1 This genitive appears sometimes under the form -es, sometimes also under the form -i, as: pernicies, gen. pernicies, progenies, gen. progenii. See the passages quoted by Schwartze, das alte Ægypten, p. 565. 2 Ας δημόσιο, δήμοιο, δήμου, comp. the nom. plural.

3 The dative or ablative in bus is sometimes found in those nouns which have e or i before the characteristic: thus we have diibus from deus (Gruter, II. 9; XXIV.6; XLVI. 9); and filibus from filius (id. DLIII. 8; DLIV. 4).

If now we compare these particular instances with the general scheme, we shall see that, taking all the varieties of the crude form, of which the above are specimens, there are only two assumptions in the general table,—namely, the original forms of the nominative and accusative plural. All the others are actually found, either in nouns or pronouns, at some epoch of the language.

With regard to the nominative and accusative plural, the assumed original forms are derived from a sound induction according to the principles of comparative philology.

And first with regard to the nominative plural. The sign of this case must have been originally -s throughout the declensions. Now it appears from general considerations, as well as from an induction of facts, that -s was also the sign of the nominative singular (New Cratylus, § 243). Therefore the -s of the nominative plural, if it was to distinguish the form from the same case in the singular, cannot have been appended to the mere crude form of the noun; for then the nominatives singular and plural would have been one and the same inflexion. It must have been formed by adding the -s (with, of course, an intervening short vowel, for the Latin language does not tolerate a double -s at the end of a word) to the full form of the nominative, and thus constituting, as the total addition to the crude form, or the real termination, the syllable -ses. If we compare lapid-ēs, patr-es, with éλπid-es, Taтéρ-es, we shall see that the long e in the Latin words cannot be accounted for otherwise than by the absorption of an s, which has probably become vocalized in i. In the Greek forms this s, like the v of the accusative, has been

dropt altogether. This view is supported, not only by the fact that the plurals vo-bis, λóyo-i-s, &c., actually stand in this relation to the singulars ti-bi, λóyw = λóyo-, &c., but even more so by the analogy of the genitive singular. For in many cases the genitive singular is identical, in its secondary form, with the nominative plural: thus familiæ, avi, are the common forms of both cases. But familiæ is actually written familiâs = familiaĕs in compounds with pater, mater, filius, &c. Hence we may presume the same original form of the nominative plural familiæ (compare dies, &c). Now the original form of the nom. singular must have been familiă-s; consequently, if, when the nom. sing. was familia, the nom. plur. was familia-ès = familiæ, it follows that when the nom. sing. was familia-s, the nom. plur. must have been familia-ses. The same follows from the form avi. The omission of s between two vowels is fully supported by Greek analogies: for if ἐλέγου is manifestly a corruption of ἐλέγεσο, ἴχθυες may well be a similar corruption of ἴχθυσες. I have preferred to treat the original form of the nominative plural as an assumption, and to support it by the arguments which I have just adduced; but if we remember that the original s of many Roman words was not changed into R till about the 4th century A.U.C. (above, Ch. VII. § 3), we might take the existence of such forms as spe-res (which occurs in fragments of Ennius), and gnaru-res (which is found in Plautus, Mostellaria, I. 2, 17; Panulus, prol. 47), as a distinct confirmation of the theory. And here again the analogy of the genitive becomes applicable, as will be seen below (§ 5). The pronouns also supply a partial confirmation of the above induction; for though in common Latin we find a genitive singular in -s by the side of a nominative plural in -i, we learn from old inscriptions that there was also a nominative plural in -8: see Senatus Cons. de Bacch. 11. 3, 7; Lex Rom. Bant. Tab. 1. 21; Klenze ad Leg. Servil. p. 12.

