffix
troy and thence the feminine trd^ The neuter form is prin-
cipally used, and, like the feminine trA, of rare occurrence,
forms substantives which express instruments, which are,
as it were, the inanimate accomplishers of an action. They
Gunise the radical vowel, and, for the most part, in accord-
ance with the Greek analogous forms in rpo, Spo, rpa, dpa^f
accent the first syllable of the word. The following are
examples : rJ-tra-my " an eye," as ** conducting," or " in-
strument of conducting" (root ni); sro^tra-m, "ear" (root
irw, "to hear"); gdrtra-mt "limb" (root g&, "to go");
vas-tra-m, "garment" (root vas, "to put on"); sds-tra-m,
"arrow" (root sas, "to slay"); ySk-tra-m, "band" (root
1/uj, " to bind ") ; d&nshPrd, " tooth " (root dans, " to

* At the end of compounds hhds-kara-s, "making brilliance," "the
sun;" bha-yan-kara-Sy "making fear," "formidable."

t It is a question whether the 6 of dpoy 6pa, is produced by the in-
fluence of the p, in analogy with the law of sounds in force in Zend (see
§. 47- )> or whether independently of the p a change or weakening of the
tenuis to the aspirate has taken place, as has become the rule in Ger-
manic languages (see §. 87.)' The latter appears to me more probable, as
the combination rp is very usual ; but 6 for an original r occurs also
before vowels, as in the suflix ^fi/=Sanscrit tas^ Latin tus (§. 421.), and
in the personal terminations of the middle and passive which begin with
aO (see §. 474.).

FORMATION OF WORDS, 1107

bite''); ydtrd, feminine, "provisions" (root yd, "to go'').
So in Greek, viTrrpo^v, irT^KTpa^v, fidtcTpo-v, KeKTpo-v (" bed," as
"means of lying**), ^aicTpo-v ("stick," as "means for going"),
fco-c-TjOo-v, aporpO'V, deKyijTpO'V, ipiKijTpo^v, Bhtrrpo'V, Bfjpa-
rpo'V, apOpor-Vf ^ddpo^Vt Kelfiffipo^v, fiafcrpa, 7r/-<r-Tpa, KoAvir-
Tpo, ^adpa, KpefiadpoL The suffix in the class of words under
discussion is, in Sanscrit, seldom accented, and still more
rarely in Greek : the most common Sanscrit word of this
kind is vaktrd-m, " mouth," as ** speaking," or " instrument
of speaking ;" so paktrd-m, " holy fire," properly " that
which cooks" (root pack from pak); dhartrd-m, "house,"
as "holding," "receiving" (root dhar, dhri); vitrd-m, "a
reed," as " moving itself" (root vi). In Greek, \ovrp6-v
and SaiTpo-v belong to this head. The latter, by its pas-
sive signification, corresponds ("the distributed") to the
Vedian ddtrdm, " gift," as " that which has been given," or
" is to be given." As respects its base syllable, how-
ever, iairpov {$al(ji}) belongs to the Sanscrit root d6 = dd,
"to cut ofi^," whence ddira-m, "a sickle." As the suffix
tdr, in Sanscrit, is occasionally preceded by an i as conjunc-
tive vowel, so also is tra, and then either the conjunctive
vowel or the base syllable is accented : the former in khan-
{•tra-Tfiy "a spade" (khan, "to dig"), the latter in vdd-i-
tra-m, " a musical instrument," properly " that which
causes to speak or utter a sound" (root vad, " to speak,"
in the causal) ; gdr-i-tra-m, " rice," properly " that which
causes to eat," "nourishes" (root gar, gri, " deglutire,^'' in
the causal). As we have above (§. 810.) compared the
Greek e of forms like yev-e^r^p with the Sanscrit-Latin
vowel of conjunction i of the corresponding jan-i-tdr, gen-i-
tor, so may also the e of ipep-e-Tpo-v be taken as the cor-

* Benfey quotes in his Glossary to the S&ma V^ds, p. 88, the follow-
ing passage of the Rig. Veda : dsi bh6g6 dsi ddtrdsya ddtd^ 'Hhou art the
Lord : thou art the distributor of alms/'

