^{page 1187 note 1} Aruṇaśūra, Aruṇeśvara, or Aruṇa-svāmin(?). See, however, n. 4 below.

^{page 1187 note 2} See Lévi's, Sylvain “Les Missions de Wang Hiuen-ts'e dans l'Inde” in Journal Asiatique, 03, 04, 1900, pp. 306–8Google Scholar.

^{page 1187 note 3} Oxford, 1904, pp. 298–9.

^{page 1187 note 4} JA., fase. cit., p. 300. In JA. for 1892, p. 337, Professor Lévi had suggested the rendering (discarded in his later paper just quoted) Senāpati Arjuna, which V. A. Smith has nevertheless adopted in op. cit. It has been suggested that A-lo-na-shun may be Amśu[-varman], a rendering which I consider unsatisfactory on phonetic grounds. From an historical point of view it is also hardly convincing, for Aṁśuvarman, the Thākurī prince of Western Nepāl who was seemingly a feudatory of Harṣavardhana of Kanauj and became supreme probably on the latter's death in a.d. 648, was not improbably the very personage who assisted Hsüan-ts'ê with troops against A-lo-na-shun.

^{page 1188 note 1} JA., fasc. cit., p. 307, note.

^{page 1188 note 2} The T'ung-chien-kang-mu (twelfth century) also states that Ka-p'i-li is a name of the Ganges (see Bulletin École Française Extrême Orient, t. iv, 272, n. 4). In the T'ai-p'ing-yü-lan cyclopaedia (pub. a.d. 983), Mr. Pelliot observes (op. cit.), it is wrongly spelled P'i-ha-li. We shall see below that it was also the name of a kingdom.

^{page 1189 note 1} A Kapili River actually exists in Assam, but is obviously out of the question. On the other hand, in Hindū classical literature mention is made of a Kapilā (tributary of the Narmadā); of Kapila-dhārā, as a name of the Ganges; and of a second Kāverī which Professor F. Hall thought to be a tributary of the Narmadā (see Wilson's, Viṣṇu Purāṇa, 1865, vol. ii, pp. 148, 151Google Scholar). It has been shown quite recently (JRAS., April, 1910, p. 442, and July, 1910, p. 868) that this Kāverī joins the Narmadā from the south about a mile above Māndhātā, and that the confluence, called Kāverāsaṇgama, was held sacred. Evidently it cannot be connected with the Ka-p'i-li of Chinese records, which is located in the middle and not in the west of Northern India.

^{page 1190 note 1} On Sheet 103 of the Indian Atlas a small watercourse, also marked “Soondee” (Sundī), is shown lower down wending its way towards Chhaprā and Cherānd. Whether it is this, rather than the other Sundī (i.e. the Dahā) the Chinese authors had in mind, I am unable to say.

^{page 1190 note 2} Cf. Cunningham's, Ancient Geography of India, p. 441Google Scholar, where the name Sāran of the actual district is traced to Skr. ŝāraṇa,”refuge,” through the legend related by Hwên Tsang that some demons converted here by the Buddha sought the “refuge” of the Buddhist triad. For this legend see also Watters, ’ Yuan Chuang's Travels in India, O.T.F., London, 1905, vol. ii, pp. 60–1Google Scholar.

Dr. Grierson kindly informs me, however, that General Cunningham later onabandoned that early theory of his on the derivation of Sāran, which has no basis in fact.

Given that the smaller watercourse marked ‘’ Soondee N.” on Sheet 103 of the Indian Atlas was already in existence at the period here treated, and that it is this the Chinese authors meant by their term Shan-lien, it is not impossible that the streamlet in question was then better known as Cherānd (if not Sāran) River, after Cherānd or Chirānd (a well-known ancient town, Dr. Grierson informs me), close to which it joins the Ganges. If so, the term Shan-lien would have to be traced to Cherānd rather than to Sāran, which would not be altogether unsatisfactory from an etymological point of view.

^{page 1190 note 1} JA., fasc. cit., pp. 323–4.

