In the far South of India, the Pallavas retained their posi- tion as the leading power virtually throughout the period cov- ered by plate III.D.2 and for some time thereafter. During periods of respite from wars with their northern neighbors they struggled against the Pā&ntod;&dtod;yas to the south and also inter- vened in the politics of Si&mtod;hala, where in 684 they helped to install one of the monarchs of the 2d Lambaka&ntod;&ntod;a dynasty on his throne after a period of internecine strife. For their part, the Pā&ntod;&dtod;yas ended the rule of the Ka&lline;abhras of the Ko&ndot;gu country around the turn of the 7th century, waged war suc- cessfully against the Perumals of Kerala, and tried repeatedly, but in vain, to replace the Pallavas as the regionally dominant state. Politics in Si&mtod;hala displayed considerable instability and localism but was also much affected by events in southern India.

Little is known of the geopolitical changes in northwest- ern India after the termination of Hū&ntod;a overlordship. Hsüan- tsang's account reveals that a number of petty principalities to the north of the Hindu Kush were tributary to the Turks in the second quarter of the 7th century, rather than to the Sassanians who had previously ruled the region. Kāpiśa (Arabic Kābul or Kābulistān) was the leading state between the Hindu Kush and the Indus. Its dependencies included for a time Lampā. Nagarahāra, Gandhāra, and Vara&ntod;a; but eastern Gandhāra (i.e., Tak&stod;aśilā and Si&mtod;hapura) had been lost to the rising power of Kāśmīra. Jāgu&dtod;a (Zābulistān) and Udyāna were also significant kingdoms. Kāśmīra, under the Kārko&ttod;a dynasty, having incorporated a number of peripheral states to its south- west, emerged as a contender for regional hegemony not only in the western Himalayas, but also on the plains of the Punjab. The large kingdom of &Ttod;akka in the Punjab held the mountain region of Parvata and the plains region of Mūlasthānapura as its dependencies; but virtually nothing is known of its political interaction with other states, even including the mighty king- dom of Har&stod;a, to whom it may for a time have been subject. Sindhu, an important state during Har&stod;a's reign, grew further during the rule of the Brahman Caca dynasty. The Cacas con- quered Mūlasthānapura from &Ttod;akka, as well as other parts of the Punjab, which brought them into conflict with the rulers of Kāśmīra and Kānyakubja, and in the west they extended their sway over the Makran coast and over much of the hilly country of what is now Baluchistan.

Well to the west, the period following the death of the Prophet Muhammad in A.D. 632 witnessed the momentous rise of a new power, the Arabs, who zealously set about the expansion of their newly acquired Islamic faith. In two deci- sive battles they brought about the collapse of the long-stand- ing and mighty Sassanian Empire of Persia, which was occu- pied by the victors between 637 and 651. The fall of Persia exposed South Asia to a new foreign menace, the Islamic Ca- liphate. But, apart from raids, the expansion into the subcon- tinent proper was not to occur until the early 8th century. In the interim, Arab advances were directed toward Central Asia and areas in what are today Afghanistan and Baluchistan, large parts of which fell under their control. Arab naval units also raided the ports of Sindhu, especially Debal (as early as 643) and those of Saurashtra and along the Gulf of Cambay. Our depiction of the early expansion of the caliphate (map (c)) shows only the major land thrusts and fails to suggest the areas they temporarily subdued or the maximum limits of their raiding.

Of the indigenous powers opposing the Arabs, Kābulistān, Zābulistān, and Sindhu are particularly worthy of note, Sindhu having successfully repulsed as many as six attacks via the main approach through the inland province of Kaikānān. A drive along the coast in 680, however, resulted in the Arab conquest of Makrān. Although various regional powers of northwestern India revealed awareness of the impending Arab threat and some exchanged embassies with China, seemingly in an attempt to enlist Chinese support against the caliphate, and though Kānyakubja also showed signs of its concern, no unified defensive alliance could be forged.

The political, economic, and above all cultural implications of the advent of the Arabs and of Islam in South Asia will be discussed in the Introduction to section IV.

