^{page 51 note 1} It is necessary to explain that this chapter was originally drafted in the year 1898. It was then deemed advisable to postpone the publication of the work till the Linguistic Survey of India should be near completion. In the meantime, I utilized the draft for the preparation of pp. 51–63 of The Languages of India, published in 1893. The chapter has now been rewritten, but so much of the original as had not become out of date was retained. Hence, much of what follows will also be found in the above work, which; however, goes into the matter in much greater detail.

^{page 51 note 2} Cf., however, yon Bradke, ZDMG. xl, 673 ff.; Wackernagel, , Altindische Grammalik, xiii, xix, xxxv.Google Scholar

^{page 51 note 3} Macdonell, , History of Sanskrit Literature, 24; cf. Wackernagel, xvi ff., xxv.Google Scholar

^{page 52 note 1} Hillebrandt, , Vedische Mythologie, i, 89, 114, 136.Google Scholar

^{page 52 note 2} Cf. von Bradke, 669 ff.; Wackernagel, xii.

^{page 52 note 3} See Risley, , Report of the Census of India (1901), i, 511Google Scholar, repeated in the Imperial Gazetteer of India (1907), i, 302 ff. According to him, the earlier Aryan invasion suggested by Hoernle, and mentioned below, was one of a tribe or tribes who brought their women with them. The later invaders represent the Indo-Aryan population of the Midland, which presents the ethnological type that might be expected to result from the incursion of a fair long-headed race that entered India by a route which prevented women from accompanying them, into a land inhabited by dark-skinned Dravidians, whose women they took for themselves. It is thus seen that Risley postulates two sets of invaders, one bringing their women and settling at first in the Central and Western Panjab, and the other coming without their women and settling at first in the Midland. It is evidently immaterial to his argument which was the first and which the second, but he assumes that the first was that with women.Google Scholar

^{page 52 note 4} Gaudian Grammar, XXX ff.Google Scholar

^{page 53 note 1} Three interesting points are on Hoernle's side. One of them is the optional change of r to I in Cūlikepaiśācika. The same change was obligatory in Māgadhī Prakrit; cf. Māhabhāsya (Kielhorn, i, 2, 1. 8), hē layō for hē, arayah, in the speech of the Asuras, which is often said to be Māgadhī Prakrit, but can be better explained as Cūlikāpaiśācika Prakrit. The second is the change of sm to s (Kš. asi, “we,’ etc.). See Hoernle, Gḍ. Grammar, 280, ñ. 1. The third is the frequent use of n both in Paiśācī Prakrit and in Māgadhī Prakrit (Hēmacandra, iv, 305, etc.; cf. Hoernle, Gel. Grammar, 11).

^{page 54 note 1} This point is discussed in detail in an Appendix to this chapter.

^{page 54 note 2} Hillebrandt, 104 ff., 109.

^{page 54 note 3} Ib. 114.

^{page 54 note 4} Hillebrandt, 110, also maintains that there was a second invasion of Aryans from the west. It is worth noting that Visvamitra called Vaśistha a Yātudhana, or Rāksasa, a form of abuse that the latter strongly resented (Rv. vii, 104, 15).

^{page 55 note 1} The kingdom of Magadha was, as a whole, hostile to the Midland ; see Jacobi, , Das Rāmāyana, 104.Google Scholar

^{page 55 note 2} Pargiter, , JRAS., 1908, 334 ff., and map; Grierson, ib., 602 ff.Google Scholar

^{page 55 note 3} Jaeobi, , op. cit., 69.Google Scholar

^{page 55 note 4} LIA^2. i, 720, 742, 743, 791.

^{page 56 note 1} Wackernagel, xxxiv.

^{page 56 note 2} The earliest examples of this are to be found in the inscriptions of Asōka (circ. 250 B.C.), and in the Māhabhāsya (circ. 150 B.C.), Bhandarkar, R. G., Wilton Lectures, 280.Google Scholar

^{page 56 note 3} Wackernagel, xxxii, xxxiii; Liebich, Pānini, 47 ff.

^{page 56 note 4} See Thomas, , JRAS., 1904, 471, 748.Google Scholar

^{page 56 note 5} e.g. Kielhorn, , i, 259, 1. 14.Google Scholar

^{page 56 note 6} JRAS., 1904, 480.

