Thus naras, "a man," consists of three parts, nar+a+s, where nar is the root, a the suffix, s the termination; nar+a=nara is the stem.

The terminations vary with every relation or connexion of the noun, and their variations constitute what are called cases. In the modern languages the terminations of the Sanskrit noun no longer exist in their original form; in some of the seven languages no trace of them is perceptible in the singular, and only weakened and half-obliterated traces remain in the plural; in none of the languages are more than slight and doubtful indications of their presence still observable. The variations of case are expressed in the modern languages by particles placed after the stem, just as in modern European languages they are expressed by particles placed before the stem. The English declension, fish, of fish, to fish, stands in the same relation to the Gothic fisks, fiskis, fiska; and French champ, de champ, à champ, in the same relation to Latin campus, campi, campo; as Hindi nar, nar kâ, nar ko, does to Sanskrit naras, narasya, narâya.

The first point then to be considered is the formation of the stem, and in this inquiry our attention is entirely concentrated on the final syllable or suffix. When treating, in the First Book, of Phonetic Changes, I exhibited, as far as the materials at my command permitted, those alterations and corruptions which have operated to change the form of the root; and the next step is to discuss the processes by which the stem, consisting of the combination of root and suffix, has been built up in its manifold developments.

Sanskrit grammarians divide nominal stems into two classes. They do not admit the possibility of a noun being a primitive word, derived from nothing else, but hold that all the words in their language are derived from verbal roots (called dhâtu), by additions and changes of various kinds. The first class, therefore, consists of those nouns derived from the dhâtu directly:

these they call kṛidanta; the second, of those derived immediately from other nouns: these they name taddhita. I shall use for kridanta the term "primary stem"; for taddhita, "secondary stem."

§ 2. Derivation is effected by the addition to the dhâtu, or root, of certain syllables called pratyaya, or suffix. These suffixes in many cases not merely add a vowel or syllable to the root, but also effect some change in the root itself, either by lengthening the vowel, eliding a final consonant, or in some other way. In order to indicate the effect the suffix has on the root, the Indian grammarians have prefixed or added to each suffix certain letters, which serve as a sort of memoria technica. Thus पाक “cooking,” would be described as पच् + घञ्, that is, derived from the root "to cook" by the suffix ; the effect of this suffix is principally to add a final to the root, but it also lengthens the root-vowel, and this effect is indicated by the letter ; it moreover changes a final palatal into a guttural, and this is denoted by the . The number of pratyayas is very great, the grammarians having multiplied them without stint, in order to give expression to the large mass of minute distinctions with which they have encumbered their writings. As, however, we are not discussing Sanskrit grammar, but only the simpler grammar of its modern descendants, it will not be necessary to consider all these suffixes in detail. They may be conveniently thrown into groups according to the vowel or syllables which they actually add to the root, the technical letters being referred to only when the distinction which they mark is of importance in its bearing on the forms of the spoken languages. Among the kṛidantas or primary stems the Indian grammarians include participles and verbal forms of all kinds, as well as abstract nouns. It will, however, be more in accordance with the principles of linguistic science to reserve the consideration of participles and all verbal formations till the

chapters on the Verb are reached, and to discuss in this place merely those suffixes which form substantives and adjectives.

Suffixes are monosyllabic or dissyllabic. Out of the vast mass of suffixes mentioned by the Sanskrit grammarians, many may be dismissed as unconnected with the present subject, inasmuch as the stems formed by them were so formed before the rise of the modern languages, and the question of their construction is a matter not of modern, but of ancient, grammar. Our seven languages took over these words ready made, and, as far as we are concerned therefore, they may be regarded as primitive words. Only such suffixes will here be introduced as have left traces in the speech of the present times; and if it be necessary to speak of certain of them which are of purely Sanskrit use, it is merely because of their bearing on phonetic or structural processes of recent introduction.

