Hindus Beyond the Hindu Kush: Indians in the Central Asian Slave Trade*

Abstract Historical analyses of slavery in India generally emphasize the escalation of this social institution during the era of Muslim domination in north India. The present study is not an exception to this rule. However, while historical records make it clear that the Delhi Sultans and Mughal emperors retained slavery in order to suit their political and economic needs, it should be emphasized that Muslim rulers did not introduce slavery to the subcontinent. Sources such as the Arthaśāstra, the Manu-smrti and the Mahābhārata demonstrate that institutionalized slavery was well established in India by the beginning of the common era. Earlier sources suggest that it was likely to have been equally widespread by the lifetime of the Buddha (sixth century BC), and perhaps even as far back as the Vedic period. Furthermore, just as slavery was common in India long before the eighth-century Islamic conquests in Sind, recent work demonstrates that the institution continued, in various manifestations, well after the decentralization of the Mughal Empire in the early eighteenth century. Still, it is argued here that the expansion of slavery in Muslim India is an important component of the medieval and early modern history of the region and, at least in terms of its role in the commercial and cultural relations of India and Central Asia, it is a subject that would benefit from further historical analysis.

transportation of people as a commodity across India's northwest frontier dates, like the institution of slavery itself, to ancient times. Alongside Buddhist Qalmaqs (Oirats), Christian Russians, non-Sunn¿ Afghans, and the predominantly Sh¿«a Iranians, Hindu slaves were an important component of the highly active slave markets of medieval and early modern Central Asia. The history of the institution of slavery in India and Central Asia is a subject that merits much more attention than can be offered in this brief study. 4 The intention here is only to demonstrate that the slave trade between India and Central Asia was signi®cant, and to suggest that the movement of considerable numbers of Hindus to the Central Asian slave markets was largely a product of the state-building efforts of the Delhi Sultanate and Mughal Empire in South Asia.
The slave population of early modern Central Asia was considerable; virtually every af¯uent household included several slaves to look after its affairs and maintain the garden, and large numbers of slaves were used to cultivate the land and watch over livestock on the plantation-style farms of Central Asia's wealthy families. 5 Slaves were also used for such purposes as soldiering, maintaining irrigation canals, working in brick factories, and many were trained in construction engineering. 6 The seventeenth-century Ma¢lab al-¢a Ålib¿n records, for example, that one Juybari Sheikh (a Naqshbandi Su® leader) owned over 500 slaves, forty of whom were specialists in pottery production while the others were engaged in agricultural work, tending livestock, and carpentry. 7 Demand was especially high for skilled slaves, and India's comparatively larger and more advanced textile industry and agricultural production, and its magni®cent imperial architecture, demonstrated to its neighbours that skilled labour was abundant in the subcontinent. 8 This accounted for the common practice of rival political powers enslaving and relocating large numbers of artisans following successful invasions. 9 For example, during Timur's late fourteenthcentury sack of Delhi, several thousand skilled artisans were enslaved and taken to Central Asia. Timur presented many of these slaves to his subordinate elite, although he reserved the masons for use in the construction of the Bibi Khanum mosque, located in his ourishing capital of Samarqand. 10 It is perhaps not surprising that attractive, young female 4 For a social analysis of slavery in nineteenth-century Bukhara, including facsimiles of Persian-language archival records, see T. Faiziev, Bukhoro feodal jamiyatida qullardan foidalanishga doir hujjatlar (XIX Asr), in Uzbek (Tashkent, 1990). 