radical (all but one, -ole, being monosyllabic stems) have -ētum in the supine (§ 692). But the great mass of the rest have -Itum (§ 693). A few omit the vowel altogether (§§ 700—709). Short I is a very frequent substitute for ě, especially in unaccented syllables (§ 234). The occurrence therefore of a short 1 in the great majority of suffixes from verbs with e stems is strongly suggestive of the shortness of the final stem e.

(2) There is a numerous class of adjectives with stems in -do (§ 816). Most of these are derived from verbs, and all but a few of these are from verbs with e stems. In all these cases there is a vowel preceding the -do, and this vowel is short 1. In no instance is there a long vowel, unless radical, preceding -do, and in no instance is the adjective derived from a stem with & or ẽ or ú or I. This again points to a connection of I-do with shortness of the stem vowel of the e verbs.

(3) The perfect of verbs with e stems which have -Itum in the supine is in -ul, never in -ēvi. And the same perfect is found in a great many other verbs of the like stems, which have no supine or other word of this formation in use. Now it seems difficult to account for the general prevalence of ui (instead of ēvi) in e verbs, compared with -āvi in a verbs, unless from the quantity of the vowels being different. The difference in quality between a and e, when these vowels come before u, does not seem of a kind to account at all for the nearly universal solution of the one vowel and maintenance of the other. Verbs which, as monosyllables and as having radical e, have the best claim on à priori grounds to e long, have evi in the perfect, accompanying ētum in the supine. But ĕ+u seems calculated to pass into eu and then into u without difficulty.

These facts together seem to me to make strongly for the shortness of the ĕ in mone- and such like verbs. Nor do I see any argument1 for its length, which is not drawn from facts which, to say the least, are perfectly compatible with this theory. I conceive the

=

1 Gellius indeed speaks (VII. VI. 15) as if 'calescit, nitescit, stupescit, et alia hujuscemodi multa' had e long, and quiescit' e short. Those who consider this a proof of the characteristic vowel of e verbs being long naturally, may explain how 'quiescit' came to be (according to Gellius) short.

length of e in parts of the present stem, e. g. monēs, monēmus,
monētis, monēre, and similar parts of the passive verb, to be explica-
ble by a contraction of the final e with the initial vowel of this
suffix, moně-ĕre = monēre. For the existence of the initial vowel
of the suffix, I refer to the consonant stems. (For Corssen's theory
respecting these consonant stems, see § 743.)

The analogy of Greek stems appears to confirm the same
view. There the e is unquestionably short, e.g. piλéw; and
wherever a long vowel appears in its stead, a contraction has taken
place.

I might refer to the quantity of the e in the half compounds, e.g. pudefacis, but the evidence is not decisive. All the instances will be found collected in § 994. The majority of them have e short, and of the dozen which are found with a long e, three (experge-, rare-, vace-) are not from e stems, one (sve-) from

a verb with radical e, four others (liqve-, pate-, putre-, tepe-) are Luer IV.

also found with e short; and the remaining four (conferve-, contabe-, 1,442 perfrige-, obstupe-) are each used once only, and that in writers

(Plaut., Ter., Lucret.) whose use in such a matter can hardly be re-
garded as decisive. The probable solution of this occasional lengthen-
ing may be sought in a wrong inference from the length of the e in
monemus, monere, or in a fancy that, e. g. perfrige-facis is contracted
for perfrigere facio. Anyhow the evidence from these compounds
on the whole inclines considerably in favour of the theory of the
final e of the stem being short.

There are a few verbs with a stems which seem to me to have ǎ
short. They will be found named in §§ 645 and 688. The
greater number of them are markedly distinguished from ordinary
a verbs by the same characteristics as have been noticed in most e
verbs, viz. a perfect in -ul (instead of avi), and a supine in -Itum
(instead of atum). Some of these show indications of having their
natural character eventually overborne by the analogy of the others.
Hence we have micui and dimicāvi, enecui and enecavi, -plicui
and -plicāvi. Dă- retains its radical short quantity throughout,
except in das; stă- is, as regards the present stem, swept into the
strong current of the derivative verbs; sonă- gives place to a verb
ōn-; or it may perhaps be held that sonis, sonĕre are really attempts

345

at preserving the proper quantity without the apparent anomaly of a short ǎ. Ră- and să- deviate in other ways. On inqvam, see $561.

The argument from the supine will be best appreciated by an examination of Book 11. Chap. xxiv. It will be seen how few are the cases in which a vowel is found before turo- in the supine, without the other parts also showing a vowel stem. (See § 698, also fruiturus and ruiturus.) Nor are the instances many more in which, if the above principles be adopted, the quantity of this vowel does not correspond with the quantity of the final stem vowel. (Corssen supposes in the case of e stems a shortening of an original ē; in the case of the a stems the coexistence of a verb of the 3rd conjugation. Ausspr. 11. 292—295 ed. 2.)

