Resistance at the Edge of Empires: The Archaeology and History of the Bannu basin from 1000 BC to AD 1200

By Cameron A. Petrie

From 1985 to 2001, the collaborative research initiative known as the Bannu Archaeological Project conducted archaeological explorations and excavations in the Bannu region, in what was then the North West Frontier Province (NWFP) of Pakistan, now Khyber-Pakhtunkhwa. This Project involves scholars from the Pakistan Heritage Society, the British Museum, the Institute of Archaeology (UCL), Bryn Mawr College and the University of Cambridge. This is the third in a series of volumes that present the final reports of the exploration and excavations carried out by the Bannu Archaeological Project. This volume presents the first synthesis of the archaeology of the historic periods in the Bannu region, spanning the period when the first large scale empires expanded to the borders of South Asia up until the arrival of Islam in the subcontinent at the end of the first and beginning of the second millennium BC. The Bannu region provides specific insight into early imperialism in South Asia, as throughout this protracted period, it was able to maintain a distinctive regional identity in the face of recurring phases of imperial expansion and integration.

• Language: English
• Publisher: Oxbow Books
• Release date: Dec 28, 2020
• ISBN: 9781785703041

Cameron A. Petrie

Cameron Petrie is the Senior Lecturer in South Asian and Iranian archaeology at the University of Cambridge (UK). He has extensive field and research experience in India, Pakistan and Iran, and has co-directed collaborative research projects in each of these countries.

Book preview

Chapter 1

At the edge of empires: the Bannu basin from 1000 BC to AD 1200

From the mid-first millennium BC to the early second millennium AD, the history of north-west South Asia is marked by the rise, floruit and collapse of a series of major kingdoms and empires. These entities were controlled from centres within South Asia, but also Central Asia, and on the Iranian Plateau, and while often large in geographical terms, the historical records, inscriptions, and coins that survive indicate that they typically comprised collections of smaller states or provinces. These smaller entities were often geographically bounded, and one appears to have been centred on the Bannu basin, which today lies in the south of Khyber Pakhtunkhwa, formerly the North West Frontier Province, Pakistan (Fig. i).

The Bannu basin never takes centre stage throughout this protracted period, but it is either incorporated into or sits at the edge of several empires in a number of guises. For instance, between c.520 and 330 BC, it was the province of Thatagush or Sattagydia under the Achaemenids and it might have been the Banagara mentioned by Ptolemy in the second century AD. Chinese pilgrims to the subcontinent referred to it as Poh-na in the fifth century AD and Fa-la-na in the early sixth century AD, and Islamic historians in the later seventh century AD refer to it explicitly as Banna. The region’s name therefore appears to be ancient.

Part of the Bannu basin’s importance lies in the fact that it is situated strategically across two of the major routes from the alluvial plains of the Indus and Punjab to the highlands areas to the west that comprise modern Afghanistan (e.g. Markham 1879: 47–50; Verma 1978: 100–101; Thomas and Knox 1994; Thomas et al. 2010). Although it appears in the historical record with some regularity, its importance is usually overshadowed by the Peshawar Valley to the north.

The extended period stretching from 1000 BC to AD 1200 was marked by phases of both stability and conflict, which included dramatic shifts in the orientation and nature of political control that impacted on the borderlands of South Asia. Long-range overland trade appeared to continue throughout this period, but there is relatively little known about the nature of the non-trade-based relationships that must have existed between the various centres of royal or imperial control and the regional kingdoms that made up their dominions. This is partly due to the focus of research on the centres of imperial power, and also to the nature of the available textual and historical documents, which are not comprehensive and often originate from those centres. It is therefore perhaps no surprise that our lack of knowledge is most acute when it comes to comprehending the impact of shifts in political control on the lives of the inhabitants of regional areas such as the Bannu basin.

The Bannu Archaeological Project conducted archaeological research in the Bannu region between 1985 and 2001 and has demonstrated that the region’s archaeological heritage stretches back to the Palaeolithic (Khan F. et al. 1991). Most of the research of the Bannu Archaeological Project focused on the early villages of the proto-historic period, with particular attention being given to the period between the late fifth and late third millennium BC (Khan F. et al. 1991, 2010a; Petrie 2010).

This volume looks at a very different part of the chronological spectrum, and presents the first systematic investigation of the archaeology and history of the Bannu basin from the first millennium BC to the early second millennium AD. It does this by combining information from primary and secondary historical sources, assessing numismatic, architectural, and sculptural remains and presenting the results of fieldwork carried out by the Bannu Archaeological Project. This fieldwork included surface surveys conducted between 1985 and 1996, and excavations carried out at the sites of Ter Kala Dheri in 1995 and Akra from 1996 to 2000 (cf. Khan F. et al. 2000a, 2000b, 2000f; Magee et al. 2005a). In interrogating this diverse range of information, this volume contextualises the archaeology and history of the Bannu basin and Akra across this extended period, both in relation to that of the borderlands as a whole and to the surrounding regions. It thus presents a ‘local history’ of this small region during the historic periods.

The period being considered is protracted, but it reflects the span of occupation at Akra as far as it has been determined. In archaeological terms, this period stretches from the appearance of the first truly sizable settlements in the region around 1000 BC, through the period during which the Achaemenids and various other entities exerted political control during the rest of the first millennium BC, and thence on throughout the first millennium AD and up to the reign of the Ghaznavid sultans. It should perhaps be no surprise that the archaeological evidence for the Bannu region does not have consistent resolution across this entire span, and attention is focused on two major chronological periods.

The first is the period during the early to mid-first millennium BC when the urban centre at Akra was established, and there is evidence of the contemporaneous use of a regionally distinct assemblage of material culture. This material is distinctive to the Bannu region, which appears to have been some type of regional polity that was subsequently incorporated into the Achaemenid Empire as the province of Thatagush. Wheeler (1962: 5) argued that the Achaemenid Empire was responsible for bringing iron, order, and civilisation to this part of the subcontinent in the first millennium BC, but it has long been known that this overtly colonialist interpretation was incorrect. The research of the Bannu Archaeological Project plays a critical role in re-evaluating the nature of the regional economy of Bannu in the centuries before the rise of the Achaemenids. It also provides insight into the relationship between the Achaemenid imperial core and its eastern periphery, and the place of Bannu within the complex local socio-economic and political milieu of the mid- to late first millennium BC. Local potentates may have been both empowered local rulers and satraps of the empire who had to negotiate political relationships at local and imperial scales, and also with neighbouring regions that were potentially beyond the reach of the empire.