^{page 56 note 7} Jacobi, , Ramāyana, 114; Muir, Sanskrit Texts, ii², 158; Wackernagel, xxxviii, n. 6.Google Scholar

^{page 56 note 8} Cf. Lévi, Sylvain, Bull. Soc. Ling., 8, pp. viii, x, xvii, quoted in Wackernagel, xxxix, n.Google Scholar

^{page 56 note 9} The Primary Prakrits plus their literary form as conserved in the Vēda correspond to Wackernagel's “Altindisch’, and the Secondary Prakrits plus their literary form to his “Mittelindisch’.

^{page 57 note 1} It is quite certain that even in the Vedic period the popular speech of at least some classes of the people already contained many words in the same stage of development as Pāli, i.e. as the earliest phase of Secondary Prakrit. Cf. Wackernagel, xviii, xxv.

^{page 57 note 2} It is always the Midland which has been behindhand in the race of development. Śaurasēnī Prakrit is less developed than Māhārāstrī Prakrit, just as the Modern language of the Midland is less developed than any of the Outer languages, including Marāthī. Is this because the inhabitants of the Midland represent the latest Aryan immigrants (see above), or is it due to the influence of literary Sanskrit—itself a Midland language? Opportunity may here be taken to warn against one common error. It has often been stated that because (e.g.) Śaurasēnī Prakrit is less developed than Māhārāstrī, it is therefore earlier in point of date. Such an argument is fallacious. It is a well-known fact that different languages of a common origin do not all develop at the same rate of progress. To take an example from the Romance languages, Italian is much less developed than French. To use Indian terms we might almost say that Italian is in the Pali stage, while French is in the Prakrit stage. Nevertheless, they are contemporary.

^{page 58 note 1} Cf. as the latest authority, Michelson, , AJP. xxx (1909), 284, 416, xxxi (1910), 55; JAOS. xxx (1909), 77, xxxi (1911), 223; also Grierson, JRAS., 1904, 725. The eastern dialect in the days of Asōka was the official imperial language, and was understood even where it was not spoken as a vernacular (JAOS. xxx, 77).Google Scholar

^{page 58 note 2} The Brahmagiri (Siddapura) Edict is written in a mixture of eastern and western forms (Bühler, El. iii, 135). But this, being in a Dravidian country, is not decisive. Cf., however, the close connexion between Mahārāstrī and Ardha-Māgadhī Prakrit. Wackernagel (xxi) considers that there were probably in Vedic times an eastern and a western dialect. The eastern, which was the language of the earlier Aryan immigrants, was then spoken on the banks of the Ganges. The literary language of the Vēda would, in the main, correspond to the western dialect. We cannot trace in the Vēda any marks of a dialect of the extreme north-west, but we can deduce nothing from their absence.

^{page 59 note 1} We must, however, credit the grammarians with expressly warning us that their rules are not universal; cf. Ho. i, 2 ; see also Bhandarkar, R. G., op. cit., 77, n., “all these rules are general, not universal.”Google Scholar

^{page 59 note 2} Pisehel, Prakrit Grammar, § 12.

^{page 60 note 1} Cf. Pischel, § 364. Regarding the changes which Prakrit has undergone in becoming literary, see ib., § 9, at end.

^{page 60 note 2} For the last, compare the change of pronunciation of Māgadhī Prakrit ś: to 8 in Bihārī, although ´s is invariably written.

^{page 60 note 3} See, for instance, Michelson, , AJP. XXX, 285.Google Scholar

^{page 61 note 1} Konow, , IA. XXXII (1903), 181.Google Scholar

^{page 61 note 2} For this division of the Prakrits, see Konow, , Māharāstrī and Marāthī, IA, xxxii (1903), 181 ff., with which I am in entire accord.Google Scholar

^{page 61 note 3} Mārkandēya, xvi and comm. to xviii, perhaps calls it Tākkī or perhaps Pascatya. Cf. Rāmatarkavagīśa in Lassen, ILP., App. p. 5, and Hoernle, Gḍ. Gr., 15, n. 1.

^{page 62 note 1} See the dates fixed in § 66, post. Apabhramśas could hardly have been a living language in Hemaeandra's time, for his grammar does not deal with one Apabhramsa, but with several dialects which he mixes up together. His very rules are frequently contradicted by his own examples. He would not have done this had he been dealing with a living language known to him. In this respect, his grammar is a compilation put together from many widely differing and mutually contradictory sources (Pischel, Pr. Or., § 28).

^{page 62 note 2} Bhandarkar, R. G., Wilson Lectures, 302.Google Scholar

^{page 63 note 1} Cf. also Mārkandēya, , Preface, 7, and xvii, xviii, and Grierson, “Vrācaḍa and Sindhī,” JRAS., 1902, 47.Google Scholar

^{page 64 note 1} That this is a justifiable assumption is shown by the fact that Mārkaṇḍēya, a late grammarian of the seventeenth century, admits the termination i as well as ē even into literary Māgadhī Prakrit (xii, 26).

