A New Archaeological Framework for Cultural Changes in Ancient Buddhism

Mingora, the modern-day capital city of Swat or ancient Mengjieli / Wikimedia Commons

By Dr. Anna Filigenzi
Professor of Archaeology
University of Naples “L’Orientale”

Towards a New Reading of Conflicting Sources

The more analytical studies highlight the artistic value, theoretical coherence, and innovative character of the rock sculptures, the more glaring appears the contradiction with literary and archaeological sources. For the period under consideration, these sources unanimously attest to a general impoverishment of the region and a marked deterioration of the ancient Buddhist monuments, which were once so numerous and flourishing.

Until now, the causes and dynamics of this crisis have not been accounted for by either archaeological findings or the scanty literary sources that are still extant. Sparse evidence has been found to support the hypothesis that the country was ravaged by a natural catastrophe, which in turn generated severe political and economic fallout. A concomitant factor in these events can be seen in a shift of the bulk of traffic to alternative routes, while the destructive impact of Hūnạ invaders postulated by previous theories nowadays finds little support among scholars.

We can easily assume that the crisis, whatever its reasons, had made unaffordable the heavy costs of the maintenance of such a complex network of Buddhist settlements. Moreover, the information provided by Xuanzang on mythical Udḍịyāna and the whole of the north-west regions of the Indian subcontinent in general shows a certain falling off in Buddhist zeal, or at any rate in what Xuanzang must have considered as such. Besides, recent archaeological discoveries are disclosing to us a less monolithic view of the religious culture of Udḍịyāna, where alternative panoramas are being highlighted. In turn, these discoveries shed more light on other pieces of evidence that thus far have been isolated and difficult to connect with any coherent framework. As a result, a more diversified cultural history of the country starts being delineated, in which Buddhism, while playing a prominent role, is to be seen from a more pragmatic perspective of an interactive relationship with other religious systems.

Buddhism and Other Religious Cultures

The doctrinal and social face of Buddhism in late antiquity was not just a result of internal developments but also the outcome of the continuous interaction with the vicissitudes of history. Over the course of time, changes were induced by necessary compromises with religious substrata, firmly rooted although not institutionalised, and by a stronger confrontation with Hinduism, a religious phenomenon that showed a certain affinity with Buddhism in that it, too, was organised in one coherent system. In the more strictly Indian territories, Hinduism gained from the confrontation, possibly because in its multifarious divinities it more successfully fused the cultural tradition of Vedic origins with the chthonic cults. Other factors probably came into play, too, including the “nationalist” flavour with which the Guptas were able to imbue Hinduism, contrasting with the favour accorded to Buddhism by foreign invaders, while the notable wealth of many Buddhist foundations may well have generated a certain diffidence, cooling popular enthusiasm.[1]

As for Swat, Buddhist art and architecture seem to have dominated the region for over a thousand years – a very long time indeed, and quite enough to have significant effect on the modes of expression and collective religious imaginings. It may even have sufficed to outlast a decline in consensus and absorb external influences without making it too evident. In the course of this millennium many things must surely have changed in the relations between the Buddhist establishment and environment, but we gain no more than partial insight into the way art reacted to or contributed to these changes. It is a matter of not only objective difficulties deriving from the partial conservation of material elements, but also, and perhaps most importantly, of the impossibility of relating them to their human context. Of the type of society that bore the cost of certain works of art, producing and maintaining them and, to some extent, being conditioned by them, we know only a few characteristic traits, and can hardly begin to gauge the psychological impact of a system of visual communication on a society subjected to far fewer such stimuli than we are today.

The evidence we have of Buddhism with regard to its socio-cultural contexts is mostly partial and disjointed, on top of which there are the notorious lacunae in our knowledge of ancient India. Generally speaking, however, as is so often the case, much can be inferred from visual arts, which show a greater openness towards the outside world than does the canonical literature. At times the need for compromise also involved religious literature, transforming spurious traditions into aetiological myths, as in the case of the legend of Hārītī; in this edifying example of evil force converted to good there survives, refined along ethical lines, the popular devotion to an ambiguous female divinity, as motherly and cruel as nature itself.[2]

Only rarely, however, do the texts account for occurrences of such conceptual syntheses, while iconography takes them in its stride as accomplished facts; it may relegate them to marginal contexts, but still fragmentary evidence remains of forms of symbiosis between Buddhism and other ways of conceiving the sacred. It is not always possible to identify the exact context from which the alien element emerges, and at times it is even more baffling to see why it should be absorbed, as in the case of the six-armed divinity of Butkara I (Faccenna 1962- 1964: II, 3: 107-108; pl. CCCXXXVIa), for which we know of no directly comparable examples.[3]

Whatever the reasons for and degree of effective integration, however, the divine and semi-divine figures that cross the sacred enclosure or come significantly close to it demonstrate that elsewhere, not too far away, there must be places and cults dedicated to them. In fact, while it can plausibly be argued that Buddhism was the major religious phenomenon in the North-West, at least in the first centuries CE, there is no telling how many other cults might have been peaceably practised alongside it, not only in symbiosis or forced relations of subjection but simply as alternatives.

We are also unable to assess the incidence of local religious traditions predating Buddhism, but we may reasonably suppose that they were never completely eradicated or waned away but lingered on in more or less latent forms within the dominant organised religions. In the specific case of Swat, today we can detect only elusive traces of these traditions, sensing them here and there in the context of the religious cultures that supplanted them – Buddhist yesterday, Islamic today. Archaeology has for its part yielded only scant evidence for study of the topic, although, as we will see below, recent studies and discoveries are now outlining new interpretive models. Swat has so far shown no traces of monumental buildings dedicated to organised religious systems other than Buddhism except for the very late period, but this does not mean there were none.[4]

On the other hand, we find ample documentation of the fact that the golden age of Buddhism in Swat did not uproot, for instance, the ancient fertility cults, commonly attested to by female and animal terracotta figurines brought to light by various excavations. Regardless of the faith professed by the household, these objects must have had their own special place in the home as traditional omens of prosperity. Nevertheless, this remains a phenomenon of marginal importance, for fertility cults are so universal that they offer only slender evidence of the religious substratum associated with them. From a purely functional viewpoint, we can draw a parallel with the superstitions widespread in the Western world, in which it is often still possible to discern the survival of ancient pagan cults even though no conscious link with them persists.

Greater significance can be attached to the fact that Swat, or Udḍịyāna, is well known as an ancient land of magic. A factor that probably contributed to this fame is the ever-active ferment of a religious substratum powerfully conditioned by the human awe of the forces of nature, whose caprices and energy could in turn be subdued with the use of magic. Quite probably it was this substratum that acted as the real antagonist of Buddhism in Swat, although, given its obscure nature, there would have been no overt conflict, if any at all, until the Vajrayāna offered it the means to be channelled back into the Buddhist tradition itself (Tucci 1977: 68-69).

The ancient religious world that preceded Buddhism and Islam still survives in the regions of Hindu Kush under the guise of popular beliefs, and from these we can gather that the sacred place par excellence was once the site where the tremendous, inexhaustible force of nature concentrated and manifested itself spontaneously thanks to the chance concourse of favourable elements. The absence of cult buildings coeval with their Buddhist counterparts and alternative to them might be due to their failure to survive because of being built of perishable material, or simply to our incapacity to recognise them as such.[5] We also cannot rule out the possibility that the local non-Buddhist religion preferred spaces for their cult spontaneously created by nature rather than resorting to architecture; these space were then left unchanged or modified imperceptibly. This is the case of the “natural sanctuaries”, whose presence in our archaeological records is growing noticeably both in number and significance. They represent the best evidence so far of rituals and beliefs living apart from the dominant cultures and, at the same time, persisting through time as a sort of underground vein passing through the latter.

