personal possession. But in Sanskrit it is only those pronouns whose personality is strongest, namely, the first and second singular, which thus tend to impart themselves to that which is connected with them by the genitive relation; and when that connection is thought less closely their dative is used for the genitive. In whichever way the genitive is expressed in these two pronouns it is thought quite differently from the ablative, and this gets its own proper expression in -at.

9. The two genitive endings -sya and -as might suggest the conjecture that the genitive element had originally a fuller form syas; and such a supposition would be supported by the Latin genitive -ius, which would correspond to -yas, also by the old genitive of second personal pronoun σεοῦς (64). There may possibly be also a trace of an original n in the genitive plural in Sanskrit, in which n takes the place of the s of the genitive (13), and this would lead to the supposition that the original form was -syans, which would be very similar to the Sanskrit comparative ending iyans, and would probably have a similar significance of production or increase. But the n of the genitive plural is more probably due to weakening and softening of the inflection in the noun by the preceding long vowel (209).

The inflection of the nominative plural, like that of the genitive singular, involves i as well as -as, which may be seen in the a- stems of the Sanskrit pronouns, and in all the corresponding stems in Greek and Latin. And this would lead to the supposition of an original ending -yas (164). Now this supposition is countenanced by the old Latin nominatives vireis, gnateis, populeis, ministris,1 and by the Greek huis, bus, from the a stems asma, yusma.

The Vedic nominatives also in -sas,1 to which the Zend correspond, suggest a further addition, and lead to the conjecture that the original ending was -syas (see also 113).

The genitive ablative ending of the feminine stems ending in a and 7 involve a thought of the noun, or, in other words, a pronominal element referring to the noun, for the a is evidently lengthened by the gender of the noun. This renders it probable that the a of the ablative ending at is pronominal also.

10. Just as the genitive and ablative relations are thought more fully with a stems masculine and neuter than with the others, so is the instrumental relation, which gets with them in the singular its fullest expression ina. With all the other stems it is reduced to ā, which seems to absorb the prolongation of feminine -a. The use of the instrumental case ending in adverbs, which express direction of motion, suggest that perhaps its original meaning is along of (225), motion according to the way defined by the stem. It expresses not only the instrument, but the manner.

11. The dative ending also has in the singular its strongest form -aya with the masculine and neuter a stems. The dative singular of the second personal pronouns suggests that the original form perhaps was ab yam, m denoting the object, and ab ya the proximate, akin

1 Bopp Vergl. Gram., sect. 228 b.

to Skr. abi to, o by; and this is confirmed by the dative dual, in which also the masculine and neuter a stems have the strongest forms, for the a- and -i of the feminine is not case, but gender.

Feminine stems in a take y before all the case endings which begin with a vowel in order to preserve their final vowel; and neuter stems in i and -u take n for the same purpose, the quiescence of n suiting their lifeless nature. But masculine and feminine stems in -i and -u do not seek to preserve these vowels; what they add to the radical or attributive part being perhaps a less important and weaker element, while the neuter attaches to it an important element, a sense of a lifeless thing.

Feminine stems ending in a vowel which they lengthen in the nominative and accusative singular to express feminine gender, lengthen also the dative and locative case endings, as well as those of the genitive and ablative (9), showing that these case endings also are thought with attention directed to the substantive so as to take up its gender, the a of ai, am, au, being pronominal. The close implication of the case ending with the substantive stem is highly characteristic of these languages. Thus the plural case endings end in s, except the genitive and locative and some of the nominatives and accusatives, and this s is evidently expressive of the plural, so that the case relation affects not the plurality but the individual, and the individual as affected with that relation is pluralised. In the dative ablative plural the i is probably due to y assimilating to itself the vowel which precedes b. In the instrumental dative ablative of dual, the ā is peculiar to the dual, and must be expressive of it, so that the first and strongest part, aby, of the compound case relation ab'yam, penetrates to the individual, and is followed by the dual prolongation of the vowel. This is like what is found in the Hyperborean languages of Europe and Asia.

12. The element of duality, -i or -au, is similar to the element of locality -i, -um, or -au, and both involve a common element of thought, juxtaposition.

The dual au is doubtless akin to dwa, the stem of the second numeral (184). And the essential element of coupling in dwa is u, the a being the substance (Def. 4) of the couple.