^{page 1191 note 1} See Cunningham, op. cit., p. 440. In connexion with a river of a similar name to Shan-lien, namely, the Shih-la-na-fa-ti, mentioned by Hwên Tsang as flowing past Kuśinagara and identified by Watters (op. cit., ii, 29) with the Siranyavatī(for Hiraṇyavatī) —the Hsi-lien or Hsi-lien-shan, , of Fa-hsien (see Legge's, Record of Buddhistic Kingdoms, Oxford, 1886, p. 23 of Chinese textGoogle Scholar) and other Chinese pilgrims—I would beg to point out that Shih-la-na-fa-ti is more probably a transcript of Sikrāna (Sikrāna-vatī?). This is one of the names of the Būrhī (or Chota, i.e. “Little”) Gaṇḍak, seemingly also known as Hiraṇyavatī, albeit this term appears to more properly denote the Little Gaṇḍak, a quite distinct stream known to this day as Hiraṇa(a contraction of Hiraṇyavatī).

Some confusion seems to have arisen regarding the correct application of the term Hiraṇyavatī; but if, as it seems beyond question, the Shih-la-na-fa-ti is, conformably to our suggestion, the Sikrāna, all doubt would at once be removed as to the real location of Kuśinagara, and this city should forthwith be looked for on the upper course of the Būrhī Gaṇḍak, i.e. in the North Bettiā and South Lauṛiya territory, where this stream just happens to be more particularly called Sikrāna. The now widely accepted location of Kuśinagara to the north of Lauṛiyā Nandangaṛh receives confirmation thereby, but would far earlier have been inferred had the Sikrāna been recognized in Hwên Tsang's Shih-lana-fa-ti, and no fanciful suggestions such as Kasiā and the like would have been put forth. For the same reasons I fear that V. A. Smith's Nepāl theory (see op. cit., p. 139, n. 3, and JRAS., January, 1902) is now untenable.

^{page 1192 note 1} See Indian Atlas, Sheet 103.

^{page 1192 note 2} In the Imperial Gazetteer of India, vol. x (Oxford, 1908), pp. 174–5, the name is spelt Chāpra. Cf. Chāparmukh on the Kapilī River, Assam: a striking parallel caseGoogle Scholar.

^{page 1193 note 1} Cf. Cunningham, op. oit., p. 441. Quite recently an archæological tour in the Sāran District yielded no very brilliant results (see JRAS. for April, 1910, p. 546). Chupra, the Chhaprā in question, is, however, the only place of a similar name appearing in the maps appended to vol. i of Holwell's, Indostan, London, 1766Google Scholar.

^{page 1193 note 2} It may be possible to ultimately fix it at Cherānd itself, and, if not, at Revelganj, either of which may have of yore borne appellations resembling Ch'a-po-ho-lo. Professor Lévi has pointed out (op. cit., p. 307, note) that the first three syllables of this name recall the term Ḍavāka; but the Ḍavāka “frontier country” conquered by Samudra Gupta in A.D. 345–80 was certainly not in this neighbourhood (see JRAS., 1897, pp. 29 and 879; also V. A. Smith, op. cit., p. 250), unless we assume the existence of a second Ḍavāka = Ātavl, Ālavi (?) in the Saran District, and locate here the legend of the demons’ conversion to Buddhism related by Hwen Tsang (see Watters, op. cit., ii, 61). Ch'a-po-ho-lo undoubtedly is a rendering of something like Ch'a-parha-ra, JDabargarh, or Ḍavarhara, and about the only approach in the area in question is, besides ChhaprĀ, DeghwĀra(for whichsee Cunningham, op. cit., p. 442).

^{page 1194 note 1} Watters, op. cit., ii, 87.

^{page 1195 note 1} The route he followed from MahĀsĀla (MasĀr ?) on to VaiśĀli is yet far from settled, but he seems on the whole to have travelled along the northern bank of the Ganges.

^{page 1195 note 2} The position at or near the confluence of two important streams like the Ganges and the Ghaghra must have been in past ages of considerable strategical as well as commercial importance, so as to justify the foundation, or spontaneous growth, of a town.

^{page 1196 note 1} Watters, op. cit., ii, 63.

