AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA (DAR AL-KHILAFA OR JAWSAQ AL-KHAQANI) BY ALASTAIR NORTHEDGE

THE MAIN PALACE BUILT BYAL-MUcTASIM AT SAMARRA AT Khilafa is more appropriate for English-language the time of the foundation of the city in 221/836 usage. is one of the most famous of Islamic palaces, although it is not in fact preserved in very good condition. Nevertheless, itis one of the few exam- Past Work on the Palace ples of a great imperial palace of (at least later) antiquity where the complete plan is also known. It was not surprising that the site should have The site is best known from Creswell's Early attracted the interest of Western visitors in the Muslim Architecture, based on the work of the nineteenth century.5 InJune 1834 the surgeon of German Samarra Expedition of 1911-13.' Al- the British Residency at Baghdad, John Ross, though Herzfeld's plan of the palace (fig. 1) is visited Samarra, and published the following care- frequently published, the reader is often left with ful description: 6 the impression of an incomprehensible and unan- alyzed grand mass of rooms, although sometimes The Kasr-el-Khalifah, or Khalif's palace, is a long T- tacit editing of the plan is done, perhaps to make shaped mass of ruins on the edge of a high bank, the plan fit the page.2 divided by three cross walls: its extreme length land- A new overall presentation of the palace is wards is about 900 paces, the breadth of face towards needed, and it is hoped that this will eventually be the river 130 paces, and of the landward face 580; and it consists of ranges of gateways, arched rooms, vaults done within the context of a proposed publica- under ground, &c., with empty areas divided by cross tion of the excavation sites of the Samarra Expe- walls. One vault excavated to a great depth, calledJibb, dition. Several constraints limit me to a kind of is pointed out as the prison: its entrance is by a narrow preliminary report on this objective. In the first shaft, and people must have been lowered into it and place the subject is a large one, requiring the hauled up by means of a rope. Another deep square correlation of the small finds, the stucco and hollow close to it is called Birket-el-Seba', or the Lions'- other decorations, and the wall paintings. Sec- den. A narrow subterranean passage is cut from the ondly the field journals have notyet been worked Jibb to the Birket, from the door of which, criminals are through, nor all the drawings prepared for pub- said to have been thrown to the wild beasts. From the lication. Nevertheless it was thought worthwhile face towards the river, an inclined platform, resting on to publish the progress that has been made, al- arches, leads down to the hawi; and outside of the palace, at its N.W. corner, stand fine ruins of a turreted though the ideas presented are not necessarily building called Hamam, or the bath. final. It should be said at the beginning that the In April 1846 LieutenantJ. F.Jones of the Indian designation of the palace-al-Jawsaq al-Khaqani- Navy also visited the site.7 However, it was only at used by Herzfeld, s subsequently by Creswell, and the beginning of the twentieth century-half a which has come to be used generally in Western century after the opening of Western archaeolog- scholarly circles, cannot be correct. The reasons ical work on the Assyrian royal cities, for exam- for this will be explained below, but it is evident ple-that Samarra became the object of more that the name used in Iraq-Qasr al-Khalifa (Pal- serious scholarly attention. ace of the Caliph)-is in fact quite close to the The first efforts were French: General L. de terminology used in the third/ninth century. Al- Beyli6 published in 1907 the results of a visit to Jawsaq al-Khaqani may be an attractive and exotic Samarra and other sites in Iraq. 8 The French name-though one which incorrectly suggests architect Henri Viollet prospected and sur- that the palace was in some way Turkish4 -but the veyed the site in 1908, and then returned to palace of al-Jawsaq must have been part of the excavate in the palace in June 1910, digging overall complex, not the entire complex itself. twenty-four small sondages in six weeks. He pub- Some name such as Palace of the Caliph or Dar al- lished his work promptly, bringing out a short but 144 ALASTAIR NORTHEDGE excellently illustrated monograph on each of the Washington, D.C., and the negatives of the two campaigns. 9 expedition's photographs, taken by Sarre, are in The German interest in Samarra began with a the Staatliche Museen in Berlin. visit by Herzfeld in 1903,10 but came to fruition Fieldwork on the complex was subsequently later in the German Samarra Expedition, which taken up again by the Iraqi Directorate of Antiq- conducted two field campaignsin 1911 and 1913." uities, although only after a long delay.'7 The Bab Herzfeld turned his attention to excavating the al-Amma was cleared and consolidated, and the palace in the second campaign. Some 11,000 m rooms on its north side were excavated sometime square (equal to about a quarter) of the square in the 1960s or 1970s. However, for the most part reception-hall blockwere cleared, and about 5,000 excavations were conducted elsewhere at Samar- m square elsewhere, for a total of 18,000 m ra, and it was only with the beginning of the square. "Project for the Revival of the Two Archaeologi- In 1923, after the First World War, Herzfeld cal Cities of Samarra' and Mutawakkiliyya" in noted his regret that the excavated remains 1981 that new work was done on the palace. In had been robbed out.' 2 This was not unusu- 1981 a certain amount of clearance work was al; the robbing of fired bricks was common done in Herzfeld's excavation area in the recep- throughout Iraq before the introduction of tion-hall block, but it was given up as unprofit- concrete construction. The city walls built able. However, the tunnel which runs under the in the 1830s were entirely constructed of Ab- great courtyard-the Great Esplanade, to use basid bricks. Herzfeld's terminology-was excavated, togeth- In 1914 a preliminary report on the excavation er with six secondary buildings on the north side of the palace was published in Der Islam.'3 In this of the courtyard.'s In 1983 work turned to clear- article the first version of the grand plan of the ance and restoration of the Small Serdab (Haw- palace was published. The plan (fig. 1) was, of iyat al-Siba'). The project was finished in 1986, course, mainly the result of surface reconnais- and published in the same year.'9 Work began in sance, but also incorporates the excavation areas.' 