Culture of the Valleys and Culture of the Mountains

The interaction between Buddhism and “Kafir-Dardic” tribes

The stunning material evidence represented by the numerous Buddhist foundations scattered over the territory – some of which are of astonishingly imposing scale – conveys a very strong impression of widespread consensus surrounding them and the doctrine that had inspired them. In this respect, however, due attention should be paid to the relationship between propaganda and effective popular response. As rightly pointed out by Callieri (2006), the entrenchment of Buddhism in the North-West of the Indian subcontinent in general and in Swat in particular was preceded by a fairly lengthy phase of great works of propaganda (finding expression in the early Buddhist foundations) supported by the wealthy and influential classes for whom political and intellectual interests coincided.[6] However, in urban settlements artefacts of unmistakable Buddhist character are practically absent from the pre-Kushan archaeological layers. Furthermore, the overwhelming visibility of Buddhist remains, though of undeniable historical relevance, risks putting various socio-cultural realities into the background, realities that nevertheless may have made a noticeable impact on certain philosophical ramifications of key Buddhist doctrines and practices.

In recent years, a number of rock shelters have been documented in Swat in the framework of the “Archaeological Map of Swat Valley” project (AMSV). Their function – certainly multi-faceted – has yet to be fully understood. In most cases they seem to have had some sort of ritual purpose, probably connected to pastoral communities living in the mountains or at the edges of the valleys (Olivieri and Vidale 2006; Olivieri 2011 and 2013, with bibliography). The graffiti and paintings that often decorate these shelters, though distributed over an extremely long period of time, namely from the Bronze Age to approximately the twelft century CE, fit into a coherent system. In relation to the dominant artistic culture of the valleys, this system evidently represents a parallel visual code, with its own syntax and social interactions, as well as a stock of knowledge, rituals, and religious and mystic beliefs.

Recurring patterns in rock paintings, in particular dominant anthropomorphic figures with outstretched fingers, have been associated with magical or shamanic practices of trance or ritual death. Already appearing in the Bronze Age, these pictograms survive in later paintings with only slight modifications (Olivieri 2010b: 21; Id. 2013: 66-67). Furthermore, most of the painted shelters are not meant for standing visitors – the spaces are so small that a person can barely sit inside – and so seem to have served the purpose of hermitage and/or some sort of initiation rituals (Olivieri and Vidale 2006: 125; Olivieri 2011: 142). It is interesting to note that, in a way not dissimilar to the Buddhist rock sculptures of late antiquity, the shelters generally display a strong sense of geomantic harmony, which translates into a symbolic interpretation of the rocks and the surrounding landscape. Occasionally they may have had anthropomorphic or zoomorphic shapes, thus appearing to be “gigantic images sculpted by nature” (Olivieri and Vidale 2006: 141; figs. 37, 45; Olivieri 2011: 142; pls. 6-7; Id. 2013: 132 ff.) (Fig..4). Evidence of winemaking in historical times, in the form of wine presses, adds a further element of cultural identity to the archaeological record connected to the users of the shelters. Just for the sake of convenience, we can call these people “Kafir-Dardic”, a generic name that nonetheless is, for the time being, the most pertinent we have.[7]

Fig. 4 – Ghirai: Gigantic rock with anthropomorphic shape (courtesy IsIAO; photo by L.M. Olivieri)

Buddhist symbols can occasionally be found in the rock paintings and graffiti (Fig. 5). According to Olivieri (2011: esp. 138-141), they might be less a reflection of direct Buddhist inspiration than an indication of how these “Kafir-Dardic” tribes, living at the edge of urban areas and their rich agricultural belt, saw Buddhism. These mountain people must have had some sort of economic ties to the Buddhist settlements, probably supplying them with forest products.[8] Thus, while not integrated into the dominant culture, they may have regarded the Buddhist establishment as a source of welfare and accordingly recorded it as part of their symbolic visual narrative. Nevertheless, another hypothesis (not necessarily in opposition to the first) is that the painted shelters were used by various people of different creeds (ibid.: 139).

Fig. 5 – Kafir-kot 2: Rock paintings with Buddhist symbols (after Olivieri and Vidale 2006: fig. 38)

In any case, the contiguity of the two worlds appears, thanks to the recent archaeological surveys and studies, to have entailed a meaningful and effective interaction. This adds credence to Tucci’s sharp intuitive ideas about the role of aboriginal beliefs and praxis in the development of the late forms of Buddhism (Tucci 1977: 68-69).

Whether or not the graffiti and paintings depicting Buddhist symbols in rock shelters may be read as specific markers, the lack of such evidence is nevertheless a mere and inconclusive argumentum e silentio. We cannot rule out, in fact, the possibility that at least some of these shelters may have been used as retreats by Buddhist hermits who had totally or partly renounced cloistered monastic life.[9] Close contact with the “parallel” culture of the mountain people may have conferred special features on a well- rooted Buddhist tradition of secluded meditative retreat, in particular aspects of an asceticism closely connected to nature, magic and the supernatural.

The Buddhist practice of meditation, implicitly calling for remoteness and solitude, is actually better known from textual sources than material contexts. However, in recent years archaeological and art historical research has started assembling sparse pieces of material evidence, and cross-case analysis has now enabled the comparison of a few particular instances. While still only small in number, they are extremely interesting. In Swat, isolated retreats have been documented uphill from big Buddhist settlements. They consist either of monastic cells (as in the case of Abba-saheb-china, Tokar-dara, Nawagai, Saidu Sharif I; on this latter see Callieri 1989: 47, figs. 56-57) or of rock hermitages. Among these latter, worthy of special note are the sites of Amluk-dara in the upper Kandak Valley (Olivieri 2010a: 358; fig. 3), Topialai near the Cherat Pass (Olivieri and Vidale 2006: 147), Qal‘a and Tangu near the Ambela Pass (Nasim Khan 2000), and the “caves” of Pānr Ị , cut into the natural bank of compact clay.[10] Moreover, in places as distant as Hadḍ ạ̄ in Afghanistan, Kara Tepe in Uzbekistan and Qizil in Xinjiang, around the fourth/fifth century CE secluded places existed within the monastic settlements that were devoted to meditation and more precisely – as the painted decoration suggests – to meditation on death.[11]

On the other hand, many other sites that are not so specific and therefore difficult to date − such as completely bare caves near monastic settlements (Fussman 2008: passim [see index, s.v. “Grottes”]; Olivieri 2010a: 358, fig. 3; Callieri 2012) − speak of a quite widespread practice whose exact relationship to institutional monasticism we do not know.

This phenomenon is still poorly represented in the archaeological record and it is certainly underestimated. In fact, if we understood it better, we would probably change our perception of the historical dynamics of Buddhism. For the time being, the still meagre evidence is sufficient to shed new light on the origins and diffusion of the practices of Buddhist hermitage and asceticism, so extensively documented in the Himalayan Buddhism from its inception up to our days.

Buddhism and Hinduism: Archaeological Evidence from the Śāhi Period

Bearing noticeable relation to the multicultural environment of late antique Udḍịyāna is also the Hindu temple, probably devoted to a Vaishnava cult, recently discovered by the Italian Archaeological Mission on the top of the hill of Bīr-kot-̣ghwanḍại (Barikot) (Fig. 6). The life of the temple covers the entire span of the Śāhi period, from the seventh to eighth centuries (Early Śāhi or Turki Śāhi) to the ninth and tenth centuries (Late Śāhi or Hindu Śāhi).[12] Thus the monument, besides enhancing our scant knowledge of the political and cultural history of Swat in general and of the Śāhi period in particular, for the first time adds the concreteness of material evidence to the vague reference made by Xuanzang to ten temples of devas and attests to a well organised presence of Brahmanical cults. Moreover, it tells us that the flourishing of the Buddhist rock sculptures at least partly overlaps the “official” settlement of Hindu cults.