1108 FORMATION OF WORDS.

ruption of i, and the said word be contrasted with Sanscrit
formations like khan-i-tra-m and rddri-ira-m. It may, how-
ever, be the case, that the e of fpip^-rpov is identical with
the class-vowel e of ff>ip'e^e, tf^ep-e^ov^ &c. ; then (pify-e^pov
would correspond to Sanscrit formations like pdt-artram,
"wing,'^as "instrument of flying;^ vdcUi-^Or-tra-m, "weapon,''
as " slaying f' krint-a-trcL-n^ "plough,'' as "cleaver" (root
krit from kaH, in the special tenses krirdt compare Ketpca) :
for which, indeed, the Grammarians assume a suffix oilTa,
the a of which, however, appears to me identical with the
inserted vowel of the first and sixth class ; thus, pat-ar-tra-'mf
like pdi-d-ti, " he flies ;" krint-a-tra-nh like krird^a-ti, " he
cleaves.**'* Thus in Greek the rj of forms like tfylTitj-rpo-v and
KopYi'Bpo'V evidently belongs to the verbal base, and is iden-
tical with that of ipi}Jf<T(a, Koprj-coi. The same is the case
with the d and 6 of the corresponding class of words in
Latin ard-tru-m, JvlgS^tru-m, fulgi-tra, verS-tru-nh where it
must be observed, that, according to §. 109*. a., the d of the
first as well as the 6 of the second conjugation are identical
in their origin with the i; of the above-mentioned Greek
forms. As, however, the i of the second conjugation is
less permanent than the d of the first and the i of the
fourth (see §. 801. Note), we cannot be surprised to find,
not mrdgi'traf mulgi'-trum, but mulc-tra, mulc4rur-m ; not
moni-tru-m, but mon-s-trufn. The s of monstrum corre-
sponds to the euphonic 8 mentioned in §.95. A similar
one is also to be found in luns-trum and fivrs-trum. Vi-trumy
" glass,'' as it were, " instrument of seeing,'' or " making
to see," has lost the d of the root We should have ex-
pected viS'trum (see §. 101.) according to the analogy of
ras-trum, ros-trum, daiis-trum, cas-trum. In the third con-

* The Indian Grammarians include the i of the above-mentioned words
in itra in the suffix.

FORMATION OF WORDS. 1109

jugation, the class syllable of which has, from the time of
the unity of language, as a rule not extended itself beyond
the present, with its derivatives, and the imperfect, the
suffix is joined, for the most part, direct to the root, e.g.
rurtrum, spec-trum. In the fourth conjugation we should
expect i'trum, in accordance with d-trum in the first, and
i'trum in the second; but hauS'trum, from hauris, is in
conformity to the other anomalies of this verb.

817. The Zend has, according to §. 47., changed the t of
the suffix tra into th, but leaves it unaltered after sibilants,
which, in general, do not admit of th after them ; hence
Aj7oAy^«bi^Aj;CL^ yaoschddihra, ** means of purification " (V. S.
p, 263), nominative accusative -thre-m (see §. 30.) : ddUhre-m,
"eye" (as " seeing '")> is connected in its root and suffix
with the Greek dearpov, although the meaning of the latter
has taken a difierent direction, since it signifies the place
which affi>rds the spectacle. The corresponding Sanscrit
root is most probably dhydi, with which Pott Q'E. I. /.'^
p. 23l) has been the first to compare the Greek deaofiat,
although dhydi signifies not ** to see '' but " to think,"' where
it is to be observed that ^ budh, " to know," has, in 2^nd,
received the meaning of " to see," as flre vidf ** to know,"
has in Latin, while the Greek root IS (etSco, otSa) unites the
two meanings. Remark, also, with Burnouf ("YagnOf'*'*
p. 372), the New Persian root di, "to see" (infinitive di-dan)*
and the contraction which the Sanscrit root dhydi has ex-
perienced in the substantive dhi (nominative dhi-s), " un-
derstanding," " insight." The following are examples in
which the suffix spoken of has preserved its original tenuis
under the protection of a preceding sibilant : vastrem, "robe,"
feminine vastra (see §. 137., Sanscrit vdstra-m, see §, 721.

* The present Mnam belongs probably to a different root, and, in fact,
to the Sanscrit vid.

lllO FORMATION OF WORDS.

Note **), and as7^j)au9 iMtra (as theme), ** the willow/' as
"growing^"* (connected in its root with the Old High
German base wahs-a'-fnonf ** shrub," " fruit,'' see §. 799.),
whence the often occurring vdstravat, "willowy,'' as also
vdstrya (nominative -yd), "farmer." The Zend uses the
formations in thTa, trof also in the sense of abstract sub-
stantives, which, according to what has been said (§. 809.)
regarding the radically connected Latin formations in tura,
cannot surprise us. The following are examples : ^g7(3j^
dar-i-thre-m, "possession," "reception," "retention" (San-
scrit root dhar, dhri, "to keep"); 5f7<3f7Aj9 mar-e-threm,
"mention" (Sanscrit root smart smru "to remember");
9^7^Au^ kh&thremj " splendour ;"f 9g7^J9jai^ khds-trem.

* I doubt not that this expresmon comes from the Sanscrit root vaksh,
''to grow/' which, in Zend, in the devoid of Guna special tenses of the
fourth class, is contracted to uc8. With respect to the suppression of
the guttural in the above form, compare the relation of the Sanscrit
chasJ^tSy "he says," to the root chaksh, and the Zend ckashman^ "eye"
(as "saying," "announcing"), to the same root, and to the cognate San-
scrit word chdkshua.

t At the end of compounds pSururkhdthra^ " having much lustre"
(see Bumouf, " Yagna^" p. 421). I consider khathra to be an abbreviation
oiMiarthra {kharethra^ according to §. 44.), and derive it from the root
Wiar, "to shine," whence, also, H^/P^as^ hharend^ "lustre.* The root
mr (from «?ar, see §. 36.) corresponds in Sanscrit. The loss of the final
consonant of the root appears to be compensated by lengthening the vowel,
as in the Sanscrit J4^<f, " bom," from jan ; khatdy " engraven," from khan.
Observe, also, the relation of the Zend j( joi^aj^ zaxdmi^ " I produce," to
the Sans. J4/anmt. Bumouf gives another derivation ofkkdthra^ " lustre "
(1. c. p. 419) dividing it into kha^ ^^stmsy" and dthra^ accordmg to which
its literal meaning would be ^' suum ignem hahens^*' suxd therefore dthra
would be connected with the word dtar^ "fire," which is used in its un-
compounded form, and the a of which is suppressed in the weakest cases ;
hence dthr-at, " igne ;" dthrahm^ " ignium" Bumouf touches also on
the possibility of the prefix ^ m, hu, " fair," being contained in khdihra^
in which case its proper signification would be '^pulchrum ignem habens."