§ 3. Stems in -a. The nouns of this class are of all three genders, and make in the nominative case of Skr., masc. -as, fem. -â, neut. -am. The s of the masc. nom., however, is not permanent; before a word beginning with a sonant letter, it, together with the a of the stem, changes into o. Inasmuch as the sonant letters in Sanskrit are more numerous than the surds, the form of the nominative case in -o was much more frequently heard than any other; and the vulgar, who are not careful of minute grammatical distinctions, appear even in the earliest times to have used the termination in -o to the exclusion of all others.. Thus Vararuchi (v. 1) gives it as a general rule that o is substituted for su (= as), the technical name of the nom. sing. ending, in all words whose base ends in a, as vachchho, vasaho, puriso, for rṛikshas, vṛishabhas, purushas. He considers that the a of the suffix has been elided, though it is more probable that the whole termination -as changed into -o through the intermediate form -ah, the labial vowel owing its origin to the involuntary contraction of the

1

lips during the passing of the soft breathing indicated by the visarga. In Magadhi Prakrit the nom. is sometimes formed in -e, and in Apabhransa in -u.

In more recent times two separate sets of stems developed themselves out of this a-stem.

The first probably ended in u in all the languages down to about the fourteenth century; since then it has ended in -a in all but Sindhi, which still retains the termination in -u. In old Hindi and Marathi the nominative in u is distinctly traceable, and in the former is even lengthened to û at the end of a verse where the metre requires a long syllable. This is especially the case in Tulsi Das's Ramayan, a work which exhibits a wide range of popular forms and phrases. Thus we have शरीरू, बीरू, कपारू, प्रचारू, at the ends of rhyming lines for शरीर, बीर, in modern H. TH, sarir, bir. The second ended in o, which in the शरीर, बीर, broad pronunciation of old Hindi sounds au, though the form in o is often used by the earlier poets; and there is no distinction between the two: it is merely a matter of manuscripts, some using one, some the other form. This o softened subsequently still further into â, in all but G. and S., which still retain o.

Concerning the cause of the parallel existence of the two forms there has hitherto been much doubt. Dr. Trumpp considers it a mere matter of accident, and thinks the presence of one form or the other is due to habit and daily usage.1 Dr. Hoernle, however, holds, and with justice, that this reasoning is insufficient, and would ascribe all words which exhibit the áo termination to the forms of nouns with an added, which are extremely common in Prakrit. To this opinion I in the main subscribe. I think myself fortunate in this second volume

1 "Eine Regel scheint bei dieser Separation nicht vorgewaltet zu haben, wenigstens habe ich bis jetzt noch keine entdecken können, sondern der tagtägliche Gebrauch scheint sich für die eine oder andere Endung entschieden zu haben."--Zeitsch. d. D.M.G. vol. xvi. p. 131.

to have hints from other and better scholars than myself, which were wanting when the First Volume was written.

There is, however, one other possible way of accounting for the peculiarity. This is the influence of the accent.1 It is vigorously denied by some that the Sanskrit accent has persisted until modern times; but it may be urged that the idea is a new one, not altogether without support in facts, nor in any way à priori impossible, and it is not quite fair to dismiss it contemptuously without due consideration. The influence of the accent is naturally confined to words of the early Tadbhava class, namely to words which have come down uninterruptedly from those times when Sanskrit was spoken, and whose form depends on the ear, not on the eye. This is all that is claimed for it it is not pretended that the Sanskrit accent has been preserved in words resuscitated from the written language by learned men, centuries after it had ceased to be spoken.

With these reservations, the reader is invited to examine the lists hereinafter given, given not in a spirit of dogmatism, but merely because I cannot satisfy myself in rejecting accentual influence in them, till I am met by something more convincing than sneers; though in the desire to arrive at the truth, which is the only object which these researches can possibly have for one in my position, I shall be ready to throw down my weapons and acknowledge myself beaten directly I feel myself fairly convinced.

In any case, whether the influence of the accent be admitted or rejected, the collection of words now given will be useful as

1 See Vol. I. Ch. I. § 6, pp. 17 ff.

2 Unfortunately little is known of the spoken accent in Skr. In the following pages, the rules laid down in Bopp's Vergleichendes Accentuationssystem have been my guide; but perhaps since that was written advances have been made by European writers. These, in my isolated position in the depths of an Indian province, I have no means of obtaining. I am, however, quite conscious of the unfavourable effect isolation has on this, and all other parts of my work, and only wish my critics would bear it in mind sometimes.