ARCHAEOLOGY INTERNATIONAL occupation gradually moved south, and by A Central Asian city on the Silk Road: the mid-first millennium BC the urban cen­ tre had shifted to the south and east, where ancient and medieval Merv it was well sited for departures on the great Georgina Herrmann trade routes. Its location reflects theimpor­ tance of contacts with Iran, India and In the 11th centuryAD, Merv was one of the world's largest Central Asia. India could be reached by and most splendid cities. Today it is a huge archaeological travelling up the Murghab Valley, and site, th e complexities of which are being unravelled by th e Merv was also well sited for the 180km march across the desert to ancient Amul In ternational Merv Project (IMP). Here th e Director of th e IMP (modern Charjew), a crossing point of the charts th e succession of cities th at have fl ourished in th e Merv mighty Oxus (modern Amu-darya). From oasis and describes th e work of th e project. Amul it was an easy stage to Bukhara and points east, or upstream to Termez. Merv's strategic location was demonstrated as erestroika and the opening of the toric urban centre of the Merv oasis in the recently as the 1980s, when it formed one former Soviet Union to foreign Kara Kum desert (Fig. 1). Since then, six of the bases for the Soviet invasion of initiatives led in 1989 to the seasons of surveys and excavations have Afghanistan. P Institute's first involvement in been completed.1 The city's location on the eastern edge the archaeology of Central Asia The Merv oasis is the child of the Murg­ of the oasis suggests that by the mid-first at the early neolithic site ofJeitun in Turk­ hab River, which rises in the Afghan moun­ millennium BC, if not earlier, the Murghab menistan (described in the previous article tains, flows north across the piedmont into had been dammed and the sophisticated in this issue). On the independence of the Kara Kum desert and deposits its rich irrigation system, on which Merv's agri­ Turkmenistan in 1992, a second collabo­ silts in a fan before drying up in the sands. culture depends, had been set in place. ration, the International Merv Project, was The bed of the river is deeply cut into the One of the aims of the International Merv agreed between the Turkmen Academy of piedmont, and the earliest occupation in Project is to study the irrigation history of Sciences, the Institute of the History of the Bronze Age was in the north of the oasis, the oasis. It has often been suggested that Material Culture (St Petersburg), and Uni­ where the river flowed nearer to the sur­ complex irrigation systems are prone to versity College London, to work at the his- face. With more sophisticated irrigation, failure, with consequent salinization and km 200 KAZAKHSTAN . ./\ 0 / / \ . .,/ \ US TYURT PLAT EAU <.; .._.."\ \ '-. I . ......;\ . \ "\ I �!J-} \; \ \ UZBEKISTAN I ) \ \ · . ._. .... -._.J TURKMENIST AN KARA KUM DESERT Ashgabat 0 ' (�� AFGHANISTAN � )'-? 52"E Figure 1 Northern Iran and western Central Asia, showing the location of Merv. 32 ARCHAEOLOGY INTERNATIONAL latest settlements within the cities. Even N more importantly, these sites have been pre­ Mosque of Yusuf Hamadani served by the Turkmen authorities who have • placed them in an archaeological park. This t opportunity of studying change through time, with remains near or on surface, applies to every aspect of our work - urban planning, military architecture, the ceramic and environmental sequences-and under­ pins our research strategy. The earliest city was the polygonal Erk Kala, 20ha in extent, whose walls still stand some 30m high (Fig. 3). It was probably founded in the Achaemenian period, c. 500 BC. The Seleucids used Erk Kala as the cit­ adel of the Hellenistic metropolis built by Antiochus 1 (who reigned from281 to 261 BC), Antiochia Margiana, which continued in use for over 1500 years in the area re­ ferred to today as Gyaur Kala. However, in the final phases of its life, much of Gyaur Kala was turned over to industry because • Standing monuments occupation had shifted to an extramural Citadel suburban development beside the Majan G Location of Gyaur Kala gates Bairam Ali Khan � canal. This became the medieval city, known <� today as Sultan Kala, which was walled by Kala � 0 1000 Abdullah-Khan Kala m the Seljuk Sultan Malikshah (1072-92). At its largest, it occupied some 630ha and was Modern Bairam Ali one of the capitals ofthe Seljuk empire. The next city, Timurid Abdullah Khan Kala, Figure 2 Th e city sites of Merv. Th e earliest, Erk Kala, was probably fo unded c. 500BC. sited some 2km to the