A

FORMATION OF WORDS. 1111

" taste." The latter Burnouf (" Va^a:' p. 220) derives,
undoubtedly with justness, from the Sanscrit root svdd : the
transition of d into s is here quite regular (see §, 102. con-
clusion) ; and khdstrem therefore resembles, both in the
euphonic treatment of the radical d and in the suffix, the
(§. 815. conclusion) above-mentioned Latin formations, as
claus-trum.

818. As regards the formation of abstract substantives
through the suffix under discussion, the German languages
admit of comparison with the Zend in several interesting
forms. The Gothic furnishes us •with the neuter base
maur-thra (nominative accusative maurthr, see §. 153.), " mur-
der," properly " the killing," the obscure root of which
leads us to the Sanscrit mar, mri, '* to die," causal mdrd-
y&mh " I slay." * Besides the above, J. Grimm (II. p. 123)
deduces from bldstrels a neuter blddr, "oblation" (theme
bldstra), which I should be glad to admit did it anywhere
occur. Nevertheless, I think its existence must be as-
sumed, and I derive from it the existing masculine bids-
trei-s, the base of which, bldstrya (see §. 135.), has the same
relation to its presupposed primitive base blddra that the
previously mentioned Zend vdstryd (theme vdstrya), " coun-
tryman," has to its primitive base vdstra, " pasture." f The
root of the Gothic base bl6s-tra is blM, " to sacrifice," " to

A derivation, however, in which khdthra would etymologically also sig-
nify what the sense requires, and according to which it would be radically
identical with a word (khareno) literally meaning "lustre," appears to

me the most natural.

* The u of the Gothic form is a weakening of a, to which, according
to §. 82., an euphonic a is prefixed. As most of the German languages
have lost the r of the Gothic maurthr, and consequently the agreement
between them in sufiix with the primitive sufl&x tra, thra, is no longer
recognisable, we should remark with care the English "murder.*

f It is a rule in Sanscrit that verbal bases terminating with a vowel
reject their final vowel before vowels or y in an annexed derivative suffix.

1112 FORMATION OF WORDS.

worship,"' whence, aceording to §. 102., blds-trOt in analogy
with the Zend hMh-tra, " taste," from hhAdrtra ; so gils-tra,
"tax,"' nominative accusative gibtr, from gild-4ra, gild-tr,
from the weakened form of the root gold, with the prepo-
sition us and frOf " to repay." The a of the Old High
German geb-tar, keh-iar, gheh-tar (Graff, IV. 194.), I take to
be an auxiliary vowel inserted to remedy the harshness of
an accumulation of consonants at the end of a word, and
which, on the annexation of the case-terminations in these
and similar words, is again dropped, hence genitive plural
gfieb'iro; so from bluos-tar, blos^tar, "oblation," dative blos^re;
from JUah'tar, " laughing," " laughter," dative hlah'tre.f
We have, therefore, in the common German expression
Ch-ldch-ter, as also in the English ** laugh-ter," analogous
forms to the Zend abstract neuter bases in thra^ tra, as also
to the Sanscrit formations in <ra, Greek in r/oo, and Latin
in tru. Thus in English also " slaugh-ter," which in its
radical part, graphically at least, is more perfectly retained
than the cognate verb " slay." Probably, also, " thun-der"
and " wea-ther" are to be included in the class of words
which are formed in Sanscrit by the suffix trOi though the
^-sound of the suffix is lost in the appellation of " thunder"
in the older dialects (Old High German drniar masculine.
Old Saxon thunar, Anglo-Saxon thunor) ; on the other hand,
in Latin we have ton-i-trust ton-i-trUf where the u of the
fourth declension is matter of surprise, as the Sanscrit a

* With respect to the interchange of t, th, and d (hUs-ira^ gih-tra,
compared with maur-thra), in suffixes originally commencing with /, I
refer the reader to §. 91.

t Whether the gender be masculine or neuter is not to be determined
from the cases which occur (accusative hlahtar^ dative hlahtre and hlah-
tere) ; as, however, the perfectly analogous blSstar shews itself, by the ac-
cusative plural bldstar^ to be neuter, I agree with Graff (IV. 1112.) in
considering hlahtar also as neuter, in accordance with the analogous Gothic
and Zend forms.

