India from east to west is 28,000 stadia, and from north to south 32,000 stadia," or 60,000 stadia altogether.

At a somewhat later date the shape of India is described in the 'Mahâbhârata' as an equilateral triangle, which was divided into four smaller equal triangles.* The apex of the triangle is Cape Comorin, and the base is formed by the line of the Himâlaya mountains. No dimensions are given, and no places are mentioned; but, in fig. 2 of the small maps of India in the accompanying plate, I have drawn a small equilateral triangle on the line between Dwâraka, in Gujarat, and Ganjam on the eastern coast. By repeating this small triangle on each of its three sides, to the north-west, to the north-east, and to the south, we obtain the four divisions of India in one large equilateral triangle. The shape corresponds very well with the general form of the country, if we extend the limits of India to Ghazni on the north-west, and fix the other two points of the triangle at Cape Comorin, and Sadiya in Assam. At the presumed date of the composition of the 'Mahâbhârata,' in the first century A.D., the countries immediately to the west of the Indus belonged to the Indo-Scythians, and therefore may be included very properly within the actual boundaries of India.

Another description of India is that of the NavaKhanda, or Nine-Divisions, which is first described by the astronomers Parâsara and Varâha-Mihira, although it was probably older than their time,† and was after

* Journ. Asiat. Soc. Bengal, xx. Wilford, quoting the Bhishma Parva of the Mahâbhârata,' as communicated to him by Colebrooke.

Dr. Kern, in preface to the Brihat-Sanhitâ' of Varâha-Mihira, p. 32, states that Varâha's chapter on Geography is taken almost intact, but changed in form, from the 'Parâsaratantra,' and must, therefore, be

wards adopted by the authors of several of the Purânas. According to this arrangement, Pánchála was the chief district of the central division, Magadha of the east, Kalinga of the south-east, Avanta of the south, Anarta of the south-west, Sindhu-Sauvira of the west, Hârahaura of the north-west, Madra of the north, and Kauninda of the north-east.* But there is a discrepancy between this epitome of Varâha and his details, as Sindhu-Sauvira is there assigned to the south-west, along with Anarta. This mistake is certainly as old as the eleventh century, as Abu Rihâm has preserved the names of Varâha's abstract in the same order as they now stand in the 'Brihat-Sanhitâ.' These details are also supported by the Mârkandeya Purâna,' which assigns both Sindhu-Sauvira and Anarta to the southwest. §

I have compared the detailed lists of the 'BrihatSanhitâ' with those of the Brahmânda, Mârkandeya, Vishnu, Vâyu, and Matsya Purânas; and I find that, although there are sundry repetitions and displacements of names, as well as many various readings, yet

considered as representing the geography of Parâsara, or perhaps yet more ancient works, " and not as the actual map of India in VarâhaMihira's time."

*Brihat-Sanhitâ,' ch. xiv. 32, 33.

Ibid., xiv. 17,

Nairrityám disi desâ

Pahlava Kamboja Sindhu-Sauvîra—

Wilford has given Varâha's list in vol. viii. p. 341, of Bengal Asiat. Researches; but he has made two divisions of Sindhu- Sauvira, and omitted Kauninda. His details, however, agree with the 'BrihatSanhitâ,' in assigning Sindhu-Sauvira as well as Anarta to the southwest.

The Nine Divisions of Abu Rihân are given in Reinaud's 'Mémoire sur l'Inde,' pp. 116, 117. Compare No. II. Map, fig. 3.

*Some of them,

all the lists are substantially the same." however, are differently arranged. All of the Purânas, for instance, mention the Nine Divisions and give their names, but only the Brahmânda and Mârkandeya state the names of the districts in each of the Nine Divisions; as the Vishnu, Vâyu, and Matsya Purânas agree with the 'Mahâbhârata' in describing only five Divisions in detail, namely, the middle Province and those of the four cardinal points.

The names of the Nine Divisions given in the 'Mahâbhârata' and the Purânas differ entirely from those of Varâha-Mihira; but they agree with those of the famous astronomer Bhâskarâchârya.† They follow the same order in all; namely, Indra, Kaserumat, Tamraparna, Gabhastimat, Kumarika, Naga, Saumya Váruna, Gándharva. No clue is given to the identification of these names, but they certainly follow a different order from that of Varâha's Nine Divisions, as Indra is the east, Varuna the west, and Kumárika the middle, while Kaseru must be the north, as the name is found in the detailed lists of the Vâyu and Brahmânda Purânas.

The division of India into five great provinces would appear to have been the most popular one during the early centuries of the Christian era, as it was adopted by the Chinese pilgrims, and from them by all Chinese writers. According to the Vishnu Purâna, the centre

The list of the Brahmânda is given by Wilford in Bengal Asiat. Researches, viii. 334,-that of the Vishnu Purâna in Wilson's translation, where, also, will be found the list of the 'Mahâbhârata;' that of the Markandeya Purana is in Ward's Hindus,' iii. 9.

+ 'Siddhânta Siromani,' chap. iii. 41.

Wilson's Vishnu Purâna,' edited by Hall, vol. ii. b. iii. c. 3. p. 132. The north Division is not mentioned in the text; but as the Hunas

was occupied by the Kurus and Pânchúlas; in the east was Kamarupa, or Assam; in the south were the Pundras, Kalingas, and Magadhas; in the west were the Surâshtras, Suras, Abhiras, Arbudas, Kárushas, Málavas, Sauviras, and Saindhavas; and in the north the Hunas, Salwas, Sakalas, Rámas, Ambashtas, and Parásikas.

In the Geography of Ptolemy the true shape of India is completely distorted, and its most striking feature, the acute angle formed by the meeting of the two coasts of the Peninsula at Cape Comorin is changed to a single coast-line, running almost straight from the mouth of the Indus to the mouth of the Ganges. The cause of this mistake is partly due to the erroneous value of 500, instead of 600, Olympic stadia, which Ptolemy assigned to an equatorial degree, partly to an over-estimate in converting road-distance into mapmeasurement, but chiefly to the excess which he allowed for the distances of land journeys over those of sea voyages.*

If the measures of distance by sea had been increased in the same proportion, or had been estimated at the same value, as the measures of distance by land, all the places would have retained the same relative positions. But the consequence of Ptolemy's unequal estimate of the value of land and sea distances was to

and Sakalas certainly belonged to the north, I presume that the north has been accidentally omitted. There is a similar omission of the name of Kumârika in this Purâna, which has only eight names for the Nine Divisions.

*The question of Ptolemy's erroneous longitudes is treated at length in Appendix C, where I have given all the data on which Sir Henry Rawlinson has founded his correction of three-tenths of the geographer's distances in easting.