Fig. 6 – The Hindu temple at Bīr-kot-̣ghwanḍại (Barikot) (courtesy MAI; photo by Luca Colliva)


Figs. 7-8 – Sculptural fragments from the Hindu temple at Bīr-kot-̣ghwanḍại (Barikot) (courtesy MAI; photo by P. Callieri)

The temple, rectangular in shape and with the main entrance to the east, approached by a monumental flight of steps, was probably dedicated to Visṇụ , as suggested by some sculptural marble fragments. The most significant piece is represented by a beautiful female figure, evidently the attendant of a larger central figure, to which the fragment of a leg larger than the natural size is most probably to be attributed (Figs. 7-8). By analogy with a well-known repertory it can readily be inferred that we are dealing here with an image of Visṇụ accompanied by his personified attributes, Gadādevī (i.e. our female figure) and Cakrapurusạ .[13] On the evidence of the few fragments recovered we cannot detect the original iconography, whether a single headed or many-headed Visṇụ . Nevertheless, among the number of stucco fi nds belonging to the architectural decoration of the temple, many fragments of animal masks deserve to be mentioned. They depict either lions or another kind of animal that, on the basis of the surviving fragments, we tentatively identifi ed as boars (Fig. 9). These bipolar animal motifs are strongly reminiscent of the Caturānana Visṇụ, whose iconographical depiction is characterised by the co-presence of human and animal features expressed by his four heads, with benign, lion, boar, and wrathful aspects respectively (Fig. 10).

Fig. 9 – Samples from the architectural decoration of the Hindu temple at Bīr-kot-̣ghwanḍại (Barikot) (courtesy MAI; drawing by F. Martore)

The construction of this imposing, eye-catching building, which once dominated an astonishingly vast horizon, must have required the support of the ruling class, the only one capable of affording such a formidable and expensive enterprise. Moreover, a further signifi cant piece of evidence is represented by the remains of a monumental platform fi rst documented by A. Foucher (1901: 167, fi g. 31) and then by Farooq Swati et al. (2002) at Hathi-dara, in the Zalam-kot Valley. Signifi cant similarities to the Bīr-kot-̣ghwanḍại temple (Olivieri and Vidale 2006: 119-120; Filigenzi 2010c: 413; Olivieri 2010a: 359) allows us to conjecture a similar function.

If we can draw any conclusion from the data at our disposal, the impression is that, as far as the religious sphere was concerned, the Śāhi rulers mostly employed their resources in disseminating or supporting the Hindu cults (Olivieri 2003: 31 ff., esp. 40; Filigenzi 2010c). The physical trajectories of competitive religious systems, well organised and politically supported, can be better evaluated when compared with other geo-cultural data, chiefly the line of military outposts created or, at least in some cases, enhanced by the Śāhis. This system, whose major lines seem to run between the Kunar Valley, in Afghanistan (Ball 1982), and Swat, was apparently based on a capillary control of the territory by means of a series of forts and watchtowers, situated at high places and crests and characterised by their imposing height (up to 10 m) and circular buttresses. Currently assigned to the Hindu Śāhi period on the basis of surface finds, they might nonetheless have had an earlier phase, although in view of the lack of any systematic research this remains mere speculation (Olivieri 2003: 31 ff.; esp. 40).

Significantly, the mapping of the rock sculptures and the military architecture chart two different geographical ranges: while the Buddhist sculptures run north-east from the Barikot area, the Śāhi forts stretch out from the Swat-Panjkora area to Buner, along the main transit roads in this crucial zone connecting the limes of Kunar in Afghanistan with Kashmir and the Mardan plain (Olivieri 2003: 42; figs. 45, 53). Thus, what archaeology is revealing about Swat’s topography in late antiquity seems to match Tibetan and Chinese sources of the eighth century, which describe Udḍịyāna as split into two parts, one dependent on the Laghman area, the other ruled by the king Indrabhūti (Tucci 1958: 324, fn. 1). Whether real or legendary, the king Indrabhūti epitomises the existence of a political entity with a defined ideological profile, where Buddhism evidently still played a relevant role. The concentration of the Buddhist rock reliefs in the north-eastern portion of Swat adds to the literary records the support of a material evidence that, though still vague and isolated, is reinforced by a symmetrical opposition: in the area around Barikot, where the concentration of Buddhist rock reliefs decreases, alien religious expressions seem to take over.

As for the Buddhist community, it probably did not receive from the Śāhi rulers any significant support (at least from the economic viewpoint), as the poor conditions of its settlements witness, but it also never suffered from them any prohibition or restriction of its activities.

The Crisis of Udḍịyāna: Facts and Fancy

In Swat, as anywhere else, reappraisal is now due of the opinion widely held among the scholars of past generations that the decline of Buddhism can be directly imputed to the invasion of the Hunnic people and their absolute lust for destruction.[14]

The most telling negative evidence in this respect we owe to Songyun, who arrived in India around 520 but made no mention of wreck and ruin. At the time of his arrival, according to the Luoyang Qielanji (A Record of Buddhist Monasteries in Luoyang), Gandhāra was ruled by the second or third Hephthalite sovereign, who was said to have a cruel disposition and often apt to massacre. The text, however, makes no mention of religious persecution of the Buddhists, although the king did not subscribe to the faith, instead showing zealous devotion to the gods of his own people. Nevertheless, just how little his personal faith affected the destiny of Buddhism can be seen in the enthusiastic description Songyun gave of the city of Shahbaz Garhi, its inhabitants and, indeed, its Buddhist foundation. Equally positive is the picture he painted of Udḍịyāna, whose king is actually described as a fervent Buddhist (Beal 1958 [1884]: xciii; Kuwayama 1989: 92-94).[15]

The very different picture that emerges from the description of Xuanzang, who records a general decline in number and wealth of the Buddhist foundations of the North-West in the first half of the seventh century, poses
challenging questions about the fate of Udḍịyāna Buddhism and its capacity to nurture significant doctrinaire
and artistic developments.

From the mere comparison of the two accounts we can only conclude that the span of time elapsing between the visits of Songyun and Xuanzang saw a significant reversal that changed the face of the region, possibly caused by some natural calamity. Being highly seismically active, the areas of Northern Pakistan have experienced many disastrous earthquakes and floods during historical times.[16] Relevant archaeological evidence comes from the urban settlement of Bīr-kot-̣ghwanḍại, still under excavation, where Period IX marks a severe decay of the lower settlement after episodes of collapsing and floods (Callieri 2010: esp. 375-377; Olivieri forthcoming).[17] Repetitions of large-scale seismic events might have worsened the situation and further reduced, or re-shaped, the intertwined networks of Buddhist foundations and urban settlements.

However, Songyun also tells us of a war that the king of Gandhāra reportedly waged for three years against Kashmir, probably not so much for purely territorial ambitions as to gain control over trade with the region of the Salt Range. The conflict with Kashmir is directly pointed to as the cause of the general discontent arising from the heavy taxes levied on the subject populations (Beal 1958 [1884]: 93; 95-97). It is in fact quite possible that the state of war with Kashmir (probably periodic and in any case always latent) also involved the bordering regions economically and politically, aggravating their progressive impoverishment. It is when the Hunnic domination came to an end (and not when it started), around the mid-sixth century, that the climate of instability and weakness – both economic and political – probably intensified in regions like Swat, which were never wholly integrated into a centralised state structure. Ruled by local chieftains on the basis of a “feudal” kind of system, they constituted small vassal states that held on to a sort of independence paying tributes to one or another of the foreign invaders. Although this particular system inevitably meant chronic weakness from the political point of view as well as weighing heavily on the economy, it did, as Tucci (1958: 282) rightly pointed out, allay the more serious hazards of violent, destructive invasions.

Again Xuanzang and the Archaeological Portrait of His Time

Although possibly hyperbolical, the statement of Xuanzang about the state of neglect and ruin which so many Buddhist foundations had fallen into suggests that there must have been a quantity of decaying monuments and monasteries dotting the landscape in the seventh century. This fact actually finds confirmation in the archaeological record. As discussed above, in comparison with the earlier centuries, the number had certainly been much reduced.

The natural disasters and economic distress that seem to have severely affected the network of Buddhist settlements in Swat must have led to efforts and resources being concentrated on a small number of sacred areas that, for one reason or another, were considered of particular significance. We know, for instance, that important sacred areas such as Saidu Sharif I were never rebuilt, while we have signs of activity (although at different times) in several others, like Butkara I (Faccenna 1980-1981: Part 1, 11 ff., Part 3, 635), Shnaisha (Qamar and Ashraf Khan 1991; Abdur Rahman 1993), Malam-jabba (Ashraf Khan 1993: 40-44; Rafiullah Khan 2011), and, to a more limited extent, Pānr Ị (Faccenna, Khan and Nadiem 1993: 130), to mention just the best-known examples. Other foundations probably also went through some sort of restoration, as indicated by additional – although less circumstantial – archaeological evidence.