Sources (in addition to those in the General Bibliography)

Epigraphy and Numismatics

J. Allan (1914a, b); V. V. Ayyar (1943); D. R. Bhandar- kar (1927–36); C. J. Brown (1920), (1922); R. Choudhary (1958); A. Cunningham (1891), (1894a); P. B. Desai (1962); W. Elliot (1896); J. F. Fleet (1970); A. S. Gadre (1943); E. Hultzsch (1890–1903); B. Indraji and G. Bühler (1880); V. V. Mirashi (1955); R. Mukherji and S. K. Maity (1967); R. S. Panchamukhi (1941); N. Ramesan (1962); V. Rangacharya (1919); R. Sewell (1932); P. Sreenivasachar and P. B. Desai (1961); V. Upadhyay (1970–).

Other Primary Sources

Indian: Āryamañjuśrīmūlakalpa; Bā&ntod;a (1 and 2); Cūla- va&mtod;sa; Da&ntod;&dtod;in; Dīpava&mtod;sa (a); Hemachandra (4 and 5); Jinasena II; Kalha&ntod;a (a, b, and c); Mayūrapāda Thera; Meru- tu&ndot;ga; F. E. Pargiter (1913); Purā&ntod;as (2, 5, 10, 16, and 17); Rājāvaliya.

Arabic and Persian: 'Alī ibn Hāmid, trans.; al-Balādhurī; al-Bīrūnī (2); H. M. Elliot and J. Dowson, eds. (1867–77); Minhāj Sirāj Jūzjānī (c); al-`Utbi (a).

Chinese and Tibetan: Hsüan-tsang (1 and 2); Hwui Li; I-ching; Tārānātha.

Other Works

V. S. Agrawala (1953); T. G. Aravamuthan (1925); P. C. Bagchi (1951), (1955); R. D. Banerji (1931); V. V. Bartol'd (1956–62), (1958); K. L. Barua (1933); R. Basak (1934); C. Bendall (1886); S. Chattopadhyaya (1968); B. C. Chhabra (1965); G. C. Choudhary (1964); P. C. Choudhury (1966); H. Cousens (1929); A. Cunningham (1924); D. Devahuti (1970); V. R. R. Dikshitar (1936); M. L. Ettinghausen (1906); J. F. Fleet (1882); D. C. Ganguly (1937); Gauri- Sankara Chatarji (1950); H. A. R. Gibb (1921/23), (1923); R. Gopalan (1928); M. S. Govindasamy (1965); S. H. Hodi- vala (1939, 1957); G. Jouveau-Dubreuil (1917), (1920); V. Kanakasabhai (1956); B. V. Krishnarao (1942); M. V. Krishna Rao (1936); G. Le Strange (1930); S. Lévi (1905–8); T. V. Mahalingam (1952); H. Mahtab (1959, 1960); R. C. Majumdar (1931); C. Minakshi (1938); V. V. Mirashi (1960–66); R. K. Mookerji (1926); G. M. Moraes (1931); K. M. Munshi (1955); K. A. A. Nilakanta Sastri (1929), (1955); K. A. A. Nilakanta Sastri, ed. (1939); R. B. Pandey (1957); K. M. Panikkar (1922); V. S. Pathak (1966); L. Pe- tech (1939); E. A. Pires (1934); H. C. Ray (1931, 1936); S. C. Ray (1970); D. R. Regmi (1969); N. K. Sahu (1958); B. P. Sinha (1954); D. Sircar (1935), (1939); B. Spuler (1960–69); A. Sreedhara Menon (1967); M. A. Stein (1907), (1921), (1933); R. S. Tripathi (1937); C. V. Vaidya (1921– 24); N. Venkata Ramanayya (1950); K. J. Virji (1955); J. P. Vogel (1911).