5 According to Mukminova, the majority of slaves mentioned in sixteenth-century waqfna Åmas were used for agricultural purposes or to tend livestock. Cf slaves commonly demanded an even higher market price than those skilled at construction engineering. 11 Because of their identi®cation in Muslim societies as ka Å®rs,``non-believers'', Hindus were especially in demand in the early modern Central Asian slave markets. 12 They were by no means, however, the only ethnic or religious group present in large numbers. It is well known that the slave markets were also stocked with considerable numbers of Iranian slaves, whose association with the Sh¿«a sect of Islam made them legitimate targets for the Sunn¿ Muslim Uzbek and Turkman slave traders. The abundance of Iranian slaves in early modern Central Asia is generally attributed to the great number of prisoners taken during the Uzbeks' wars with the Safavids and, in later years, to repeated Turkman raids on the poorly protected villages of northeastern Iran. 13 It is also well known that Christian Russians were similarly abducted from their frontier settlements by Qalmaq, Crimean Tatar, Nogay and Bashkir pastoralists and brought to the markets of Khiva, from where many were further transported to Bukhara and Balkh. 14 Despite ample evidence demonstrating the presence of a considerable population of Indian slaves in medieval and early modern Central Asia, this aspect of the Central Asian slave trade has received much less attention. A survey of available sources reveals that a Bukharan waqfna Åma (a letter indicating a religious endowment) written in 1326 repeatedly lists Indian slaves together with slaves from various other parts of Asia. 15 A similar document dating to 1489 from the archive of the great Naqshbandi Sheikh Khwaja Ahrar (1404±90) mentions a group of Indian slaves working as agricultural labourers and artisans on an estate near Bukhara. 16 Furthermore, the presence of Hindu slaves even among the Turkic pastoral groups in early modern Central Asia is mentioned in an account of the Uzbek ruler Shibani Khan's victory over the Qazaq ruler Tanish Sultan. 17 According to this account, in the winter of 1509±10, a fourteen-year-old Indian slave escaped from his cruel master in a Qazaq qishlaq (winter encampment) near the city of Turkestan and, while wandering through the steppe, fortuitously came across the encampment of Shibani Khan and informed the Uzbeks of the location of the enemy Qazaq qishlaq. This enabled the 11 According to the account of Baron Meyendorff, in the 1820s in Central Asia a slave skilled as a craftsman was valued at approximately 100 tilla, whereas an attractive young female slave could fetch as much as 150 tilla. E. K. Meyendorff Uzbeks to defeat Tanish Sultan's forces, for which the slave earned the favour of Shibani Khan, who renamed him``Khush Khabar'' (``Good News'').
Signi®cant numbers of Indians were taken to the slave markets of Central Asia in a variety of ways. It is a regular, persistent feature in the commercial history of the region that many slaves were exported by caravan merchants, who either purchased them outright or received them in exchange for other commodities in demand in India, especially horses. 18 While travelling from Lahore to Kabul in 1581 the Portuguese Jesuit missionary Father Antonio Monserrate reported that one tribe in the Punjab, identi®ed as thè`G accares'' (Ghakkars), had made mediating the trade of Indian slaves for Central Asian (``Turk¿'') horses such a regular practice that they had even become associated with the proverb,``slaves from India, horses from Parthia''. 19 The Central Asian slave markets swelled with Indians following Shah Jahan's aborted annexation of Balkh in 1646±47 and the capture and enslavement of many of the retreating Indian soldiers. 20 According to the Uzbek chronicler Muhammad Yusuf Munshi, following a four-month siege of the Indian army in Balkh during an unusually severe winter, Shah Jahan recalled his starving army to India and, during their retreat, the Central Asian``wolves'' captured the¯eeing Indiaǹ`s lave-sheep'' from every direction and took them to Samarqand, Turkestan and Tashkent. 