to the verbal stem of the sentence and not to the noun. We shall take for example the verbal stem a'nu to throw.

nux' anu'kux' a stone he throws.
nux' anūsa'kux' a stone he flings with.

When the direct object is named, the pronoun ñan, i'kin, or ñin to him, to them both, or to them, is added to the instrument.

sax' nux'-ñan anūsa'kux' the bird a stone to it he flings with.
Here is another example:
suku'qiñ I take.
sū-sa-ku'qiñ I take with it.
cañ sūsaku'qiñ my hand I take with it.

When the direct object is named and to the instrument a personal pronoun is added, the latter agrees in number with the direct object. For instance: 1. qax' cañ-ñan sūsaku'qiñ a fish I take with my hand.

. 2. qax cañ-i'kin sūsaku'qiñ two fishes I take with my hand. 3. qan cañ-ñin sūsaku'qiñ fishes I take with my hand Literally these sentences have to be translated as follows: 1. The fish my hand it (i.e. the fish) with it (i.e. with the hand) I take. 2. Two fishes my hand them both with it I take. 3. Fishes my hand them with it I take.

Thus the instrumental element sa suffixed to the verbal stem (su) refers to the instrument can my hand, while the pronoun (ñan, i'kin, or ñin) refers to the direct object (qax', qax, qan). We wish to add that the dative of the personal pronouns (ñan, i'kin, ñin) takes the place in the above sentences of the accusative, as in other cases with verbs. In nouns the absolute case appears as the nominative-accusative, while in personal pronouns the absolute case is wanting. For the third person pronouns are used, which are combined with an adverbial element showing the distance of the third person from the speaking person. For instance, a'man he who is far from the speaking person, but not out of sight; i'ñan he the near one; aʼkan he the upper one; u'knan he the lower one; uskan he the inside one, etc.

Nouns and verbs are formed from the same stem. When the stem becomes a noun, its instrumental element will be si (instead of sa in verbs). For instance, tu'ga expresses the notion beat: tuga'-kux' he beats; tugā-sa'kux' he beats with it; tu'ga-x' the beating; tuga'-si-x' the instrument for beating (stick, club). asu'ga is the stem dig: asu'ga-x' the digging; asūga'-si-x' a spade, i.e. 'the instrument for digging.'

The comitative. The comitative is expressed by means of a'six, a participle form of the stem a- to be: a'six being with (somebody or something).

a'dañ a'six hwagana'qiñ 'my father being with (me) I came to this place,' i.e. I came to this place with my father.

But the same locution may be expressed by the instrumental element sa suffixed to the stem. hwagana'qiñ I came to this place; hwa'ga is the stem, na element of past tense, qiñ personal ending. a'dañ hwagāsa-na'-qiñ my father I came with, or I brought my father. Thus we see that the instrumental element sa may transform an intransitive verb into a transitive.

THE -TT- IN LATIN QUATTUOR

ROLAND G. KENT

UNIVERSITY OF PENNSYLVANIA

Numerals share with pronouns the honor of being the best criteria of linguistic relationship; and yet among the names of the digits we find more exceptions to the principle of regular phonetic development, probably, than anywhere else in an equal group of unquestioned cognates. The exceptions arise, in the main, from the fact that the words are in use so closely associated with one another that they affect one another's forms.1

One such puzzling exception is the -tt- in Latin quattuor. The conventional theory is that it is a doubling, or more strictly a lengthening, of the consonant at the junction between two syllables, as in the occasional Iusstus, solluit, etc. of inscriptions, and in the Italian febbre, acqua from Latin febris, aqua.2 But in quattuor the doubling is admittedly much earlier than that in the Italian words, and as compared with the Latin examples, it is regular and not sporadic. For the length of the initial syllable of quattuor is assured in the time of Plautus, by its use in Pseudolus 1303; and the spelling quattuor, with two t's, is found in CIL 12.587.ii.18, 21, an inscription of 81 B.C.3

The only circumstance in which, in Latin, a single consonant became doubled, is that of its standing after a long vowel of the accented syllable, at the same time that it, the consonant, belongs to the following syllable, and is immediately followed by the vowel of that syllable. Thus older Iupiter regularly became Iuppiter, and older litera became littera. The inference is unescapable, that quattuor represents an older *quatuor, in which the -u- had already become vocalic. Can a long vowel in the initial syllable of this word be explained on any reasonable basis, in the face of the short vowed attested by all other languages?