FORMATION OF WORDS. 1113

would lead us to expect only the unorganic u of the second
declension (see §. 116.)* The corresponding Sanscrit root is
stan, "to thunder/*" whence stan-ayi-tnilrs, "the thunder."
"Weather"' belongs to the Sanscrit root vd, "to blow,''
whence also the Lithuanian we-trch " storm." To return
to the Gothic; fd-dr, "sheath" (theme /d-dra), and huH-s-tr,
"veil" (theme huli'S-tra), belong to the class of words here
discussed. The latter proceeds from the verbal base hvi-ya;
its if therefore, is the contraction of the syllable ya, as in
the preterite hd-i-da. I regard the s as an euphonic affix,
as in the Latin IvrS'trurm, flu^s-trum (see §. 815. conclusion),
capUs'trum, The following nominal derivatives are ana-
logous : avi'-s-trf " sheep-cote," as " place of the sheep,"
from the lost primitive base am (= Sanscrit am, Lithua-
nian awi) ; and navi-s-tr, " grave," as " place of the dead,"
from miuSf theme navat with the weakening of the a to i,
as in the genitive nam-s (see §. 191.) Observe that the
Greek and Latin languages very frequently transfer the
suffixes of verbal derivatives to nominal derivatives. JPd-dr,

* Ay]a the character of the tenth class, and itnu the suffix, which forms
adjectives with the signification of the participle present and masculine ap-
pellatives ; as, harshayitndLSj '' rejoicing," and as a substantive masculine
"son," as "the causer of rejoicing" (so nandana, "son," from nand "to
rejoice"). The % is evidently merely a vowel of conjunction, as in the
future stannoy-i-shydti, " it will thunder :" there also exists, as well as
i-tnu^ a more simple suffix tnuy as in hatn&s, masculine^ ''sickness," and
**a weapon," as "slaying," from Aaw, "to slay." The t oftnu and itnu
may be regarded in the same light as the euphonic t mentioned above
(§. 797. Note) ; so that, therefore, only nu would be left as the true suffix,
as appears in thdnd-a, " sun," as " giving light." The circumstance that
the Latin ton-i-iru-8, fon-t-^ru, stands in the class of words under discus-
sion in a very isolated position, owing to its t« of the fourth declension,
may lead us to compare it, with respect to its suffix also, with the San-
scrit stanayiin&s, by assuming an exchange of the liquids ; so that tru
would stand for tnu, just as in the Latin puUmd {for pln-md) an / stands
over against the Greek nasal ofirvev/juov (compare ^. 20.)

4 c

1114 FORMATION OF WORDS.

" sheath/' theme fd-droj in its obscure root corresponds to
the Sanscrit pA, " to receive," and in its entire form to
pdtra-m, " vessel," as " keeping." With respect to the
Gothic df for the tK which was to be expected, compare
fa-drein, "parents," with the Sanscrit pi-tdrdu (for pa-),
which is also radically connected with fd-dr (see §. 812.).
The Old High German y8-tor, fuo-tar, " fodder" (for fd-tr,
Anglo-Saxon, fS-dr, fd-dher, fo-ddar, fo-ddur) is identical in
root and suffix with the appellation of *' sheath," which
" supports," but only in a different manner from that in
which "fodder" does. To this class of words may be
added, with more or less certainty, a few other Old High
German neuters which end, in the nominative and accusa-
tive, in tar or dar : viz. fla-dar, " float," from the root flu
(ss Sanscrit pZu), which has generally assumed the affix of
a z (seel09^l•); fld-dary " fluorf""* from the same root;
ruo'dar, *' rudder," apparently as " making to flow or navi-
gate," in root and suffix akin to the Latin ru-trum, and
Greek pe-dpov {pia} from <Tpe{F)<i^f Sanscrit srdvdmif from the
root srUf " to flow," causal srdvayX and radically, perhaps,
also with re^mus.'f Perhaps, too, we ought to class here
wundaVf wuntar, " wonder," and fDuMaVf " glory," t as deri-
vatives from roots now unknown.

819. To the Sanscrit feminine suffix trd, as in ddnshtrd,
"tooth" (see §. 815.), corresponds the Gothic thld, in nethld
(nominative and accusative nStJda), "needle," as "instrument
of sewing ;" as in the Greek cuciarpa, but with Z for r ; which,
according to §. 20., cannot surprise us, particularly as the
Greek suffixes tAo, 0\o, tKyi, SKyj (see Pott, II. p. 556), are

* The Sanscrit form for flu-dar^flS-dar, would beplS-tra-m {6=au)*
t Graff, II. p. 493, presuppoBes a root rod ; but the Anglo-Saxon rovan^

reavaUy revan, " rendgare" mentioned by him, proves the contrary, and

answers to the Sanscrit causal base trax&y,
\ Gothic vulthus, probably with thu, = Sanscrit tu^ as suflSx.

FORMATION OF WORDS. 1115

likewise evidently to be referred to the Sanscrit tra, trd ;
as in ^-e-T\o-i/, yy^Xo-Vy dv-tr-OXo-v, e)(^€'T?\rf, yev-e-OTsaj,
"Ox-e-Tho-v, in a Sanscrit form would be perhaps vah-i-tra-m,
or vah-a-'tra-m. With regard to *^eve&hyi as an abstract
substantive, I must remark, that in Sanscrit also the femi*
nine suffix tr& is occasionally used to form abstract sub-
stantives ; thus, the y&trA mentioned above (§. 815.) means
also "gait.*" In Old High German the word for "needle"
exhibits in the nominative and accusative, in different
writers, nA-dla, ndrdih, nd-^la, and ndrdal: the Anglo-
Saxon form is wa?-di We have further to mention, in
Gothic, hleithra (theme -thrdX *' a tent,"" which has retained
the old r, though its root is obscured : it belongs, in my
opinion, to the Sanscrit sri from kri, "to go"" (compare
v^sman, " house," from vis, " to enter"), whence A-sraya-s,
** asylum," ** house," and in Gothic also hliya, masculine,
(theme -^yan), " a tent." To this root belongs also, among
other words, the Old High German hlei-tara (for hkitra)^
(which, on account of its suffix, also belongs to this class),
Anglo-Saxon hkedre, Mce-der, German Lei-ter, "ladder," as
" instrument of mounting."