III.D.3. Purā&ntod;ic India

Our maps of Purānic India reconstruct the geography of South Asia on the basis of the sacred Hindu texts known as the Purā&ntod;as, which constitute an encyclopedic corpus of lit- erature on Indian history, geography, religion, and culture. Purāna literally means "ancient lore" or "old narrative," but it has also become a generic name for a class of literature and in some cases deviates from its original meaning, as is indi- cated by the title of the Bhavi&stod;yat Purā&ntod;a, or Ancient Lore of the Future, an apparent contradiction in terms. Traditionally, the eighteen major Purā&ntod;as, designated as the Mahā-Purā&ntod;as, are distinguished from more than one hundred Upa-Purā&ntod;as (minor or subsidiary Purā&ntod;as). The former allegedly devel- oped from a single original compendium of tradition and leg- ends. Unlike the Vedas, which are immutable, the subject mat- ter of the Purā&ntod;as developed over time and reflected the varying objectives of their compilers. The most recent stage in the de- velopment of the Purā&ntod;as, for example, is marked by its sec- tarianism and its didactic and propagandistic character.

In utilizing the Purā&ntod;ic evidence for the study of Indian history, geography, and culture, the question of the age of the texts is particularly important. For, whereas the original com- pilation of a particular Purā&ntod;a may be attributable to the 8th or 7th century B.C., the extant versions may have been com- piled many centuries later and may have continued to be en- larged and altered even after the mention of the eighteen Mahā-Purā&ntod;as in the Mahābhārata redacted in c. A.D. 400. Some of the historical and geographical material of the extant Purā&ntod;as is indeed very ancient, and several scholars have ar- gued that the first draft of the geographical list (bhuvanakośa) of the Matsya Purā&ntod;a was compiled during the 3d century B.C., whereas comparable materials in such other works as Brahma, Vi&stod;&ntod;u, Kurma, Mārka&ntod;&dtod;eya, and Padma Purā&ntod;as belong to the period between c. A.D. 400–600 and others, such as the Garu&dtod;a, are of even later date. It follows from the above that the geographical matters portrayed on our maps do not relate to any specific point in time. However, most of the materials were known between the 4th and 7th centuries, the time span covered by subsection III.D. It should be apparent that any attempt on our part to indicate on our map stages in the growth of knowledge of peoples, regions, and specific places on the basis of Purā&ntod;ic evidence would be futile. But a comparison with maps of preceding subsections will reveal the early knowl- edge of many of the names we locate.

The problem of chronological interpretation of Purā&ntod;ic names is more complex than simply determining the dates of the compilations. In transmitting traditional knowledge, the Purā&ntod;ic compilers exercised considerable freedom and not only included information acquired over preceding centuries by peoples of regions where particular Purā&ntod;as were compiled, but also replaced less known peoples of earlier ages with the better known ones of more recent times or even of their own age. For example, the original Ür&ntod;a was corrupted into Hū&ntod;a and the Śatadruja or Śataduja (people of the Śatadru River) were transformed into Śatadravas, Sakad&rtod;has, Śakadruhyas, Śakah&rtod;adas, Śakabhadras, and, finally, owing to some copyist's carelessness, into Śaka-Hū&ntod;as, since they were a well-known people of the northwest when the texts were revised. This type of vitiation of the data obviously requires that corroboration from other sources be obtained before using Purā&ntod;ic evidence in historical geographic reconstructions.

The foregoing caveats notwithstanding, the value of the Purā&ntod;as for reconstructing ancient history and geography is immense. A number of the Purā&ntod;as deal with cosmography and contain, as has been noted, a section called the bhuvana- kośa, or geographical list, comprising janapadas (peoples/ tribes and their territories), mountains, and rivers. In some Purā&ntod;as these lists are comprehensive, in others condensed. The longer lists of janapadas are available in the Matsya, Vāyu, Brahmā&ntod;&dtod;a, Vāmana, and Mārka&ntod;&dtod;eya Purā&ntod;as, while the shorter ones are found in the Vi&stod;&ntod;u, Brahma, Kūrma, Śiva, and Agni Purā&ntod;as. The bhuvanakośa texts of the Garu&dtod;a and Padma Purā&ntod;as differ from the preceding ones. The account of the Padma is identical with the Bhī&stod;ma-Parvan section of the Mahābhārata, and the Vis&ntod;udharmottara Purā&ntod;a, which is excluded from the canonical list of the Mahā-Purā&ntod;as, gen- erally follows the Garu&dtod;a's version. The longer texts are iden- tical in the five aforementioned Purā&ntod;as, which also include most of the data of the shorter lists. Therefore our depiction of janapadas in map (a) generally draws upon the longer lists.

