Page images
[ocr errors]

Of these, only dies has i before e, and the 1 here is a vowel of the root, not part of a derivative suffix, as in notities, &c. As for the rule concerning the quantity of the e, diei alone when dissyllabic has always an e long (as indeed a short e between two l's would

be utterly unstable in Latin): rei is used with e long in Plautus 1.688, and Lucretius, with ě short in Plautus, Terence, and Horace: fidei 2.112

has e long in Ennius, Plautus, and Lucretius; ě short in Manilius

and Silius. 2 §6.

There are, so far as I can find, no other instances in verse of a genitive and dative singular in ei. The dissyllabic nature nekaner, of ei can be shown only by express mention or by verse. 1.978

Now, putting together the following facts, (1) that at least in IV 885

many words the stems in e are collateral to stems in a; (2) that an antique genitive of -a stems, in ūī, was preserved in poetry by occasional usage for some time; (3) that in Cicero's time the genitive and dative of the -e stems were written either with e or with i; (4) that el was an accredited spelling of either an intermediate sound between e and i, or of long i: (5) that the use of any genitive or dative sing. of these stems is decidedly rare, except in three or four words, and that Quintilian regarded the form, at least as regards progenies, as either non-existent or disputed ;—putting these facts together, we may conclude that while ei may very possibly have been one mode of spelling the ending of the genitive and dative, it was probably monosyllabic, except in poetic and antiquarian writers. There is, however, no reason to doubt that, after Gellius' times, this was the ordinary spelling, and possibly, under the deceptive influence of diei, fidei in the phrases bonæ fidei, and plebei (in tribunus plebei, plebeiscitum), and the monosyllabic stems re-, spe-, the ei was regarded as dissyllabic. I have given in the paradigms of the declension ($ 343) what I suppose Cicero or Livy would have given.

It may perhaps be the most convenient course in modern times to continue to write ei, but we should pronounce it as a diphthong ($ 267), and use such forms as little as may be. It is impossible to suppose, considering the words', that the rare occurrence of the genitive and dative is not in some degree the result of a felt difficulty: and some of the instances which do exist are probably

e. g. acies. I have not hit upon any place in Livy where the genitive or dative of this word is used.


due to copyists who restored the ordinary spelling of their time, not to the writing of the authors of the gold or silver age themselves.

Noun-stems ending in i and in a consonant.

In determining which are i stems and which are consonant stems, I have followed principally the clue given by the genitive plural, and, in the case of neuter substantives or of adjectives, that of the nominative and accusative plural also. But I have also taken into account, especially where evidence on the above points was either non-existent or vacillating, the use of -is in the nom. or accus. plural of masculine and feminine nouns, and of course, in the few nouns which exhibit it, -im in the accus., and the more frequent i in the abl. singular. Many writers have considered words like ars, mens, &c., which do not exhibit the i in the nominative singular, as having, either in this case or in the singular number generally, passed into the consonant declension, or as having two stems, a consonant stem and an i stem. But the thorough-going distribution of the words of the third declension, adjectives included, between consonant stems and i stems, and the enumeration of all the words (except very numerous derivatives), with mention of any peculiarities they may show, have not, so far as I am aware, been done before. And this has brought into light two important points, stated respectively in § 406 and in § 408, compared with § 435.

The first of these points is that the difference between retaining or omitting the i in the nominative singular is due to phonetics and not to etymology. The 1 was evidently so weak in this final syllable, that, with rare exceptions, it was retained only when the nature of the preceding consonants was such as to be powerfully affected by an adjoining 8, Thus stems in -mi, - vi, -qvi, -gvi, -ni, -i, -ri, -si, retain the 1 with rare exceptions. The exceptions show the extent to which the stem would have been disguised, if this protective influence had not been exerted. Thus nix is hardly recognizable as of the same stem as nigvis or nivis; præcox, though looking very different, really stands to præcoqvis in


