ration of the affixes by which the bases of nouns are formed should precede their inflection. The early student, however, may satisfy himself by a cursory observation of the eight classes under which these affixes are distributed. Some of the most uncommon, which are only applicable to single words, have been omitted. Moreover, in accordance with the practical character of the present Grammar, the servile and indicatory letters of Indian grammarians, under which the true affix is often concealed, if not altogether lost, have been discarded. For example, the adjective dhana-vat, rich,' is considered in the following pages to be formed by the affix vat, and not, as in native Grammars, by matup; and the substantive bhoj-ana, food,' is considered to be formed with the affix ana, and not, as in native Grammars, by lyut.

In my explanation of the inflection of the base of both nouns and verbs, I have, as before, treated both declension and conjugation as a process of Sandhi; that is to say, junction of the crude base, as previously formed from the root, with the terminations. But in the present Grammar I have thought it expedient to lay more stress on the general scheme of terminations propounded by native grammarians; and in the application of this scheme to the base, I have referred more systematically to the rules of euphonic combination, as essential to a sound acquaintance with the principles of nominal and verbal inflection. On the other hand, I have in the present work deviated from the Indian system by retainings as a final in the declension of nouns and conjugation of verbs, for the practical reason of its being more tangible and easy to apprehend than the symbol Visarga or h, which is imperceptible in pronunciation. (See the observations under changes of final s, pp. 32, 33.) Even in native Grammars those terminations, the finals of which are afterwards changed to Visarga, are always regarded as originally ending in s;

b 2

and the subsequent resolution of s into h, when the termination is connected with the base, is a source of confusion and uncertainty. Thus s is said to be the termination of the nominative case; but the nominative off agni, ‘fire,' would according to the Indian system be written : agnih, which is scarcely distinguishable in pronunciation from the base agni. In the following pages, therefore, the nominative is given agnis; and the liability of agnis to become agnih and agnir is explained under the head of changes of final s (at p. 33). This plan (which is that of Professor Bopp) has also the advantage of exhibiting the resemblance between the system of inflection in Sanskrit and Latin and Greek.

The difficulty experienced in comprehending the subject of Sanskrit conjugation has led me to give abundant examples of verbs conjugated at full. I have of course deviated from the Indian plan of placing the third person first. I have, moreover, deemed it advisable to exhibit the English equivalents of Sanskrit words in the principal examples under each declension and conjugation, knowing by experience the thankfulness with which this aid is received by early students, not thoroughly familiar with the Devanágarí character. The numerous examples of verbs, primitive and derivative, will be found to include all the most useful in the language. In previous Grammars it has been usual to follow the native method of giving only the 3d pers. sing. of each tense, with an occasional indication of any peculiarities in the other persons. The present Grammar, on the other hand, exhibits the more difficult tenses of every verb in full, referring at the same time for the explanation of every peculiar formation to the rule, in the preceding pages, on which it depends. This is especially true of the 2d and 3d preterites, as these constitute the chief difficulty of the Sanskrit verb; and I have constantly found that even advanced students, if required to write out these tenses, will be guilty of inaccuracies, notwith

standing one or two of the persons may have been given for their guidance.

In the chapter on compound words I have again endeavoured, without ignoring the Indian arrangement, to disembarrass it of many elements of perplexity, and to treat the whole subject in a manner more in unison with European ideas. The explanations I have given rest on actual examples selected by myself from the Hitopadesa and other standard works in ordinary use. Indeed this chapter and that on syntax constitute perhaps the most original part of the present volume. In composing the syntax, the literature as it exists has been my only guide. All the examples are taken from classical authors, so as to serve the purpose of an easy delectus, in which the learner may exercise himself before passing to continuous translation. The deficiency of native Grammars on this important subject is only to be accounted for on the supposition that their aim was to furnish an elaborate analysis of the philosophical structure of the language, rather than a practical guide to the study of the literature.

The exercises in translation and parsing, in the last chapter of this volume, will, it is hoped, facilitate the early student's first effort at translation. Two fables from the Hitopadeśa are given, as before, with a translation and grammatical analysis; but I have thought fit to omit the story of Vedagarbha and the selections from Manu, which I appended to my first Grammar. The Sanskrit of the former is too modern and interspersed with Bengálí idioms, while that of the latter is too advanced. I have therefore substituted for the one some easy sentences selected from classical sources; and for the other, a few simple fables from the Pancha-tantra, the book from which a great part of the Hitopadeśa itself is drawn. Every word in these selections is explained either by notes at the foot of the page or by references to the preceding pages of the Grammar.

The separation of words by the free use of the Viráma, and the employment of a dot underneath to mark the division, whenever the blending of vowels or the association of crude bases in a compound make junction unavoidable, may offend the eye of the Oriental scholar, if habituated to the Indian system of writing; but the beginner can scarcely be expected to know which is the final and which the initial letter of words thus joined together. Why, therefore, refuse him a clue to guide him in his search for the word in the dictionary? and why, by uniting those parts of a sentence which admit of separation, superadd an unnecessary source of perplexity to the necessary difficulty, unknown in other languages, resulting from the blending of vowels and the composition of words? It may be quite true that, according to native authorities, the Viráma ought only to be employed when no Sandhi takes place; and that, according to the strict interpretation of the word Sandhi, actual contact ought to ensue whenever a law of euphony comes into operation. But does euphonic connexion necessarily imply contact? and may not words be mutually affected by euphonic laws, without being actually joined together?

The system of uniting words which are really distinct may commend itself to the natives of Hindústán, as tending to reduce the labour of writing; but in Europe, where abundant punctuation is deemed essential to facilitate reading, the absence of spaces must always be regarded as productive of unnecessary hindrance. The student has already sufficient obstacles to surmount in the Deva-nágarí character and the rules for the permutation of letters. The changes required by these rules will cause no embarrassment, provided separation be permitted, in accordance with the European method. Thus the Latin scholar, if acquainted with the laws of permutation, would not be embarrassed by the sentence Uby ad Dianæ venerir itav at sinistram

(euphonically changed from ubi ad Diane veneris ito ad sinistram); but he would, to say the least, be unnecessarily hindered if this permuted sentence were linked together into two words, thus-Ubyaddianæ veneriritavatsinistram. Nor is it easy to understand why the slight spaces between the words in the first case should be deemed incompatible with the operation of euphonic laws. If such separation, therefore, is only to be effected in Sanskrit by extending the legitimate functions of the Viráma, the facilities afforded by modern typography ought to leave us free to do so. The only cases in which it is undesirable to separate distinct words, acted on by Sandhi, are when two vowels blend into one, and when final u and i are changed into their corresponding semivowels v and y.

In regard to the general scope of the book, it remains to state that my aim has been to minister to the wants of the earliest as well as the more advanced student. I have therefore employed types of two different sizes. The larger attracts the eye to those parts of the subject to which the attention of the beginner may advantageously be confined. The smaller generally contains such matter as offers no claim to immediate consideration.

Under the conviction that the study of Sanskrit ought to possess charms for the classical scholar, independently of its wonderful literature, I have taken pains to introduce in small type the most striking comparisons between this language and Latin and Greek. I am bound to acknowledge that I have drawn nearly all the materials for this important addition to the book from the English translation of Bopp's Comparative Grammar, by my friend and colleague Professor Eastwick.

One point more remains to be noticed. The want of an Index was felt to be a serious defect in my first Grammar. This omission is now supplied. Two full Indices have been appended to the present work, the one English, and the