^{page 219 note 1} JA., 1918, i, pp. 120–1Google Scholar.

^{page 219 note 2} For a conspectus of the MS. evidence about the verse in question see Ruben, , Studien zur Textgeschichte des Rāmāyaṇa, Stuttgart, 1936, pp. 77Google Scholar and 132.

^{page 219 note 3} See the lists in Kirfel, W., Kosmographie der Inder, pp. 74 and 82Google Scholar, and the same author's Bhāratavarṣa (Stuttgart, 1931), pp. 46, 60Google Scholar. In the latter work the common geographical matter of the Purāṇas has been put together and critically edited, though the notes to the translation do not always propose the best identifications of localities.

^{page 219 note 4} Gait, , History of Assam, pp. 12, 15, and 30Google Scholar, identifies Gauhati with Prāgjyotiṣapura, and in some of the inscriptions Assamese kings call themselves lords of Prāgjyotiṣa. Kālidāsa (Raghuvaṃśa, iv, 81) places it on the further side of the Brahmaputra, but, if Mallinatha construes the following verses correctly, he distinguished it from Kāmarūpa.

^{page 220 note 1} The vulgate reading of this verse, adopted in the critical Poona edition, v, 19, 15, viz. Tasya Cīnaiḥ, Kirātaś ca kāñcanair iva saṁurtam Babhau balam anādhṛṣyaṁ karṇikāravanaṁ yathā, is unsatisfactory. Either one should amend to kāñcanair eva, or, following the K tradition, read Kāmbojaś caiva saṁvṛtam. The golden colour of the Kirātas is often referred to in literature, and the first suggestion is probably the sounder.

^{page 221 note 1} Index to the Names in the Mahābhārata, p. 113 b.

^{page 222 note 1} This edition has not yet reached most of the verses discussed in the previous section about Bhagadatta; some modification of my remarks may be necessary, when Professor Edgerton's edition of the Sabhāparvan appears, but it is not probable that any alteration in my conclusions will be required.

^{page 222 note 2} It appears only in the later Nepali MSS., not in the oldest one; see Sukthankar, , Ann. Bhandarkar ORI., XIX, 204Google Scholar.

^{page 223 note 1} When Dr. Tarn, p. 458, n. 2, gives Patañjali as a reference, he has been misled by scholars carelessly copying their predecessors, instead of checking their references. I have not troubled to trace the mistake further back than Bhandarkar, D. R. in Ind. Ant., 1911, p. 12Google Scholar.

^{page 223 note 2} It is curious that the wording of the sūtra excludes Kauśāmbī; did Pāṇini not know of its existence ? The name occurs in the Pāṇinean gaṇa to iv, 2, 97, which is probably much later. Though Aśvaghoṣa takes Kuśāmba to be a ṛṣi, he is named as a king in the epics and Purāṇas.

^{page 224 note 1} See his vṛtti on his own grammar, iii, 1, 67 (ed. Liebich, , AKM., XIV, p. 188)Google Scholar. I have failed to trace the rule in the Kātantra.

^{page 224 note 2} For full discussion of the evidence see Lüders, , SBPAW., 1930, pp. 52 ffGoogle Scholar.

^{page 225 note 1} The latest discussion is in Indian Culture, IV, pp. 435 ff., which has some useful points. The evidence from Buddhist sources, discussed by me in the Journal, 1929, pp. 77–89, can now be fortified from a recently discovered work of Nāgārjuna who, it is generally agreed, probably lived in the second century a.d. Ch. ii of his Ratnāvalī (Journal, 1936, pp. 237 ff.Google Scholar), stresses the duty of a king to follow dharma as the highest policy (nīti). Verse 30 runs:—

Parātisaṁdhānaparā, kaṣṭā durgatipaddhatiḥ|

Anarthavidyā duṣprajñair arthavidyā kathaṁ kṛtā‖

The verb atisaṁdhā in the sense of “overreach”, “ruin by deceitful means” is a euphemistic term which occurs frequently in the Arthaśāstra (Shamasastry, , Index Verborum, pp. 22–4)Google Scholar, but does not appear to have been used in this sense by any earlier writer; the similar use in Jātakamālā, p. 53 (cf. Journal, 1929, p. 82), in a passage evidently referring to the Arthaśāstra may be compared, and I do not think there can be any reasonable doubt that Nāgārjuna here has Kauṭilya's doctrine in mind and employs arthavidyā as a synonym for arthaśāstra.