^{page 1196 note 2} Ibid., 39.

^{page 1196 note 3} Cunningham (op. cit., 438–9) identified it with Ghāzīpur (ancient Garjapur, Garjana-pati?); but it seems to me that this town is not far enough from Benares to suit as a site for the capital of Chan-chu. If this term really be a translation of Yuddhapati, such a name may still survive in Juddoopoor(Yuddhapur ?), a place marked above Korantadih in the Garha pargana(Balliā District) on Sheet 103 of the Indian Atlas.

^{page 1196 note 4} And, it may be of interest to notice, to the territory of the later mediaeval Jaunpur kingdom, the name of which (variously recorded in the forms Jaunpur, Juānpur, Jawanpur, Jamanpur) is notoriously of uncertain origin, but may have existed from a far earlier period in connection with the older town, which, judging from ancient remains, stood on the present site of Jaunpur. If so, Chan-chu might refer to this territory, being a transcript of Juan-pati or something of the sort. See, however, note 3 on p. 1199 below in connexion with Chao-no-p'o.

^{page 1197 note 1} See JA., fasc. cit., 307, note.

^{page 1197 note 2} In the P'ei-wén-yün-fu the following account of it occurs, culled presumably from the Sung Shu:—”The Indian [T'ien-chu] State of Ka-p'i-li. In the 5th year of Yüan Ghia[a.d. 428], the King sent an envoy with a letter and offering, of diamond finger-rings and Mo-lei [Marak = Marakata, Marakta = ‘Emerald'] gold rings” (see China Review, xiii, 339). Mas'ūdi (a.d. 915) states (Les Prairies d'Or, t. iii, p. 47, Paris, 1864Google Scholar) that the Sindān(= Sanjan) and Kambāya(Cambay) Districts produced a sort of emerald (= beryl from the Jaipur District in Rājputāna ?). Beryls from the Hazāribāgh District in Western Bengal are, however, presumably meant in the passage quoted above.

^{page 1198 note 1} Cf. Hwen Tsang's, Yileh-kuang = Chandraprabha (Watters, op. cit., i, 244), and , Ch'u-ai- Udayana (ibid., i, 368).

^{page 1198 note 2} V. A. Smith, op. cit., p. 257.

^{page 1199 note 1} The classing of Ka-p'i-li in Eitel's, Handbook of Chinese Buddhism (Hong-Kong, 1888, p. 70)Google Scholar among the alternative Chinese transcripts of Kapilavastu; the Sanskrit equivalent Kapila adopted for it by ProfessorSchlegel, in T'oung Pao, x, 160Google Scholar, and so forth, are evidently absurd and, topographically, unjustified in so far as they are made to apply to Kapilavastu, a city which had long ceased to exist: Fa-Hian in 405–11 found it deserted. If Ka-p'i-li is at all to be regarded as a transcript of the Sanskrit term Kapila, it can only apply to the Ganges, also known as Kapila-dhārā, as we have pointed out above; and this view finds its confirmation in the Chinese texts, which tell us that Ka-p'i-li was likewise aname of the Ganges.

And, though far less probably, to Hardwār, which according to tradition was also named Kapila after the sage of that name who is said to have had his hermitage there. See, however, the note on next page as regards Kampilla.

^{page 1199 note 2} Were we to accept for Ka-p'i-li the seemingly wrong spelling P'i-ka-li occurring in the T'ai-p'ing-yil-lan, we might locate it at Bikapur, a village 1 mile to the east of Ballia, where Cunningham (op. cit., 439) places Hwen Tsang's Aviddhakarna monastery. The correct spelling unquestionably is, however, Ka-p'i-li, which occurs in all other known Chinese texts onthe subject.