4 1986-87 on the excavation of the Large Serdab; Herzfeld's reading of the surface remains seems by early 1989 the excavation was fairly complete, to have been good, and there is much drawn and the restoration was finished before my visit in which would be difficult to recover today. The early 1990. Unfortunately the publication in the main failure seems to have been in the northeast journal Sumer, the manuscript of which has been corner, where he only drew parts of the small completed, has been delayed by the effects of the palace-type units at H283 and H294 (see fig. 2 for Kuwait war. Work has also recently been done on building numbers).5 However, after the graphi- the range of rooms on the south side of the Bab cally primitive version published in the report of al-'Amma, and on a poorly preserved octagonal 1914, the final version, as published by Creswell, fountain or pavilion in the great courtyard. was "corrected," with a number of buildings squared up, where air photographs show the original version to have been more accurate. The Description of the Site polo maydan at the east end is in fact a parallelo- gram, not a rectangle. The palace complex is located on the conglom- In 1923, in volume 1 of the Ausgrabungen von erate of the east bank of the Tigris, which was laid Samarra,Herzfeld published the wall decorations down by earlier river beds in Tertiary times. The uncovered by the expedition; in 1927, in volume relative softness of this material made possible 3, the wall paintings; and in 1948, in the posthu- the cutting of underground structures on a large mous volume 6, his analysis of the historical scale. On the west side of the palace, there is a topography.' 6 However, as is well known, the steep descent of about 10 m to the flood plain of volumes planned on the architecture of Samarra the Tigris, and the garden on this side is located were never published: to be more precise, it was in the flood plain. Although the frequent move- the final report on the excavation itself that was ments of the Tigris bed within the flood plain are not published. Creswell, in his EarlyMuslim Archi- not capable of carving away much of the con- tecture, straightforwardly quotes Herzfeld, using glomerate, the river has removed the southwest the preliminary report of 1914. Unpublished part of the western garden since the third/ninth drawings and the field journals are in the Herzfeld century. In the middle of the 1950s a barrage was Archive at the Arthur M. Sackler Gallery in constructed at Samarra to divert flood water from AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 145 the Tigris into the Tharthar depression. The lake its original state, although there are many such behind the barrage flooded the low-lying land of buildings in early Islam; rather over the years units the flood plain, including all parts of the com- may have been added, rebuilt, or abandoned. plex west of the Bab al-Amma. By the end of the The first interpreter of the palace,Viollet, start- 1980s deposition of silt carried by the Tigris had ing from the Bab al-'Amma, saw the complex as turned the lake area into a marsh, and the area of consisting of the square building, the great court- the garden has reemerged as wetland. It is uncer- yard, and attached buildings-the two Serdabs, tain whether it will ever be possible to recover the maydan, and racecourse 2. It was Herzfeld material from this area, where nothing is visible who included the northern palace building in today. the complex, and called it the treasury. Buildings The layout is based on two architectural palace H326 and H327 in the southeast corner, although units. The first, on the south side, is composed of included in Herzfeld's plan, probably have no a square building of 180 x 200 m (H343), contain- connection with the palace. ing the Bab al-Amma, the throne halls, and the The area defined by Herzfeld is 175 ha. The harim, with a large courtyard on the east side (the dimensions are 1,346 m east-west from the pavil- Great Esplanade; H302). This layout of square ion on the Tigris to the spectator's lodge building and courtyard can be paralleled at overlooking racecourse 2, and 1,160 m north- Balkuwara, 20 and the palace at P16 at - south from the north palace to the south gate. siyya.2 1 On the north side a second palace struc- ture, termed by Herzfeld the "treasury" (H293), The western garden (H339; fig. 3). The western with a smaller square reception-hall block formal garden appears to have led down to the and residential apartments, is situated within Tigris, whose exact course in the third/ninth an outer enclosure wall, as one sees at al-Ukhay- century is not known. A flight of monumental dir22 and the Dar al-Imara at Kufa.? steps, 60 m wide, leading down from the Bab al- On the west side of the southern unit is a large 'Amma cannot be seen today, but can be seen on formal garden; there is probably another garden aerial photographs (fig. 5). On the west of the of a different type on the south side. Between the steps was a square ornamental pool 115 x 130 m. two palace units there is a large circular sunken From the pool a pair of walls 220 m apart basin, termed by Herzfeld the Large Serdab, and defining a longitudinal axis run directly west to in Iraq the Birka Handasiyya (H301). Further to the end of the garden as it survives. Perhaps this the east is a complex including what Herzfeld was a formal avenue for the approach of the termed the Rotundabau (H353), and, at the east caliph to or from a landing stage on the river. end, a smaller square sunken basin, termed the Such a procession is mentioned on the occasion Small Serdab, and in Iraq Hawiyatal-Siba' (Lions' of the investiture of the heirs of al-Mutawakkil in Den) .24 On the north and south sides of the basin 235/849-50.27 are courtyards with two pavilions, and lines of parallel halls, which Herzfeld called the "stables." When al-Mutawakkil appointed his heirs from among On the east side there is a maydan with a specta- his children, he rode at Samarra in a procession, finer tors' lodge, and the start line of racecourse 2, than which had never been seen. The heirs rode in stretching away to the east.5 On the south side of front of him, and the Turks in front of them, and their the main courtyard is an iwan pavilion in an children marching in front of al-Mutawakkil, with belts of gold in their hands ... then he descended into the enclosure (H313), adjacent to what must have water, and sat in it, and the army with him in originally been the main entrance of the com- jawankhiyyat28 and the rest of the boats, and he came plex leading to the city, and a further square until he stopped in the qasrwhichis called al-'Arus, and building, adjacent to the main reception-hall he gave permission to the people, and they entered block (H311). The palace may be described as a into his presence.9 complex of architectural units. Construction was evidently begun in 221/836, At the far western end a square raised terrace 48 and the last reference to it as an occupied build- x 40 m appears to represent the site of a garden ing dates to 269/884.26 The known occupation pavilion. lasted forty-eight years. It is obvious that a consid- A pair of walls 47 m apart also define a north- erable process of change must have taken place south axis on the west side of the pool. It is possible during this time. The palace is not one which was to make out the plan of several small buildings designed, constructed, and then abandoned in between the walls. 146 ALASTAIR NORTHEDGE