1 I give a minimum of references, citing here collectively A. Walde, Lat. etym. Wtb., s. vv.; E. Boisacq, Dict. étym. de la langue grecque, s. vv.; K. Brugmann, Gdr. d. vergl. Gram. d. indog. Spr. 22.2.1-82 (1911); F. Sommer, Hbd. d. lat. Lautu. Formenlehre2 (1914); F. Stolz, Lat. Gram.5 291-5 (1926), revised by M. Lehmann, in I. v. Müller's Hdb. d. Altertumsw. II. 2.

2 Leumann-Stolz, op. cit. §152.

3 Sommer, op. cit. 203.

Perhaps the point of connection may be found in the series, put in the pre-Plautine values, oinos duo trēs quatuor quinque, in which the characteristic vowel-if we may grant that the characteristic vowel of duo is the ō rather than the u, as a doublet form *dvõ may have been used in rapid speech-is without exception either a diphthong or a long vowel. Note further that the next five cardinals agree in having a short vowel in the corresponding position: sex septem octo novem decem. So I incline to think that the long vowels of oinos duo tres quinque infected *quátuor and remade it to *quátuor. But another form may have contributed to the long vowel in quatuor: the pre-Latin word for 'fourth' seems to have been *kwatwortos, whence *kwawortos with dissimilative loss of one t, and quārtus by contraction. The length of the vowel in quärtus is assured by inscriptional writings with the apex, and there is the parallel instance of Māvors Mārs, though here the prior vowel is long before the contraction occurs. To these two influences, then, that of the long vowels in the other cardinals of the series and that of the long vowel in the ordinal quārtus, I attribute the lengthening of the first vowel of *quátuor, which made possible thereafter the doubling of the consonant with a conincident shortening of the vowel.

There is the objection that quater 'four times' and words beginning with quadr- (quadrāgintā, quadringentī, quadrupēs, quadrīgae, etc.) are proved by the meters of poetry to have a short a in the initial syllable. But quater goes back to an adverb ending in -trus, and the other words have a peculiar -dr- after the vowel, with an unexplainable -d- before ---; the differences in the consonantal groups may have been the decisive factor in preventing the spread of the long vowel from quätuor.

The length of the vowel in quinque also is not original, as was recognized long ago, but came back from the ordinal quinctus into the cardinal; for short vowels were lengthened before -nct- in early Latin.

6

4

* Mon. Anc. iii. 22;CIL 3.4959; cited by Sommer, op. cit. 472.

5 For quater, see Vergil, Aen. 1.94, Ge. 1.411, 2.399. For quadr- the evidence is more difficult to locate; for while in Plautus and in Terence a stop plus a liquid does not make position, substitutions of long syllable for short and of two shorts for a long are freely made, and in later verse the consonants may make a closed syllable. The surest warranty for a short vowel seems to be in quadrupedem, Ter. And. 865. But to keep the shortness of the initial syllable, editors are obliged to emend the manuscript reading quadruplari, Pl. Persa 62, to quadrupulari, and quadruplator, ib. 70, to quadrupulator. If the initial syllable of quadrilibrem, Pl. Aul. 809, be short, one must accept a hiatus after the next word, instead of an elision.

• By R. Thurneysen, KZ 30.501-2.

6

And yet the -c of quinctus itself remained only by the reciprocal influence of quinque, since the group -nct- tended to become -nt- in early Latin.? No parallel to this association of the first five digits in their vocalic length can be found in Oscan and Umbrian, where the words for '4' and '5' were petora and * pompe, amply attested by glosses and by derivatives; and therefore the phenomenon which we have examined is distinctly a Latin phenomenon. At the same time, it must be recognized as very early, since Verrius Flaccus remarks on the long vowel in old Latin quincentum '500.'8 Converging evidence indicates that the

' length in *quätuor and in quinque became established at not far from the same time, and therefore increases the likelihood of the explanation of quattuor which has been offered here.

? For citations of spelling with -c-, cf. Neue-Wagener, Formenlehre d. lat. Spr.: 2.310.

8 Ap. Fest. 254 M. 338 Th.: Quincentum et producta prima syllaba et per c litteram usurpant antiqui, quod postea levius visum est, ita ut nunc dicimus, pronuntiari.