21 Although this event was unique in the history of the region, it resulted in the in¯ux of a large number of Indians into Central Asian society. Thus, whereas in 1589 a thirty-three-year-old male Indian slave in good health was sold in Samarqand for 225 tanga, after the Mughal retreat in 1647 the Central Asian slave markets were¯ooded and the price of an Indian slave dropped as low as 84 tanga. 22 Smaller numbers of skilled slaves were also commonly included in the gifts sent between the rulers of India and Central Asia. For example, in the sixteenth century, Badr al-Din Kashmiri reported that four slaves skilled in masonry were included in the gifts that the Mughal emperor Akbar had dispatched to the Bukharan ruler «Abd Allah Khan II. 23 A century later, Shah Jahan is reported to have sent 100 Indian slaves to the Ashtarkha Ånid rulers Imam Quli Khan and Nadir Muhammad. 24 Also, in addition to their regular employees, Indian merchants venturing to Central Asia commonly brought slaves with them, many of whom were likely to have been sold in the Central Asian markets. Sources further demonstrate that, while traversing dangerous caravan routes, some unfortunate  Probably the greatest factors contributing to the increased supply of Indian slaves for export to markets in Central Asia in this period were the military conquests and tax revenue policies of the Muslim rulers in the subcontinent. 28 Already during the early Arab invasions of Sind at the beginning of the eighth century, the armies of the Umayyad commander Muhammad bin Qasim are reported to have enslaved tens of thousands of Indian prisoners, including both soldiers and civilians. 29 The policy of enslaving conquered Indians continued in a much greater magnitude under the Turko-Afghan Ghaznavid invasions of subsequent centuries. For example, according to the Ta Å»r¿kh-i Firishta, following the Ghaznavid capture of the Indian city of Thanesar in the year 1014,``the army of Islam brought to Ghazna about 200,000 captives (qar¿b do s¿t haza År banda), and much wealth, so that the capital appeared like an Indian city, no soldier of the camp being without wealth, or without many slaves''. The same source also mentions that, several decades later, the Ghaznavid ruler Sultan Ibrahim led another raid into the Multan area of northwestern India and returned to Ghazna with 100,000 captives. 30 While it is important to note that Firishta's history was written several centuries after the Ghaznavid period, his assertions are supported by the reports of contemporary observers. For example, in his early 25 See, for example, the account of the Indian merchant «Ala al-Din Khan, who was captured while returning to India from Balkh in 1646. Living as a slave in Bukhara, and then in Khiva where he was owned by a Tatar woman, he eventually escaped to Cherno Yar where he was arrested by Russian authorities in 1661. He was then sent to Astrakhan where he applied to the Tsar for permission to convert to Christianity. K. A. Antonova, N. M. Gol'dberg and T. D. Lavrentsova, eds., Russko-indiiskie otnosheniia v XVII v. (Moscow, 1958) eleventh-century Ta Å»r¿kh al-Yam¿n¿, the Arab historian al-«Utb¿ recorded that in 1001 the armies of Mahmud of Ghazna conquered Peshawar and Waihand,``in the midst of the land of Hindustan'', and captured some 100,000 youths. Several years later, following his twelfth expedition into India in 1018±19, Mahmud is reported to have returned to his capital with such a large number of slaves that their value was placed at only two to ten dirhams each. This price was apparently so low that, according to al-«Utb¿,``merchants came from distant cities to purchase them, so that the countries of Ma Å wara Å» an-nahr (Central Asia), «Ira Åq and Khura Åsa Ån were ®lled with them, and the fair and the dark, the rich and the poor, mingled in one common slavery''. 31 Later, during the Delhi Sultanate period (1206±1555), references to the abundant availability of Indian slaves at low prices are common. This should be attributed primarily to the vast human resources of India, which boasted a much larger and denser population than its neighbours to the north and west. 32 Many of these Indian slaves were reserved for use in the subcontinent, but their availability in substantial numbers greatly contributed to their affordability, which likewise increased their demand in international markets.