The second period of focus examines the gradual spread of Islam towards South Asia in the mid- to late first millennium AD, and the impact that this dispersal had upon the predominantly Hindu society and economy that appears to have existed at that time in the borderlands of north-west South Asia. It has traditionally been assumed that the Bannu basin was a part of the Turki and Hindu-Shahi kingdoms from around AD 666 up until the time when it was apparently ravaged comprehensively by Mahmud of Ghazni in AD 1000, during one of his early campaigns into South Asia. Nevertheless, the detailed analysis of archaeological material from Akra and the Bannu basin and the relevant historical evidence that is presented here demonstrates that the process of interaction during the mid- to late first and early second millennia AD was far more nuanced than previously assumed. This volume demonstrates that Akra continued to be occupied up until c. AD 1200 and that a complex regional economy persisted up until at least that point. By providing evidence of the continuity of existing cultural practices into the Ghaznavid period and potentially beyond, it is possible to challenge the prevailing assumptions that the arrival of Islam in South Asia shattered the society and economy of the entire frontier. Rather than destruction, it appears likely that the new Ghaznavid rulers imposed a form of hegemonic control on some of their new domains, which were thus able to go some way towards resisting imperial domination.

This model of hegemonic control during the Ghaznavid period comes from a substantivist approach to economy within empires, where dominated regional populations are viewed as being capable of maintaining their integrity and resisting a cultural takeover. It is argued that this model fits with James C. Scott’s (1990) assertion that all subordinate groups resist in ways similar to peasants, with the existence of everyday resistance showing that the dominated have not consented to dominance. Scott (1990: 2–16; see also Scott 1985: 284–289) differentiates public transcript from hidden transcript, where domination is accepted in public, but questioned in private. A version of this framework can be applied to the investigation of the nature of political control and resistance in the Bannu region across the entire period that is being considered (see Chapters 4 & 11). It is also argued that these concepts provide a useful model for framing discussion of the relationship between rulers and the ruled in the borderlands as a whole. Importantly, the Bannu region lies at the edge of but very much within what Michaud (2010: 188, Figs 1–2, 5, also 2018) refers to as Zomia+, which is essentially van Schendel’s (2002) extension of the concept of Zomia that was formulated in 2007, but not published. Taken together, Zomia and Zomia+ are upland areas that lie between the regions of East, South-east, Central and South Asia and have been occupied by populations that have resisted dominance by large-scale states and empires (van Schendel 2002; Scott 2009; Michaud 2010, 2018).

In addition to considering long-term politico-historical change, this volume explores the specifics of the organisation of the regional economy in Bannu through a detailed analysis of three ceramic assemblages from the site of Akra, two from the first millennium BC and one from the late first and early second millennium AD. In particular our analysis considers what the evidence for ceramic production and distribution contributes to our understanding of the operation of a productive economy, and thereby the regional economy at this time.

1.1. The borderlands of western South Asia and the Bannu basin

The regions that comprise the modern province of Khyber Pakhtunkhwa (Pakistan) have for millennia been a borderlands area. They are part of the geographical transition zone where the plains of the Indus and Punjab and the ridges of the Salt Range that make up the north-western corner of South Asia, meet the far eastern edge of the Iranian Plateau to their west, and the Himalayas to their north (Fig. i). Khyber Pakhtunkhwa is the northern extension of this borderland region, while the southern extension lies within the modern province of Baluchistan. In this location, Khyber Pakhtunkhwa is situated at the major junction between the South, West, and Central Asian landmasses, and is both a geographical and a modern political frontier.

The Bannu basin is a small topographically defined region that lies to the south of the Peshawar Valley, to the north of the Gomal plain, and almost immediately to the west of the Indus River (Fig. 1.1; see notes on terminology). The Bannu basin has been described as being climatically marginal, as it is situated between western winter and eastern summer precipitation zones and does not receive regular rainfall from either regime (Thomas 1986; Thomas et al. 2010b). It is likely that this climate placed specific constraints on the subsistence strategies and social and settlement systems of the prehistoric and historic inhabitants (Thomas 2003; Thomas et al. 2010a). It is notable that throughout the history of settled occupation in the region, there are indications of continuous contact with the highland areas immediately to the west. There was also variable and intermittent contact with Baluchistan, the Gomal, and the Indus Plain to the south and east, and more distant parts of Central Asia to the north-west and west (Thomas and Knox 1994; Thomas et al. 1995; Khan F. et al. 2010b; Magee et al. 2005a).

Figure 1.1. Topographic map showing the location of Khyber Pakhtunkhwa and the Bannu basin in the borderlands of South Asia.

1.2. Previous research on the borderlands in the historic periods

The western borderlands of South Asia have been of interest to Indologists and antiquarians for over 150 years, although much of this attention has focused on the Peshawar Valley (e.g. Cunningham 1871; Marshall 1904). The Bannu basin is considerably less well known, but it was certainly investigated during the same period.

Of all the ancient settlements known in the region, the site of Akra has received the most attention as it is physically the most prominent site, being similar in size to the Bala Hissar at Charsadda (Khan F. et al. 2000a: 9–10). Its prominence attracted the attention of a number of European officers stationed in the district in the late nineteenth and early twentieth centuries, including Sir H. Edwardes (1851) and S.S. Thorburn (1876). H.G. Raverty (1883), Sir M. Aurel Stein (1905), D.H. Gordon (1943: 170), A.H. Dani (1968), and F. Khan (1986) all described the site and/or aspects of its historiography (Khan F. et al. 2000a: 45–52). Akra has long been referred to as the ancient capital of the Bannu basin (e.g. Stein 1905: 10; Khan F. 1986: 184), and is probably the most important historic-period site in the region (Khan F. et al. 2000a: 9).

It is argued that the interpretation of the historic periods in the borderlands generally, and the Bannu region in particular, has suffered in two key areas: first, there has been no coherent reconstruction of long-term political, social, and economic development from the first millennium BC to the early second millennium AD; and secondly, for most of the latter part of that period, there has been minimal attention paid to the archaeological record and its relationship to these factors.

