328 LATIN LANGUAGE an s between two vowels passes regularly (probably through the in termediate stage of a sibilant pronounced like z) into r between two vowels, as in uro for an earlier uso. The loss of a y (i) between vowels is not uncommon, e.g., sedo for scdaio through sedao ; it is less frequent after consonants, as in o bex for ob-iec-s ; under similar circumstances w (u) disappears, as in amasti for amavisti, canis for cuanis. AVith regard to inflexion, the following may be noted as the chief developments subsequent to the stage described in vol. xi. p. 131 : 1. In substantives there was a considerable extension of the class of f-stems, due partly to new creations, partly to the transforma tion of stems belonging to other classes. Thus a primitive garu-s, Gr. fiapv-s, in Latin is grau-i-s, a primitive kmn-s becomes cani-s. Very few of the Latin i-stems have corresponding z -stems in Sans krit or Greek. In some cases the i appears to have been originally an a ; comp. imbri- and opfipo-. 2. The final -d of the ablative was retained, and (in Latin at any rate) the -frits of the dative plural ; on the other hand, the instru mental in -bhi (Gr. <pi) does not appear at all on Italian soil. 3. The dual number was lost both in nouns and in verbs, as in the later Greek. 4. An entirely new system of inflexion for the reflexive tenses (the middle, or, as it subsequently became, the passive voice) was created by the use of the reflexive pronoun se as a suffix. (Whether this system is common to the Italian and the Celtic languages, or whether the apparently similar formations in the latter are of different origin, is a question not yet definitely settled. ) 5. In many verbs the compound aorist with an s element was made by the action of analogy into a perfect in si. 6. Numerous verbs adopted for their perfect tenses a suffix in -vi or -ui. This has been commonly supposed to represent a new pro cess of combination with the root bhu instead of cs ; but weighty objections have recently been brought against this explanation, and it can no longer be propounded with confidence. 7. The root bhu was employed to form a past imperfect in -bam, and a future in -bo ; but in the case of consonant verbs and z-verbs the latter formation was usually replaced by an optative form used as a future. 8. Imperfect and pluperfect tenses of the subjunctive were formed apparently by compounding the present and perfect stems with the optative of the root cs, to be. " 9. The infinitive and participle system received a considerable expansion, especially by the formation of gerundives and supines, which, however, were differentiated in. usage in the various Italian dialects (see below). 10. The pronominal elements, though for the most part the same as in Greek, were commonly used in composition one with another, and thus acquired a different form. 11. The it-class also was extended by the more common use of the suffix -lit for verbal nouns. With regard to the vocabulary, very extensive additions were made, probably in many instances from Celtic sources. Many of the most common Latin words are entirely without demonstrable cognates in the other Indo-European languages ; and, even when the common root may be suggested with considerable plausibility, the particular Latin word has evidently behind it a long and independ ent history, during which its meaning and usage have been greatly modified. Hence all attempts to deal with the etymology of the Latin stock of words are confronted with a residuum which the materials at our command do not allow us to deal with satisfactorily. 1 Dis- The principal distinctions between the Latin branch of the tinctive Italian group and the Umbro-Sabellian are the following: features 1. Neither Umbrian nor Oscan had any character for o ; for this of Latin, the former language used v = u, e.g., puplumpopulum; the latter u (perhaps approximating to o), e.g., pud = quod, or sometimes u, as in arac/etiid^argcntod. 2. The Old Umbrian did not distinguish between surds and sonants in the case of gutturals and dentals, having no g or d; but both these letters were used in Oscan. 3. Oscan distinguished between i and i e , the latter a sound pro bably intermediate between i and c. 4. In Umbrian d, when occurring between vowels, or at the end of a word after a vowel, was replaced by r, in later Umbrian by rs, e.g., asam ar-=aram ad, i.e., ad aram, dupursus = biped ibus. K before c and i acquired a palatal sound, not existing in Latin until long afterwards, which was denoted by <j, i.e., f; e.g., qcsna^cena. 5. The Umbro-Sabellian dialects agreed in retaining the earlier genitive in s (becoming afterwards in Umbrian r), e.g., tuta-s, " of a city," molta-s = muUac, while the Latin has in the case of a-, c-, and o-stems substituted for this a form in i, probably a locative. 