The revenue system of the Delhi Sultanate produced a considerable proportion of the Indian slave population as these rulers, and their subordinate iq¢a Å«da Års, ordered their armies to abduct large numbers of Hindus as a means of extracting revenue. 33 While those communities that were loyal to the Sultan and regularly paid their taxes were often excused from this practice, taxes were commonly extracted from other, less loyal groups in the form of slaves. Thus, according to Barani, the Shamsi``slave-king'' Balban (r. 1266±87) ordered his iq¢a Å«da Års in Awadh to enslave those peoples resistant to his authority, implying those who refused to supply him with tax revenue. 34 Sultan «Ala» al-Din Khalji (r. 1296±1316) is similarly reported to have legalized the enslavement of those who defaulted on their revenue payments. 35 This policy continued during the Mughal era. 36 An even greater number of people were enslaved as a part of the efforts of the Delhi Sultans to ®nance their expansion into new territories. 32 The early modern populations of Uzbek Turan and Safavid Iran were not likely to have been much more than ®ve million each, while at the end of the sixteenth century, the population of the Mughal Empire is estimated to have reached between sixty and ninety-eight million. See Dale, Indian Merchants, pp. 15±21. 33  the ®rst of the Shamsi slave-kings) invaded Gujarat in 1197 and placed some 20,000 people in bondage. Roughly six years later, he enslaved an additional 50,000 people during his conquest of Kalinjar. 38 Later in the thirteenth century, during Balban's campaign in Rantabhur, he is reported to have defeated the Hindu army and to have subsequently takeǹ`c aptives beyond computation''. 39 K. S. Lal's assertion that the forcible enslavement of Indians due to military expansion``gained momentum'' under the Khalji and Tughluq dynasties is supported by available ®gures. 40 At the beginning of the fourteenth century, Barani suggested that Sultan «Ala» al-Din Khalji owned some 50,000 slave-boys, and had an additional 70,000 slaves engaged in architectural construction. Later that century, Sultan Firuz Tughluq is said to have owned 180,000 slaves, roughly 12,000 of whom were skilled artisans. 41 It should be noted that many of the very large number of slaves owned by Sultans «Ala» al-Din Khalji and Firuz Tughluq were likely to have been military slaves who were not used as labourers or domestics. Still, despite earlier traditions of maintaining an army comprised of both Hindu soldiers and Turkic slave-soldiers (ghula Åms, mamlu Åks) from Central Asia, the early thirteenth-century rise of the Mongol Empire disrupted the movement of Turkic Mamluks to the Delhi Sultanate and forced the Delhi Sultans to turn to local Indian populations to satisfy their demand for both military and domestic slaves. 42 By and large, the enslavement of Hindus and their exportation to Central Asia continued unhindered throughout the Mughal period. Although the Mughal emperor Akbar (r. 1556±1605) attempted to prohibit the practice of enslaving conquered Hindus, his efforts were only temporarily successful. 43 According to one early seventeenth-century account, «Abd Allah Khan Firuz Jang, an Uzbek noble at the Mughal court during the 1620s and 1630s, was appointed to the position of governor of the regions of Kalpi and Kher and, in the process of subjugating the local rebels,``beheaded the leaders and enslaved their women, daughters and children, who were more than 2 lacks [200,000] in number''. 44 Whether agriculturalists or pastoralists, following their enslavement many of these individuals were sent in large numbers to markets beyond India's northwest frontier, far from their family support systems. Recognizing that the ®gures presented in the chronicles and other accounts are likely to be exaggerated, it still seems reasonable to accept the estimation that, over the years, Mughal military expansion in India accounts for the enslavement and exportation of hundreds of thousands of individuals. This included not only those men who militarily resisted the Mughal armies, but also vast numbers of women and children. The slave population was signi®cantly augmented by the considerable number of children who were sold into slavery by ®nancially destitute parents, a factor that increased dramatically in times of famine or other economic hardship. 45 As noted above, mediatory traders, such as the Ghakkars, purchased many of the conquered and destitute and marched them to India's northwest frontier to be exchanged for Central Asian horses.