It has been already said (11) that in the instrumental, dative, ablative, a expresses duality (V. 51), and in the Veda a occurs as the ending of the nominative dual instead of au, probably pronominal; but u added to a expresses it more fully, just as in Arabic the element of the second numeral n is added to a to express duality. The vowel i is itself significant of juxtaposition or proximity, as may be seen in the Sanskrit prepositions ad'i, api, ab'i, pari, prati. And in one application of this idea i might be a dual ending, while in another it is a locative ending, the dual requiring always its final vowel to be long or diphthong. In the locative of some of the pronouns it is strengthened with n, the ending being in, which reminds of the preposition in siv. Another locative ending appears in its full form -swa in the locative plural in Zend. In some old words of kindred swa or x'wa appears as if it were a preposition signifying with, and akin

to sam, x'am, which signify with; thus in Sanskrit swag'ana cognatus, x'wax'ura socer, Goth. swaihra Teveegós, Sanskrit swasri soror, Lat. sobrinus. From swa a locative ending ău might come, and from the other form, sam, might come another locative ending -am. For the case endings take the most essential element of roots which best express the relation with the nominal stem. And a pronominal a prefixed as in genitive and ablative would give au and am.

Now, as in the dative dual, the duality is between two particles of relation; so in the locative dual of the a- stems, the locative i gets between the stem and the dual o, but needs to be confirmed by the addition of s, a fragment of swa. In Sanskrit the locative relation to a dual noun coincides with the genitive. In the locative plural the plurality s is similarly between the particles of relation, namely, the more general particle i and the more particular swa; but it destroys the locative expressiveness of s, and this has to be supplied by wa or u.

Bopp derives the Greek dative plural -10 from the Sanskrit locative -isu, but iv, iμiv, opiv, opion, indicate a nasal, such as belonged to the Sanskrit dative, not only in the dual, but originally also in the singular (11); and it would be analogous to the dative dual to suppose that the original ending of the dative plural was b'yasam, the weight of which caused an abbreviation, and afterwards an obliteration of the second syllable with nouns; and the Latin dative -bus is from the Sanskrit dative. Bopp admits that the dative singular in Latin corresponds to the Sanskrit dative, though he strangely supposes the dative singular in Greek to be the Sanskrit locative. But both are in Greek, as dixos Sanskrit vex'è locative, díxwvxaya dative.

13. The ending of the genitive plural of the demonstrative pronouns is -sām, which corresponds to Latin -rum, and in the substantives it is -nām or -ām. In the first and second personal pronouns the ending is -kam, in which perhaps, as Bopp suggests, k is borrowed from a possessive formation, but it cannot be the neuter of such a formation as he conjectures. If k is possessive in -kam then probably s is genitive in -sām, and -ām is plural, the a being lengthened by strong sense of plurality; the final a of masculine neuter demonstrative stems being changed to ē, seems to indicate syām (11, 156). This analysis of -sām may be confirmed by an analogy. The ordinals of the higher numbers are formed with -ma or -tama, expressing that special one of the number reckoned, which completes it as an aggregate (82). Also Sanskrit ēkatara means one of two, hatama means one of many, -tar being expressive of the step of transition from one to another in an alternative of two, as in uter, alter, &c., or in a relation of kinship, -tam expressing the step from many to one, in thinking which, the many are massed in an aggregate, which m expresses as in oμou, äμa, &c., Skr. sam with, sama all whole. Another analogy is in Latin multesimus, one of many parts, a small fraction.

If this analysis of the ending of the genitive plural be correct, then in it too case has got inside number; the genitive element being

1 Bopp. Vergl. Gram., sect. 177.

2 Ibid. sect. 340.

reduced to s or n, and the reduction compensated by lengthening the final vowel of the stem, and the plural being ām.

14. In the nominative, vocative, and accusative plural of neuter stems the plural ending is i, and there is a tendency to lengthen the preceding syllable as if to increase the substance by massing into an aggregate rather than by noting the individuals.

It is to be observed that in some substantives and adjectives, masculine and feminine, the stem has a fuller form in the nominative, accusative, and vocative singular and dual, and in the nominative and vocative plural than in the other cases, because in these other cases thought is attracted from the stem by the stronger subjoined element so as to reduce the sense of life in the stem.

