English
In cases where an initial vowel is obscured by sandhi or otherwise, it often writes that initial over the line : thus As may be guessed from these examples, this Bengali fashion of writing the — has often been the occasion of errors creeping into a text by misdivision. Trend of modem usage in respect of division.— The very first Sanskrit work printed in Nagari was Carey’s Hitopadesa, Seramporo, 1804, with an introduction by Colebrooke. That great scholar, in speaking (p. xv) of * the peculiar difficulties of the language said: ‘ To lessen one of those difficulties, which aiises from the frequent permutation of letters at the beginning and close of words, the editor has marked, by a dot under the syllable, places where the elision of a letter is found, or any other permu¬ tation, that is not obvious. So the Samkhya-pravacana-bhasya of Seram¬ poro, 1821. Coming to recent prints, Shankar Pandurang Pandit, in his four quarto volumes of the Atharva-Veda (1895-8), and in the second ed. of his Malavika, separates the words quite thoroughly. This he does by using virama with utmost freedom ( 3 ^^ 3 ^% I 7 ^^ I ^ 77^7 377 ^ | &c.), and by even disregarding the sandhi (^ | &c.). This is no hastily considered scheme of a novice, for he had printed an elaborate edition of Raghuvausa years before (1869-), with close adherence^ * VerzeichnisB, ii. 151. * Abb. d. Gottinger Gesell., 39. 10. ^ But ho does not shrink from using ^ at the end of an odd pada. xxxvi Notes on the Externals of Indian Boohs to the traditional continuity of writing. No one will deny his great learning as judged by high native standards. Pischel, in the preface to his Hemacandra (1876, p. xi), laments the necessity of printing the text in Roman letters. 'And yet,’ he adds, ‘ this way of publishing Sanskrit works has the one advantage of making the comprehension more easy by the more frequent separation of the words.’ His text shows very numerous violations of the rules of sandhi, but he is convinced that many of them go back to Hemacandra himself. Jacobi, Kalpasutra, 1879, p. 29, says: ‘The intei-pretation of Prakrit texts pre¬ senting numerous difficulties, I have thought it expedient to separate the parts of compounds.’ In his Maharastri Tales (1886, p. ix) he says: ‘ I have indicated the members of compounds by dots. . . . This will greatly facilitate the understanding of the text. The Hindus themselves, feeling the need of this help, occasionally separate the words by little strokes. It is the briefest kind of a commentarlus 'perpetuus^ and is, for simple texts, quite sufficient.’ ♦ Word-division in this volume. — 1 , By spacing, — The results of word- division may be considered under three heads, according as they are attained, (1) by simple spacing; (2) by free use of virama; and (3) by use of a vertical stroke above the line or of an inverted v beneath it, to show the cuts between the members of compounds and the points at which crasis has taken place.—1. The Bombay MBh. of 1878 prints ii. 39.4 a thus ; The same line in the ed. of 1889 is printed ^ ^ Not even the most conservative object to this. Bohtlingk ^ lamented that Bhatevadekar had run together the words which he could have separated by simple spacing. So did the Hindu, and spaced them in the second edition. 2. Word-division by use of virama.—No alphabet renders the sounds of a language with entire precision. Compare the beggarly outfit of half-a- dozen letters with which the English alphabet has to render the many vowels and diphthongs of spoken English. But the question whether we shall write or cTT^ ^^^n one of orthography (the pronunciation being the same in either case); it is simply a question of the habits of the scribes. These are notoriously inconsistent, even within the limits of the same MS.^ To illustrate, let me give the reading of a line or two of the MBh., first from my MS. of 6alya-parvan of 1620, second from the Bombay ed. of 1878, and last from the Bombay ed* of 1889. ^ Melanges Asiatiques, viii. 203 (1877). = Cf. the methods of marking Vedic accents, HOS. vii, p. exsi. By the Editm'^ of this Series It will be seen that it is the modern print and not the old MS. that avoids the virama and uses But the fact is not of the smallest significance, for in the print near by we find &c. Thus at ix. 8. 2: 1889 h It is in the highest degree significant that the scribe did not think of -71 ni~ as a single syllable: he thought of sarvdn as a separate word and finished writing it; and not till then did he begin nihanisyati ! For a like reason he has neglected the sandhi of rathdn^ although wnting it—with beautiful inconsistency—in the next word but one I Virama in the MSS.