While a detailed study of slavery in early modern Central Asia has yet to be written, a survey of available sources reveals the presence of signi®cant numbers of Indian slaves in the region. For example, the Uzbek ruler Kuchkunchi Khan (an uncle of Shibani Khan and ruler of Samarqand from 1510±30) is reported to have used ninety-®ve Indian slaves for the construction of irrigation canals in the vicinity of the Central Asian city of Turkestan. 46 According to Muhammad Talib, author of the Ma¢lab al-¢a Ålib¿n, the late sixteenth-century Juybari Sheikh Khwaja Sa«id, son and successor of the great Sheikh Khwaja Islam (c. 1492±1563), had in his possession``1000 slaves of Indian, Qalmaq, and Russian origin'' many of whom tended his ®elds and animal herds, while others were engaged in construction and in household services. 47 In 1558, the Englishman Anthony Jenkinson visited Central Asia where he noted the vitality of the Bukharan slave market and observed that Indian and Iranian merchants who visited Bukhara commonly dealt in slaves. 48 Unfortunately, there is no means by which to determine precisely how abundant Indian slaves were in early modern Central Asia. It is, however, possible to establish a rough estimate of the proportion of slaves of Indian origin in relation to those of other regions, at least in terms of the slave population of late sixteenth-century Samarqand. A survey of seventy-seven letters regarding the manumission or sale of slaves in the Majmu Å«a-i-watha Å»iq reveals that slaves of Indian origin (hind¿ al-a Å¡l ) accounted for over 58 per cent of those whose region of origin is mentioned. 49 It must be emphasized that the Majmu Å«a-i-watha Å»iq provides a relatively small, restricted sample. It would be irresponsible to suggest that, based on the information elicited from this source, 58 per cent of the population of slaves in all of Central Asia was of Indian origin. However, the predominance of Indian slaves in the early modern markets of Central Asia is further supported by information found in the Khu¢u Å¢-i 45 Kidwai,`Sultans, Eunuchs and Domestics', p. 88. 46 P. P. Ivanov, Khoziaistvo dzhuibarskikh sheikhov: k istorii feodal'nogo zemlevladeniia v Srednei Azii v XVI±XVII vv. (Moscow, 1954), p. 24. 47 Muhammad Talib, Ma¢lab al-¢a Ålib¿n, fol. 81a±b. Unfortunately Talib does not mention the proportions of each. Although Indian slaves are not mentioned speci®cally, it is reasonable to assume that they were part of the contingency of roughly 3,000 slaves used by Yalangtosh Bey, the vazir of the Ashtarkha Ånid ruler «Abd al-«Aziz Khan (r. 1645±80), to construct the Sherdar and Tillakari madrasas in Samarqand. Cf. Mukminova, Sotsial'naia differentsiatsiia, p. 125; P. P. Ivanov, Ocherki po istorii Srednei Azii (XVI±seredina XIX v.) (Moscow, 1958), p. 71. A signi®cant number of Indians were also likely to have been included in the hundreds of slaves owned by the Juybari Sheikhs, mentioned in note 7, above.
48 Jenkinson, Early Voyages and Travels, I, pp. 88±89. According to Morgan and Coote, editors of Jenkinson's account, the Bukharan slave market was the largest in the region. Ibid., I, p. 89, note 2. 49 See Majmu Å«a-i-watha Å»iq, fols 3a±50b. Of the seventy-seven slaves referred to in these entries, which date from 1588±92, twenty-nine are identi®ed as being``born of the house'' (kha Ånahza Åd ) with no information given as to their ethnic identity. Twenty-eight of the remaining slaves are identi®ed as being``of Indian origin'' (hind¿ al-a Å¡l ); ®fteen as being``of Afghan origin'' (a Åfgha Ån¿ al-a Å¡l ); four as being``of Russian origin'' (ru Ås¿ al-a Å¡l ); and one as being`o f Badakhshani origin'' (badakhsh¿ al-a Å¡l ).