Of special relevance to the understanding of the historical background of the rock sculpture, and particularly with regard to the chronological concurrence between the rock sculpture and refurbishment of ancient sacred areas, are the data provided by the excavation of Shnaisha. The fortuitous position of this religious foundation – open to the north, towards the valley, and protected to the south by the Tharkana heights – together with the grandeur of the main stūpa, the quality of the sculptural decoration and the presence of numerous other remains in the immediate vicinity all point to a complex that must once have enjoyed exceptional prestige.

Striking in this context is an exterior addition to the body of the main stūpa made at a later date – a simple chapel abutting on the south-east corner built with perfunctory technique using mostly recycled material from the site itself (Qamar and Ashraf Khan 1991: 185; Abdur Rahman 1993: 15, 16, 20 and pl. VIIa). The chapel housed a stela (Qamar and Ashraf Khan 1991: 18 and pl. 13; Abdur Rahman 1993: 20, 22 and pls. XXa, XXVIIb; here, S179; II: Figs. 73a,b; GS 26) which, in terms of typology and style, clearly belongs to the rock sculptures class.[18] The concise and occasionally contradictory preliminary excavation reports do not allow any precise assessment of the find. It would be useful to ascertain whether the chapel was the place where the stela was originally installed; the fact that the missing pieces have not been found raises some doubts on this score. In particular, the stela, broken off in the lower part (the figure is apodal), rested on a base that was found in the collapsed material, but the authors give only the barest details of it. Qamar and Ashraf Khan (1991: 185) offer a brief description of it that does not, however, tally with the photo later published by Abdur Rahman (1993: XXVIIIa). Only close examination could indicate whether the base is to be considered an integral part of the sculpture, or whether – as seems equally likely – it was a pastiche combining a late piece (the stela) with an earlier one (the base) retrieved from discarded materials at the site (Taddei 1998: 178-179).

Be that as it may, the fact remains that the stela – whether specially made for the place where it was found or transported there from elsewhere – was utilised in the limited restoration of a monument already in a state of abandonment or serious decay. The prominent position of the sculpture in a newly built chapel makes the matter all the more significant, especially considering that the renovation seems to have been made somewhat clumsily and hastily, with the help of reused sculptural material.[19]

This remains to date the only case known to us of a sculpture of this type introduced into a newly built structure in a pre-existing religious monument. Nevertheless, a different but still significant piece of evidence is offered by Butkara I, where a stela representing Padmapānị (C27; II: Figs. 28a,b) was found in the immediate vicinity of the sacred enclosure. Although a clear stratigraphic relationship is lacking here, we can dare to conjecture that also in this case there must have been a significant relationship between the rock sculpture and the last refurbishment of the sacred area.[20] Moreover, settings of the sort featured by these two sites may have been more frequent than archaeological research could ascertain to date and are still waiting to be brought to light.

Changing Patterns: Some Indications from the Archaeological Record

The image of decay portrayed by Xuanzang’s description of Udḍịyāna also conveys a notion of cultural stagnation, if not regression, a fact that we usually read literally and take for granted without questioning the possible cultural bias of the witness. In fact, we cannot help suspecting that his judgement derived from some diffidence towards the emerging Vajrayāna, of which Swat represented one of the four most prestigious centres (Lévi 1915: 105; Tucci 1958: 280). For instance, vague hints by Songyun about Udḍịyāna’s fame for magic spells (Beal 1958 [1884]: LXXXIX, XCVIII) are reinforced by Xuanzang, who was dismayed by Udḍịyāna monks no longer being able to grasp the true message of the Buddha and (as we infer from the context) dabbling in magic. Xuanzang makes a point here of mentioning those who forbid it (ibid.: 120-121). It can be assumed that such statements correspond to the true state of affairs, but they could also be a clue that a different form of Buddhism – namely a proto-Vajrayāna – had evolved in the region. Evidence in support of the latter hypothesis is provided by the rock sculptures, which, in addition to the general characteristics discussed above, conform to a new iconographic lexicon in which we can detect the formative stage of a vajrayanic – or proto-vajrayanic – orientation of the doctrine.

The above-mentioned considerations lead to the question of whether, or how, the complex and sophisticated aesthetics of the rock sculptures – indeed the only consistent evidence against cultural regression – might help in better interpreting the somewhat disjointed picture of their historical framework. One wonders whether, having in mind the earlier “Gandharan” patterns of visual art and architecture, we fail to recognise in the objectively scanty and controversial archaeological record a watershed process that accomplished the transition from old to new models and media.

Excavations at Bīr-kot-̣ghwanḍai, the ancient Bazira of the classical sources, have moreover provided us with new pieces of information that invite us to reassess already collected archaeological records with a new critical approach. In trench BKG2, in an area partly occupied by buildings of uncertain nature (religious?),[21] the layers corresponding to the Hindu Śāhi Period display a noticeable change in architectural techniques and materials. The large-scale use of brightly painted, thick clay coats has been attested not only by remains still in situ, but even more by plentiful traces in the archaeological layers they melted into. In some cases, these coats of clay evened out the surfaces of rough masonry. This was in sharp contrast to the accurate, clean-cut stone masonry that, though still in use, seems either to represent the survival of earlier traditions observed in the lower settlement prior to the crisis of Period IX, or else was distinct for some functional reason.[22]

A transition towards decreasing the use of cut stone and increasing the use of clay and/or stucco was also observed in sacred Buddhist areas. We still have a blurred picture, however, of the causes, effects, and time of this process. Although it is a matter of record that stucco and clay are the predominant media in “late-Gandharan” Buddhist sculpture, we still know all too little about their coeval architectural settings. The new trends in sculpture and architecture pose challenges to archaeological interpretation, since their real magnitude and impact are extremely difficult to verify. In most cases, changes are only insufficiently documented by partial additions that overlap existing layouts. Moreover, as the durability of structures depends on the durability of their materials old installations in stone have often survived where later additions made of more short-lived materials have almost disappeared.

Without regular maintenance clay, stucco and wood decay rapidly. Therefore, one must consider that very little of any late additions and repairs made with such materials could survive centuries of total abandonment in a country that long ago acquired a different religious identity. It is clear that even the most careful investigation will not be able to fill all the gaps completely. Hasty excavations, often carried out in the framework of rescue archaeology, further aggravate this problem. Under such circumstances, traces of ephemeral buildings that may have succeeded the solid masonry of earlier times can easily be unnoticed or wrongly interpreted. This can be seen in the case of the rescue excavation of the sacred area of Nawagai, where the remains of the latest phase were only summarily recorded, and attributed to “non-believers” since they were not in keeping with the traditional “Gandharan” standards (Qamar 2004: esp. 185).[23]

A further exemplifying case is represented by the Buddhist site of Safi Abad, in the Mardan District, which, after being heavily looted and damaged by illegal diggings, was submitted to a rescue excavation in the early 1990s (ul Wahab 2012). Though insufficient, the documentation published by the Archaeological Department allows us to hypothesise that the site was renovated or possibly founded – which would be even more significant – in the late- or post-Kushan period. If excavated and documented under better circumstances, the site would probably have provided a precious instant picture of that extremely interesting phase of co-existence of stone and clay/stucco which preceded a large-scale transition from the former to the latter. Hence, it seems that it is not only a matter of objective difficulties but also of modus operandi. A change of cultural mentality is advisable, which may induce us, archaeologists and art historians, not to overemphasise\ the “classical” Gandharan art and architecture at the expenses of its still unknown – and prejudicially under-evaluated – aftermath.