Again, in regard to the accusative plural, which in all the above instances ends in -s preceded by a long vowel, we must infer that -s is the termination of the plural as such, from considerations of the same nature with those which have just been brought forward. We should also have no difficulty in supposing that the long vowel indicates the absorption of some consonant. This consonant can only be the -m of the accusative singular; for not only is this most probable à priori, but it is

the only supposition which explains all the phenomena. Let us take the Greek, Latin, Sanscrit, and Gothic forms in a particular word; and we shall see that, while the Gothic alone preserves the outward marks of such a derivation of the accusative plural from the accusative singular, the only possible explanation of the other forms is the supposition that they were originally identical with the Gothic. Thus, Xuko-v, lupu-m, vrika-m, vulfa-n, are the accusative singular of synonymous words in these four languages. The plural of the Gothic vulfa-n is simply vulfa-n-s, whereas all the other forms strengthen the final vowel of the crude form, and drop one of the concluding consonants: XÚkov becomes Aúkovs, lupum is converted into lupós, and vrikam into vrěkán. The comparison of odous, &c., with dens, &c., shows us that Xúkovs may stand for λύκους ; and the analogy of τύπτων = τύπτον[τ]ς is sufficient to explain the change of vrikans into vrěkân. The Umbrian also has shown us both the original formation and the corruption of the accusative plural: for while we have abron-s exactly corresponding to the Gothic vulfan-s, we have also abrof, which, as I have shown (above, p. 91), must have proceeded from abrom-habrom-s. If we add to this, that when the accusative singular has lost its final consonant, the plural accusative merely adds -s to the existing form of the singular (as in ἄνδρα[ν], τύπτοντα[ν], sing., ἄνδρα-ς, τύπτοντᾶ-s, plural), we have, it should seem, the most satisfactory evidence which the subject admits, in support of the assumed original form of the accusative plural.

Having thus justified the only hypothetical forms in the above scheme of cases, it will be desirable to make some remarks on the most striking peculiarities in the existing inflexions.

§ 5. Existing forms-the Genitive.

In the general scheme, the genitive singular is characterised by the terminations -is, -sis, or jus; the gen. plural by the ending -rum, where the r is generally dropt, except in the a, e, and o declensions, which constantly retain it. The difficulty here felt is, to connect the plural form with the singular. Struve's assertion (über die Lat. Decl., 3, 15), that the r is merely euphonic, would tend, if we assented to it, to complicate and increase this difficulty in no small degree. The comparative philologer cannot doubt that the original form of the genitive

plural in the Indo-Germanic languages was that which is preserved in the Sanscrit -sám=2QM (see Müller ad Varron. L. L. VIII. § 74, p. 192). This form, after the fourth century A. U. C., would appear in Latin as ROM, which was afterwards softened into RUM. The Indians wrote -nám for -sâm in many of their words where the n represents the s, as in vrěkán for vrěkás=vrikăm-s; but in the pronouns, which generally preserve the authentic forms longer than the nouns, we have tá-sâm-istá-rum. The Greeks very often omitted an σ- between two vowels in a case like this; and as they wrote ἐλέγου for ἐλέγεσο, ἴχθυ-ες for ἴχθυσ-ες, 50 they gave us δήμοιο, or ultimately δήμου, for the original δημόσιο, and μουσάων, or ultimately μουσῶν, for μουσάσων = μουσα-σιον-ς. That -rum is the proper and genuine form of the Latin genitive is proved not merely by the fact that the Romans actually wrote -um for -orum when it suited their convenience1, thereby showing the reason for the omission of the r in the other declensions, but also by the fact that the r is found in the pronouns, the oldest and most immutable parts of speech, and that in the more ancient state of the language even nouns of the other declensions retained the r: thus we hear of such words as boverum, Joverum (Varro, L. L. VIII. § 74), lapiderum, nucerum, regerum (Cn. Gellius apud Charisium, I. 40). This evidence receives very striking confirmation from the analogy of the genitive singular. The most common characteristic of the genitive singular is the termination -is.

There are reasons, however, which may induce us to doubt if this is the full and original form of the genitive-ending. The Sanserit vrikasya compared with λύκοιο, and the possessive δημό otos by the side of dnuó-to, might lead us to suspect that the termination commenced with an s, which was subsequently absorbed ; and this suspicion is confirmed by the fact, that there are, in old Latin, genitives ending in -ris-sis where the r=s is not part of the crude form. Thus we have sue-ris for suis in the fragment of Plautus quoted by Festus, s. v. Spetile, p. 330: "Esto pernam, sumen sueris, spetile, callum, glandia." Compare Varro, L. L. V. § 110, p. 44. And from the extant forms of the nominative plural in -res we may fairly infer that the genitive in -ris-sis was not uncommon. The Latin possessive adjectives end in -ius or -eus, e. g. prætor-ius from prætor, virgin-eus from

1 On this abbreviation, see Cicero's remarks in Orator. c. 46, § 155.