820. Let us now consider somewhat more closely the
perfect passive participle, which we have already had oc-
casion to mention more than onccf Its suffix is, in San-
scrit and Zend, usually ta (masculine and neuter), feminine
id, and is, I have no doubt, identical with the demonstrative
base ta (see §. 343.). There is no ground, therefore, in the
word itself for a passive signification, except, perhaps, in
the accent ; for while, according to §. 785. Remark, the ac-

* Graf (IV. p. 1115.) quotes for the nominative the forms leitra,hleitar^
leitera, letter^ genitive hleitra. It admits of no doubt, that the forms in r
have lost a final a, and that they cannot be classed with muotar^ tohtar,
sttestar, of which the proper termination is r.

t See §§. 513. 588.

4fC2

1116 FORMATION OF WORDS.

tive forms require the most powerful accentuatioD, i.e. the
accent on the first syllable ; in the passive participle under
discussion the suffix receives the accent: hence we have
^ paktdst ** codvsr accusative paktdm, standing similarly op-

posed to pAx^han, " coquenSf^'' pdchantam, " coqaerdem^ as
above (§. 785. Remark) suchydd, *' purificaiurj''' is opposed
to iuchyalii *' purificaC Greek verbals in ro-r, which, as
scarce needs to be noticed, are identical with the perfect
participles passive of the cognate languages, have retained
the old accentuation, and thus we have TroTo-^^ ttot^, itorov,*
standing in the same relation to vorrog, "the drinking"
(compare §• 785. Remark, near the end), that, in Sanscrit,
piydtS, " bOntur:' has to piycii (Class 4, middle), " bOnC
The paroxytone or proparoxytone accent of abstracts in to
appears to be preserved principally where, together with
the abstract, the passive verbal is actually in use, and where,
consequently, there is the more ground for bringing the
abstract meaning prominently forward by the accent ; whilst
otherwise the abstract follows in its accentuation the pre-
vailing example of verbals with passive signification ; hence,
indeed, woToy, aporos, aixifro^^ Tpvytfros, efiCTo^, oKero^, op-
posed to TTOToj, apoTos, dfujTos, TpvyfjTog, e/x6TQf, oAerof (o?^-
rov) ; but not icoTreTOf, iccokuto;, oXoj/toj, but Koveros, KcuKtrrog^
dKofyrog^ as these abstracts have no oxytone passive verbals
to match them. There are, however, some isolated abs-
tracts, or words which express the time of an action,
which have the accent thrown back, as ^toro^, Se/Tri/jy-cr-Toj".
821. The participial suffix K ta is either joined direct to
the root or by a vowel of conjunction ». To the first kind
of formation belong jud-tds, " known " = Greek Yvco-r^r ,

« Compare the Sanscrit pttds^ pttd, pttdm, from the root pd, ^^ to
drink ;" which, in the passive, has the d weakened to I. There is also a
middle root p$ of the fourth class.

FORMATION OF WORDS. 1117

Latin (g)n&-lu-s, i-gnd-tiirs ; dat-td-s, "given/'* Zend dd-t6
(theme ddta), Latin da-tws, Greek So-to-^ ; sru-td-s, ** heard/'
Greek icAu-to-?, Latin clu-tu^s ; bhu-td-s, " been/' " being,"
Greek ^v-ro^ ; bhri-td-s (from bhartas, see §. 1.), " borne,"
Zend heretd (theme -to), Greek (^c/o-to-j) a-t^e/o-TO-y, Latin
fer-tus, " bearing," " fruitful ;" strUd-s, " extended " (from
startds), Zend fra-stdreto (fra preposition), Greek arpa-To-g, ^

(transposed from o-ra/o-Tor), Latin strd-tu-s; pak-tds, '* cooked,"
Greek TreTr-ro-y (root ttctt from we/c, Sanscrit pocl^, from pak,
Latin coct from poc), Latin coo-tts^s ; uk-td-s, "spoken" (irre-
gular for vaktds), Zend udo {huctd, " well-spoken " (from hu-
uct6) ; f/uk-td-s, " bound," Greek feuK-ro-j, Latin junc-tus ;
bhrish'td-s, '* roasted " (from bhrashfds, and this from bhrak-
tds), Greek ^pvK-To-^, Latin fric-tus ; bad-dhd-s, " bound "
(euphonic for badh'td-s, root bandh), Zend bas-tS j'^ lab-dhd-s,
obtained " (euphonic for labhtds), Greek Xtjir-^o-g ; jd4d-s,
bom" (root yan), Zend zd-td, Greek Ye-ro-r, in the com-
pound Tt]\vyeTO£ ; % matd-s, "thought" (root man), Zend matd,
(compare fxev-e-Tog); dish-td-s, " shewn" (euphonic for dish-
tds, from dik-tds, see §, 21.), Greek (SeiKTog) dvairoSeiKTo^^
X^ipoSeiKTo^, &c., Latin dic-tus ; dash-td-s, "bitten" (eupho-
nic for das'tds, from dak-tdi), Greek (S)7k-to-j), aSrjicrog, KapSto-
SrjKTo^; drish-td'S, "seen" (from darshtds, and this from