Given the paucity of accurate archaeological records, the data provided by the excavation of the sacred area of Butkara I, identified by Tucci (1978: 60-61) with the splendid Talo/Tolo visited by Songyun and the Dhumat‘ala of the Tibetan pilgrim Orgyanpa (Id.: 1971: 369 ff., 396-399), is particularly valuable (Fig. 11). The long archaeological sequence, stretching from the third century BCE to the tenth/eleventh century CE, and carefully recorded by Domenico Faccenna, allows a focused investigation into both the continuity and the changes in sculptural and architectural patterns.[24]

Fig. 11 – Butkara I (courtesy MAI)

At Butkara I, five main building periods have been detected, corresponding to the construction and four successive reconstructions of the main stūpa (GSt 1-5; Fig. 12). A shift towards plastic materials and related techniques can be observed on a large scale during the period of GSt 4 (end of third/early fourth century-seventh century CE), especially in Phase 5 of it (fifth-seventh century; Faccenna 1980-1981: Part 1, 77-120; Part 3, 649-64, 676-693). Of particular interest is a chronological clue provided by a secondary deposit of coins in one of the niches of GSt 4, which accompanied the re-positioning and restoration of a relief panel sealing the niche that had probably fallen down. This episode, which occurred in Phase 4, took place in the framework of extensive building and restoration activity after widespread collapsing and damage, most probably caused by an earthquake (Faccenna 1980-1981 Part 3, 635). The terminus post quem for Phase 4 is offered by the latest coins in this deposit, which were issued by Kavād 1 and dated by Göbl to about 356/360 CE.[25] Thus, the largescale renewal of the site reveals a departure from earlier artistic traditions, and thanks to a reliable archaeological sequence we can place this event within a specific time span.

Fig. 12 – Butkara I: Isometrical view showing the five periods of the Great Stūpa (after Faccenna 1980-1981: III.1, pl. V)

Indeed, the archaeological evidence shows that in both Butkara I and Bīr-kot-̣ghwanḍại, the increasing use of media such as clay and most probably wood can be connected with periods of economic distress, which might well have obliged cheaper building options using low-cost materials and processing techniques. Nevertheless, it must also be taken into consideration that there may have been other triggering or concomitant factors, such as the spreading of new ideas and tastes radiating from Afghanistan, Southern Central Asia and Xinjiang, where the large-scale use of clay and wood in architecture, architectural decoration and sculpture is a well-rooted tradition.

Certainly the more direct connection of Udḍịyāna to Afghanistan, prompted by the changes in the geo-political scenario and the entering of Swat (or at least a part of it) first into the orbit of the Kidarites, and then into that of the Śāhi of Kabul (Filigenzi 2010c: 409 ff.), must have contributed to an easier flow of ideas, models and techniques. Thus, we can infer from the available evidence that the trend of artistic and architectural transformations we detect at various times in late antique Udḍịyāna is to be regarded as a multifarious phenomenon, one not only of an economic nature but also (and possibly even more) of cultural relevance. In particular with relation to Buddhist art, one must recognise that in the above-mentioned areas strong and captivating artistic forms developed precisely because of the malleable materials, such as clay, being used. The Central Asiatic artistic expression based on coroplastics contributed significantly to a new aesthetics in Buddhist art, with gigantism, pathos and polychromy as distinctive features. On the other hand, this new aesthetic is evidently patterned on important developments in Buddhist heuristics, gnoseology and praxis. The balanced match of forms and contents gave birth to a new and widely-shared indexicality, of which we only know disconnected parts. A broader reflection on cultural contexts is indeed an unavoidable challenge that cannot be answered by excessive localism of studies, or the one-to-one relationship between iconographic lemma and specific literary references.

Nevertheless, for the time being steps in this direction must be taken from working hypotheses. As for late antique Udḍịyāna, one of these concerns the aesthetic dimension of the Buddhist monasteries. If, as archaeology shows, the old architecture and sculpture in stone was largely replaced by other, more perishable materials, comparative diachronic analysis lets us conjecture that the appearance of the new buildings, with either civil or religious functions, must have closely resembled the appearance of buildings in rural Pakistan still being built today, or more closely parallel the specifically Buddhist architecture we know in the Himalayan countries.[26] However, with respect to the latter, the construction in late antique Udḍịyāna was probably on a much smaller scale since, as mentioned above, the new trends were mostly reflected in minor additions to pre-existing monuments.

On the other hand, at the synchronic level we can only investigate other incomplete archaeological evidence. The closest comparisons might be offered by Afghanistan, provided we are able to get fresh data. If we do, we will also have a better chance of retrieving overlooked details from earlier documentation. Encouraging clues are being provided by recent discoveries in Mes Aynak.[27] Among the many exceptional finds from this site, special mention should be made of wooden architectural elements, decorated in relief that still preserves abundant traces of a polychrome painted finish (Fig. 13). Wood is an essential component in clay architecture, but generally we can only guess about its use in ancient times. In addition to customary practices of re-use, the preservation of wooden artefacts in archaeological contexts is hampered by their perishable nature itself. As an organic material, wood normally decays under combined biological and chemical degradation when buried in earth. Only in desert sites, such as in Xinjiang, does the extremely dry climate allow long-term preservation. In other areas – including Swat – we have little more than impressions or negligible fragments. However, thanks to these crucial finds from Mes Aynak we can now reduce the distance between reality and hypotheses. Thus, despite the negative or incomplete evidence, we can try to imagine what the surviving monasteries – their number is difficult to estimate – of late antique Udḍịyāna looked like, with their smooth lines of clay, the bright colours of the surfaces, and the visually powerful interplay of architectural elements made of wood.

Fig. 13 – Wooden architectural elements from Mes Aynak (courtesy DAFA)

But if for the monasteries this was a time of change and not only of mere regression, less static interpretations of Buddhist monasticism might be explored as well. One must wonder whether and how the monastic system itself was affected by internal changes. Dissenting or alternative views may have led some of the Buddhist intelligentsia to different practices or a hermit way of life away from the monasteries, thus laying the groundwork for the siddha path.[28] Once again, how far this social phenomenon can be traced through the archaeological record depends on the receptiveness of our investigative tools.

Rock Sculptures, Ritual Practices, and Pilgrimage Paths

As a new picture emerges of late antique Udḍịyāna, one can no longer accept the idea that the Buddhist rock art was simply an economical way to mark out ruined sacred places and restore them to life. Too close and sophisticated is the relationship between sculptures and the local topography – a relationship that is not confined to the physical proximity to holy monuments, but rather encompasses the landscape around the sculptures as well as the singular tie between the images and their material support. The illusion of the images being svayaṃbhū created by the artists is a way to express in fully manifest forms the strong geomantic character of the landscape and re-establish the earlier sacred topography through a sort of “revelation” or “rediscovery” of its immanent sacredness. Hence, one must wonder whether the underlying concept of these sculptures represents a mere – albeit insightful − interpretation of the svayaṃbhū, or rather a conscious and congruent visual counterpart to doctrinal issues connected with the “revealed” nature of tantric texts.[29] Such a conceptual coincidence appears even more likely – and significant as well – in the light of the particular role played by Udḍịyāna in the genesis of Vajrayāna and its textual elaboration. According to the tradition, the Guhyasamājatantra, the king – and earliest – of all Tantras, was revealed to the king of Udḍịyāna, Indrabhūti, by Śākyamuni himself or, in a different version of the story, passed on to Indrabhūti by Vajrapānị , who had heard the revelation when travelling in Udḍịyāna with Śākyamuni (Tucci 1949: I, 121, 212-215; 1977: 68-69; Roerich 1988: 359).[30]

Thus, in this particular cultural and geographic environment a common background must have existed between the rock sculptures and the textual tradition. If this is the case (whatever the linguistic expression in the semantic memory of these places), we might dare consider the intertwined phenomenon of literary and artistic experience as a sort of forerunner to the rang byung and gter ma traditions, i.e. the visionary revelation so deeply embedded in Tibetan mysticism.[31]

Moreover, the topographical and iconographical uniformity of the sculptures raises a number of general questions about how to read their physical environment. The available evidence suggests a close association with pilgrimage practices and their metaphorical re-enactment of the spiritual journey.[32] If this is the case, then they must have been related to pilgrimage routes whose existence and dimension can only be inferred through archaeological data, not any direct textual evidence or a surviving tradition. One might imagine a local road network serving marginalised Buddhist communities from across the Swat valley and leading to dilapidated sacred areas. However, our perception of things might be misled by the deceptive appearance of macroscopic data. In fact, if one begins to question what seems obvious, some alternative hypotheses take shape that fit the sparse and apparently conflicting pieces of evidence even better.

