lay down any rule for its insertion or omission, unless it be that where the initial letter of the second member is weak, â is inserted; where it is strong, omitted. Compare H. "twentytwo” for ब + वीस with बत्तीस “ thirty-three” for ब + तीस ; but even this rule hardly carries us through all the series. The Prakrit had originated these compound forms long before the modern languages came into existence, and the secret of the composition must therefore be looked for in that stage of speech; and the data for Prakrit numerals are unfortunately so defective, that as yet I have been unable to formulate any rules on the subject. Marathi introduces a joining vowel e in the series with "forty," and in the "seventy," "eighty," and "ninety" series; but the reason of this is not easy to find.

Above "one hundred" composition ceases, and the words are written separately, except in M. G. and S., which, from "onehundred-and-one" to "one-hundred-and-nine" can combine the lesser numeral with the Skr. "above," thus:

“One-hundred-and-one, ” M. एकोत्तर शें, S. एकोतर सौ, G. एकोत्तर सो.

§ 27. The ordinals of all the languages are formed each on its own basis for the first four numerals, but after that follow a regular system of genuine Sanskrit origin.

"First" has the following forms: H. uf, P. id., M. id., S. पर्यो, पिहियों, पिहिरों, G. पेहेलो. O and B. generally use the Skг. ч, but the H. f is gaining ground, and is heard among the lower orders constantly. If we derive these words from ¤, we are met by the difficulty of admitting the change of intoor, and I therefore suppose that they come from an unrecorded comparative, which would stand in the same relation to the superlative as Latin prior does to primus. Bopp (§ 293) has already established the fact that is the superlative of Я, and similarly (§ 321) he shows that all the cognate languages use the superlative form, as primus, πрwтos for Tрóтатоs, and eristo erst, from er eher.

So also our own "first," in Old-English fyrst, is the superlative of "fore." There must also, one would imagine, have been a comparative, and this could be nothing else than Ot.

“Second” is H. दूसरा, P. दूजा, दूआ, S. बिओ, बीजो, G. बीजो, M. दुसरा, 0. दुस्रा, but O and B often use द्वितीय. It is probable that the rude uncivilized Oriyas and Bengalis did not possess the idea of an ordinal till late times, when they naturally had recourse to Sanskrit to supply the want. Old-H. has also, which, together with the Panjabi form, are from Skr. द्वितीय, Pr. दू, also दोच्च, but the modern दूजा refers to the form, formed by hardening the into, and

then dropping one (Vol. I. p. 249). The S. and G. come from another Pr. form fafa, reduced by dropping the ; this is further shortened into . For the H. and M. दू and दुसरा, see under the next paragraph.

“ Third.” H. तीसरा, P. तीजा, तीआ, S. ट्रिओ, ट्रीजो, G. तीजो, M. तिसरा, O. and B. तिस्रा is in use, but तृतीय is common also. This ordinal is parallel to the last, and is derived from the Skr. by the same process as the words for "second." The form with its analogous seems to

be a recent compound of the numeral themes, at, with some word meaning progression, and probably connected with the root "to go," but there are no certain facts on which to found an opinion.

"Fourth" is regularly derived from the Skr., see Vol. I. p. 144.

From "fifth" onwards the ordinals are regularly formed by adding H. at, Old-H. Яt, P. at, S., G. ♬, M. a. B. and O., having no ordinals of their own, use the Sanskrit terms when necessary. All these terminations come regularly from the Skr. a, the termination of the ordinals.1 Hindi uses for "sixth", the regular Prakrit form of Skr. 48, and with this agrees G., but all the rest are regular.

1 See for further discussion of this question Chap. III. § 54.

§ 28. Besides the cardinal and ordinal series, the languages have other sets of numerals formed in ways peculiar to themselves. Such are the fractional numbers, in which all the seven languages are rich.

.