There is less distinction of case relations in the dual than in the plural, and in the plural than in the singular. For the relation is less distinctly thought when the transition is to different objects at the same time; and in the dual this cause of indistinctness is greater because the twofold individuality is fully thought, whereas in the plural the individuals are more merged in the plurality; in neither is the relation so distinct as with a single object.

It is to be observed that the nasal which expresses the accusative relation, -m singular, -ns plural, is in the plural preserved only with the masculine -a, -i, -ri, and -u stems; because the relation is more strongly thought with masculine nouns (143), and with these stems it is expressed only by n, with the other masculines by a for euphony.

15. The Sanskrit verb shows a remarkable sense of the process of the being or doing; for this is what the conjugational elements express. They are confined to the present parts of the verb, namely, the present tense, the potential, which is a potential present, the imperative, which is an imperative present, the imperfect or past present, and the present participle. These differ from the other parts of the verb in thinking the act or state as going on or in its process, and it must be this element variously thought according to the idea of the act or state, which the various conjugational formations express (III. 93). Now in about two-thirds of the primitive verbs of the language this element is taken up into the root, so as to suggest a comparison with the Syro-Arabian languages whose special characteristic is their expression of the process within the root. The difference, however, between these languages and Sanskrit is at once apparent when it is seen that the Sanskrit root takes up only an abstract sense of process which is partly expressed outside the roots affected with it, and that many verbal roots and forms in the language are not affected with it at all, whereas all the Syro-Arabian verbs take it up in all its fulness.

The form in which the process is for the most part taken up by the root in Sanskrit is Guna of the vowel of the root (2, compare IV. 108); but this cannot be applied if that vowel is a, or if it be followed by two concurrent consonants, or if it be a long vowel, unless it be final. In each of these cases Guna or incorporation of a would give excessive length, suggestive rather of quiescence than

of movement (3), but with a long final vowel this effect is escaped by the vowel turning into a semi-vowel before the a which follows. The first conjugation is of those roots, about 1000 in number, which take Guna and subjoin a.

The second is of about 70, which take no conjugational element. The third is of about 20, which reduplicate the initial consonant, using for it in the reduplication syllable the unaspirated consonant corresponding to it if it be an aspirate, and the corresponding palatal if it be a guttural; but if the root begin with s, followed by another consonant, it is the second that is reduplicated; the vowel of the reduplication syllable is the short vowel corresponding to that of the root; i is used for yi and sometimes for a.

The fourth conjugation is of about 130 roots, which subjoin ya. Many roots form neuter verbs in the fourth conjugation, which in another conjugation form transitive verbs.

The fifth includes about 30, which subjoin nu.

The sixth includes about 140, which subjoin a.

The seventh includes about 24, which insert n before their final consonant.

The eighth includes about 10, which subjoin u; 9 of them end

n or n.

The ninth is of about 52, which subjoin nī, or before vowels n.1

The tenth conjugation is that of several roots as verbs simply active, and of all causals. It Gunates the vowel of the root when not final, Vriddhies it when final, and generally when it is a between two consonants, and subjoins aya, before which p is inserted if the root ends in a, or in ē, ai, ō, changeable to a, and therefore incapable of Vriddhi; other roots in ai also insert p, but most others in ē or ō insert y. This conjugation differs from all the others in this respect, that the affection of the root and the subjoined addition to it are not confined to the four conjugational parts of the verb, but are carried throughout it except in the precative Parasmai and the aorist (27. 7), which drop aya; the final a, however, of aya is dropped before the i which is taken in all the non-conjugational parts. It is a derived verb rather than a conjugation, and can be formed on any verb.2 The pinserted after a seems to belong to the causal element, being brought to light to preserve a and a; paya is perhaps akin to the root of ποιέω. The causation enters into the root, increasing its vowel unless when this would make its length excessive, as when that vowel is a followed by two consonants. When it produces Vriddhi it makes itself felt as dominating the root, which is passive to it.

16. The process which is expressed throughout the conjugational or present parts of the verb is the process of being or doing which the subject realised. In the third conjugation it is probably thought in its totality as the complete process of accomplishment, being expressed by reduplication. But there is another aspect of the succession of being or doing which gets expression in the Sanskrit verb; this is the going on or process thought as of the life of the