4(

«

* From daddtas^ with irregular retention of the reduplication of the
special tenses.

t See §.102., and compare Greek analogous form, as Keardg, maros.
With regard to the Latin form of this participle in roots with a T sound
see §. 101.

t It is a rule in Sanscrit that before formative suffixes beginning with
t, which require no Guna augment, the n and m of the root are rejected ;
jarij " to produce, to bear," and khan, " to engrave," lengthen their vowel
in doing this. From han, " to smite, to slay," comes Mtds, with which
we may compare the Greek ~<l>aTos, as <^£NQ {<f>6vosy hrcffivov)^ like 6vrj<rKa>y
most probably belongs to the Sanscrit root han, from dhan {nidhana,
"death").

1118 FORMATION OF WORDS.

dark'tds), Greek (SepicTos), eirtSepicTo^ ; ush-td-Sf " burnt,'' La-
tin m-tus. The following are examples with the conjunc-
tive vowel i: prat-i-td's/' extensus''' (root inprath, whence
pritkw-Sy ** broad," from prathii-s, Greek TrAarJ-j, Lithuanian
pla-tu-s) ; anchr-i-td-s, " eredus,^ pat-i-td-s, " qui cecidUr So
in Latin, dom-i-fiLS, mon-d-ttis, mol-i-tus, genr-i-tus. In Greek
the e of forms like fxev-e-roSf <r#ce\-€-Toy, e/OTr-e-roy, corre-
sponds, where we again leave it undecided whether this t
be a corruption of an i or an a.'f

822. The Latin forms in idits, springing from neuter
verbs, and for the most part of the second conjugation, as
paHri'duSy ferv-i-^us, frig-i-^us, torr-i-duSi tim-i'dust tep-i-dm,
splend-i'dus, nitH-dus, luc'i'dris, fidg-i-diLS, mv-i-dus, sap-i^dus,
flu'i-dus, correspond to the participles in td in Sanscrit,
which spring from neuter verbs, and have an active signi-
fication, and especially to those which have a present
meaning ; as, tvar-i-tds, " hastening,'! sthitds, ** standing,"
suptds, "sleeping" (also "having slept"), kaktds, "being
able," J yat'tas, "striving," bhi-tds, " fearing," hri-tds, "being

* Regarding the active signification of this participle in neuter verbs
see §.513. conclusion; so, in Greek, orardy, "standing," = Sanscrit sthi-
tds (weakened from sthdtds), which likewise signifies present time : on
the other hand pra-sthitds means both ^^ proficiscens*' and ** profectus."

t Compare §. 815., and Curtins " De Nominum Grcecorum formatuyne"
pp. d8, 60. Indian Grammarians assume a suffix (uxtddi) aid, the a of
which, however, is most probably only a class-vowel, with which the
Greek c might be compared ; thus, ifyir-e-ros (compare ifm-c-re) like
pach-a-tds^ "fire/* as "cooking." The abstracts Bdv-a-ros, "death," and
Kdfi-a-Tos^ "fatigue," have preserved the conjunctive vowel in its original
form^ and thus correspond to the Sanscrit mar-a-td-Sy " death ;" where,
however, we must observe that the Sanscrit root mar, mriy " to die,"
in its verbal conjugation, does not belong to the first or sixth class any
more than the Greek roots 6av and jca/i.

X The form with the conjunctive vowel (iak-i-tds) has a passive signifi-
cation, so yat'i-tds, " obtained by efibrts, sought for," compared with i/at-
tds, " striving." In Latin, vice versd, rap-i-dtis^ active, opposed to rap-tus,

passive.

FORMATION OF WORDS. 1119

ashamed ;" and to the Greek oraTo^, " standing ;" fxevcrosy
" remaining ;" efntero^^ ** creeping."' The opinion, there-
fore, elsewhere stated, appears probable, that the d in the
Latin forms just mentioned is only the weakening of an
original tenuis, just as in quadraginta, quadrupbiSf quadrur-
plex, for qimtraiginta, &c. An active and present meaning,
though in a transitive verb, and with the retention of the
old tenuis, occurs in the participle spoken of in fertits,
" bearing," " fruitful,"' which corresponds in form with the
Sanscrit bhritas, from bhartds, "borne,"' Zend beretS, and
Greek -0€|t)Tos (see §. 818.)-