The garden may be the Bustan al-Khaqani of rooms has been excavated in recent years on mentioned in an event of 254/868, and which the south side of the Bab al-'Amma. 3 3 apparently lay between the Jawsaq and the At the east end of the central axis was a cruci- Tigris: form plan, more normally found in the center of such a square building. The centerpiece was a Bugha came to the bridge [al-Jisr] in the first third of dome chamber with four pillared halls. On the the night, and when the boat approached the bridge, east, west, and south sides were courtyards be- those entrustedwith it sent for those in the boat ... , and yond the pillared halls; on the east side a large Bugha disembarked into the Bustan al-Khaqani, and a transverse hall, which led straight out into the number of [the guards of the bridge] attached them- great courtyard (Grand Esplanade). It is possible selves to him ... and he rushed to al-Jawsaq. He asked that this last hall is a subsequent addition, as the permission of al-Mutazz [to enter], and al-MuCtazz gave permission to him.30 wall thicknesses do not match those of the re- mainder of the square building. The courtyard on the north side of this central The Bab al-Amma and the square reception-hall dome chamber has not been excavated, but those block (H343; figs. 6-7). The main reception-hall on the west and south sides were excavated by block of the southern palace is nearly square in Herzfeld. That on the south side constitutedwhat form, measuring 180 x 200 m. The Bab al-cAmma Herzfeld called the harem. On the south side of is the only part still standing; elsewhere only this courtyard there was a square room with four fragments of walls remained above floor level, piers, and this contained a circular basin carved and in places the bricks have been robbed down from a single piece of Egyptian granite, which was into the foundations. named Kasat Fir'un (Pharaoh's cup), and is In the center of the west side stands the main thought to be of ancient Egyptian origin.3 4 The outside entrance, the famous triple iwan known remainder of the surround of the courtyard con- as the Bab al-'Amma. This consists of a main sisted ofa warren of small rooms, which Herzfeld central tunnel-vaulted iwan, and two shallow side says had been repeatedly rebuilt. iwans, each of which has a semi-dome carried on The harem was richly decorated, notably with squinches, which lead into closed tunnel-vaulted wall paintings published in Herzfeld's Malereien halls behind. von Samarra.35 Its identification as a harim rests The photographs taken by Viollet and those principally on the figural character of the paint- taken by Sarre for the German Expedition show ings, which is comparable with the art of the quite clearly a block of masonry projecting from private palaces of the Umayyad period. Without the top of the building. This suggests that there doubt these paintings should belong to the last was a second story. A ramp has also been excavat- period of occupation of the palace: Herzfeld ed on the south side of the gate. Presumably there notes the rebuilding, and elsewhere at Samarra was a majlis (formal reception room) in this up to three layers of wall painting can be seen on second story, with a view over the Tigris and the fragments." By that time, in the reign of al- flood plain (fig. 8), on the pattern of similar Mu'tamid, the area of the complex occupied may rooms described over the gates of the Round City have been reduced. This part of the palace may of Baghdad,3 and the rooms found over the gates only have become a private residence at a late of Umayyad desert qu.siir.32 However, there are no date in its occupation. indications that the remainder of the square building had two stories. The great courtyard (H301). On the east side of After the Bab al-'Amma, the visitor passed the square building the principal feature is the through five rectangular halls placed transverse- great courtyard, a rectangle measuring 360 x 186 ly, to reach, at the center of the building, a m, and called by Herzfeld the Great Esplanade. courtyard with a basin. The courtyard was The courtyard is surrounded by a wall of mud excavated by the German expedition, together brick with arched niches and stucco moldings. with a bath on its south side. On the north side of The wall was topped with stepped merlons. 3 7 this axis was a courtyard building with a central The courtyard was divided in two by a depres- circular feature 31 m across, almost certainly a sion, which Herzfeld thought in his preliminary basin. There are several underground vaulted account to be a canal.? Excavations in 1982 showed rooms in this area. The south side of this western this to be a tunnel leading under the courtyard, part has not been clarified, although a sequence the roof of which had fallen in.3 9 Clearance of the AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 147 north wall at this time also produced evidence of later to west of the wall, and building H323 may six small secondary courtyard buildings. These be a late construction. were well aligned with the enclosure wall and At any rate the idea of a double gate is given were evidently constructed before the abandon- support by an event in 256/870: al-Muhtadi "left ment of the palace. 40 On the west side of the through Bab al-Masaff, until he went out through tunnel, Herzfeld's plan indicates two circular the gate known by the name of Itakh, then to features: recent excavation has shown the eastern Suwayqat Masrur, then Darb al-Wathiq, until he one to be a poorly preserved octagonal foun- came out to the Bab al-'Amma. "46 Al-Muhtadi's 4 tain. 1 The plan also shows a series of rectilinear itinerary is to leave the palace by the south gate, lines in the eastern part: photographs confirm turn west, and then go north to the Bab al- the existence of these lines, but it is not known 'Amma. The house of Itakh was next to the south what they represented, and they have now disap- gate,4 7 and evidently the name Bab al-Bustan has peared. Herzfeld thought theywere canals, based here been replaced by the Gate of Itakh.48 on his supposition that the transverse depression Building H313 on the west side at the north end was a canal.4 Perhaps they were walls delimiting of this proposed passageway has a building with a a reduction in the size of the courtyard to accom- single iwan in a courtyard and small courtyard modate the secondary buildings mentioned buildings around the periphery. If Herzfeld's above. plan is correct, 4 9 then the building corresponds with a pattern quite common in Iraqi Abbasid The south side of the complex. The area on the architecture. At Ukhaydir there is a separate south side of the great courtyard is one of the single iwan structure between the inner building most difficult to explain. There are a great num- and the outer wall.5 At al-Musharrahat a similar ber of different buildings, and no excavation has separate building is to be seen in the northeast taken place in this part, except for a small area corner of the enclosure, with a slightly variant excavated by the German expedition on the north plan. 51 There is a further separate building with side of building H313. The area is demarcated by an iwan and a mosque by the entrance of Sur'Isa an exterior enclosure wall running east-west 390 at Samarra.5 2 These buildings give the impression m south of the great courtyard (H363), and of being the majlis of someone responsible for divided by a north-south wall running on the west the building itself, or for its security. One is side of building H323 (H364). reminded of the situation in the Round City of A monumental avenue, 65 m wide and nar- Baghdad, where the central circular courtyard rowed by later overbuilding to awidth of about 12 contained a building for the guard (aras) and a