Politics, Economy, Market, and Trade Routes

Tibetan and Chinese sources of the eighth century attest to the strategic importance of Udḍịyāna for controlling the routes that connected Central Asia and northern India (Tucci 1958: fn. 1; 1977: 75 ff.). One wonders, therefore, whether and to what extent the region might have played an active role in this road network. Once again, however, we move among dispersed portions of data. Any evidence that can be retrieved is valuable, although we still have the problem of finding the right place for it within so many blank spaces.

The cultural richness that characterised Swat, especially at the acme of its splendour, during the Kushan period (Tucci 1977: 67, who cites the evidence of the Běishǐ [Pei-shih]), rested on a flourishing economy that was in turn fuelled by a wide range of resources. Agriculture was thriving and must have afforded conspicuous uous exports to the bordering mountain regions, valuable and marketable minerals including gold and silver were available (Tucci 1958: 280-281), and the region’s craftwork was renowned for its quality and in demand far and wide, as in the case of the famous Swati blankets, or kambalas, an industry whose ancient origins are recorded in Sanskrit sources (ibid.: 281 and fn. 16; Stein 1930: 63; Srinivasan 1990). Swat’s economy must also have benefited significantly from active participation in the trade prospering on the caravan routes connecting India with China and Central Asia. However, business was particularly vulnerable to vicissitudes, both political and environmental, that could gravely penalise the region’s economy. Probably the role Swat played in this respect was also at its peak during the Kushan period, when strong state unity was able to guarantee a trade traffic monopoly. Swat, connected with Central Asia through Chitral and the roads joining Laghman to the Indus, enjoyed an extremely advantageous geographical position, but as great powers – such as the Kushan Empire first and the Hūnạ rule afterwards – came to an end and the course of historical events became more episodic when not explicitly adverse, Swat would presumably have taken a more peripheral position in relation to the great trade routes.

Traffic on the route that Swat was directly connected with – the road running from India to Central Asia via Kunar – was probably significantly reduced by the competition of alternative routes created or simply enhanced both to satisfy the needs of a more complex trade network and to exploit the higher safety standards they could offer. One of these alternative routes was developed in the first half of the sixth century through Tokharistan, Kāpiśī, and Bāmiyān. Records of traffic in this direction can be found in the accounts of famous pilgrims; although such attestations are numerically very scant, they find indirect confirmation in the archaeological evidence. Traffic along the caravan routes traversing Afghanistan can in fact be considered, as Kuwayama argues (1987: passim; Id. 2006: 125), a decisive economic incentive in the rise of Bāmiyān and more generally of that region, attested to by the religious foundations that flourished there.

Nevertheless, the shift in caravan traffic to the west is hardly likely to have been so drastic as to do away with the eastern routes traversing Gilgit. Indeed, recent archaeological evidence suggests that the latter were extremely vital, especially in the period between the fifth and eighth centuries. Field research and studies conducted by the German-Pakistani mission led by Karl Jettmar and his successor Harald Hauptmann (Antiquities of Northern Pakistan 1-5, with complete bibliography of the preceding contributions) are increasingly shedding light on the history of the northern regions of Pakistan (Gilgit, Chilas, Baltistan and the northern side of the Indus-Kohistan).

The large number of petroglyphs and inscriptions found confirms that the kingdom of Palur/Bolor, including Gilgit and Baltistan (possibly to be identified respectively with the Little and Great Palur/Bolor of the sources), constituted the most powerful political unit in the region of Karakorom in the period between the fifth and eighth centuries thanks to the strategic position on the route connecting the Tarim basin to Kashmir and, through the latter, to India – a route that was moreover already in use in the first century CE (Jettmar 1989b: passim). Of great historical and cultural significance is also the distinctive production of bronze sculptures that can now be definitely related to the rulers of the Palola/Patọ la Sạ̄ hi lineage.[33] Besides witnessing to the Buddhist affiliation of the Gilgit aristocracy, these sculptures, as remarked by von Hinüber (2004: 9-10), also had substantial bearing on the development of bronze casting in Tibet. Their high artistic level, especially in consideration of their earliness with respect to the “Kashmiri” school, is a positive evidence for the exellence of local workshops, as well as for the country (or the local intelligentsia) being at a leading-edge of the Buddhist world. However, as will be seen later, the bronzes from Gilgit cannot be considered an isolated phenomenon. Rather, they are to be seen as the expression of a broader cultural (and somehow political) entity, as proved by kindred artistic productions in Swat, not only bronze sculptures but also, indeed, our rock sculptures.

Though the exact configuration of Palur/Bolor and adjoining areas is still open to discussion, south of Palur/ Bolor lay the territories of the Daradas, which for some centuries fore and after the Common Era constituted a political unit or, as better said by Jettmar, a “system”, still in existence between the fifth and the eight century CE (Jettmar 1977: 421). Even if distinct identities, Palur/Bolor and the Daradas’ territories were certainly in close contact with each other, in virtue of geographical contiguity, routes, economic interests and cultural ties. Swat/Udḍịyāna was part of this system, as one of those principalities that, though formally belonging to this or that kingdom, basically maintained some sort of independent status. It might be useful to recall here what said before about the (legendary?) king Indrabhūti and his alleged sovereignty over one part of Udḍịyāna, which might correspond to the above mentioned division of the Swat territory into two different political units, marked by the presence/absence (or better, rarity) of Buddhist rock sculptures. The part ruled by Indrabhūti (whom we are tempted to consider here as the name – real or fictitious it matters little – of some Darada lord) would thus represent a zone of cultural continuity – and privileged interaction – with Palur/Bolor. This would also add substance to the still vague connections between material evidence and transmitted traditions concerning the active role of Buddhist and Bonpo masters from Udḍịyāna and Gilgit in the further Himalayan developments.

In the route network connecting the Tarim Basin with India, a position of great prestige was enjoyed by Chilas – which seems at the time to have been included as a more or less independent frontier district in the kingdom of the Daradas (Jettmar 1989a: XIX) – since the shortest and easiest way from Gilgit to Kashmir passed through it (Biddulph 1893; Tucci 1977: 81-84). Indeed, it is in Chilas, along the course of the Indus, that the most interesting concentrations of petroglyphs and inscriptions of the area – in terms of both quantity and variety – have been found.[34]

On the evidence of their distribution and the environmental characteristics of the complexes housing them, Jettmar draws certain conclusions of a historical-cultural nature, including the likelihood that there was an important market centre at Shatial, created mainly to offset a presumed interdiction forbidding populations from the west (in this case Sogdians above all) to encroach on Kashmir territory (Jettmar 1989a: XLII-XLV). Jettmar also conjectures that in ancient times – as it still is today – the stretch of the Indus between Chilas and Harban was much frequented by gold-seekers (ibid.: XXXII). The activity may also have attracted the interest of merchants stranded here often for periods of months by adverse climatic and environmental conditions, the northern passes being viable only in the late autumn, the southern ones only in the summer. The fact that there was a great emporium and the appeal of some remunerative activity during prolonged stopovers would have accounted for the constant presence of travellers, and with them the concentration of petroglyphs and inscriptions in an area stretching westward from what were considered the most beaten tracks – places apparently insufficiently hospitable as to be chosen as a place to encamp.

Compared with the route via Kunar, the eastern way via Kashmir and the route further west, via Kāpiśī, must have enjoyed greater protection, guaranteed by the control of China, Tibet and Afghanistan. The Pamir regions, on the other hand, despite having an interest in maintaining an advantageous control over the caravan routes through alliances now with Tibet, now with China (Tucci 1977: 79), must have periodically borne the consequences of the conflicts between the two powers, with negative repercussions on their internal stability. As for Swat, it seems to have adopted a policy of friendly relations with China, as attested by Chinese sources that refer to contacts through ambassadors on missions from and to Udḍịyāna dating back as early as the sixth century. In 720 China sent through its ambassadors investiture to the king of Udḍịyāna and to other neighbouring states (Beckwith 1987: 91) but, with a subsequent diplomatic mission dated 745, it recognised the sovereignty of the king of Ki-pin (Kapiśa) over the region, possibly on account of a change in the policy of Swat, favouring Tibet (Tucci 1977: 75; 84-85; see also Prakash 1970: esp. 23 and 28; Kuwayama 1991: 283-284). Interestingly, this period roughly corresponds to the arrival in Tibet of Padmasambhava and other Indian masters, a fact that might also be interpreted as a consequence of the growth of the Tibetan political supremacy over much of these areas (Beckwith 1987: 91; 162, fn. 119).