Various attempts have been made to reconstruct later prehistory and early history of the borderlands during the first millennium BC, and there has been a significant shift in academic opinion, particularly over the last fifty years (e.g. Wheeler 1962; Dittman 1984; Magee et al. 2005a; Coningham and Ali 2007; Petrie and Magee 2007, 2012, 2013; Petrie et al. 2008; Magee and Petrie 2010). The annexation of the borderlands by the Achaemenid Empire was initially given a colonialist spin with suggestions that it led to the introduction of iron technology and order to regions that ’were the battlefields of jealous local régimes‘ (e.g. Wheeler 1962: 5). In contrast, it is now argued that the appearance of urban centres and distinctive material cultural assemblages in these regions during the late second and early first millennia BC potentially indicate that regional polities existed throughout the borderlands several centuries before the Achaemenid Empire even existed (e.g. Magee et al. 2005a; Coningham and Ali 2007; Petrie and Magee 2007, 2012, 2013; Petrie et al. 2008; Magee and Petrie 2010). There is no doubt that the Achaemenids gained control of the borderlands, and it was also through this region that invaders and migrants as diverse as Alexander the Great, the Indo-Greeks, the Indo-Scythian, the Indo-Parthians, the Kushans, and the Turks all made their way into the subcontinent. The region also came under the control of the Mauryan Empire as it expanded to the west and into Afghanistan and was a critical bulwark under the Turki- and Hindu-Shahis as they resisted the expansion of various Islamic states. The archaeological and historical evidence for this great sweep of history is, however, patchy.

Attempts to understand and hypothesise about the political and economic dynamics that were in operation during the mid- to late first millennium BC in the borderlands have typically been patched together using inscriptions, distant and often later historical records from the classical world and East Asia, and a rich, enigmatic, and challenging body of numismatic material (e.g. Errington and Curtis 2007). Disappointingly, consideration of other aspects of the archaeological record, particularly ‘ordinary’ objects (cf. Smith M.L. 1999, 2010), has been limited.

Similarly, there have been few attempts made to create coherent models for the social, political, and/ or economic developments that transpired in the borderlands during the first and early second millennia AD (e.g. Dani 1989). When reference has been made to social, political, and economic factors it has again been common for scholars to use textual, epigraphic, or numismatic evidence in isolation, with little or no reference to the archaeology of ordinary people. Furthermore, most of the archaeological investigation that has actually been done on this period was conducted before chronometric dating and other methods of scientific analysis were commonly used in South Asia (e.g. Wheeler 1962; Mughal 1967b; Rahman A. 1968–69). In general, it has been the norm for preconceived conclusions derived from the historical material to shape the interpretation of the archaeology and as a result, archaeologists investigating the historic periods have often used top-down models, and searched for the material correlates of what is implied by the historical record, rather than interpreting the archaeological material on its own merits.

The sources that can be used to inform discussion of the period from the first millennium BC to the early second millennium AD in the Bannu basin are similar to those that relate to the borderlands as a whole. Combinations of historical records, numismatic material, preserved architecture, and chance archaeological finds all provide insight into the occupational history of the region (e.g. Khan F. et al. 2000a). The historical records themselves indicate that from the mid-first millennium BC onwards, the Bannu basin was a part of various kingdoms and empires that were governed from centres as far away as Persepolis, Kandahar, and Bactria, while at other times it was dominated by centres as close as Kapisa, Kabul, Hund, and Ghazni. While it is possible that the Bannu region was self-governing at certain times, it was never the centre of a far-reaching kingdom in its own right. As such, on top of its climatic marginality, the basin could easily be described as being historically marginal and should be quite properly viewed as a peripheral area dominated by a range of external centres.

In addition to the limitations of the available evidence, there have been few systematic attempts to create a proper synthesis that provides an impression of political, social, and economic development and change in the borderlands over this protracted period (e.g. Errington and Curtis 2007). From the extant historical records it is evident that the mid- to late first millennium BC and first millennium AD are notable for major shifts in political control, the use of local kings and governors based in specific regions to rule parts of larger kingdoms, and for the spread of a number of the world’s major religions. The latter is particularly significant as this period includes the proselytisation of Buddhism under Ashoka, its revival under the Kushan kings, the resurgence of Hinduism under various kings including the Guptas, and numerous incursions led by Muslim caliphs, governors, and otherwise independent leaders that resulted in the eastern spread of Islam. The quality of the historical evidence for this extended period is, however, inconsistent and scholars have tended to focus on specific periods of interest; as a result, a coherent socio-political or historical narrative has thus far been elusive. The absence of such a synthesis is also in some measure a reflection of the breadth of the geographical and typological scope of the data that is available, and the fact that evidence that has dramatic significance continues to be discovered on a regular basis (e.g. the Rabatak inscription; Göbl 1999; Cribb 1999; Sims-Williams 1999).

The discussions of economic developments in Khyber Pakhtunkhwa during the first or early second millennium AD that have previously been ventured have similarly been piecemeal. When scholars have discussed the nature of economic organisation, they have typically selected pieces of information from preserved historical sources to comment on the type of control exerted by particular ethnic groups. For example, while there has been an active theoretical debate over the rise and nature of feudal states and administrative systems in South Asia during the first millennium AD (e.g. Sircar 1966; Mukhia 1981, 1985, 1995; Sharma 1985, 1987, 1995; Kulke 1995), this debate has predominantly focused on the northern, central, and southern areas of the sub-continent. The scope of the debate has barely extended further west than Taxila, which lies on the eastern edge of the borderlands. On the rare occasions that scholars have commented on the social, political, and economic organisation of Khyber Pakhtunkhwa proper, they have proposed that there was a gradual move away from a centralised satrapal system of administration towards a more ‘feudalised’ system of vassal kingdoms during the mid-first millennium AD; in other words, there was a regionalisation of what was a formerly centrally administered system (e.g. Dani 1989: 140–141). Expanding on the model of developing ‘feudal’ control, it has also typically been assumed that governors or kings of regional areas came to obtain direct control of their lands and its produce, and were thereby able to enforce a degree of local autonomy, while continuing to show fealty to a higher power (e.g. Biswas 1973: 135–138; Dani 1989: 140). Such assertions are, however, usually supported by equivocal evidence and typically lack archaeological substantiation.

Thus far, no models have been proposed to reconstruct social and political variation throughout the first millennium AD and its effect on the economy of the Bannu basin. There are historical descriptions of the region that indicate that Buddhist monks were living there at various times and that there were numerous Hindu temples there in the early seventh century AD. There are also reports of Muslim incursions into the region as early as AD 644. Our understanding of the status of the Bannu region within the larger social, political, and economic milieu of the frontier and the neighbouring regions of South and Central Asia during this period has thus far, however, been relatively simplistic. The same can be said for our knowledge of the significance of the site of Akra.