m, approached the point where this north-south portico for the (shurta), but it is not possi- wall joined the outer enclosure wall (fig. 9). It is ble to be certain.5 3 To the east of this north-south evident that this avenue is not to be identified alignment, buildings H324, H325, H337, and with the Grand Avenue (Shari' al-A'zam) de- H357 seem to be store and service buildings. scribed by al-Ya'qubi, which passed by the palace, At the west end of the south side, directly south but with the Shari' Abi Ahmad, which ended at of the square reception-hall block, a building 70 the Bab al-Bustan (Gate of the Garden) and the m square has been built on a slightly different qu6sr al-khatfa (palaces of the caliph).4s As this alignment (H31 1), and this has been linked with avenue, the most monumental of Surra Man the south side of the reception-hall block. Ra'a,44 is not highly regarded by al-Ya'qubi, it Between this building and the north-south must belong to the early period at Samarra, the alignment of the entrance from the south gate original layout ofal-Mu'tasim.45 Therefore a main the terrain is uneven, cut by a wadi with three south gate probably stood at this point, with a branches. There are three groups of lines of passage into the great courtyard. On Herzfeld's rooms with courtyards on different alignments- plan, on the west side of the north-south wall, a H316, H317, and H358. These buildings are in- passage can be seen turning west around build- distinct on the aerial photographs, and were ing H313. However, on the aerial photographs apparently not of solid construction. In fact there the avenue definitely approaches the east side of are two references in the texts that suggest that the wall, through the space occupied by building the area was once a garden-firstly the name of H323, and to an inner gate by courtyard the gate discussed above as the Bab al-Bustan, H314 (H374). If this latter situation were the and secondly a reference in 248/862 to going case, then the entrance may have been moved from al-'Umari "through the gardens."5 4 This 148 ALASTAIR NORTHEDGE

interpretation is supported by the plan of al- objects which needed to be raised above ground Ja'fari (fig. 18), which also has an empty enclo- level for security. sure on the south side. It would evidently be a The sunken center of the structure is cruciform garden of a different type from the formal layout in shape and 115 m across. At the center there is on the west side. a circular basin 65 m in diameter, with amphorae The building complexes at H316, H317, and set into the brick walls, in which fish bones were H358 can probably best be explained by noting found. Straight monumental staircases lead down that the palace was the center of disturbances and to the interior from north and south. The basin is military operations during the 860s, notably in fed by a qanat coming from the east, which is not 248/862, 55 252/866, and 256/870.56 visible on the surface; and a further drain qanat exits to the west. At all four points of the cross On the first day of Rajab, 252 [866], a battle took place there are service ramps at right angles to the between the Maghariba and the Turks. .... They over- monumental staircases, although not all were in came the Turks at al-Jawsaq and expelled them, saying, use: on the west side one was used for a rubbish "Every day you kill one caliph, depose another, and kill dump and found full of ash, pottery, and glass. a vizier." ... When the Maghariba expelled the Turks The rooms around the basin were arranged radi- from al-Jawsaq, and overcame them at the treasury (bayt ally, although based on a four-iwan plan. In the aLmal), they seized fifty mounts from them.57 southeast corner there is a large bath, with fur- At times the palace was evidently an armed camp, nace and hypocaust. and temporary pis6 barracks would no doubt The rooms surrounding the basin are exten- have been put up in any empty space to accom- sively decorated with stucco dadoes, which are all modate a garrison. of the vine-leaf ornament, style A. While the sequence of Samarran stuccoes needs new work, The Large Serdab (H301; fig 10). On the north it is certain that this style was already in existence side of the great courtyard the main feature, lying before the foundation of Samarra in 221/836, between the two palace units, is the Large Serdab. but also continues later than Samarra.6? The It is not a genuine sirdab-that is, an under- buildings at Samarra, which are extensively dec- ground room for use in summer-but rather a orated with the vine-leaf ornament, that is, the sunken basin cut into the conglomerate with Bab al-'Amma with its mixed style A and style B surrounding rooms intended for the same pur- ornament, 61 and the upper and lower palaces at pose of escaping the heat of summer. A number al-Huwaysilat, one of which is to be identified of the surface buildings were excavated by the with the Qasr al-Juss of al-Mu'tasim, 6 2 are normal- German expedition. The sunken area was exca- ly attributed to the beginning of the Samarra vated by the Iraq Directorate of Antiquities be- period. The dating evidence of the stuccoes, though tween 1987 and 1990, and called the Birka Han- weak, suggests that the Large Serdab is an early dasiyya (Geometric Basin). building, part of the original plan of the palace. There is an outer buttressed wall 180 m on a side, with courtyard buildings on the inside at the The Rotundabau (H353; fig. 12). To the east of surface level. Herzfeld noted that some of these the Large Serdab, there is a further complex of rooms contained stores: Chinese pottery, materi- courtyard buildings (H333, H353). The recep- als for pavements, and luster tiles.5 8 In the south- tion room was excavated by Herzfeld and called eastcorner Herzfeld excavated halfofa rectangu- by him the Rotundabau or Rundsaal. The plan lar building composed of five thick parallel walls. was never published.' It consists of a typical Obviously these were supports for a thick-walled Samarran grand house-a reception block fol- elevated building. Al-Ja'fari has a similar building lowed by a courtyard and a further iwan. There is type: a central building supported on heavy walls, a further building with rooms around a courtyard surrounded by courtyard buildings (building A12; on the west side (H333), and four more indistinct fig. 11 ). Itis apparently a traditional Mesopotami- enclosures on the east and north sides. an building type, for a similar structure exists in The reception block itself consists of a trans- the northeast corner of the South Palace at Baby- verse hall measuring 18 x 7 m projecting into the lon, constructed by Nebuchadnezzar at the great courtyard, which seems to be an addition to beginning of the sixth century B.C., called by the original structure. Then there is a further Koldewey the "vaulted building."5 9 The build- transverse hall 13.8 x 4 m, similar to the portico of ing type is presumably a storehouse for valuable a T-shaped iwan, and this is flanked by two rooms AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 149 with niches. The niches were subsequently cov- the remaining two 96 x 11 m. There are five ered up by beveled-style, style C, stucco dadoes. entrances into the adjacent courtyards. Herzfeld The principal hall is circular, 9.4 m in diameter, identified these galleries as the stables for the with four curious small hexagonal rooms with low polo maydan.?8 Although it may seem surprising vaults on the east and west sides. The plan indi- that the stables should have been built adjacent cates that the circular form is secondary and that to a pleasant spot intended for the caliph's re- the hall had originally been square, with door- pose, with the obvious inconveniences of smell, ways into side rooms decoratedwith engaged piers. he may have been right. The form of the galleries The building and its associated dependencies resembles a building plan which exists also in a give the impression of being the majlis of an group of buildings at al-Ja'fari (A208-16; fig. 15). official closely associated with the functioning of At al-Ja'fari the galleries surround a central court- the palace, one who had need of work space and yard, and the group of buildings is placed at a service buildings. 