  1. A phenomenon much like the case in China, arising on account of the huge donations and great privileges accorded the Buddhist clergy under the Eastern Jin (Demiéville 1970: 1260 ff. [repr. 1973: 376 ff.]).
  2. I would very tentatively like to suggest that this may be, as it were, a first-level explanation. In fact, this exemplary tale became a very popular “fable with a moral”, alluding to the far more challenging theme of inversion and transfiguration which, perhaps mistakenly, we confine within the specific domain of Tantric doctrines.
  3. For G. Gnoli (1963) it is a representation of Śiva (but see the later reconsideration in Gnoli 1992). Agrawala (1966) and Taddei (1966), on the other hand, see it as a divinity of martial character, although they take different lines here; for the former the image takes inspiration from the Indian prototype of Skanda-Kumāra, while the latter holds more probable derivation from a Syrian prototype. Without departing from the Gandharan environment, however, see the possible interpretation of the presence of Skanda in certain contexts (Filigenzi 2005c), to which, although indirectly, the six-armed divinity from Butkara I is probably connected (Filigenzi forthcoming b).
  4. My reference is to the Hindu temple of the Śāhi period discovered on the top of the hill in Barikot (see below). Even more interesting, given their unusual features, are other religious buildings of still unknown nature that were brought to light on the slope of the hill (Filigenzi 2010c: 410 ff.) and, more recently, in the residential area of the plain (Olivieri 2014: 101 ff.; Id. forthcoming).
  5. On the probable existence of actual non-Buddhist sanctuaries in Gandhāra and the neighbouring regions, see Fussman 1988: 7-8¸ and 1991: 162. In the two works dedicated respectively to a statuette of the goddess Śrī and a brass mask of Śiva, the scholar makes marginal reference to the question, stressing the risk of ignoring or underrating it, drawing hasty conclusions from arguments e silentio.
  6. As archaeological investigations and epigraphic records coherently show, in Swat this already happened in pre-Kushan times, under the local dynasties of Odi and Apraca. Being relatively new (and crucial), this issue has been under close scrutiny since the first publication of the Senavarman inscription (Bailey 1980). For some of the most important contributions, see Fussman 1982, 1993, 2003-2004; Salomon 1986, 1997; Falk 2003b; Callieri 2002; von Hinüber 2003; Schopen 2005: 79 and fns. 64-67; McDowall 2007. On the implantation of Buddhism in the North-West, see also Fussman 1994d.
  7. For the meaning and implications of this name, I refer the reader to Tucci 1977; Jettmar 1986; Klimburg 1999; Cacopardo and
    Cacopardo 2001. As for wine production and consumption in Swat/Gandhāra see Falk 2009 and Filigenzi forthcoming a.
  8. Similar modes of interaction between Buddhist monasteries and (Dardic) tribes have been documented in modern-day Ladakh
    (Olivieri 2011: 138, fn. 26).
  9. Of particular relevance to the question of hermitage is also the depiction, at several sites, of triśūla symbols and of a figure holding a triśūla (Olivieri and Vidale 2006: esp. 140; Olivieri 2010a: 359; fig. 7; Id. 2013: passim). I would cautiously venture to read these as a reference to Shaiva hermitage practices, whose presence would be plausible in the cultural environment we are tentatively reconstructing here. In this regard, it seems opportune to recall the several hints, scattered throughout Tucci’s works, of the strong connection between Swat (as part of a broader area including Gandhāra, the northern regions of the Khyber Pakhtunkhwa, up to the Indus, central and eastern Afghanistan) and Shaiva philosophies (Tucci 1958: 283-284; 1963: passim; 1968; 1977: 68).
  10. On the caves of Pānr ̣I see Faccenna et al. 1993: 115-124, where these are tentatively interpreted as store-rooms or dwellings (ibid.: 121). According to Callieri (2012), instead, the caves might have been used by the monks for dwelling or meditation, and could be dated to a quite early period in the life of the sacred area, possibly to its very beginning (mid-first century CE). The Buddhist rock shelters of Qal‘a have been only recently discovered (personal communication by L.M. Olivieri). As for the rock hermitage of Tangu, an element of extraordinary interest is its being decorated with polychrome mural paintings (Nasim Khan 2000: 35-68; reviewed by Olivieri 2002). The surviving fragment, showing a preaching Buddha flanked by two seated bodhisattvas, has been attributed by Nasim Khan (ibid.: 67) to a “late” Gandharan horizon (fourth/fifth century; ibid.: 36), although the sixth century is to be considered, more likely, the terminus ante quem non. A sixth/seventh century attribution is also suggested by Lo Muzio (2012: 329).
  11. On this topic, I refer the reader to Lo Muzio 2005: 486 ff., where other stimulating hypotheses of possible Gandharan antecedents are also suggested. See also Quagliotti 2012.
  12. For the temple of Bīr-kot-̣ghwanḍại, see Callieri et al. 2000, Callieri et al. 2000-2001; Callieri 2005; Filigenzi 2005c,d; Filigenzi 2010c. For the possible presence of an earlier stūpa, see above, p. 26, fn. 17. For further discussion related to these monuments, see Chap. 7.4.
  13. The figure is standing against a sort of tapering pillar at her back which, according to a current iconographical convention, can be recognised as the gadā (for a more detailed discussion, see Filigenzi 2005d).
  14. A new and more detailed historical panorama is being worked out by scholars dealing with the Hunnic period, especially with regard to the different waves of peoples which appeared across Central Asia/India starting from the fourth century CE – often improperly referred to as Hephthalites – and their non-destructive impact. The number and importance of scholarly contributions is by now too big to be fully accounted for here. Some of the most relevant to this study are Kuwayama 1989; Alram 1996, 2007; Callieri 1999, 2002; Grenet 2002; de la Vaissière 2005, 2007; Melzer 2006; ur Rahman, Grenet and Sims Williams 2006; Vondrovec 2008. Re-examination of archaeological sources also provides negative evidence of the havoc attributed to the Hunnic waves; see for instance Thakur 1967: 260; Stavisky 1993/1994; Dani 1986: 6, 148; Tarzi 2009; Filigenzi 2010b; Pidaev, Annaev and Fussman 2011: 17.
  15. A recent manuscript on the history of Kalam (Mankiralay 1987), translated from the Pashto and based on sources probably including local traditions passed down orally or in manuscript, speaks of the peaceful and prosperous reign of Kala, a Hephthalite sovereign closely related to Mehr Gul (Mihirakula, nephew or grandchild of Kala according to the manuscript). He is said to have wrested the throne of Swat from the cruel local sovereign Raj Singh, with the approval of the population. A follower of Buddhism, of amiable and just nature, he is reported to have boosted the urban development and economic prosperity of Swat and to have assigned cultivable land to the Huns, who had been exiled from the territory controlled by Mihirakula. This resulted in the creation of numerous colonies, above all in northern Swat, that had previously been scantly populated (ibid.: 298-299).
  16. Just recently, in 2005 and 2010, we have had tragic demonstrations of what earthquakes and floods can inflict on these areas. For a brief assessment of the seismotectonics of Pakistan and earthquake recurrence, see http://www.pakmet.com.pk/SeismicReport_PMD.pdf.
  17. The chronology of Period IX is being now better detailed after new excavations. The suggested dating is late third-early fourth
    century CE (Olivieri forthcoming). Analogous episodes can be inferred from the archaeological record, probably also around the fourth century CE, for instance at Butkara I (Faccenna 1980-1981: Part 3, 635; see below) and Saidu Sharif I (Faccenna 1995: 482, 488; pls. 42, 46-48; Filigenzi 2010b: fn. 5).
  18. For a more detailed discussion of this stela and its archaeological context, see Chap. 5.2.
  19. The most important aspects of this phase of restoration from both the technical and symbolic point of view seem to have escaped the authors of the excavation, but they are given close examination by Taddei (1998: 182; 2006: 48- 49). He points out that the criterion used for selecting the material to be re-used shows little regard for the formal coherence of the result and is based, rather, on the relevance to the selected decorative theme, thus favouring subjects of an “iconic” nature.