The verb atisaṁdhā in the sense of “overreach”, “ruin by deceitful means” is a euphemistic term which occurs frequently in the Arthasastra (Shamaśāstry, , Index Verborum, pp. 22–4Google Scholar), but does not appear to have been used in this sense by any earlier writer; the similar use in Jātakamālā, p. 53 (cf. Journal, 1929, p. 82), in a passage evidently referring to the Arthaśāstra may be compared, and I do not think there can be any reasonable doubt that Nāgärjuna here has Kauṛilya's doctrine in mind and employs arthavidyā as a synonym for arthaśāstra.

^{page 225 note 2} A reference, not usually noticed, is at Petavatthu, ii, 9, 13–15, where Minayeff reads Bheruva, but the Burmese MSS., apparently correctly, have Roruva, the usual Pali form of the name.

^{page 226 note 1} The text gives Trigupto Hanumātīre for the immediately preceding place, for which Lüders with some probability prefers the variant reading Trigupto hy Anūpatīre; and he then takes Anūpa to refer to the sea-coast, thus ruining the geographical order observed in the work. Anūpa usually implies dwellers in a valley along a river; thus the Anūpas of Rudradāman's inscription are the same as the Anūpas of Kirfel, , Bhāratavarṣa, p. 48Google Scholar, who dwell on the slopes of the Vindhya, i.e. along the Narmadā, apparently in the neighbourhood of Mahiṣmatī. See also p. 221 above for a similar use. The rule is not invariable; for the Anūpa district of the Sindhu king at Harivaṁśa, 6408 ff. is the land where Dvāravatī was built, and therefore indicates the sea-coast there. In the present case the reference is presumably to the Indus valley, if the amendment is correct.

^{page 227 note 1} Ruben, op. cit., p. 125.

^{page 227 note 2} Jacobi in Festschrift Wackernagel, p. 124, followed by Keith, , History of Classical Sanskrit Literature, p. 33Google Scholar. Dr. Tarn, p. 235, has misunderstood the latter when he says that Patañjali places the Abhīras in Sindhudeśa; the Mahābhāṣya does not say where they lived.

^{page 228 note 1} Dīgha, II, 236: Jātaka, III, 470.

^{page 228 note 2} B.M. Catalogue, Coins of the Andhra Dynasty, pp. xxxii ff., Ix f., and cxviii f.

^{page 228 note 3} Personally I see no reason for differing from Bühler's, conclusion, Ind. Ant., XVII, p. 192Google Scholar, that the people referred to are the Gurjjaras and that the capital is identical with Bhillamāla, the native town of the astronomer Brahmagupta, now called Bhinmal (fifty miles north-west of Mount Abu).

^{page 229 note 1} See Allan, , B.M. Catalogue, Coins of. Ancient India, pp. cxlvii ffGoogle Scholar.

^{page 230 note 1} I do not deal here with the Jain evidence, which is of no great value; the Abhidhānarājendra sums it up by describing the Sovīra country as lying along the Indus. But a late Jain story translated by Meyer, (Hindu Tales, p. 109)Google Scholar tells how Udāyaṇa, king of the Sovira and Sindhu country, crossed the desert to attack Ujjain, and this is hardly the route he would have taken if he had been in command of the coast.

^{page 231 note 1} Cf. Raverty, , JASB., 1892, p. 219Google Scholar.

^{page 232 note 1} The best known Afghan breed, the Bāhlīka, appears among the middling horses; this list is clearly in geographical order, but in an inverse direction to the list of the best class. Hiuan Tsang refers, however, to the Kapiśa breed of horses.

^{page 232 note 2} It is possible that it was adventurers from this tribe who in the tenth century a.d. founded a short-lived dynasty bearing this name at Priyagu, in Bengal, Ep. Ind., XXII, pp. 150 ffGoogle Scholar.

^{page 233 note 1} Allan, op. cit., pp. cxxiii–cxxv.

^{page 233 note 2} See ASR., VI, pp. 190 ff., by Carlleyle, who quotes Cunningham's views.

^{page 234 note 1} Davids, C. A. F. Rhys, Psalms of the Sisters, pp. 132–3Google Scholar.

^{page 234 note 2} Senart, E., Ep. Ind., VIII, p. 90, No. 18Google Scholar.

^{page 235 note 1} ASWI., IV, p. 38. Dr. Tarn's reference, p. 254, n. 5, to SirMarshall's, J. views as expressed in CHI., I, p. 637Google Scholar, suggests a failure to realize the fact that there are a number of caves at Nasik of varying dates. The passage he alludes to is, when properly understood, in consonance with the views I express above.