4 distance from the inhabited parts of the palace- al-Ja'fari being later, this placement was possibly The east end and the Small Serdab (fig. 13). The the result of an unpleasant experience with the east end of the great courtyard is marked by a complex under discussion here.` This arrange- group of structures: a smaller, square sunken ment of galleries around a central courtyard is in basin (the Small Serdab, or Hawiyat al-Siba': fact the same as the usual arrangement of the H29), six long galleries, two pavilions, a polo may- stable galleries in Iranian caravansarays, also found dan, spectators' lodge, and racecourse. in Iraq. The Small Serdab (H29; fig. 14) was first inves- Al-Yaqubi, however, describes the caliph's pub- tigated by Viollet6&and then by Herzfeld, who lic and private stables as being located in the cleared the surface entrance and one staircase center of the city on the Sharic al-Sarija (= Shari' al- and prepared a plan. The Iraq Directorate-Gen- A'zam) .70 The site can be identified approximate- eral of Antiquities and Heritage began work on ly, and there is no sign of this type of building (fig. the site in 1983 and cleared and reconstructed 9). Nevertheless it is evident that stables located the site, finishing it in 1986.6 in the center of the city, five kilometers from the The entrance was situated in the center of the palace, were concerned with the long-term ques- east wall of the great courtyard. Herzfeld de- tions of raising and provisioning mounts for scribes this entrance as a square room in which the palace and the army, rather than the day- was found a frieze of painted stucco representing to-day stabling of animals available to the pal- two-humped camels.6 7 It has been reconstructed ace, and particularly the polo ponies and race- as a domed chamber with niches, much as a horses. typical entrance to one of the grand houses at North and south of these courtyards with sta- Samarra. On the surface level the structure mea- bles, there are two courtyards (H321 and H331) sures 60 x 54 m, and is represented by Herzfeld as with pavilions (H322 and H332). The north pavil- a series of rooms around the basin. It has been ion is constructed of mud brick and is still partly reconstructed as an arcade, with the arches per- standing to the transition of the dome of the pendicular to the sides of the basin. side iwan. The building was excavated by The basin itself is functionally the same as the Herzfeld (H332; fig. 16), but never pub- Large Serdab, but smaller and square (21 m a lished. 7 side), and sunk about 8 m into the conglomerate. The building consists of a central T-iwan, of There is a triple iwan on each side of the basin, which the front portico measures 16 x 5 m, and and all are decoratedwith stuccoes of the beveled the iwan itself 5.5 x 8.5 m. In addition there are style, style C. A qanit enters by the north central two shallow side iwans, measuring 4.2 x 3.2 m, iwan, and the water drained by a further qanat which were covered by semi-domes mounted on towards the south. One staircase descended squinches. These side iwans obviously resembled from the south side of the main entrance, those of the Bab al-'Amma, but had no door and a second descended to the northeast through into the hall behind. The exterior is corner of the basin. rectangular, measuring 38.5 x 22.5 m, and has round buttresses. The southern pavilion (H322) The stables and pavilions. On each side of the seems to be similar. The two pavilions would Small Serdab there are three long galleries (H330, appear to be intended for inspections of the H320); four of the six measure 106 x 11 m, and caliph's polo ponies and racehorses. 150 ALASTAIR NORTHEDGE

The polo maydan (H328) and the spectators' the existence of the maydan. So the maydan is lodge (H329). From the Small Serdab a contemporary with this first group, in spite of the passageway led through into a square courtyard change of alignment. The maydan is aligned with and then to a long rectangular maydan. The the racecourse; the only reason it has been possi- gateway into the maydan was excavated by ble to detect why the racecourse should have a Herzfeld.? The long rectangular maydan with different alignment from the axis of the palace in half-round buttresses measures 525 x 66 m. Ac- the flat terrain is that, had it followed that original cording to the evidence of robbing, the walls were axis, it would have cut a musalla (Y6) in the of fired brick. steppe, used for the festival prayers of 'Id al-Fitr On the far side of the maydan there is a specta- and c'Id al-Adha (fig. 9). Y6 is the mufalla closest to tors' lodge (H329), which looks out both onto the palace, and was perhaps sufficiently popular the maydan and the start of racecourse 2. This to force a change of alignment in the racecourse.78 building measures 28 x 45 m and is constructed of It is not certain that racecourse 2 is part of the fired brick, with two buttresses on the eastern same constructional program as the maydan, but corners. The plan is one of a chamber running it seems probable. There is some relative dating through east-west, which represents the passage- on racecourse 2: it is later than racecourse 1, way of a gate. On each side there are five cham- which it cuts.7 9 bers arranged north-south. This is obviously the The implication would be that this east end substructure of an upper story, where the viewing group is a secondary addition to the palace. platform would have been. There are some minor points which tend to The maydan is identified by Herzfeld as intend- support this idea: the east end group is not per- ed for polo.7 3 Thisjudgment is obviously correct, fectly aligned with the great courtyard. The Small for the arrangement of a long rectangular may- Serdab is decorated with beveled-style style C dan and pavilion with a viewing platform on the stuccoes. Evidently the group is a sports and upper story has a close parallel in the Safavid leisure complex, for racing, polo, and passing the Maydan-i Shah in Isfahan (524 x 159 m), built in summer days comfortably. 1590-1604, with its pavilion, the 'Ali Qapu.7 4 Ibn Qutayba confirms that polo grounds in the Ab- The barracks. In the northwest corner of the basid period were much like the maydans of palace complex, Herzfeld included a group of Samarra by remarking that "the width of the three blocks of courtyard buildings (H286, H288, maydan is made 60 cubits [31 m] so that [the H291), enclosed within a compound. H286 con- spectators] who are sitting on its wall will not be sists of 62 units round three sides of a courtyard, interfered with or assaulted." 