The verbs like facio, capio, &c. are generally regarded as having an inorganic i inserted in some parts, whilst in others what is considered its real consonant stem is shown. I have ventured to consider these verbs to be vowel verbs with stem ending in -1. For, as far as I understand the laws of vocalization in Latin, the phenomena are exactly those which would be found, if they had this stem ending: I would maintain its place before a labial vowel (o or u), and would be omitted before 1; comp. adice for adjice, &c. (§ 144). But when s becomes r, I would of course become ě, and this completely accounts for what otherwise seems such strange variation as capio, capis, capit, capiunt, capiebam, capias, capies, capere, caperem1, &c. The imperative singular cape from a stem capi- is evidently analogous to mare from a stem mari-, and may be accounted for in the same way, whatever that be (see § 196). It may be remarked that a final ĭ is very rare in Latin words (see §§ 280, 243, 4). Such instances as do occur are all due to poetic shortenings of original long vowels.

Some of these verbs exhibit this i short in the supine. In others it is omitted, as it is in many vowel verbs. Any short vowel in this position would almost inevitably have become I, and the omission of I in, or adjoining to, suffixes is far from being uncommon. I

1 Comp. Grassmann in Kuhn's Zeitschrift, XI. p. 50.

am not confident as to the quantity of the final stem vowel in such verbs as senti- (sentio, sensi, sensum). I have sometimes marked it long as usual. It may be, these verbs are instances of a semi-perversion by the analogy of more regular 1 stems, e. g. audio, audīvi, auditum; or the 1 is here distinctly realized as a suffix of inflexion only, a mark of the present stem, instead of the verb stem. In verbs which have reduplicated perfects, or perfects in -si, the same distinction is practically recognized.

Concluding Remarks.

I have stated in different parts of the book such obligations as I thought necessary to mention in a book of this kind, which can rarely be formed by independent research from the original authorities directly. But I desire here expressly to recognize the debt I owe to Ritschl, Corssen, Neue, and Curtius, to all of whom I hope, at a future time, to express renewed obligations for further information. Many of the statements about Latin inscriptions of the Republican period are taken from Ritschl, and taken with the confident belief that, though they may not prove always right, it is exceedingly improbable that I should be able to correct him. Some of his writings on Inscriptions are not easily accessible. I look forward with much interest to their republication in his Opuscula, as well as to the new edition of his Plautus, and the promised Grammar of old Latin, if indeed the last is not put off to the Greek Kalends. The statements about later inscriptions, and some respecting republican inscriptions, are chiefly founded on statements by Corssen or Brambach (Die Neugestaltung der lateinischen Orthographie, 1868). These of course cannot claim anything like the weight of Ritschl's statements, which are the outcome of years of skilled and careful labour. To Corssen I am the more anxious to acknowledge my frequent obligations, because his very prominence in the field of Roman phonetics has made it necessary for me, in some cases, to express and vindicate my dissent from his views. The second volume of the new edition of his Aussprache did not reach me in time to make much use of, except in occasional reference and correction. Curtius' very careful identification of Latin and Greek

roots has been followed almost implicitly to this extent, that I have rarely suggested an identity which he has not approved, though I have frequently omitted some which were either superfluous for the purpose in hand, or appeared to me to admit of some doubt.

Neue's Formenlehre (1300 closely printed pages without an index) has enabled me to give a more complete, and at the same time a briefer, account of Latin inflexions than will be found in other Grammars. It seemed to me useless, as a general rule, to encumber my book with references to the passages where a particular form occurs, when this work has been done exhaustively already, and the result can be easily obtained by any scholar who seeks to test a matter himself. On the other hand Neue's book is quite unreadable by the majority of students, and is, in fact, not so much a grammar itself, as a quarry from which grammars will be built. I hope greatly to improve my own 1st and 3rd Books when the corresponding parts of Neue's work are published. It may be useful to add that, being mainly a collection of references, it is accessible to a great extent by students who have little knowledge of German. I have tested his references in a great many cases, and rarely found them inaccurate. Of course, later critical editions of authors will

sometimes alter his results.

Madvig's Grammar (3rd Germ. edit.) has not been of so much service to me in this volume, as in the Syntax. In that my obligations to him are paramount to all others. To Key's Grammar I certainly owe much in the way of suggestion, but how much I cannot tell, as I have often used it for many years, and in such a case it is impossible to distinguish between ideas which have been more or less borrowed, and those which have been obtained by independent inquiry with eyes fixed in the same direction. But there is no recent Latin Grammar, that I know of (except Madvig's in the Syntax), which is based on so fresh a study of the facts, or has done more in awakening a more scientfiic treatment. I have also read some of his other Philological papers, and sometimes got useful hints even from those with whose general arguments and conclusions I am quite incompetent to deal.

Gossrau's elaborate, but not, as I think, very happily conceived Grammar, and Schweizer-Sidler's Formenlehre, were not published till my first two books were in print. And two English books,