rakshati, ‘again he preserves.' Analogously, gí-ratha (i. e. fTM gir ratha), 'epithet of Brihaspati.'

f. Analogously to 69. c, + éatur+taya becomes a ćatush-ṭaya, ‘the aggregate of four.'

72. Prefixes such as nir and dur must be treated as originally ending in s; see nis, dus, 69. a.

73. r preceded by a vowel may optionally double a consonant immediately following; thus निर् दय nir daya may be written either निर्दय nirdaya or निर्द्दय nirddaya, ‘merciless;' excepth and a sibilant followed by a vowel, as in 71.6; but karshyate may be written karshshyate. In doubling an aspirated letter, the aspiration of the first is rejected, as (for w).h is said to have the same effect in doubling a consonant immediately following; thus brahman may be written brahmman; but for the sake of simplicity it is better to avoid doubling in both cases, and write always nirdaya and brahman.

a. The doubling of consonants, when they come in contact with others, is constantly allowable in Sanskrit, though not usual in practice, Thus, in any conjunction of two (or even more) consonants preceded by any vowel, especially if a semivowel be the last letter in the compound, the first letter, provided it be not Tor, may be doubled (Páņ. vIII. 4, 47); thus may be written for y, "ha for मध्वत्र, इत्त्याकर्ण्य for इत्याकर्ण्य, but the more simple form is preferable.

b. Again, any one of the first four consonants of any class may be doubled before the nasal of its own class, and if this takes place the middle consonant is called the yama of the preceding; thus in kkniti (Pán. 1. 1, 5) the second k is the gama or twin letter.

c. It should be noted that by Pán. VIII. 4, 65, there is an optional rejection of one of two homogeneous consonants after any consonant, so that fa may be written f.

The following table exhibits the more common combinations of consonants at one view. In the top line of initial letters the aspirated consonants have been omitted, because it is a universal rule, that whatever change takes place before any consonant, the same holds good before its aspirate,

H

48

ák kg gk ég jk t g dn

át kdgé

nk pg b n m g yg rg lg v k

bjjt td dn nt pdb n mdydr l l d vé

sk sg gh

ch t

s d dh

43. a.

n. pn bn

mny nrl in v

n

sn h

nt ch

55

n is only doubled if preceded by a short vowel.

† A final n before & and j is often allowed to remain unchanged.

CHAPTER III.

ON SANSKRIT ROOTS AND THE FORMATION OF
NOMINAL STEMS.

BEFORE treating of the declension of Sanskrit nouns (náman or sanjňá), it is necessary to point out the peculiar method of forming

the stem from the root.

74. Every Sanskrit noun (including substantives, adjectives, pronouns, and numerals) has at least two distinct states prior to the formation of the nominative case; viz. 1st, a root (dhátu); 2ndly, a stem (prátipadika or anga*) formed directly from the root or from a modification of the root, generally by the addition of a suffix (pratyaya); which stem becomes a complete word (pada) by the addition of a case-ending (vibhakti)†.

a. The root is of such importance in Sanskrit that it should be clearly defined before another step is taken.

A root (dhátu) is to language what the primitive elements are to chemistry; it is that primitive part of a word which, being incapable. of grammatical decomposition, is supposed to contain the primary meaning antecedent to any addition or modification. When a root has been developed in any way by the addition of letters or syllables or by internal change it becomes a stem, which again is subject to further development by the addition of letters or syllables called case-endings or inflexions (vibhakti), whether nominal or verbal. Thus dána and dadá are stems (the former nominal, the latter verbal) developed out of the root dá, but dána and dadá are not fully

According to Pán. 1. 4, 13, the term anga is used for the stem when speaking of some suffix (pratyaya) or termination which is required to be added to it, whereas prátipadika is a general term for a stem without reference to its suffix.

†The process of forming a complete word (pada), in the case of nouns, may be shewn, as it were algebraically, thus: Root (dhátu)+Suffix (pratyaya) = Stem (prátipadika); again, Stem (prátipadika) + Case-ending (vibhakti) = a complete word (pada); e. g. in the word jan-a-s, ‘a person,' jan is the root, a is the suffix, and s is the masculine termination for the nominative case.

developed until they have received terminations or inflexions, when they become complete words (pada); thus dána-m, ‘a gift;' dadú-ti, 'he gives' (cf. Lat. do-nu-m, Gr. di-dw-01).

b. There are in Sanskrit about 2000 roots, and every one of these conveys some simple idea, conveniently expressed in English by the sign of the infinitive 'to,' as in ad, 'to eat,' though it must be noted that the simple root ad only denotes the idea of eating,' which appears under different modifications in its derivatives (see 76. a). The following are a few of the commonest roots, with the leading idea conveyed by each (omitting ‘to'):

75. A cursory glance at the above list of common roots will serve to shew that they are all monosyllabic. In other respects they differ. Some consist of a single vowel only; some begin with one or two consonants, and end in a vowel, but none end in either a orau; some begin with a vowel, and end in one or two consonants; and some begin and end with one or two consonants, inclosing a medial vowel; so that a root may sometimes consist of only one letter, as i, 'to go;' and sometimes of four or more, as skand, to move.' Roots consisting of simple letters, such as I, 4, 3, fa, 39, &c., are probably primitive; and those which have compound consonants, such as &c., are in all likelihood developed out of more primitive forms*. Those with cerebral letters, such as 'to roll,' have some of them been formed by adopting sounds from aboriginal dialects.

a. The few polysyllabic words recognized as roots have probably resulted from a constant habit of joining some particular preposition with some particular monosyllabic root till it has at length come to be regarded as part of the root; e. g. in sangram, to fight,' q avadhír, 'to despise,' the prepositions sam

and ava have combined thus with the root. A few other polysyllabic roots are the result of the constant habit of reduplication; (as, ¿f daridrá, to be poor;'

6

jágri, 'to be awake;' ćakás, to shine;' veví, 'to go,' 'pervade ;') and a few are derived from nouns; as, to play,' from THIT kumára, a boy.' Most of the latter are of the 10th class, and may be regarded as nominal verbs (see

288. b).

* Thus

séyut (also written séut), 'to drop,' beginning with three consonants, was probably merely developed out of rts. éyu, éyut, a sibilant and dental having been added (cf. 51, 53, 84. III).