the same relation that cocus does to coquos. Ci generally drops i, but sci retains it, clearly because fascis would otherwise have been confused with fax. Ti generally dropped it, notwithstanding that this occasioned the loss of the t also. I presume, the close affinity of the continuous dental sharp s to the explosive dental sharp (t) rendered the former a sufficiently clear symbol of the real stem. But this clearness could not last, if other consonants were also to be absorbed by the nominative suffix; and therefore sti and -di retain the i, and thereby retain their distinctive consonants; restis is not allowed to become res, nor pedis to become pes. Assis, semissis, bessis (cf. App. D. p. 449), are found both in the full form, and as as, semis, bes, the abbreviation being the natural result of constant usage. Again, where t is preceded by a short vowel, the omission of the i would confuse stems having a short vowel, with stems having a long vowel. Hence nătis does not become nas, because nas would presume a stem nāti-; sitis is, by the retention of its 1, preserved from an identity with the commonly occurring word sīs. On the other hand, intercus, compos, compared with cutis, potis, show the tendency allowed to operate, in consequence of the desire of shortening a long word overpowering the risk of confusion—a risk which is indeed less when a word has a prefix than when it is a simple stem. But the confusion is evident, where such principles have been disregarded. Princeps may fairly enough represent principis, but then præcipitis should not have been allowed to sink into an apparently analogous præceps. Ennius indeed, and another old poet, seem to have been deceived by the nominative, and used præcipim, præcipe, for accus. and abl. Clear evidence of the antipathy of n, 1, and r, to an adjoining final s is afforded by the nominative of such consonant nouns as had stems ending in these sounds. It would not have been well to cut all such words down, as supellectilis was cut down, simply through this, to (supellectils, supellects) supellex. Who could have borne messis becoming mes, tussis becoming tus?

Corbis and orbis retained their i, probably because otherwise they might be confused with p stems. Thus urbs was doubtless

1 This is by no means the only instance in which the very early poets (Greeks by origin) seem to have simply blundered.

pronounced urps, but there appears to have been an unwillingness so to write it, lest the last evidence of the b stem should vanish. For, it must be remembered, though the Romans knew nothing of the modern theory of stems, yet they were struck by the apparent anomaly of writing, e.g. urps in the nominative and urbem in the accusative.

It is probable that the 1 has a very different origin in some of these words from what it has in others; in some it may be original, in others a weakened a (or o or e); in others it may have been inserted in order to give more distinctness and independence to a puny stem, and ward off the dangers of an overbearing s. This appears to be the case in canis, juvenis. Senex found another way out of this difficulty.

The stems with nom. in -es, I have thought best to class with the i stems, as those with which they have most resemblance. I am well aware that they are often supposed properly to have their stems, not merely their nominative case, in -es (cf. e.g. Schweizer-Sid. Lat. Gr. $ 50, and see Leo Meyer, Corssen, &c.), but this appears to me far from certain (see § 405). And in a case of obscurity I have preferred to be guided in my arrangement by the balance of objective facts.


[ocr errors]

In SS 408, 435, I have pointed out some striking differences between the words which have i stems, and the words which have consonant stems. While fully admitting the probability of some of both classes of stems being as original as stems in a and o, I am inclined to regard the second class of nouns as on the whole of later birth than the first class, and the majority of these stems as being weakened forms of o and a stems, the so-called i stems having been for phonetic reasons arrested at an intermediate stage, the consonant stems showing the latest and furthest stage. As the words increased in length by the addition of derivative suffixes, they under the influence of the Latin accentuation first thinned the final vowel, then dropped it altogether. This final vowel was, it is true, originally very important as the sign of gender, but as the language grew older, the imagination which saw sex in inanimate objects grew duller, and first the distinction of male and female became unimportant in such matters, and then the distinc-,

tion of sex and no sex, The new derivatives which were the offspring of the rational faculty were names of abstractions, not of things, and they were by the process of their formation descriptions, not pictures. Thus the gender became masculine or feminine according to some distant analogy, instead of present vision; and it was recognized not by one special and invariable suffix for each sex (o or a), but by the character of the derivative suffixes themselves; e.g.-on masculine, -lön feminine ; -tör masculine, -tric feminine, &c.; -ús or -ěs neuter. So again some suffixes were confined primarily at least to adjectives, e.g. -tili: others to substantives, e.g. -on, -1on.

Gossrau (Lat. Gr. $ 86, p. 92) has called attention to the connection of the genitive plural with the accent, and proposed the following rules: •(1) All pari-syllables, as belonging to the i de• clension, have -lum. (2) All words, which with the ending in •-ium need not draw the accent forward from the syllable on which it falls in the genitive singular or nominative plural, have ium; 6 others have um. Or the rule may be thus stated: all words which in the genitive singular have the penultimate syllable long have

-lum, those which have it short have -um. This rule,' he adds, • is good also for all adjectives.' But there are some considerable exceptions, as he acknowledges, to these rules.

In my opinion the only truth, contained in these rules, is what I have before referred to; viz. that the consonant stems are to a considerable extent stunted i- stems, the Roman law of accentuation exerting a constant influence to shorten the word at the end, and this particularly, when the penultimate syllable is short.

Verbs with vowel stems.

Some readers will probably be surprised at seeing the final vowel of some verb-stems marked as short; e. g. domă-, mone-, faci-, and others of the classes to which these belong. My reasons for regarding them as short are these.

To take first the case of e stems? (1) A few verbs with e

? A very competent comparative philologer, Grassmann, has already taken a similar view, and on much the same grounds (Kuhn's Zeitschrift, XI. p. 89).

« PreviousContinue »