mamhu Åra bemahr-i qa¥a Åt-i Bukha Åra Å, a smaller collection of judicial documents from early eighteenth-century Bukhara that includes several letters of manumission. 50 Again, over half of these letters refer to slaves``of Indian origin''. 51 It is revealing that, even in the model of a legal letter of manumission written by the chief qa Å¥¿ for his assistant to follow, the example used is of a slave``of Indian origin''. 52 The Central Asian slave trade continued at an active level throughout the eighteenth century, although during this period there were considerably fewer Indian slaves exported to Central Asian markets. Rather, from the beginning of the eighteenth century, the slave markets of Bukhara and Khiva appear to have been increasingly stocked with Iranians. Thus, whereas John Mouraview, an Armenian merchant from Derbend, observed in 1813 that Khiva was home to some 30,000 Iranian slaves and 3,000 Russian slaves, his account does not mention any Indian slaves. 53 In 1821, Baron Meyendorff visited Bukhara and estimated that there were in excess of 25,000 ± and perhaps as many as 40,000 ± Iranian slaves in that city. 54 Despite the prohibition of the slave trade according to the Russo-Khivan treaty of 1873, a Russian report from the mid-1870s estimates that there were still approximately 10,000 Iranian slaves in Khiva at that time. 55 The Central Asian slave trade remained active up to Russia's nineteenth-century expansion into the region. According to the Russian academician A. A. Semenov, when Russia annexed the Samarqand vilayat in 1868, there were still some 10,000 slaves in just that district. 56 The reliability of this rather high ®gure is supported by the account of an Indian merchant who, in the early 1860s, reported to British authorities that at that time the Bukharan Amirate was home to 100,000 slaves, 20,000 of whom lived in the city of Bukhara. 57 In the 1880s, the British Lt Col C. E. Stewart similarly reported that, although the slave population in Central Asia had signi®cantly diminished, prior to the Russian prohibition of the slave trade, the total number of slaves in the combined Khivan, Bukharan and Turkman territories was in excess of 100,000. 58 Muzaffar Alam has attributed the decline in the exportation of Indian slaves in the eighteenth century to both economic and social factors, specifying that, as India produced more textiles for export, it was no longer necessary for Indian merchants to trade slaves for Central Asian horses. 59 Alam's assertion about the magnitude of Indian textile production and its demand in Central Asian markets is well founded. This does not, however, fully explain the unwillingness, or inability, of merchants to continue exporting large numbers of Indian slaves to Central Asia, presuming the availability of an adequate supply at an agreeable price. With this in mind, it seems reasonable to suggest that the supply of Indian slaves dwindled as the Mughal Empire decentralized and its military expansion came to an end. This was compounded by the general exclusion of slaves from the tax-revenue systems of the successor states and the growing commercial and cultural separation of India and its neighbours to the north and west under the British Raj. The combination of these factors resulted in a general decline in the exportation of Indians to the Central Asian slave markets. This left Central Asian slave traders little recourse but to look elsewhere for a viable source to satisfy the market's substantial demand. It was for this reason that, from the beginning of the eighteenth century, the infamous Turkman slave-raiders looked to the numerous comparatively close, and poorly defended, Iranian cities and villages bordering their territory for their unfortunate merchandise.
That is not to say that slavery in the subcontinent ended. Despite the best efforts of the slave-holding elite to conceal the continuation of the institution from the historical record, slavery was practiced throughout colonial India in various manifestations. 60 Furthermore, the movement of Indians and Afghans to the Bukharan slave markets did not come to a complete halt. Smaller numbers of Indian slaves continued to be sold in the markets of Bukhara well into the nineteenth century. Turgun Faiziev has uncovered several nineteenth-century records documenting the presence in Bukhara of slaves of Indian origin, some identi®ed as``Hindu'' and others as``Chitrari'' (i.e. from Chitral, a region deep in the Karakoram mountains of far northwestern India, modern Pakistan, bordering Afghan Badakhshan). 61 Writing in the early nineteenth century, «Abd al-Karim Bukhari also mentioned that Balkh and Bukhara regularly sent weaponry and clothing to Badakhshan in exchange for``black-faced slaves'' who, Bukhari reported, came from Chitral and were not Muslims. 62 In the early nineteenth century, Josiah Harlan observed that frequent slave raids into Chitral were organized by Murad Beg, the Afghan ruler of Kunduz and a``great wholesale dealer in this unholy merchandise''. According to Harlan's account, the Central Asian slave trade``opens an insatiable outlet for the disposal of Muraad's insubordinate subjects, thousands of whom, with the useful and inoffensive Hazarrahs and the natives of Chitraul, are sold into distant and irredeemable bondage''! 63 The pro®t to be made in the slave markets of Bukhara at this time was so great that Murad Beg reportedly required his subjects to pay their taxes in slaves, a demand which exacerbated wars, feuds and slave raiding.