There is, however, one historical ‘event’ that has persisted as a virtual constant in the interpretations of the history of Bannu and Akra. For over 100 years, antiquarians and historians have claimed that Mahmud of Ghazni either destroyed or caused the abandonment of the city at Akra during his initial campaigns against the Hindu-Shahi kings in AD 1000–1001, and that this irrevocably disrupted the society and economy of the region (e.g. Khan M.H. 1867: 676; Thorburn 1876: 13; Gazetteer of the Bannu District 1883–84: 26; Khan S.M. 1894: 266; Stein 1905: 8).

In many ways, the status of our knowledge on the archaeology and history of the Bannu region at the end of the first millennium AD is a product of the way in which the region has been investigated, and the nature of the modern political situation. Beginning in the nineteenth century AD, a number of British antiquarians made attempts to reconstruct the region’s history. These men were often employed as local commissioners for the colonial administration and were particularly drawn to the large site of Akra, which appears to have been a regular destination for outings (e.g. Edwardes 1851). It is in their accounts that there are the first references to the destruction of the site by Mahmud, and the subsequent replacement of the Hindu population by Muslim hill tribes. These claims, however, are not based on any known or provenanced historical texts, so they should at best be considered apocryphal. It must also be remembered that these antiquarians were firmly rooted in a Eurocentric/colonialist paradigm, which was not averse to viewing the impact of Islamic invaders as being negative and disruptive. In contrast, more recent studies by Pakistani scholars have used historical records and inscriptions to suggest that in fact the Hindu population of the Bannu region may have been under the control of Muslim rulers from as early as the mid-ninth century AD onwards (e.g. Dani et al. 1964), effectively pushing back the history of Islamic control over parts of Khyber Pakhtunkhwa before the domination by Mahmud. In many ways, the existing reconstructions of the history of Bannu in the late first and early second millennia AD have been relatively opportunistic and oriented towards the support of specific agendas. Archaeological evidence has rarely been utilised in these arguments. Chakrabarti (1997: 53) has pointed out that it will only be when the Eurocentric imperialist and colonialist attitudes that characterise much of the commentary on South Asia are altered that a non-sectarian and multi-lineal perspective of the past in the region can be achieved, and this volume endeavours to attain such a perspective.

1.3. The Bannu Archaeological Project

The Bannu Archaeological Project was involved in the archaeological survey, excavation, and investigation of the Bannu basin between 1985 and 2001, building upon work that was begun by a number of the project members in the 1970s (Khan F. et al. 2010c). During the seventeen years of the project’s operations, investigations were made into the Neolithic, Chalcolithic, Bronze Age, Iron Age, and historic phases of occupation in the basin (Khan F. 1986; Khan F. et al. 1986, 1987, 1988, 1989, 1990, 1991a, 1991b, 2000a, 2000b, 2000c, 2000d, 2000e, 2000f, 2001, 2004a, 2004b, 2010a, 2010b, 2010c, 2010d, 2010e; Thomas 1986, 2003; Thomas and Knox 1994; Morris J.C. and Thomas 2002; Morris J.C. 2004a, 2004b, 2005; Magee et al. 2005a, 2005b; Petrie 2002, 2005a, 2005b; Petrie and Magee 2007, 2012, 2013; Petrie et al. 2008, 2010a, 2010b, 2010c; Petrie 2010; Magee and Petrie 2010; Morris J.C. and Khan F. 2010; Petrie and Morris 2010; Thomas and Cartwright 2010; Thomas et al. 2010a, 2010b).

Surface collections were made at sites throughout the region (e.g. Khan F. 1986; Khan F. et al. 2000a: 81–100), but it was only in 1996 that the Bannu Archaeological Project specifically focused its attentions on the historic periods and the site of Akra. Surveys conducted at the site in 1995, 1996 and 1998 indicated that more than 80 ha of occupation deposits were preserved, making it the largest known archaeological site in the Bannu basin. The size of Akra and its pre-eminence in the basin have been the primary factors behind the claim that it was the capital city of the region throughout the Iron Age and historic periods (e.g. Khan F. et al. 2000a). As this volume will show, the historical records and archaeological evidence that relate to the site and the greater Bannu region have shown that there is great potential for fruitful analysis (Khan F. et al. 2000a: 13–35; Petrie 2002), though much of that potential remains unrealised.

Although formal archaeological excavations only began at Akra in 1996, records exist to indicate that the site has been under fairly constant informal excavation for over 120 years, and this has created difficulties on two levels. First, it has been calculated that at least 7 m of archaeological material have been removed from extensive areas of the site, and entire occupation mounds have been removed in other areas. This degree of illicit digging has meant that there are substantial problems with the preservation of secure archaeological deposits and creates inherent difficulties for reconstructing the complete chronology of occupation at the site. Secondly, although material collected from the surface and possibly through excavations at Akra has been known for many years, it has only been possible to conduct four official excavation seasons at the site, and these were of very short duration. The fact that only a small number of short field seasons were possible has meant that the amount of well-provenanced material is limited. Subsequent field seasons in the Bannu basin initially planned for late 2001 and 2002 were indefinitely postponed due to political instability.

Much of the work of the first three excavation seasons at Akra was directed towards the investigation of stratified deposits that have been dated to the first millennium BC (Magee et al. 2005a), with sporadic attention being paid to other periods of occupation (Khan F. et al. 2000a: 102). The first-millennium BC deposits on the Lohra mound revealed wellstratified occupation spanning much of the first millennium BC, including extensive occupation in the early to mid-part of the millennium, and then a transition marked by the arrival of Achaemenid influence. In 2000, excavations were also conducted on a mound known locally as the Hussaini Boi Ziarat Dherai (or HBZD), and several phases of substantial architectural remains and a sizable corpus of well-stratified ceramic material were revealed. Preliminary analysis of the architecture and ceramics suggested that the three uppermost phases of occupation that were exposed probably dated to the mid- to late first and early second millennia AD (Khan F. et al. 2000c: 115–156). These structures directly overlay deposits dating to the early to mid-first millennium BC, which also included a corpus of stratified ceramic material (Magee et al. 2005a, 2005b).

Although limited, the excavations at Akra provide a unique opportunity to utilise archaeological evidence to address specific issues related to the archaeology, history, and economy of the Bannu basin during the period spanning from the first millennium BC to the early second millennium AD. One PhD thesis and a range of papers have been published making use of this material (Petrie 2002, 2005a, 2005b; Magee et al. 2005a; Petrie and Magee 2007, 2012, 2013; Magee and Petrie 2010), and this volume draws on all of our previous work to present our definitive assessment of life in the Bannu basin across this extended period.