75 There are twelve and H288 consists of 34 such units. H291 has 8 examples of similar maydans at Samarra, all smaller units with rooms on opposing sides of the court- than H328 and all located in or adjacent to palaces. yard, making a total of 104 units. In addition there are three smallmosques (H287, H290, H292). Racecourse 2. Racecourse 2 is an out-and-back Herzfeld identified this area as the barracks racecourse, which stretches east in a bottle shape of the palace guard. This seems possible; from the spectators' lodge.76 The sides of the there are many different patterns to military course diverge from the pavilion, and then on the accommodation at Samarra, the only unifying south side there is an S-bend after 1,510 m. After factor being that accommodation is always in the the bend, the sides are more nearly parallel with form of courtyard houses, rather than barracks of one another, and there is a straight of 2,200 m single rooms. Some confirmation is to be seen in before the curve. The track is 80 m wide; the the striking fact that there are three mosques, length seems to have been 10,420 m. 77 implying three separate groups. Presumably these were ethnic groups, who had to be kept apart, These units discussed at the east end seem to such as the Turks, the Faraghina, and the Maghari- represent a single coherent entity. The Small ba. The pointed provision of mosques seems to Serdab, stable galleries, and two pavilions are be part of an ostentatious display of Islam for the approximately symmetrical. There is a change of lightly Islamized steppe Turks, and is noticeable alignment at the polo maydan, but the north again at Sur Ashnas (see below). and south walls of the maydan retain the original alignment, creating a parallelogram, Al-Hanunam (H345). To the south of the bar- and the stables and pavilions take account of racks area and northwest of the square reception- AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 151 hall block, at the edge of the slope down to the east side there appears to be a further courtyard flood plain, there is a small square building with of similar dimensions, separated by awall. On the heavy walling and three round buttresses. The west of the central courtyard, there is a long hall building was identified by Viollet with that de- about 25 m long and 12 m wide, followed by a scribed as al-Hammam by John Ross. Two son- transverse hall.8 3 Then there is a square courtyard dages were dug in the area by Viollet, and exten- and the approach to the raised terrace. On the sive stucco decorations were recovered, of styles north and south sides of the central courtyard, A and B, similar to those of the Bab al-'Amma. 80 there are courtyard buildings which appear to be The building appears to be an elevated belve- residential apartments. All the construction of dere. Similar structures are known on the east major reception rooms was in fired brick, and has and west sides of the Qasr al-'Ashiq. been robbed out; the residential apartments ap- pear to have been constructed in mud brick or pis. The north palace (H293; fig. 17). The major In the area where the enclosure was extended, structure on the north side of the complex is a to the southeast of the main building, Herzfeld's palace building within a buttressed outer enclo- plan marks two further building complexes (H295 sure. The remains have never been excavated, and H296); their plan was not elaborated by him, and were not included in the complex by Viollet. nor are their remains very clear on the aerial Herzfeld included it, because he correctly thought photographs. However, in the northeast corner that there was no logical northern limit to the of the compound, there is a very clear plan of a palace complex without it. Regrettably many cru- residential building, apparently a small palace cial details are obscured by the front-line trench- (H294), not on Herzfeld's plan. It has the form of es dug by the British Mesopotamian Expedition- a parallelogram, measuring 169 x 90 m. There is ary Force in 1917. a square reception block in fired brick-which The rectangular outer enclosure, oriented east- has been robbed out-adjacent to the north wall, west, is 330 m wide and can be traced for a and a further reception room on the south side of distance of 462 m eastwards. The wall is equipped a central courtyard. with massive half-round buttresses about 11 m in Outside the enclosure of the complex at the diameter.8 ' The remains give the impression that northeast corner, there is a further small palace the east wall was later demolished, and replaced (H283), measuring 106 x 170 m. This has a with a new wall which is aligned with the maydan central square reception block in fired brick, with (H328). courtyards on the north and south sides. At the west end there is a square reception-hall block (H352), measuring approximately 120 x Discussion of the archaeological evidence. In its 130 m, with a buttressed exterior. It is raised on a latest form, then, the palace consisted of two terrace, and seems to have been of the type of major palace structures, each of which is larger plan with a domed chamber at the center. How- than other Samarran palaces, with the exception ever, a correct plan is unlikely to be recovered of al-Ja'fari. The south palace had a square recep- without excavation. tion-hall block with a large courtyard, but very On the east of the reception-hall block, the little residential accommodation in the form of main body of the building (H297) is 220 m wide, courtyard houses, normally an important feature and can be traced for a distance of 285 m to the in early Islamic palaces. The north palace, on east. The east end is not clear. If the building were the other hand, consists almost entirely of symmetrical, it would have been about 340 m residential accommodations. long. If it had originally been of this length, it Although it is not possible at the moment to would have corresponded to a rectangle with date all the structures described here, it seems the proportions 2:3, as found in the Samar- that the original elements of the palace included ran mosques, but would have been cut by the square reception-hall block and great court- building H294. Like the outer enclosure wall, yard-and possibly the western garden. Attached it appears to have been truncated by demo- to this, the Large Serdab (H301) and al-Hammam lition at the east end, to make way for new (H345) seem to be original on the basis of their construction. 2 decorations. The north palace must also be orig- Apparently at the original center, there was a inal, partly because of its plan which recalls that courtyard measuring 80 x 55 m, approached by of earlier palaces, but also because the Large two streets on the north and south sides. On the Serdab, itself apparently early, has monumental 152 ALASTAIR NORTHEDGE