The Purā&ntod;ic geographical lists do not exhaust the toponymic data of those works. Important information is also available in sections dealing with cosmography, sacred places, sacred riv- ers, mountains, and forests, as well as in the historical sections. Chapters 61 and 81 of the Garu&dtod;a Purā&ntod;a, for example, deal exclusively with places of pilgrimage and their mahātmyas (glories or significance).

The recognition of the importance of the Purā&ntod;as and their geographical data has grown since the publication of Wilson's translation of the Vi&stod;&ntod;u Purā&ntod;a in 1840. Burgess suggested as early as 1885 that a list of Purā&ntod;ic toponyms contained in the bhuvanakośa texts should be prepared to better elucidate the ancient geography of India, and Pargiter undertook such a pioneering work in his translation of the Mārka&ntod;&dtod;eya Purā&ntod;a (1904), providing annotated identifications of the names of bhuvanakośa chapters. Notable works by Pargiter on the Pu- rā&ntod;as include The Purā&ntod;a Text of the Dynasties of the Kali Age (1913) and The Ancient Indian Historical Tradition (1922). Other Indologists have critically studied Purā&ntod;ic cos- mological and geographical materials, but few have attempted a satisfactory cartographic representation of their findings or those of fellow scholars.

In our reconstruction of the maps and diagram (lower left) of plate III.D.2, we have utilized both the original Pu- rā&ntod;ic texts, already cited, and numerous modern works (see "Sources," below). Although the critical texts of Kirfel (1927), (1954) and Sircar (1945), (1967) were utilized for their physical geographic data, we have generally based our recon- struction of major cultural regions on Lewis (1962). For iden- tifying toponyms the sources on which we have drawn include a variety of non-Purā&ntod;ic Indian literature, accounts of foreign travelers, and inscriptions.

Purā&ntod;ic data on cosmography and geography are extensive, confusing, and largely mythological. Both the ancient Brah- manical four island continent conception of the earth or uni- verse (depicted on plate III.A.1) and a seven island continent conception (not depicted in this atlas) were known to the Purā&ntod;ic compilers. In either one, Jambudvīpa, or Bhārata (greater India), was placed to the south of Mount Meru, the pericarp of the earth. The diagram in the lower left corner of plate III.D.3 shows yet another enlarged cosmographic con- ception of the Purā&ntod;ic authors. Jambu-Dvīpa, the World, in- cludes Bhārata as its ninth division. Although, as noted, some Purā&ntod;ic cosmographies regard Bhārata as equivalent to Jambu- Dvīpa, most observed a distinction between the two.

Apart from "Bhārata," "Bhāratavarsa," or "Jambudvīpa," Purā&ntod;ic names for South Asia/India include "Sāgarasa&mtod;v&rtod;ta" and "Kumārīdvīpa." In noting the size and shape of this por- tion of the world, the dimensions cited and the descriptions vary widely. One view, extracted from the Vi&stod;&ntod;u Purā&ntod;a, ap- pears at the upper left corner of our map. The most common Purā&ntod;ic conception of Bhārata was of a "bow-shaped" region lying between Himavat (the Himalayas) to the north and the sea in the southern portion of Jambu-Dvīpa. The range of views, however, is suggested by that of the Mārka&ntod;&dtod;eya Pu- rā&ntod;a, which likened the shape of Bhāratavarsa to a tortoise lying outstretched and facing east.

The Purā&ntod;as traditionally divided Bhāratavarsa into nine subdivisions: Indra-Dvīpa, Kaserumat, Tāmrapar&ntod;a, Gabhasti- mat, Nāgadvīpa, Saumya (or Ka&ttod;āha), Gandharva (or Kumara