“A quarter” is H. पाव or पाउ, P. id., S. पाउ, G. पात्र, M. पाव, O पाए, पा, B. पोया (पत्र) ; the origin of all these is Skr. पादिक, P. पाइअ “a quarter,” the compound सपादिक = “with a quarter,” becomes in Pr. सवाअ, and in H. सवा, P. सवा, सवाइ, सवाइआ, S. सवाई, G. सवा, M. सव्वा, O. सउयाइ, B. सख्या (sau'd). Thus H. "three and a quarter," and so in all the rest. On the other hand, the deduction of a quarter is expressed by derivatives of Skr. 7, i.e. + “a quarter less." a127a, 912+ H. पौने, P. पउण, पडणे, S. पौणो, G. पोणा, M. पाऊण (nearer to the Skr. than any), O. utâ, B. id. Thus at “two and three quarters," or "a quarter less than three." In H. P. 0. and B. we appear to have the locative form, in the others a nominative.

one and

Two other numbers also are peculiar to this group: a half," H. from Skr., concerning which see Vol. I. p. 238, and "two and a half," H., P., S., G. अडी, M. अडीच, 0. अढाई, B. आडाइ, the origin of which seems to be Skr. +, to which M. adds the conjunction. For "three and a half," "four and a half," and the rest, the languages add to the complete numeral the word H. P. về, S. साडु, G. साडा, M. साडे, 0. साढे, B. साडे, from Skr. स + अर्ध ="with a half;" thus "three and a half" is H., "four and a half", and so on.

The other species of numerals, such as "once," "twice," and the rest, will more appropriately be described under the head of Adverbs.

CHAPTER II.

GENDER.

CONTENTS.-§ 29. NATURAL AND GRAMMATICAL GENDER.-§ 30. USE OF GENDER IN THE SEVEN LANGUAGES. 31. TYPICAL TERMINATIONS OF THE ADJECTIVE. 32. TERMINATIONS OF THE MASCULINE GENDER.-§ 33. TERMINATIONS OF THE FEMININE.-§ 34. TERMINATIONS OF THE NEUTER.-§ 35. FORMATION OF FEMININES FROM MASCULINES.-§ 36. GENDER OF WORDS ENDING IN CONSONANTS.- 37. DECAY OF GENDER-ITS CAUSE IN BENGALI AND ORIYA.

§ 29. GENDER is of two kinds, natural and grammatical. Natural gender is that which refers to living beings, and is threefold: there being one form for males, a second for females, and a third for mankind or animals regarded merely as such without reference to sex. The human mind has, however, not rested content with this simple and natural use of gender, but has, by an effort of imagination, extended the distinctions of sex to inanimate objects, abstract ideas, and, in short, to all nouns of every kind. All languages are not alike in this respect: some retain all three genders, others only two, and some have had the good fortune to emancipate themselves entirely in the noun, and to a great extent in the pronoun, from these awkward and cumbersome swaddling-clothes of speech. This is happily the case with our own beautiful and practical language, and is an advantage for which we ought to be deeply thankful to our Norman ancestors, whose keen common sense led them to reject much that was useless and unwieldy in the speech of our English forefathers.

The older languages of the Indo-European family have all

VOL. II.

10

three genders; those of the Semitic family got on very well from the beginning with only two, having never thought of developing the neuter. In the middle and modern IndoGermanic languages the German still retains all three genders, while most of the others have only two, and some none at all.

Natural gender exists in all languages, grammatical gender only in a certain number. In those which do not observe grammatical gender, natural gender is frequently distinguished by having two separate words for the two sexes; in those which acknowledge both kinds of gender, the distinction is affected by a variation in the form of the word, most frequently in the terminating syllable or vowel. In consequence of this habit, it has come to pass that the form of the word has created the gender; thus, in deciding what gender should be ascribed to a word which indicates an object from its nature incapable of classification by sex, the form of the word is the only guide; and if it happen to be of a form similar to that which is appropriated to the male sex in living beings, it will be classed as masculine; if to the female sex, as feminine; and if to neither, as neuter. Both classes of gender exist in the languages we are considering, but in a very different degree. Gender is in all of them indicated to a great extent by variations of form, especially of termination; but it must be observed that whereas substantives have, as has been shown in the last chapter, a very great range of terminations, adjectives and the participial forms of verbs have very few. It is principally by observing which form of an adjective or participle is used with any given substantive, that we can tell what its gender is. For instance, H. saṛak, "a road," is the name of a thing in itself incapable of natural gender; it is only by noting such phrases as bari saṛak, “a big road," saṛak banti, "the road is being made," that we discover it to be feminine. Hence it follows that in those languages which use very few participial forms, or whose adjectives have no distinctive forms for gender, we are unable to trace the