6. They retain also the future compounded with cs, e.g., Umb. 1 The Celtic element in Latin has been discussed by Professor Xewman in his Regal P^omt, and more satisfactorily by Mr Words worth in an appendix to his Lectures on Early Roman Literature ; but the question still requires further examination (comp. also Cuno s Geschichte Italiens). hcriest, Osc. her est = volet, replaced in Latin either by the optative or by a new form in -bo. 7. Botli Oscan and Umbrian allowed the velar guttural (q) to pass into p, as in the Gallo-British branch of Celtic and in Greek, while this is never the case in Latin; comp. qui-s and^is, Quintius with its Samnite equivalent Pontius = Pompciiis. Three clearly marked stages present themselves in the Stages history of the Latin language : (1) the archaic stage, pre- iu tbe vious to the development of literature ; (2) the stage of h j, stor y literary culture, during which the popular spoken language Latin runs, as it were, underground, giving but few traces of its Ian- existence ; (3) the stage at which the popular language re- appears as colouring literature, and finally recasting it in its own mould. The archaic stage is known to us almost wholly from The inscriptions, and from isolated forms and words quoted archaic by the grammarians ; although a careful study of the phenomena of the diction and especially the metre of the early Roman dramatists reveals to us many of its charac teristic tendencies. It may be said to have lasted until the time of Ennius (d. 1G9 B.C.), whose growing influence is intimated in the epitaph composed for himself by Nsevius (d. 204 B.C.) : " itaque postquam est Orci traditus thesauro obliti sunt Romai loquier Latina lingua." Perhaps the oldest specimen of the Latin language pre served to us is to be found in two fragments of the Carmina Saliaria preserved by Yarro (De ling. Lat., vii. 26, 27), and one in Terentianus Scaurus, but unfortunately they are so corrupt as to be quite unintelligible without the help of very extensive conjectural changes in the reading (<-f. Jordan, Krit. Beitrage, pp. 211-224). More valuable evidence is supplied in the Carmen Fratrum Arvalivm, which was found in 1778 engraved on one of the numerous tablets recording the transactions of the college of the Arval brothers, dug up on the site of their grove by the Tiber, 5 miles from the city of Rome ; but this also supplies many points for discussion, and even its general meaning is by no means clear (ib., pp. 203-11; <-f. Wordsworth, Fragments and Specimens, pp. 157, 158, with the notes). The text of the Twelve Tables (451-450 B.C.), if preserved in its integrity, would have been invaluable as a record of antique Latin; but it is known to us only in quotations, and it is doubtful whether any accurate reproduction of the laws in their primitive form was accessible to our authorities. Hence the language has been much modernized, and any archaic forms which have been preserved are due rather to the citations of the grammarians than to continuous quotations. Schoell, whose edition and commentary (Leipsic, 1866) is the most complete, notes the following traces, among others, of an archaic syntax : (1) both the subject and the object of the verb are often left to be understood from the context, e.g., 111 it antestamino, igitur cm capita; (2) the imperative is used even for permissions, si volet, plus data, if he choose, he may give him more ; (3) the subjunctive is apparently never used in conditional, only in final sentences, but the future perfect is common; (4) the connexion between sentences is of the simplest kind, and conjunctions are rare; ast ( si) and igitur (*-tum dcmum) have a different force from that found in later Latin. There are of course numerous isolated archaisms of form and meaning, such as calvitur, pacunt, cndo, escit ; but on the whole the diction cannot have been accu rately preserved. In the case of inscriptions there is rarely any question of their faithful reproduction of the language at the time at which they were made ; but there may be a difficulty about determining their date. Perhaps the oldest fragment of Latin preserved in this way is furnished by a vessel dug up in the valley between the Quirinal and the Viminal early in 1880. The vessel is of a dark brown clay, and consists of three small round pots, the sides of which are con nected together by short broad pipes, so that there is easy com munication from one to the other. All round this vessel runs an inscription, in three clauses, two nearly continuous, the third written below ; the writing is from right to left, and is still clearly legible ; the characters include some signs not belonging to the Latin alphabet proper, but to the other Italian alphabet, e.g., q for R, and I for Z, while the M has five strokes and the Q has the form of a Koppa.