823. The Sanscrit verbs of the tenth class, and the
causals identical with them in form, have all of them the
conjunctive vowel i ; hence pid-i-tds, " pressed," " pained ;"
v&S'i-td'S, ** made to enter." The circumstance, however,
that the said verbs extend their character ay (in the special
tenses aya) to the universal tenses also, and a great part
of the formation of words, gives room for the conjecture
that the i of forms like ptd-i-tdSf vii-i-tds, is not the ordinary
vowel of conjunction, but a contraction of ay ; or that such
forms in i't6-8 have been preceded by older ones in ay-i-tas,
^ according to the analogy of the infinitives, as pid-dyi-tum.
As, then, Latin supines like am-d-tum, aud-i-tum, are
related to pid-dyi-tum, just such is the relation of am-d-tuSf
aud't-tuSf to the presupposed pidrdyi-tas. Although the
Latin second conjugation also belongs here, and, for ex-
ample, moneo corresponds to the Sanscrit causal mdn-dyd-
mi and Prakrit mdit-^-mi (see p. lio), I would nevertheless
prefer to identify man-i-tus with mAn-i'tds in such a way

passive. Observe, also, the active cup4-dus together with the passive
cupA'tus, These, however, are only arbitrary usages, which rest on no
general principle.

* Influence ofProrumns in the Formation of Words, pp. 21, 22. Pott is
of a different opinion, E. I. M, p. 567.

1120 FORMATION OF WORDS.

that I could thence infer the existence of similar forms in
the time of the unity of language, while I would prefer as-
suming a casual coincidence in the similar abbreviation of
a common element. In Greek the >/ or to of forms like
{piK-yj-Tog, rtfi-fj'To^ (from rtyL-orro^)^ ^e/p-w-Toj, corresponds to
the character of the Sanscrit tenth class, and therefore to
the Latin A and i of am-d-ftis, aud't-tus. In Gothic, where,
as generally throughout the German languages, this parti-
ciple remains regular only in the so-called weak conjuga-
tion, the old tenuis, instead of, in accordance with §. 87.,
becoming an aspirate, has sunk down to a medial, in such-
wise, however, as that before the s of the masculine nomi-
native, and in the accusative, which has lost the final vowel
of the base and the case termination, a th for d enters
(compare §. 91.). According to the difference of the con-
jugational class, an i (from t/a), A or at, i.e. the three dif-
ferent forms of the Sanscrit character of the tenth class (ay,
see §. 109*. 6.) precedes ; hence the bases tam-i-daj " domi-
tus r frty^'dOi^ " amatus f" ga-yuk-ai^da, " svbjugatus T
nominative masculine tamiths, friydths, gayukaiths ; accusa-
tive tamithy &c. ; genitive tamidi-s, &c. (see §. 191.). The
direct annexation of the participial suffix occurs in Gothic
only in certain irregular verbs, and in such a manner that,
according to the measure of the preceding consonant, either
the original tenuis is preserved, or has become d (see §§.
626. 91.). Thus the base bauhta,t " purchased "" (bugya,

* Compare Sanscrit dam-i-tds (from dam-ayi-tdsT) from dam&ydmi,
causal of the root dam^ *^ to tame," but of the same meaning as the primi-
tive and the Latin dom-itus.

t It may be regarded as the denominative of the Sanscrit jpny a, " dear,"
^^ beloved ;" and it is also,, radically and in its formation, akin to the
Greek (^(X-i;.rdff (from </>iX6(i>, denominative of <^tXof, transposed from
<^Xtoff), the rj of wliidi has sprung, like the Gothic ^, from a.

X Euphonic for buhta (see §. 82.), and this from bttkta, from the root
bug.

FORMATION OF WORDS. 1121

" I purchase '"), corresponds to Sanscrit forms like bhuktd,
** eaten *"' (root bhtif from bhug), Greek like ^pvKTo, and Latin
like jundu ; mwnrdxu " believed," answers to the Sanscrit
ma-td, " thought," " believed," for man-td, as the feminine
substantive base ga^mun-di (nominative -n-ds) does to the
Sanscrit md{nytU " meaning."

824. In Lithuanian the participial suffix spoken of is
retained quite unaltered in form, and, indeed, in all verbs^
so far as they have a passive. In the nominative mascu-
line ta-s corresponds to the Sanscrit td-s; e.g. seMa-s,
** followed " = Sanscrit saktd-s (root sach from sah "to fol-
low," compare Latin sequ(yr)\ seg-ta-s, " fastened " = San-
scrit sak-td'S for sag-td-s (root ^^ san^, from sang, " to

; deg-ta-s, ** burnt " = Sanscrit dajr-dAd-s. In the
nominative feminine sekta, segta, degta, correspond to the
Sanscrit sakta, dagdhd, only with the a shortened, as in
Grothic, Latin> and Zend, forms like bauhta (genitive
bauhtd-s), juncta, As^di^^ basta (see §. 137.) : to the Latin
juncta corresponds literatim the Lithuanian yunkta, from
yungiu, " I yoke (the oxen) :" kept-as, kepta (from keppu, ** I
bake," see §. 501.), corresponds to the Sanscrit pak-tds, td,
Greek ireirro^, rfj, Latin coctu-s, ta. Forms like wes-ta^s,
"conducted" (root wed), correspond in a euphonic respect
to Zend like bas-td, "bound" (root baudh), ins-td, "dead"
(root irith), and Greek like 7r<(r-Toj, Keo'^Tog (see §. 102.). To
the Gothic participles of the weak conjugation correspond
the participles of those Lithuanian conjugations, which we
have above (§. 506.) compared with the Sanscrit tenth class ;
thus, mylri'taSf "beloved;" pen-e-tas, "nourished ;" laik-y-tas,
" held."