1.4. Approaches and methods

From the outset, it is important to reiterate that although the scope of the volume considers the archaeology from the first millennium BC up to the early second millennium BC, the new archaeological material presented here dates to two specific periods within that frame. Much of the analysis of this material focuses on the reconstruction of aspects of the economics and context of ceramic production and distribution in those two periods. It is not, however, possible to do this effectively without establishing the socio-economic and political context of the Bannu basin in each of these two periods and considering the period in between.

The insights into the two ends of a chronological spectrum are used to assess the likelihood that the Bannu basin was an economically, administratively, and potentially culturally distinct region throughout the first millennium BC and the first millennium AD. In order to gain a proper understanding of the cultural context of the Bannu region at these two end periods, a history of the region during the intervening period is reconstructed by drawing on the full range of relevant historical and archaeological evidence. This historical reconstruction has been coupled with an examination of the region’s administration and economy during the same period and an assessment of the relevance of the existing theoretical debate on the nature of the contemporary administrative and economic structures in South Asia. The archaeological evidence from Akra is then presented and its place within this broad historical context is outlined.

The ceramic material from the stratified deposits at Akra has been analysed with the aim of investigating the organisation and execution of ceramic production and distribution as evidenced by the corpus, and interpreting the significance of this for the understanding of the economy and culture of the Bannu region at this time. Various studies have highlighted the benefits of adopting approaches that utilise materials science techniques to investigate the different stages involved in the production, distribution, and consumption of ceramic material (e.g. Tite 1999; Sillar and Tite 2000; following Lemonnier 1993). These same studies, however, have also emphasised the importance of ecological approaches that take consideration of the overall environmental, historic, economic, socio-political, cultural, ideological, and technological context in order to interpret these processes, the main rationale being that it is not possible fully to comprehend the processes of the production and distribution of cultural material unless the context in which they occurred has been characterised. Such a holistic methodology is in keeping with the ceramic ecology approach (Matson 1965; Kolb 1989; Tite 1999; Sillar and Tite 2000).

There have certainly been criticisms of the limitations of a ceramic ecology approach (e.g. Stark 1993; Roux in Cumberpatch et al. 2001). To attempt to overcome these criticisms: a) environmental, physical, and cultural factors; b) the operational or behavioural sequences (chaînes opératoires) that are involved in the processes of production, distribution, and use; c) the need to identify the properties and performance characteristics of both the raw materials and the finished products, which are likely to affect each of these processes; and d) the interrelations between these elements have been taken into account (Tite 1999, 2001; Sillar and Tite 2000; see Chapter 6.1). It is also recognised that there are inevitably differences in what can be envisioned and what can be achieved in the analysis of variation in material remains (cf. Schiffer and Skibo 1997: 44).

The adoption of such an approach has a number of advantages. First, it encourages the use of varied types of information to aid the process of interpretation, be they geographical, historical, archaeological, or ethnographic. This is particularly suitable for the investigation of the historic periods of South Asia, which commonly offer a range of historical, epigraphic, or numismatic material that can be used to inform discussion. A holistic approach also facilitates the analysis of the interrelationship and interdependence of these different types of evidence. Perhaps the most beneficial aspect is that a holistic approach can be multi-directional, in that while an understanding of the contextual factors is required to inform the interpretation of ceramic production and distribution, the interpretation of these processes provides specific information for the discussion of socio-politics and economy.

The economics and context of ceramic production and distribution in the early to mid-first millennium BC and the later first and early second millennia AD presented here incorporate materials science and statistical techniques to assess the physical evidence of raw materials, technological processes, organisation of production, and nature of pottery distribution evident in the ceramic corpus. These factors are of particular interest because they are specifically related to the local economy and the nature of economic organisation. There is a range of evidence available to assist the reconstruction of the context of the first millennium BC and the first and early second millennia AD in the Bannu basin. However, as might be expected, the level of information relating to each of these factors varies in scale and detail. When it exists, the historical and epigraphic material that relates to the Bannu region is often very specific, but more usually it is highly selective. As will be shown, the history of some periods, dynasties, kingdoms, or events can be reconstructed and discussed in great detail while others can only be reconstructed by implication. In order to achieve some clarity, a number of levels of historical information are used simultaneously in the process of reconstructing the region’s historical context, thereby providing perspectives on the history of the Bannu basin in the surrounding areas in the first millennium BC and the first and early second millennia AD, and facilitating the interpretation of the Bannu region’s place in a broader context.

Rather than there being evidence for comprehensive changes in material culture that coincide with changes in political control, the inhabitants of these regional kingdoms appear to have maintained the use of regionally distinct varieties of locally produced utilitarian material culture. Although it is difficult to be specific about the significance of this evidence, it appears that during this extended period of political change, manifestations of regional identity were maintained in the face of more overarching systems of control and the existence of broader religious traditions. It is argued that the creation and maintenance of locally distinctive types of material culture represents a specific manifestation of both local identity and long-term resilience of both those who held local political authority and the general population.

While it is possible to explore specific relative and absolute correlations with other excavated sites in Khyber Pakhtunkhwa and surrounding areas, it is important to emphasise that there are only a small number of other contemporaneous sites that have actually been excavated, and these have usually been studied in much less detail. Despite the differences in excavation methods used, however, it is still possible to compare the material and cultural contexts of Akra and these other sites.

There have been inherent limitations to the types of analysis that could be undertaken in the Bannu region, particularly with regard to security, which has meant that wide-ranging raw material sourcing was not viable. Nonetheless, the integrated approach to ceramic analysis adopted here incorporates a number of different analytical techniques and data types, including an assessment of the geological information from Bannu and neighbouring regions, and mineralogical and elemental analysis of representative ceramic samples. The results of this materials science analysis are then contextualised according to a comparative typological analysis, and the concordance of this data is used to determine the organisation of production as evidenced by the corpus. The interpretation of this ceramic data is then contextualised by the information for the overall socio-political context outlined in the other sections, which facilitates the investigation of issues relating to the relationship between regional administrative and economic organisations and the impact of distant centres of power.