  20. This renovation – perfunctory as it evidently was − corresponds to Period [of Great Stūpa] 5, dated by Faccenna to the end of the seventh to eighth century (Faccenna 1980-1981: Part 1, 127).
  21. Filigenzi in Callieri et al. 1992: 37 ff.; Filigenzi 2010c: 411.
  22. By contrast, nevertheless, a magnificent religious and military stone architecture flourished in Swat under the Śāhis. I refer here to the above-mentioned defence system and the Vaishnava temple of Bīr-kot-̣ghwanḍại.
  23. The poorly built structure leaning against the stūpa, attributed by Qamar to a phase of de-sacralisation of the site, actually closely resembles the chapels of the late phase at Shnaisha (see above). In spite of its different connotations, some analogy can also be detected with Shrine [17] at Amluk-dara (Olivieri 2014: 357, fn. 13).
  24. For an overview of the literary sources related to Butkara I and a stratigraphic/chronological synopsis of the site, see Faccenna 1980-1981: 171-174. I refer to Filigenzi 2010b: esp. 389-391 and Filigenzi 2012b: 113 ff. for a more detailed summary of the cultural, chronological, and historical questions connected with this site.
  25. See Faccenna, Göbl and Khan 1993: esp. 106; I also refer to this work for the circumstances of the discovery. Here I would like to limit myself to remarking that the substantive argument of the coins’ dates perfectly matches the chronology that had already been proposed by Faccenna (1980-1981; Part 3, 635) on the basis of the entire archaeological evidence. For a summary, see Filigenzi 2010b. I take the occasion here to amend a mistake on p. 392 of that article, where it should read that the deposit of coins dates to the construction of Phase 4, not GSt 4.
  26. As for Pakistan, traces of prestigious public architecture in wood and clay are represented by wooden mosques, of which relatively few samples survive, especially after the wave of modernisation that flowed into the country in the 1980s and ‘90s. A project to document surviving wooden mosques was carried out in the 1980s by the Islamic team of the Italian Archaeological Mission in Pakistan, under the directorship of Umberto Scerrato (see Scerrato 1980, 1981, 1983). The project has recently been resumed under the supervision of Maria Vittoria Fontana (“The Wooden Mosques. An IsIAO Architectural Project in Pakistan”); a further project (“The Wooden Artifacts. An IsIAO Ethnographic Project in Pakistan”; director Ilaria E. Scerrato; IsIAO/ Comitato Ev-K2-CNR; funding institution: SEED [Social, Economic, Environmental Development]) is presently documenting the traditional wooden architecture in Baltistan. See also the essential contributions by Jettmar (1960), Dani (1989), Klimburg (1997, 2005). For a recent re-examination of the wooden architecture of the Hindu Kush/Pamir area, see Schadl 2009: esp. fn. 44.
  27. The site of Mes Aynak lies in Logar, about 30 km south-east of Kabul. Once an al-Qaida training camp, it is home to ancient
    Buddhist settlements of astonishing beauty (provisionally dated between the fifth and ninth century), which were probably
    founded there due to the site’s mineral wealth. As evidenced by abundant archaeological traces, the resources of the site (the
    world’s second-largest copper reserve) were already exploited in ancient times, likely by the monastic communities themselves. In 2007, a thirty-year lease was granted for the copper mine to the China Metallurgical Group Corporation (CMGC). Given the economic importance of this project ($3.5 billion, with an expected revenue of $880 million for the Afghan Government even before production begins), the area is soon destined to become a giant open-cast mine. Racing against time, the Afghan Institute of Archaeology, with the support of the Délégation Archéologique Française, launched in 2009 a rescue excavation of this impressively large site, which had been explored only in part. Following an agreement with the CMGC and the Ministry of Mines and thanks to Chinese and Afghan funds, the initial project has recently been extended. Except for some short summaries (Paiman 2010; AAVV 2011; Litecka et al. 2013), the results of the excavations are still largely unpublished. I am grateful to my colleagues and dear friends Philippe Marquis, Nader Rassouli and Abdul Qader Timori for their generosity in granting me access to the relevant documentation.
  28. Though based on somewhat different reasons, similar considerations have recently been expressed by K. Behrendt (2010). On the possible presence of a siddha among the characters depicted in the rock sculpture of Swat, see Chap. 6.
  29. I am grateful to Giacomella Orofino for this suggestion and, in general, for being unconditionally available for exchanging ideas and viewpoints.
  30. Scholars do not unanimously agree on the identification of Udḍịyāna with modern-day Swat and, consequently, on the close connection of Swat with the emergence of Vajrayāna. For an overview of the different hypotheses, see for instance Donaldson 2001: 8 ff. Nevertheless, most of the contrasting opinions are mainly based on textual evidence and do not take archaeological sources into account.
  31. Famous rang byung (self-existent) images on rock surfaces are venerated at Dentig, where there is also a cave blessed by Padmasambhava (Ricard 1994: 35, fn. 5); another famous image of miraculous self-formation is that of Hayagriva in the Agang Monastery (ibid.: 35, fn. 7). For a summary, mainly focused on textual materials, of the gTer ma (“rediscovered treasures”) tradition, see Gyatso 1992, 1996. As for Swat, it is not by chance perhaps that according to local lore the Buddhist rock carvings would mark the presence of hidden treasures nearby (Khaliq 2013), a belief which is likely to trace back to an old concept no longer understood but originally much the same as the Tibetan gTer ma. I would like to make clear that, even if this hypothesis is correct, we still may not think in terms of a one-way flow of ideas, but rather of a polycentric world that exchanged experiences, viewpoints, and reflections on topical issues, as was the case in the development of Buddhist Tantrism, mysticism and esotericism. Archaeological evidence from more or less coeval sites that are quite distant from one another helps to establish a sort of horizontal sequence, thus significantly expanding the historical framework of certain central concepts of Buddhist metaphysics (see below).
  32. Although no textual evidence can be directly correlated to this specific network of pilgrimage routes and their mystical character, the underlying concept must have been not much dissimilar from living practices in the Tibetan Bonpo environment (cf. Ramble 2007), whose possible connection with the aboriginal beliefs of the northern regions of Pakistan could be now re-evaluated on more firm grounds after the recent discovery in Swat of a huge amount of non-Buddhist monuments (see below). As already mentioned, the contribution of masters coming from modern-day Gilgit to the transformation of the primitive Tibetan Bön in a codified system was nonetheless already pointed out by Tucci (1958: 279, 282; Hoffmann 1969).
  33. The basic lines of the history of the reign of Palur/Bolor, the Patọ la Sạ̄ hi dynasty and the strategic importance of Gilgit had already been provided independently but with fairly close analogy by Tucci (1977) and Jettmar (1977). The studies on the Gilgit manuscript by von Hinüber (1983; 1986/1987) represent the most authoritative confirmation of the reconstruction proposed by the two scholars. For a more detailed dicscussion on these bronzes and their relevance to the present study see Chap. 3.
  34. For a critical reappraisal of this question, see Neelis (2002; Id. 2010: 257 ff.), who stresses the importance of the direct route (or better, a capillary network of routes) to eastern Central Asia passing through Chitral, Swat and Upper Indus Valleys. See also Sinha 1971, where the uninterrupted connections between Swat and Gilgit are particularly stressed.

From Art and Landscape: Buddhist Rock Sculptures of Late Antique Swat/Uḍḍiyāna, by Anna Filigenzi (Österreichische Akademie der Wissenschaften, Wien, 2015), 31-51