825. The Sclavonic languages have, if the opinion ex-

* Dh euphonic for t, see §. 104. In Irish, eUighaim^ " I bum," corre-
sponds to the Sanscrit ddhdmi; and dagte^ " burnt," to the passive parti-
ciple dagkdd-Sy Lithuanian degtas.

1122 FORMATION OF WORDS.

pressed in §. 628. be well based, transferred to the active
voice the passive participle here spoken of — with the re-
tention, however, of the meaning of past time — and have
weakened the original t to I, probably by changing, it in
an intervening stage, to d. In the former point they cor-
respond to the New Persian, where the participle in ques-
tion has, at least generally, an active signification : in the
latter point they agree with the Georgian, where ^s5j)Qpo
jam-U'li signifies "eaten"' (Sanscrit jam, **to eat''), and
coi^oAosofio thbob-i'lh ** warmed" (Sanscrit tap, ** to burn ").

•«^ «^.

J' J' J' J' Jf Jl J^

WORKS ON THE EAST.

^^S999StSI

I.

DRY LEAVES FROM YOUNG EGYPT,

Bj an Ex-political, being a notice of some of the transactions which
preceded the annexation of Sindh bj Sir Charles Napier, post 8vo.
with illustrations. 7s. 6d.

II.

TRADE AND TRAVEL IN THE FAR EAST,

OR,

Recollections of 21 years in Jara, Singapore, AustraHa, and China, by
G. F. Davidson, Esq. post 8vo. 10s. 6d.

III.

THE VALLEY OF THE NILE,

Illustrated by E. Prisse, containing 32 Lithographs in the first style of
art, and numerous Wood Cuts, In a handsoii^ imperial folio volume
£5 5s. plain*. £10 lOs. tinted. £15 158. fuUy colored.

IV.

OTIA iEGYPTIACA.

Discourse* on Egyptian Archaeology and Hieroglyphical Discoveries, by
George R. Gliddon. 8vo. cloth. 7s. 6d.

V.

LEGENDS OF THE AFGHAN COUNTRIES

In verse, with various pieces, original and translated, by Charles Massooi
Author of Travels in Beloochistan, Affghanistan, and ^he Punjab,

VI.

ENTERPRISE IN TROPICAL AUSTRALIA,

By G. Windsor Earl. 8vo. Cloth 6s.

JAMES MADDEN, 8, LEADENHALL STREET.

J. Wertheimer ft Ck>., Frinten, Circiu Place, London Wall.

J' J' J' J'. ,

if )r- IT tr »rir ir \\- - ■ • — J^jcjcjc

}r J^ J^ iT iT iT
Jt JC JC Jt J€ Jg

Jft Jt Jt Jt Jt Jt
ft jt jt jtjijt
jc jt jt jt jt jt

jcjejejejejt

Jr JF JF JF if^ JF^
JF jF JF JF JF iT

jT iT Jr Jr JF JF

jejtjejtjejt

jtjtjtjtjeje
jtjtjtjtjtje
jtjtjtjtjejc

WORKS FOR INSTRUCTION IN
• HINDUSTANI,

Bt Db. J. R. Ballanttne, of the Bbnarbs College.

^B»

HINDUSTANI GRAMMAR,

With brief notices of the Braj and Dakhani dialects. Second Edition,

royal 8to. 6s.

A KEY to the above. Royal 8vo. Is.

HINDUSTANI SELECTIONS.

Royal 8vo. 6s.

PERSIAN CALLIGRAPHY.

Second Edition, Lithographed. 4to. 6s«

ELEMENTS OF HINDI & BRAJ BAKHA.

4tO. 8s.

MAHRATTA GRAMMAR.

■ 4to. lithographed, lOs. 6d.

PRACTICAL ORIENTAL INTERPRETER,

OB,

Hints on the Art of Translating readily from English, into Hindu-
stani and Persian. Royal 8vo. 5s.

CATECHISM OF SANSKRIT GRAMMAR, Is.
CATECHISM OF PERSIAN GRAMMAR, Is.
DICTIONARY, ENGLISH & HINDUSTANI,

With the Hindustani character, and Roman pronunciation, by Captain

R. S. Dobbie. 12mo. 8s.

THE EASTERN TRAVELLERS'
REMEMBRANCER,

JT iT iT JT i

jFjtjc

J^ jT iT IT i

irJrJFjFjFJ^JI^i^

JFJfJFJFJFJ^J^i

J^ Jf^ J^ J^ J^ J^ J^ i
J^ J^ JF JF JF JF jT i

j^ ^ jhf jf^ jf ^

jt je jt JF JF jt jt

lOs.

Contidning Dialogues, Familiar phrases, &c., in the Arabic and Roman 11 v Jjt Jt'jt Jt Jt Jt
characters, for the use of Travellers by the Overland Route. ^2™o-\\ j^ jf j^ 3tjtjtjl

Jjt JF jt jt jt jt ji

^

JAMES MADDEN, 8, LEA.TJE^HM'^ STREET.

J. Werthdmer * Co.. Printer., Ci»!o«W»<«A<»4^]5^