This study therefore represents the first attempt to contextualise the economics of the production and distribution of historic period ceramic material from the Bannu basin in relation to archaeological, chronological, environmental, geological, historical, ideological, political, social, and technological parameters. The holistic approach that has been adopted does not seek to prescribe explanations for particular phenomena. Rather, it intends to be multi-faceted and multiscalar in its attempt to hypothesise a reconstruction of the economy of the Bannu basin from the first millennium BC to the early second millennium AD.

1.5. Chapter outline

Following this introductory chapter, Chapter 2 presents information on the geography, environment, and historic period settlement of the Bannu basin, including a discussion of the site of Akra and other contemporary sites in the surrounding region. In order to provide a long-term perspective for the socio-political and socio-cultural context as it existed in the basin in the first millennium BC and the first and early second millennia AD, a review of the political and social history of Bannu and the surrounding regions is presented in Chapter 3. The relationship between this socio-political context and the economy is reviewed in Chapter 4 through a discussion of the historical evidence and the existing interpretations of socio-economic organisation and administration in the north-western borderlands. This discussion is then contrasted with the theoretical discussions on the nature of the Early Historic economy in northern, central, and southern South Asia. This review facilitates a reconstruction of the economy of the Bannu region in the first millennium BC and the first and early second millennia AD.

With the environmental, political, social, and economic context of Akra and the Bannu basin having been established, Chapter 5 presents the excavations carried out at Akra and Ter Kala Dheri, including the description of the structural phasing and the determination of the relative chronology of those phases. This information is then contextualised in terms of contemporaneous occupation at other sites in Khyber Pakhtunkhwa, Punjab, and eastern Afghanistan (Fig. 1.2). Chapter 6 assesses the evidence for the relative and absolute chronology of the deposits excavated at Akra there, and Chapter 7 discusses the small finds. In Chapter 8, the theoretical approach and the materials science and statistical analysis of the ceramic fabrics at Akra are presented. These are used in the reconstruction of various aspects of the production and distribution process. Chapter 9 presents a typology of the Iron Age ceramic material from Akra and Ter Kala Dheri. In Chapter 10, a typology of the Middle Historic ceramic material from the Hussaini Boi Ziarat Dherai is accompanied by a discussion of the degree of standardisation evident in the corpus and the evidence of the organisation of production. Chapter 11 synthesises the overarching themes of the volume and draws together the results of the analyses to reconstruct the regional economy during the first millennium BC and the economics and context of ceramic production and distribution in the late first and early second millennia AD. This reconstruction places the organisation of the regional economy in relation to the broader context of fluctuating control from different distant centres of power.

Figure 1.2. Topographic map showing the distribution of major archaeological sites discussed in this volume.

Chapter 2

Environment and settlement in the Bannu basin from 1000 BC to AD 1200

‘The valley has yielded finds of great archaeological value, among them being coins with Greek or pseudo-Greek legends. The Akra mound near the town is reputed to be of great antiquity’ (Bazmee Ansari 1960: 1020).

‘It need not surprise us that there is at present only one locality known within the Bannu oasis where ancient remains are found. This is the site of Akra, some seven miles to the south-south-west of the town and cantonment of Bannu’ (Stein 1905: 7).

A range of geographical and archaeological data gives insight into the Bannu basin between the early first millennium BC and the early second millennium AD. Much of our knowledge about the environment in the region has come through ethnographic observation and the investigation of the later prehistoric periods of occupation, and the most comprehensive presentation of these data has been by Thomas et al. (2010b; see also Thomas 1986, 1999, 2003; Thomas and Knox 1994; Thomas et al. 1997). Many of the conclusions derived from those investigations are of particular relevance to the investigation of the historic periods. The relevant data will be reviewed briefly to provide a broad environmental context for the subsequent description of Akra, the historic period remains there, and the other contemporaneous sites in the region.

2.1. Environment – geography, subsistence, and settlement

The Bannu basin lies immediately to the east of the major plate junction of the north–south wrench fault of western South Asia, which represents a continuation of the Baluchistan Arc and the Kirthar-Sulaiman shear belt (Rendell 1986: 1; see Fig. 1.1). The geographical confines of the basin are delimited by a series of substantial ranges. These include the Sulaiman Range, which forms the highlands to the west and north-west, the Bhittani Range, which separates the basin from the Gomal region to the southwest, the Marwat and Khaisor Ranges, which separate it from the Indus plain and river to the south-east, and the Shingar and Surghar Ranges, which serve the same purpose in the east (Fig. 2.1; Hemphill and Kidwai 1973: pl. 5; Rendell 1986).

The present climate of the basin is semi-arid, and it has a low and unpredictable annual rainfall (Thomas 1986: 13; Rendell 1986: 5; Thomas et al. 2010b). Thomas (1986: 13; Thomas and Knox 1994: 95; Thomas et al. 2010b) has previously emphasised that the basin lies in a climatic tension zone, between the winter rainfall zone to the west and the summer monsoonal zone to the east. It is near the limit of penetration of both of these weather systems and as it often receives little rain from either system, it sees a significant variation in annual precipitation rates (Thomas 1986: 22; Thomas and Knox 1994: 95; Thomas et al. 2010b). Two perennial rivers flow through the basin, the Kurram and the Tochi, which bring water from the hills to the north-west and west. The Kurram River is named in the Rig Veda (2.75) as the Krumu (Stein 1905: 4). In addition to these rivers, torrential precipitation in the catchment area can cause ephemeral discharges via a number of nullahs or streams, which dissipate into the plain (Thomas 1986: 15; see Fig. 2.1). The two major rivers flow some way through the basin before they join together to form the Gambila River (Fig. 2.1). The combined torrent then exits the basin via a breach between the Marwat and Shingar Ranges, which is known as the ‘Kurram Gap’ or the ‘Darra Tang’ (Rendell 1986: 5; Thomas and Knox 1994: 94; after Stein 1905: 10).

Figure 2.1. Topographic map showing geological formations and the location of known ancient settlements, including the site of Akra (after Rendell 1986: Fig. 1.1.2).

A number of major hydrological factors appear to be in operation in the basin (Thomas 1986: 17; Thomas et al. 2010b). The bed of the Kurram River is deeply incised where it clears the Sulaiman foothills, but the river spreads out in the lower basin and frequently floods the surrounding landscape. It has also been observed that seepage from both the Kurram and Tochi river channels increases the moisture in the neighbouring soils, and this is particularly marked in the area between the rivers in the north-west of the basin, which has typically been described using the term doab, literally meaning ‘two rivers’. Evaporation also has a significant impact on this hydrological system such that in the lower doab, near the Kurram-Gambila confluence, the soils are highly saline due to the evaporation and seepage that has occurred as the water crosses the plain. The soils in the south and east of the region are predominantly blown sands and silts. They receive little or no seepage from the ephemeral torrents and lie in an area that is considerably more arid than that to the north and west. In contrast, the northern and western areas are characterised by rich silt/clay soils in the doab, and well-drained gravels in the alluvial fans at the base of the western foothills. Although these soils are fertile, the high evaporation rates in summer mean that agriculture is virtually impossible in many areas without irrigation (Thomas and Knox 1994: 95).