^{page 65 note 1} See § 29, n. 2, ante.

^{page 65 note 2} Bhandarkar, R. G., 27, 286; Wk. xlii.Google Scholar

^{page 65 note 3} Bomb, , ed., p. 47, 11. 6, 7.Google Scholar

^{page 65 note 4} Rājataranginī, vii, 610.

^{page 65 note 5} Stein, tr. Rājataranginī, i, 13, and footnotes.

^{page 66 note 1} Kāšmīrī was certainly in existence in Kalhana's time, and possibly so far back as the tenth century; see Stein's, tr., RT. V, 397–8n (I, p. 228).Google Scholar

^{page 66 note 2} Described by Bhandarkar, R. G. in Report on the Search for Sanskrit MSS. in the Bombay Presidency for 1887–91 (Bombay, 1897).Google Scholar

^{page 66 note 3} On this point cf. Bhandarkar, R. G., 302. He puts the commencement of Apabhraṃśa at the sixth or seventh century A.D.Google Scholar

^{page 66 note 4} See Bhandarkar, R. G., 21, for the change from the verbal to the nominal style of Sanskrit; cf. Wackernagel, xliv. For dialectic variations, ib., li.Google Scholar

^{page 66 note 5} Wackernagel, xlii.

^{page 67 note 1} Wackernagel, lii.

^{page 67 note 2} Some later Prakrit writers, e.g. Rājaśěkhara, borrowed Sanskrit words very freely ; cf. index to Konow's edition of the Karpūramañjarī.

^{page 67 note 3} It stands to reason that the modern distortion of a Sanskrit word may often have a result different from that of the gradual development of a Primary Prakrit word. This accounts for many of the so-called irregular Prakrit words noted by the grammarians. To quote an example, He. ii, 104, gives a number of irregular forms, sirī (for śrī), hirī (hrī), kiriā (kriyā), which are really distorted Tatsamas, not Secondary Prakrit. The true secondary form of kriyā iskiā (104). So also in the following sūtras.

^{page 67 note 4} Regarding the subject discussed in this paragraph, see Pischel, , Pr. Gr., § 8.Google Scholar

^{page 67 note 5} For the use of Tss. in Prakrit, cf. Bhandarkar, R. G., 15Google Scholar, and Wackernagel, liv. For the origin of sTss., cf. Bhandarkar, , 298. On 69 he gives an account of the so-called Gāthä dialect, which is germane to the present subject.Google Scholar

^{page 68 note 1} Pischel, Pr. Gr., § 9; Bhandarkar, R. G., 107, 131.Google Scholar

^{page 68 note 2} Śaurasēnī Prakrit, which developed in the Midland, is naturally that Prakrit which is freest from Ds. words ; cf. Pischel, § 22.

^{page 68 note 3} For Tss. and sTss. in Indo-Aryan Vernaculars, see Beames, Cp. Gr. ii, 11; Hoernle, Gṇ. Gr., xxxviii; Bhandarkar, , 131.Google Scholar

^{page 69 note 1} Many Primary Prakrit words which have survived unchanged into the Indo-Aryan Vernacular, and which are hence Tbh., are liable to be confused with Tss. Thus, the Primary Prakrit kara- remained kara- in the Secondary Prakrit, and is still kar(a) in Hindī. As kar(a) is also a pure Sanskrit word, it is generally looked upon as a Ts. in Hindi, but it can equally correctly be looked upon as a Tbh. In a book called Thēṭh Hindī-kā Thēṭh, by Ayodhya Singh Upâdhyāy, from which the author designedly excludes all Ts. words, many honest Tbh. words have also been excluded owing to this misapprehension. Nevertheless it, and another work by the same author, Adhkhilē Phūl, are invaluable records of Tadbhava Hindī.

^{page 69 note 2} For these specializations see Bhandarkar, B. G., 13.Google Scholar

^{page 70 note 1} There are a few exceptions to this. In Kš. and M., for instance, under the influence of analogy, borrowed nouns can be declined synthetically, but the above holds true as a general rule.