^{page 1199 note 3} The Ka-p'i-li State is still mentioned (retrospectively no doubt) in the Hsiang-chiao-p'i-pien, a well-digested Buddhist cyclopaedia of the Ming dynasty, in the following passage: ‘’ Pang-ko-la[Bangala, Bengal] is in the east of T'ien-chu[N. India]; Chao-no-p'o[Jaunpur?] in the middle;t Magadha in the south; Ka-p'i-li in the west; and Ka-shS-na [Kasia, Kashlpur, Kusinara, (Northern) Kosalaf] J in the north” (cf. JRAS. for July, 1896, p. 496, note. I have here, however, suggested new identifications for all the above Chinese toponyms except the first one). It will be seen from thecontext that Ka-p'i-li lay to the west (or rather, north-west) ofMagadha (i.e. Bihar), and cannot therefore be Kapilavastu, nor, more probably, Kapila (Hardwār). No better chance seems to be offered by other similarly named places (e.g., the Kāveri-sangama, Old Kālpl, the famous stronghold on the Jamna, etc.). Kampilla, now Kampil (long. 79° 14'E., lat. 27°35'N.), the ancient capital of Northern Panchāla, may, however, be intended after all. Owing to its proximity to Kanauj the Gupta sovereigns may have made it their residence in the fifth century. Not unlikely it is the place alluded to by HwSn Tsang as Ka-pi-t'a, (= Kalpita, Kapida, Kapi[stha]la? see Watters, op. cit., i, 335). If so, we would have the equation Ka-p'i-li = Hwen Tsang's Ka-pi-t'a= Kampilla= Kampil, from which latter the Ganges may have been termed the “Kampil [Ka-p'i-li] River”. We prefer anyhow to adhere to the location of Ka-p'i-li proposed above, namely, between the Ganges and the Ghaghra.

† Cf., however, n. 4, p. 1196. Chao-no-p'o cannot obviously be, despite toponomastic resemblance, Chendb(the Chanab of Babar, Memoirs, see JRAS., 1898, pp. 803–4), nor Sanab[-pur](the former name of Multan). I would not even think of connecting it with Ptolemy's Sannaba, which, if not actually Sankisa, must have stood not far from it. For the location assigned to Ohao-no-p'o to the west of Bengal and to the north of Magadha (Bihar) argues a site somewhere between Jaunpur and Bhagalpur. The Chinese characters of the original text not having been given in the extract quoted above from the JRAS., the identification of the toponyms occurring therein is rendered far more difficult than would otherwise be the case.

After writing the above I came across a further reference to Chao-no-p'o(in the form Chao-na-p'u-Srh) in the following passagetranslated by Dr. Bretschneider from the Ming Shih(concluded 1724) in China Review, iv, 388:—”Jao-na-pu-r. This country lies west of Bang-k'o-la(Bengal) and is also known under the name Chung In-du(middle Hindu [ = Madhyadeia]). In ancient times it was the kingdom of Buddha. This is, I think, the same as the Zitangpur on Fra Mauro's map (fifteenth century; see Yule's Cathay, cxxxviii).” And further on (op. cit., p. 389):—”. This country adjoins Jao-na-pu-r[see above]. . . . It seems that Di-li denotes Delhi “—in which I fully agree.

This fixes the position of Chao-na-p'u-erh or Jau-na-pu-r(evidently the same as Chao-no-p'o) between Bengal and Delhi (politically the historical continuation of Ka-p'i-li), and confirms our suggested identification of it with Jaunpur. As Magadha (Bihar) lay inthe south of Chao-no-p'o, this state evidently embraced the country to the north of the Ganges between Jaunpur and Bengal. Ka-p'i-li, adjoining Chao-no-p'o on the west, must then be the Ayodhyā (= Kampilla?)—later Delhi—State, as we have surmised.

‡“Gazna” (Ghazni) is suggested as an equivalent in JRAS., loc.cit. But this relatively modern place, or some town in its neighbourhood, is referred to by Hwen Tsang in the form Ho-hsi-na, and seems too far away in the north-west to suit the purpose. Nor would Hwen Tsang's Ka-shang-na (Kasanna= Kesh), Ka-shi-pu-lo (Kasapura, in the TJnao district ?), and Ku-sM-ka-lo-pu-lo (Kusāgara-pura). Appearances are therefore in favour—till the original Chinese characters are placed before us—of either Kāsia, Kāshlpur, Kusinārā, [Northern] Kosala, as proposed above, or even Kesariyā.