The vegetation of the basin falls within the tropical thorn forest zone as classified by Champion et al. (1966) and Beg (1975) and includes varieties of thorny trees and shrubs (see Thomas 1986: 19; Thomas et al. 2010b). Thomas (1986: 24) has suggested that the vegetation patterns visible in the basin today reflect an extensive history of human exploitation, which has resulted in a lessening of both the number of floral and faunal species, and the distribution of those species (also Thomas and Knox 1994: 95; Thomas et al. 2010b).

Modern population statistics show that the present occupation of the basin is focused in the north-western area. The doab between the perennial Kurram and Tochi rivers provides an ideal locale for settlement, and the factors of seepage and evaporation encourage the placement of settlements close to where the rivers discharge into the plain (Rendell 1986: 11). It is significant that over twenty-five years of archaeological surveys conducted in the basin showed that prehistoric and historic sites are also clustered in the north-west, while they appear to be largely absent from other areas, although this distribution may reflect the intensity of survey across the region (Figs 2.1, 2.10; Rendell 1986: 11; Thomas and Knox 1994: 97; Thomas et al. 2010b: Fig. 3.3).

Due to these similarities in the location of settlement concentrations in the region through time, Thomas (1986: 26) argued that the present-day constraints on subsistence activities provide insight into the circumstances faced by prehistoric inhabitants. This likelihood is emphasised by the fact that although present-day agriculture has been transformed by the significant extension of irrigation schemes, particularly in the doab (Fig. 2.2), subsistence practices continue to be mixed and include the exploitation of a range of domesticated animals that are used for both meat and secondary products (Thomas 1986: 26–27; see also Rendell 1986: 5). The extensive use of irrigation in the north-west doab region has been facilitated by the presence of the perennial rivers, and double cropping of vegetable and other cash crops is practised alongside the intense pastoral activities (Thomas 1986: 26–27). The drier areas to the east and south only permit single cropping and are therefore suited to more extensive pastoral practice, and a more nomadic lifestyle (Thomas 1986: 27).

Despite the more favourable conditions in the north-west of the basin, without irrigation agricultural subsistence in this area is still susceptible to the climate, variation in precipitation, over-exploitation of the landscape, salinity, and flooding (Thomas 1986: 30). Thomas (1986: 30; 2003; Thomas et al. 2010b) has suggested that in the past, strategies aimed at avoiding these hazards are likely to have included the diversification of resource exploitation and landscape use, storage of produce, trade and/or exchange, and an increase of mobility and its relationship to the use of the landscape, including transhumance and nomadism. These strategies may also have been assisted by the proximity of the basin to the western hill frontier, which potentially provided access to summer pasturing, various wood resources, and goods brought by migratory groups (Thomas and Knox 1994: 102; Thomas et al. 2010b). Analysis of the wood charcoals excavated in the region have shown the exploitation of local timbers in conjunction with wood being brought from low and high hill zones, as well as lower mountain zones to the west (Thomas and Knox 1994: 101; Thomas and Cartwright 2010). It is important to note, however, that while these strategies are appropriate for relatively small populations, it is unlikely that they would be universally suitable to support larger ‘urban’ centres.

Although the Bannu basin is separated from its surrounding areas by a series of hill and mountain ranges, contact with these areas is facilitated by a number of passes that have been cut by the perennial and ephemeral rivers and streams (Thomas and Knox 1994: 94). The Kurram and Tochi Passes both lead directly into the Bannu basin from the Sulaiman Range (Figs 2.1, 2.2) and are considered to be two of the five major passes that provide access from Khyber Pakhtunkhwa and Baluchistan into the highlands of Afghanistan; the others being the Khyber to the north, and the Gomal and Bolan to the south (Markham 1879: 47–49; Verma 1978: 100–101; Thomas and Knox 1994: 93).

The Kurram Pass has been described as a twin to the Khyber Pass, as it provides an alternative route to Kabul (Verma 1978: 101–102; see Fig. 2.2). It was used by both Chingiz Khan and Tamurlane in their invasions of South Asia and has been referred to as ‘a much popular route in pre-Mughal times’ (Verma 1978: 101–102, 270; Thomas and Knox 1994: 93). The Kurram Pass also served as the major route between the Bannu basin and Ghazni, and as early as 1878, British analysts listed it as one of the most important passes between India and Afghanistan (Markham 1879: 47–49; Raverty 1883: 317–323). Stein (1905: 7) suggested that the route leading through the upper Kurram Valley must have been of considerable commercial and political importance for communication between Kabul and the central part of the Indus Valley.

The Tochi Pass is known to have been used by Babur on his journey from the Bannu basin to Kabul in AD 1505 (Verma 1978: 104–105; Thomas and Knox 1994: 93), and F. Khan et al. (1992; see also Thomas and Knox 1994: 100; Thomas et al. 2010b) have suggested that it was an important route for trade, exchange, and socio-cultural contact with Central Asia from as early as the fourth millennium BC. The Tochi Pass was also known as an important route between Bannu and Ghazni (Verma 1978: 104–105, 273; see Fig. 2.2).

The other lesser passes formed by the Khaisor and the Shaktu torrents also allow favourable access to the Waziri hills to the west (Thomas and Knox 1994: 93; cf. Markham 1879; Raverty 1878). To the south, the Pezu gap permits relatively easy access to the Gomal plain and Dera Ismail Khan District (Thomas 1986: 30; Thomas and Knox 1994: 94), and to the east, the Darrah Tang provides access between the basin and the Indus River plain (Stein 1905: 10; Rendell 1986; Thomas and Knox 1994: 92, 94).

Figure 2.2. Major communication routes used from the tenth to the fifteenth century AD (after Verma 1978).