^{page 70 note 2} The late Sudhākara Dvivēdī (Rāmakahānī, p. 7) gives an amusing instance of the difference between literary and colloquial Hindi. A friend wrote to him a letter as follows: —āp-kē samaāamârtha maī yata-divasa āp-kē dhāma-par padhārā. Grha-kā kapāṭa mudrita thā, āp-sē bhēt, na huī. Hatâia ho-har parâvarltita huā, i.e. “Yesterday I went to your house to see you. The door of the house was shut, and I did not meet you. I returned home disappointed’. Shortly afterwards Sudhākara met the writer of this letter, who, not knowing that it had been received, said: kal maī āp-sē milnē-kē liyē āp-kē ghar-par gayā-thā. Ghar-kā darwāzā band thā, āp-sē bhēt nahī huī. Lācār hō-kar lauṭ āyā. This, in conversational Hindi, has exactly the same meaning as the letter in Sanskritized literary style, yet both came from the same man. As Sudhākara observes, the feeling of a pen in the hand of such a person makes him Sanskrit-drunk, and prevents him from using his own mother tongue.

^{page 71 note 1} Cf. the list of Dravidian words said to be borrowed by Sanskrit on pp. xiv ff. of Kittel's Kannada-English Dictionary. See also Linguistic Survey of India, iv, 278.Google Scholar

^{page 71 note 2} Most common and longest preserved in the folk-speech, i.e. Ap.

^{page 71 note 3} See Konow in Linguistic Survey of India, iv, 279 ff.Google Scholar, for details. R. G. Bhandarkar (81) attributes the development of Pāli and Prakrit to the mispronunciation of Sanskrit words by alien (i.e. Dravidian) races. I am unable to agree to this. The development, as a whole, exactly followed the same course as that of the Romance languages from the Latin dialects. See Brandreth, , “The Gaurian compared with the Romance Languages,” JRAS., 1879, 287, and 1880, 335. At the same time I readily admit that Dravidian had some influence on their development.Google Scholar

^{page 72 note 1} e.g. krtē > kahū > kō for the accusative-dative, as compared with the Dravidian ku.

^{page 72 note 2} In Old Gujarātī.

^{page 72 note 3} In Kāšmīrī.

^{page 72 note 4} So exact is the parallel that both in Sanskrit and Dravidian the verb substantive is not added to the third person, although it is added to the other two persons.

^{page 72 note 5} It is to be noted that the Modern Pi"aca languages, which apparently did not fall to the same extent under Dravidian influence, differ altogether from the Indo- Aryan vernaculars in this respect. In them the order of words is nearly the same as in English or as in Modern Persian. For the whole of this subject, see Languages of India, 62, and Konow in Linguistic Survey of India, iv, 279 ff.

^{page 73 note 1} LSI. III, i, 273 ff.Google Scholar

^{page 73 note 2} Cf. Konow, , LSI. iv, 9.Google Scholar

^{page 73 note 3} e.g. the use of the agent ease for the subject of all tenses of the transitive verb and the creation of a new impersonal honorific conjugation.

^{page 74 note 1} e.g. the Sanskrit and Prakrit sāhi, a king, not derived from the Musalmān Persian Šāh, but preserving the i of the Old Persian xšāyatiya-; see Stein, “Zoroastrian Deities on Indo-Seythian Coins,” Oriental and Babylonian Record, August, 1887.Google Scholar

^{page 74 note 2} Hence the spelling “Hindōstan”, not “Hindūstān”, is correct in India.

^{page 75 note 1} e.g. a well-known Hindi work, written in the last century, was called Kahānī Thēth Hindī-mē, or “Tales in Pure Hindī”. This does not contain a single Persian word, and yet Hindū writers class it as Urdū on account of the order of the words. The author was a Musalmān.

^{page 76 note 1} See Grierson, , “Linguistic Relationship of the Shāhbāzgarhī Inscription”: JRAS., 1904, 726.Google Scholar

^{page 76 note 2} The Paisaci Prakrit of Vararuci differs from Hēmacandra's Standard Paiśāci Prakrit in important particulars, and has, like Hemacandra's Cūlikāpaiśācika, a closer relationship with the north-west; cf. Grierson, , The Piklca Languages of North-Western India, 6.Google Scholar

^{page 77 note 1} e.g. Khō-wār ispa. Waxī spā, our.

^{page 77 note 2} Grierson, , “Étymologies Tokhariennes”: Journal Asiatique, 1912, 339.Google Scholar

^{page 77 note 3} “Le Dialecte des Fragments Dutreuil de Rhins”: Journal Asiatique, 1912, 331.

^{page 77 note 4} It is of course well known that other compound consonants occur in Kharosthi.

^{page 77 note 5} Cf. Grierson, , JRAS. 1913, 141 ff., for many other examples.Google Scholar

^{page 78 note 1} So also in Dravidian languages.

^{page 78 note 2} This is also common in Kāšmīrī, but the final short vowel is also preserved.