^—The MSS. make no difference, in their way of writing, between a medial and a final consonant. If they divide an aksara containing several consonants, the division is made according to the same principles whether within the word or at the end. Usually, indeed, they write the first consonant in the pada-form. But—to meet an objection that has been made—there are frequent examples of virama with the sarhhita- form, and that from my best MSS., bh and 4': thus, 50. 16; 52, 5; 59. 26 ; ^ 74. 8 ; ^5^® 76.13; 77. 10; 89. 1; 99. 6; 110. 20; 113.11; 123. 27 ; and so on. Indeed, we find occasional combinations like (120, 10) and (128. 28). At all events, these examples show that the method of division adopted in this text does not offend against the spirit of the Nagari writing. It is certain that the scribes of MSS. bh and 4^ were learned pandits, and men of conscience in matters phonetic. The editor of a printed text (unlike the scribe) must decide upon some consistent course. It would have been most culpable, in cases where even the native scribes leave us the choice between two ways of writing (both phonetically correct), to choose the less clear, simply out of deference to what is indeed the preponderating habit of the scribes, but of scribes who gave little heed to the matter. * From Dr. Bertel’s letter of Jan. 20,1908. xxxviii Notes on the Externals of Indian Boohs Virama before an initial vowel.—The syllables ta-nia-pa are pronounced alike, tam apa, whether written or In the latter method, one must fii’st annul the inherent 3 T of by virama,^ and then undo the annulment by wi'iting an initial By the former or Hindu method both processes are saved. But to a non-Hindu the ‘ saving* is as illusory as was that effected by use of the logotypes in the founts of old-time compositors. Even the Asoka inscriptions show initial ct after a final nasal. Virama after sounds not permitted as finals in the pause.—Certain sounds are forbidden in the pause: so c, j, y, r. It is doubtless for this reason that the Hindu shi-ank from writing vpn although precisely identical in sound with The c, y, and r aie not really in the pause (avasdne) at all, and are only apparently so to the eye. The spacing sepai-ates them only for the eye, not for the organs of speech, and is only the logical extension of a system of division which the Hindus themselves have adopted in part. The spacing may be a departure from the fashion of the scribes: yet no one condemns an editor for printing a ^ in words which the scribes write (and pronounce) with or visarga for upadkmdnlya ov jilivamuliyaj or for neglecting the doubling of a consonant after r. 3, Word-division at points of crasis.—The difficulty here is part of the essence of the language. Word-division by resolution of the crasis would yield a group of sounds disagreeing with those of the true text. The resolution must not be made; it may only be indicated. For this purpose the MSS. often use various marks, a stroke or an avagraha, or a combina¬ tion of both (li), over or under the line, and the avagraha (sometimes doubled) m the line. The avagraha is not to be commended for crasis. If employed commonly for cases of elision, as in it would seem to be the less suitable for cases like where neither xr nor ^ is elided. Strokes and avagrahas are written over the line in north-western MSS,, under the line in Kashmir MSS. The strokes are used for separating words or stems, the avagraha and its combination with the stroke (|^) only for separating words. In order clearly to distin- * It IS incidental to the partly syllabic character of the Nagaii alphabet that the two sounds, m and a, of the one letter may belong, the one to one word, the other to the next. I can cite nothing analogous from English but a line from the ‘Whimsey Anthology* of Carolyn Wells (New York, 1906, p. 52): ‘I’m sorry you’ve been 6 0 (=sick so) long; Don’t be disconsolS.* The sounds represented by the numeral 6 (=6ix=sick s) belong, part to the word sick and part to the word so. B\j the Editor of this Series XXXIK guish separated words from separated stems, Dr. Hertel employs in his text the stroke over the line exclusively for the separation of stems, and an angle UTider the line to separate words (e.g. Wherever the angle could not, for typographical reasons, bo employed under the line, he puts it over the line (e.g. For the sake of beginners, most of the stems (but not the prefixes) have been separated in the easy tales, I. v, vi; and with these a student might begin reading. In the rest of the text, stems are separated only in cases which ai*c liable to puzzle beginners who have overcome the first difficulties. Word-division has been given throughout the whole of the text. Word-division and sandhi. —The reader should consult the illuminating observations of Wackemagel in his Grammar, § 262. In the modem living use of Sanskrit in India, the sandhi is often simply ignored where the clearness of expression might suffer thereby.