Although there has been limited investigation of the Bannu region in the historic period, the material assemblages discovered at prehistoric sites in the basin have suggested that it was far from isolated. Rather, the inhabitants of the region were in continuous contact with the inhabitants of the hill areas to the west, and intermittent contact with areas as diverse as Baluchistan, the Gomal, the Indus, and various parts of Central Asia (e.g. Thomas and Khan F. 1994: 99–100; Khan F. et al. 2010d; Thomas et al. 2010a). This pattern fits with historical evidence that suggests that in the twentieth century, nomadic tribes from Afghanistan, called powindahs, moved into the piedmont zone to avoid the mountainous winter, while settled Afghans once also migrated to the lowlands for employment, with both groups returning in the spring (Thomas and Knox 1994: 96). Furthermore, until recently transhumant pastoralists from Waziristan continued to visit the plain for trade and employment. While it is presently difficult if not impossible for these practices to continue or for settled inhabitants from the basin to conduct these same practices in reverse, it is very likely that the transhumance of settled flocks during the summer occurred in the past (Thomas and Knox 1994: 96; Thomas and Cartwright 2010).

2.2. Akra – the ancient capital of Bannu

The mounds at Akra have long been regarded as comprising the most important of all the historic period sites in the Bannu region (Figs 2.1, 2.12; cf. Chapter 1; Khan F. et al. 2000a: 9). The ancient settlement has been visited and described in some way by various individuals. These visitors have included European officers such as H. Edwardes (1851) and S.S. Thorburn (1876) who were stationed in Bannu district in the late nineteenth and early twentieth centuries AD, antiquarians such as H.G. Raverty (1883), and archaeologists such as Aurel Stein (1905), D.H. Gordon (1943: 170), A.H. Dani (1968), and F. Khan (1986) (see Khan F. et al. 2000a; Petrie 2002). Some of the mounds at Akra are also documented on Survey of India 1" to 1-mile map sheets from the early twentieth century (Fig. 2.3a & 2.3b).

On the basis of geological observations that have been made of the region, it would appear that the mounds of Akra are situated on a series of Younger Terrace Deposits, which date to the Pleistocene (Hemphill and Kidwai 1973: pl. 2; Petrie and Thomas 2012). The flow of the rivers and torrents through the basin incised these terraces during the later Pleistocene and earlier Holocene, and the landscape has subsequently been filled by further sedimentation. Even more recently, the modern watercourses have incised these sedimentary deposits, and the preserved area of Akra now stands as a series of occupation mounds with interspersed agricultural fields, located close to the central part of the Tochi and Kurram doab (Fig. 2.4–2.5).

Figure 2.3a. Section of the Survey of India 1" to 1-mile map from 1908 (Map No. 38 L/9: B2), showing the area of the Bannu basin to the south of Bannu city and with villages and irrigation canals indicated. The mounds of Akra are visible in the centre of the lower part of the image (reproduced with the permission of the University Library, Cambridge).

Figure 2.3b. Close-up of the Survey of India 1" to 1-mile map from 1908 (Map No. 38 L/9: B2), showing the location of the mounds at Akra, which are shown on the eastern side of the Baran nullah (reproduced with the permission of the University Library, Cambridge).

The site is composed of multiple mounds that are divided into two groups of unequal size by a stream (Fig. 2.4), which is known locally as the Lohra nullah, although it is referred to as the Baran Nullah on Survey of India 1"-to-1-mile maps (Figs 2.3a, 2.3b). The nullah is fed by snowmelt, which flows from the hills encircling the basin to the north-north-west. Edwardes (1851, I: 284) observed that a stream that he referred to as the Luhoruh, and which ran between some of the mounds, was very shallow at the time of his visit and must have been formidable in the rainy season, as attested by the steep ravine banks. Stein (1905: 9) reported that at the time of his visit the nullah was dry, indicating that it may have only flowed seasonally in former times (Khan F. et al. 2000a: n. 41). In the recent past, the Baran dam, located in the foothills of Waziristan, has ensured that this stream flows constantly (Khan F. et al. 2000a: n. 41), but it is likely that in the past the Lohra nullah was an ephemeral torrent rather than a reliable water source. The unpredictability of the nullah is, however, partly balanced by the presence of a number of freshwater springs within the mounded area (Stein 1905: 9; Khan F. et al. 2000a: n. 41), which are still used today by the local inhabitants as a source of fresh clean water. In combination with the nowreliable flow of the nullah, these natural springs provide an abundant source of irrigation water for the surrounding villages, and their presence in the immediate area must go some way to explain why this particular location was selected for the placement of a major settlement.

The fossil bed of the nullah is quite broad, being up to 100 m wide at some points (Fig. 2.4), but the modern nullah that flows through this bed is relatively small, being only 10–20 m in width. Stein (1905: 9) posited that the nullah bed might once have been a natural course for the water that is now carried by the Uch Baran (or Kach Kot) canal that passes to the north of Bharrat village, which is situated 1 km to the north-east of the site (Fig. 2.3b). The diversion of the water in modern times may partly explain the width of the Lohra’s fossil bed, but it is also likely that as an ephemeral torrent the nullah would have carried abundant water during periods of heavy rain.

Figure 2.4. Google Earth™ image (31 October 2015) showing the main mounds of Akra and the course of the Lohra nullah.

When the Buddhist pilgrim Xuanzang visited the Bannu region in AD 644 he noted that the chief site of the region was 20 li or 5.36 km in circuit (c.130 ha; Beal 1884, II: 281–282; Chapter 3.4). In 1995, the Bannu Archaeological Project commissioned a theodolite survey of the preserved mounds at Akra, which revealed that at least 31.7 ha of the main mounded area of the site still remain (Khan F. et al. 2000a: 46). In 1996, the theodolite survey was extended, and outlined a further 48 ha of mounded area and material scatters on the right bank of the Lohra nullah to the north of the main mounds, making a total preserved mounded area of c.80 ha (2000a: 46; see Chapter 5). The precise relationship of these peripheral mounds to the main mounded area has not yet been established, but it would appear that if all or even more than one of the mounded areas were occupied simultaneously, Akra would have been sprawling.

Of the two groups of mounds mentioned above, the group on the left bank of the nullah is traditionally described as ‘Akra’ (Fig. 2.3b, 5.4), as it is here that the highest preserved mound is situated. As with many other large sites with occupation dating throughout the historic periods in South Asia, however, the actual limits of occupation extend beyond the main mound (see Chapter 5).

If Xuanzang’s estimate is accurate and it is reasonable to it relate directly to the mounds that are visible today, then there has been a substantial loss of the preserved area of the site in the intervening 1400 years. While this is not the most reliable method for estimating site size and loss, it is notable that the numerous