The many thousands of Indian slaves sold in the markets of early modern Central Asia affected that society in many ways. Their contribution to Central Asian civilization cannot be measured solely in terms of their commercial value, which must have been immense, or even their role as skilled craftsmen, architectural engineers, and labourers on the plantation-style estates of Central Asia's great dynastic families. 64 One must also consider the impact these individuals had as they eventually earned, purchased, or were otherwise granted their freedom and became part of the ethnic landscape of Central Asia. In his description of early nineteenth-century Bukhara, Alexander Burnes repeated the popular tradition that``three fourths of the people of Bokhara are of slave extraction'', implying that, although most people were not themselves slaves, they were likely to have had antecedents who were. 65 Burnes attributes this to the great number of Iranian slaves brought to Bukhara and the limited numbers of whom ever returned to Iran. It can safely be assumed that the same could be said of Indian slaves in earlier centuries.
Although the life of a slave was probably often a dif®cult one, slaves in Central Asia were regularly manumitted for any of a number of reasons. Sources such as the Majmu Å«a-iwatha Å»iq and the Khu¢u Å¢-i mamhu Åra demonstrate that it was common for slave owners to stipulate in legal contracts that their slaves would be set free after a certain number of years of indentured servitude, or to include a clause in their wills granting freedom to those slaves who outlive their masters. 66 Perhaps most frequently, slaves were manumitted after they reached a certain age, usually around ®fty. The language of these documents suggests that the slaves were emancipated as a religiously meritorious act. However, it is perhaps not unreasonable to suggest that, as the expense of clothing and feeding slaves began to outweigh the work they could do, less altruistic motives contributed to such manumissions. Finally, it should be noted that, although most manumitted slaves lived out their days in poverty, slaves could, and some did, purchase their freedom, own property, and even become prosperous citizens of their new homeland. 67 It is an underlying thesis here that the escalation of the slave trade during the era of Muslim dominance in South Asia should be understood in the context of the general intensi®cation of Indo-Central Asian trade in this period. The history of slavery in India long predates even the early Ghaznavid invasions. However, in the centuries that followed, ®rst-hand accounts report of tens ± even hundreds ± of thousands of slaves owned by the Turko-Afghan rulers of the north India. While many of these individuals were enslaved as a result of the expansionist efforts of the Delhi Sultans and Mughal emperors, others were forced into slavery to satisfy the tax demands of the state treasury. Still others were motivated to sell their children, or themselves, in an effort to avoid starvation during times of famine or other economic hardship.
The majority of these slaves lived out their lives in South Asia. However, their comparative affordability and availability in large numbers made them an attractive commodity for the international markets of the time. While some Indian slaves in Central Asia are likely to have worked as domestic labourers, those who were skilled engineers were put to work building early modern Central Asia's grand imperial architecture. Still more could be found working in such labour-intensive jobs as manufacturing bricks and textiles, building and maintaining roads, digging irrigation canals, and tending to the crops and herds on the plantation-style farms of Central Asia's great dynastic families. The comparatively advanced nature of Indian craft production most certainly added signi®cantly to the demand for Indian slaves in foreign markets.
The presence, perhaps even dominance, of Hindus in the Central Asian slave markets continued up to the early eighteenth-century decentralization of the Mughal Empire. Although subsequent years did not witness an abrupt end to slavery in India, or even to the forcible relocation of some Indians to Central Asia, the enslavement of large numbers of conquered peoples did come to an end. Furthermore, slaves do not appear to have been included in the tax revenue system of the successor states, or under the British Raj. This resulted in a dramatic decrease in the supply of Indian slaves and their movement to Central Asian markets, which forced the Central Asian slave traders to turn their attention to the conveniently located, and poorly defended Iranians for a viable supply of this commodity.
This discussion has employed a rather broad historical approach in an effort to illustrate a number of larger issues related to the role of Indians in the Central Asian slave trade. Still, many of the questions addressed here would bene®t from further investigation, and even more remain unanswered. For example, no attempt has yet been made to estimate the magnitude of Central Asia's pre-colonial slave population; virtually nothing is known of the movement of foreign slaves from Central Asia to other, distant markets; and beyond our limited sample from late sixteenth-century Samarqand, little has been made known regarding the ethnic proportions of Central Asia's slave population. These are only a few of the many questions yet to be addressed regarding this important and intriguing, yet poorly understood aspect of Asian history.