stairways leading to both the north and south palace complex in Baghdad from the reign of al- palaces. Mutadid (892-908) onwards.9 6 It is clear that the The major later additions seem to be, firstly, modern traditional name of Qasr al-Khalifa is a the east end of the south palace, including the descendant of this term. Small Serdab, the north and south pavilions, the This group of names may have represented an maydan, and probably racecourse 2. Secondly, overall designation of a complex which had a there is the east extension of the north palace. number of palaces within it, as was the case for the The one identifiable building in this extension is Dar al-Khilafa in Baghdad; but it is perhaps more a palace (H294), which perpetuates the residen- likely that they were simple synonyms of Dar al- tial character of the compound. It is strange that 'Amma; al-Ya'qubi in fact explains Dar al-Khalifa part of the original palace would appear to have in this latter way.97 been demolished, but the evidence is not clear The Dar al-'Amma was the public palace in enough to be certain what happened. There are which the caliph sat in audience on Mondays and also many other structures whose relative dating Thursdays. 8 The caliphs also seem to have con- within the complex is uncertain. ducted a large part of their business there. Only the caliph had the right to make appointments there,9 the oath of allegiance was taken to al- The Topography of the Palace in the Texts Musta'in there in 248/862, and al-Muhtadi sat in the court ofjustice (mazalim) there. l° ° The most recent traditional name for the pal- It is specifically stated that the Dar al-'Amma ace has been Qasr or Bayt al-Khalifa, 8 4 which is was built on the site of the monastery which had still used today in Iraq.8 5 However, by the time of previously been there, and the monastery build- his preliminary report of 1914, Herzfeld was call- ing became the treasury (bayt al-mal). l All texts ing the site al-Jawsaq al-Khaqani, possibly relying link the bayt al-mal with the Dar al-'Amma, nota- on the work of Schwarz." In 1983 al-cAni pointed bly an occasion in 231/845-46 when "thieves out that there are two different palaces men- made a hole into the bayt al-milwhich is in the Dar tioned in the texts, al-Jawsaq and the Dar al- al-'Amma in the heart of the palace, and took 'Amma, and that Herzfeld could not be right in 42,000 dirhams." 0 2 One may suppose that as a applying the name of al-Jawsaq to the palace result of this event the monastery was replaced complex.8 7 The crucial text is from the Tafikh of with a more secure building, for there is no sign al-Ya'qubi: 88 of a monastery building on the ground. Also closely associated with the Dar al-'Amma is He [Muctasim] stopped at the site on which is the Dar the Bab al-'Amma. In the texts, the Bab al-'Amma al-Amma, and there was there a monastery of the is the site for formal arrivals at the palace. 0° 3 For Christians, and he bought the land from the people of example, when the rebel Babak was brought to the monastery, and he laid out [buildings] on the site, Samarra in 223/838, "the people came to look at and he went to the site of the palace known as the Jaw- him from al-Matira to Bab al-'Amma, and he was saq on the Tigris, and built there a number of palaces. brought into the Dar al-'Amma to the Prince of Believers."'0 4 In 241/855-56 al-Qummi "stood at All the texts confirm that two buildings are in the Bab al-'Amma with some of the Bujja, 70 question, but other texts mention events at both ghulams on riding camels. " 05 These events were together, as though there was no great distance triumphal processions, reminiscent of Roman between them.89 triumphs. However, the Bab al-'Amma is also identified The Dar al-CAmma. At the top of the range of with public punishments. In 226/841, they "cru- terminology are expressions which call the pal- cified [al-Afshin] on the Bab al-'Amma so that the ace the House of the Caliph or House of the people should see him.... The body was burnt, . Al-Ya'qubi uses Dar al-Khalifa, 90 Tabari and the ashes taken and thrown in the Tigris."'06 uses Dar al-Khilafa, 91 Dar al-Sultan,9 2 and Dar In 256/870 the head of Salih b. Wasif "was hung Amir al-Mu'minin. 93 Dar al-Khilafa could also up at the Bab al-Amma for an hour."'0 7 In 259/ have the abstract meaning of the caliph's house- 872-73 "they beat [a Christian secretary of Kan- hold. 94 In 256/870 the complex is simply called jur] ... 1000 lashes at Bab al-'Amma, and he al-Dar. 95 This type of terminology predated died. " 08 It seems that the sinister reputation of the the use of Dar al-Khilafa for the caliphal palace in the modern tradition (in the description AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 153 of John Ross above) derives from these events, it was possible for a guard to walk around although it has spread from the Bab al-'Amma to underneath it."9 other parts of the palace.'° 9 Al-Jawsaq is clearly signaled in the texts as the In the texts the Bab al-'Amma is localized in the private residence where the caliphs lived. Al- west of the palace-one could go out from the Mu'tasim was buried there in 227/842.120 Al- Bab al-'Amma towards al-Haruni, which lay in the Wathiq (227/842-232/847) built and moved to flood plain of the Tigris"°--and in relation to the al-Haruni, where al-Mutawakkil also lived for most "avenue," that is the Shari' al-A'zam or Shari' al- of his reign (232/847-247/861). But al-Mutawak- Sarija."' One could see the Bab al-'Amma and kil settled his son al-Muntasir at al-Jawsaq,' 2' and the Dar al-'Amma from the avenue.1 12 So the according to Ibn A'tham al-Kufi (d. 314/926) al- traditional identification of the Bab al-'Amma Muntasir was buried there in 248/862.122 Al- with the triple iwan on the west facade of the Musta'in, al-Mu'tazz and al-Muhtadi all lived in al- palace must be correct, and the square reception- Jawsaq,'23 and the latter two were buried there, hall block of the south palace must be the Dar al- together with al-Muntasir. 24Al-Mu'tamid lived in 'Amma. al-Jawsaq, until a move to al-Ma'shuq, and re- It seems strange that the main avenue should turned there for a last time in 269/884.125 In 290/ be said to have passed through the formal, pre- 903 al-Muktafi decided to reestablish the capital sumably private, garden of the caliph, and the at Samarra, but seems to have found al-Jawsaq a identification of the Bab al-'Amma has been doubt- ruin, for he was forced to camp there. 2 6 ed for this reason." 3 The solution appears to lie in The limited evidence for women in the palace a chronological sequence. The archaeological is related to al-Jawsaq: in 255/869 Qabiha, the evidence of the western garden is compatible mother of al-Mu'tazz, "brought out the money, with, although it does not prove, the hypothesis jewels, and valuable possessions in the stores that it is early in date and belongs to the original within al-Jawsaq" to pay the Turks.'2 7 In 256/870 period of construction at the palace.11 4 The first aletterwas published, which awoman had brought mention of the Bab al-'Amma as a place open to "from the area adjacent to the Qasr al-Ahmar the public dates to 223/838, and the description (normally associated with al-Jawsaq)."' 2 8 of the Shari' al-A'zam by al-Ya'qubi dates to after Al-Jawsaq was used tojail distinguished prison- the death of al-Mutawakkil in 247/861.15 It may ers. The first in 225/839-40 was al-Afshin Khay- not have been possible to keep private what was dar b. Kawus al-Ushrusani, for whom a special planned by the architects as a private garden, prison was built. 