south, was built by Later it became the citadel of th e Hellenistic city, Antiochia Margiana, known today as Shah Rukh in 1409 and lasted up to the Gya ur Kala. Th e medieval city, known as Sultan Kala, was built on th e other side of the Russian annexation of Merv in 1884. main canal; and the post-medieval city, Abdullah Kh an Kala, was built in 1409, with an extension, Bairam Ali Khan Kala. The ancient cities2 The great size of the city sites of Merv loss of agricultural potential. Because there their size and form. It is usually possible to means that traditional archaeological tech­ is a series of adjacent cities with different identify gateways, principal tracks, and niques are of limited value, not only be­ archaeological periods near the surface, the courses of some of the many canals that cause of finite time and finances, but also which can be sampled relatively easily, watered the cities. From the air it is possi­ because of the sheer scale of the problem. Merv provides a unique opportunity to ble to identify some of the patterns of the We initially planned to confine our efforts study long-term land-use history. In addi­ tion, colleagues working elsewhere in this large oasis have allowed us to sample their sites, so in snapshot form our archaeobot­ anists have established a sequence of some 2500 years. Two particular questions that we are seeking to answer are the date of the introduction of cotton, long a staple of the Merv economy, and ilie degree of devasta­ tion caused by the invasion of the Mongols in AD 1221-22. A succession of cities The complex long-term pattern of urban development was one of the principal rea­ sons for selecting Merv (Fig. 2). It is as if you had Roman London, Shakespeare's Lon­ don and eighteenth-century London laid out on separate sites. Occupation has been continuous there from c. 500 BC to the present day. Not only do we have discrete and datable areas of settlement, but we have cities built on a flat alluvial surface, where topography has little effect on urban planning. Furthermore, the cities were Figure 3 Aerial view of the ancient and medieval cities. Erk Kala, with massive mud­ walled, and so even at the grossest level brick walls, is in the foregro und with th e medieval city behind. Th e twelfth -centuryAD they provide us with information about mausoleum of Sultan Sanjar can be seen top righ t. 33 ARCHAEOLOGY INTERNATIONAL fourth-century kings, such as Shapur nand steel by the eo-fusion method, where Ill, but with a few from the fifth century. wrought iron and cast iron are heated to Analysis of the ceramics of Merv poses some 1200°C in refractorycrucibles. Accord­ particular problems. Most wares were made ing to Arab writers, such as the twelfth­ fr om local clays, so their fabric does not century al-Biruni, this produced excellent change, and until the Islamic period they steel with attractive Damascus or watered were unglazed. Shapes do change but not patterning. This technique has not previously dramatically and not across the range, and been documented in the archaeometallur­ it is one of the many truisms for archaeol­ gical record. ogists of the oasis - and of other areas of this time period - that ceramics cannot be The medieval city closely dated for over a millennium. How­ Working only in the ancient city failed to ever, based on our stratigraphic sequence take advantage of the unique opportunities and on statistical analyses, changes be­ of the long near-surface timescale, which tween the fifth and sixth to seventh centu­ the adjacent city sites ofMerv offer; nor did ry ceramics are current!y being established it address the time ofMerv's greatest splen­ 0 cm for the first time. These will allow more dour, in the Seljuk period. The medieval complete revision and re-analysis of earlier city is roughly square and much the same Figure 4 Th e cross on this mould, made published and unpublished data from ex­ size as Gyaur Kala, except for the suburban fr om a re-used jar handle, was fo und dur­ cavations at Merv itself, in the oasis and extensions to north and south (Fig. 2). ing excavations of a sixth to seventh cen­ from our own 1992-94 surface artefact sur­ However, the walls of Sultan Kala are less turyAD house in Erk Kala. Itprovides the vey. It is a long-term aim of the project to regular and they reflect a different pattern first archaeological proof of the presence of establish a ceramic sequence across the time of growth, as the city began life as an extra­ Christianity at Merv, well known fr om the range of the cities. mural suburb and was not walled until the literature.