^{page 235 note 2} The latest consideration of the palæography (Stein, O., Indian Culture, I, 351Google Scholar) gives the last part of the first century a.d.

^{page 236 note 1} pp. 416–18. Dr. Tarn uses this with other arguments to suggest that the Milindapañha derives from a Greek original, which has left its traces in that work. He has proved that the type of literature to which it belongs, while unknown otherwise to India, was known to Hellenistic Greek. This is a point of considerable importance, but his further conclusion is difficult to follow. Besides the use of Yonaka, he relies on the passage describing how the guardian of a city would from the central cross-roads see a man coming from any of the four quarters (Milindapañha, p. 62), as showing that the author had in mind, not an Indian, city, but a four-square city of the Hellenistic type. The evidence available does not allow us to draw decisively any such conclusion. Hopkins, (CHI., I, pp. 271–2)Google Scholar holds the epics to describe cities as laid out in squares, and Arthaśāstra, ii, 4, 1, lays down that a town shall be divided by three roads running east and west and three at right angles; ibid., 10, places the king's palace to the north of the heart of the town (vāstuhrdaya). This is all that can be gleaned on t he subject from the earlier works, and suggests serious misgivings about the Hellenistic origin of the simile. Excavation so far has proved inconclusive, and later theory in Southern India certainly envisaged central cross-roads, as may be seen in plates xliii–xlv of Ram Raz's Essay on the Architecture of the Hindus, Acharya, P. K., Indian Architecture, p. 39Google Scholar, translates the Mānasāra as contemplating straight streets across a fortified village or town forming cross-roads with a temple or hall in the middle; this would probably allow a central viewpoint. How far theory and practice agreed at any time is no doubt hard to determine, but I do not think any inference can be safely drawn from the simile. As regards the unexplained name, Sabbadinna, of one of Menander's councillors (cf. Dr. Tarri, p. 422), I am not satisfied that it may not be Indian, seeing that Sarva is a recognized name of Śiva and Sarva in later literature is included among the names of Kṛṣṇa. A Mathurā inscription of Huviṣ;ka's time has preserved the name Śavatrātā, (Lüders, , Journal, 1912, p. 158Google Scholar), and the Ghosundi inscription has the curious name Sarvatata or Sarvatrtāa, (Ind. Ant., 1932, pp. 203 ff.)Google Scholar; Sabbadinna might therefore be equivalent to Śivadatta or less probably to Kṛṣṇadatta, though what conclusion should be drawn from such a hypothesis is another matter. That there is any real evidence for a Greek “Questions of Menander” seems to me doubtful.

^{page 237 note 1} See Mahāvaṃsa, critical apparatus on xii, 4.

^{page 238 note 1} Stein, O.. Indian Culture, I, 343–357Google Scholar, touches on the point.

^{page 238 note 2} IHQ., XIV, p. 465, to be read with p. 477, n. 178. Despite the disputes about this passage, its general meaning seems to me clear. Khāravela advanced from Kalinga through the hills and, emerging in the Gaya district, stormed the fort of Gorathagiri, which lay in the Barabar hills (see Jackson, V. H., JBORS., I, pp. 159 ff.Google Scholar); he then ravaged the neighbourhood of Rājagṛha, evidently without taking it. On “the report of this heroic deed” (kammapadṛnasaṁ (or pa) nādena) the Greek king retired to Mathurā with his army. Khāravela then, according to his own account which may well conceal the real truth, never came to grips with the Greeks and did not make any serious impression on Magadha; the presumption is that he found the Greeks too strong for him to dare attempt more than a hasty raid, and that it was only after he had retired that the Greeks withdrew to Mathurā, an event which he chose to put down to fear of his forces. The record of his twelfth year, though incomplete in detail, shows that Magadha after the Greek abandonment of it was no longer able to offer the same resistance.

^{page 239 note 1} The latest work on proper names in Sanskrit is Velze, Van, Names of Persons in Early Sanskrit Literature, Utrecht, 1938Google Scholar; but it gives no help in this case.

^{page 240 note 1} Yet another hypothesis is possibly admissible. Assuming, as I think we must, that Dāttāmitrī existed before Demetrias in Arachosia was founded, popular speech may have found the latter name too difficult and adopted the known form from a neighbouring country. Such a confusion cannot be proved from the nature of the case, and under this theory one could not take Dattāmitra in the epic and grammarians as standing for Demetrius; it would not be possible then to say which town was meant by the Nasik inscription.