129 In 248/862 al-Musta'in impris- when the palace itself was dedicated as a public oned the two sons of al-Mutawakkil, al-Mu'tazz building. At any rate the most logical place for the and al-Mu'ayyad in a room in al-Jawsaq. 130 Re- avenue to have crossed the garden is between the leased in 251/865, al-Mu'tazz was made caliph, north-south walls (figs. 3 and 9). and reimprisoned the unfortunate al-Muayyad, The bayt al-mal is associated with the Bab al- who ultimately died in prison.'3 ' In 256/870 al- 'Amma and the avenue in an event of 248/862, Mu'tamid was brought out of prison in al-Jawsaq and thus may be building H311 on the south side to be made caliph.' 3 2 of the square reception-hall block."6 If this were When one approached the palace by the Shari' not so, one would have preferred to identify the Abi Ahmad, according to al-Ya'qubi, one ap- bayt al-mal with the vaulted building (H338) in proached the Bab al-Bustan and the qusiir al- the enclosure of the Large Serdab, which seems khalifa (palaces of the caliph, a phrase which more secure, and could be described as a Meso- appears to mean the private residences of the potamian version of the raised bayt al-mal in the caliph). Al-Ya'qubi does not mention al-Jawsaq or Umayyad mosque of Damascus. al-Jawsaq al-Khaqani in his description of the avenues of Samarra, and one must presume that Al-Jawsaq. According to al-Ya'qubi al-Jawsaq be- it is here subsumed among the residences of the longed to the original construction of Samarra, caliph.l 3 In some way the residences of the caliph and was called al-Jawsaq al-Khaqani, after Khaqan lay at the east end of the palace. In 256/870, also, 'Urtuj Abu al-Fath b. Khaqan, who was responsi- Musab. Bugha "took to al-Hayr' 34 ... until he came ble for its construction and who was assigned a to the gate of al-Hayr which is adjacent to al-Jawsaq qa.'a adjacent to it."7 An iwan is mentioned in and al-Qasr al-Ahmar."'-5 Al-Jawsaq is described as it."8 There was also a tower called al-Lu'lu'a (the looking out over al-Hayr-to the east, though Pearl), which was built as a prison for al-Afshin; also mentioned as "al-Jawsaq on the Tigris."' 3 6 154 ALASTAIR NORTHEDGE

The obvious candidate for the identification of comparison with the apparently piecemeal ar- al-Jawsaq is the north palace in the complex rangements of the Dar al-Khilafa. According to (H293). It is the only building of the appropriate Tabari, there was a Bab al-'Amma at al-Ja'fari, scale to serve as the residence of the caliphs-it which is not yet precisely identified,' 4' but it is has residential apartments, which the square build- obvious that the role of public palace was, firstly, ing does not. The cantonment to the north (site integrated into an overall design, and, secondly, X) is a good candidate for the qaFi'a of Khaqan far less important than at the Dar al-Khilafa. Al- 'Urtuj, which is mentioned as having been adja- Mutawakkil had been caliph for twelve years when cent to al-Jawsaq. S3 7 he began al-Mutawakkiliyya in 245/859, and had The palace at H294, in the added eastern ex- no doubt grown tired of the hurly-burly of poli- tension of the compound, could be identified tics.'4 2 with the palace of al-Kamil, built for al-Mu'tazz by This failure on the part of al-Mutawakkil em- al-Mutawakkil within al-Jawsaq, and perhaps add- phasizes al-Mu'tasim's political vision in construct- ed to in his own reign by a building designed by ing a prominent public palace. Al-Mu'tasim had 8 his mother. 13 Al-Qasr al-Ahmar, closely associat- his relationship with his people foremost in his ed with, but separate from, al-Jawsaq, and located mind when he laid out the Dar al-'Amma. There near an east gate into the complex, could be are only two other extant early Islamic palaces identified with H283.'3 9 where orientation towards the public is regarded as important; one is the four-iwan reception hall at the entrance of the Umayyad palace in Am- Discussion man, and the other is the palace of Ashnas at al- Karkh, Sur Ashnas, where the main entrance leads The dual nature of the palace, in both the into a courtyard with a mosque in its center.'14 textual sources and the archaeological evidence, It is striking that it is the public palace which is quite striking. On the one hand, there is a received the attention in terms of additions, square palace with no residential accommoda- improvements, and quality of construction, al- tion, facing onto the Sharic al-A'zam and the though the two structures are of similar dimen- garden on the west and onto a grand courtyard to sions. The square palace is entirely constructed the east. On the other hand, there is a palace with of fired brick; the north palace is partly construct- residential accommodation, enclosed within a ed of mud brick and pis&. Several complexeswere massive buttressed wall. In the textual evidence either built originally or subsequently added to there is the Dar al-'Amma, a public palace where the great courtyard of the south palace, including caliphs are made and unmade' 4 0 and sit in audi- the sports and leisure complex at the east end. ence and judgment. On the other hand, there is The only addition made to the north palace was a private residence, al-Jawsaq, where caliphs live, the small palace H294, whereas part of the orig- die, and are buried, and which seems to be the inal structure appears to have been demolished. domain of the women. This disparity could be explained if one consid- These features are confirmed by the similari- ers the public palace as the domain of the men ties and differences in the plan of al-Ja'fari (fig. and the private palace as the domain of the 18). Without doubt the caliphal palace of al- women. No doubt the caliphs spent their days in Mutawakkiliyya, built in 245/859-247/861, was the Dar al-'Amma, not merely the two days of intended to match, and was to a certain extent a public audience.'4 4 The sporting facilities are copy of, the caliphal palace of Surra Man Ra'a. attached to the public, not the private, palace. It The area of the main palace, 176 ha., is almost is common for more to be spent on the facilities identical to that of the Dar al-Khilafa, but the for those who control the finances. building is separated from the city by a space of Lastly there is no doubt that some of the more between 1,100 m and 2,300 m, which can only peculiar features can be attributed to the com- have been intended for the caliph's peace and plex history of the palace. In particular, towards security. The reception-hall block on the Tigris is the end, under al-Mu'tamid (256/870-279/892), smaller (125 x 125 m), and the body of the palace who had little power or influence, occupation is dominated by blocks of courtyard houses, obvi- may have been reduced, and the square building ously used as residential apartments. To the north may have been the last to be occupied, which may and east, the large area and organization of explain the private nature of the wall paintings in storehouses and workshops is impressive, by the so-called haim. AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 155

Notes dans le bassin du Tigre," Revue archiologique, 4th ser., 10 (1907): 1-18. I am grateful to the Iraq Directorate of Antiquities, notably its Director-General Dr. Mu'ayyad Damerji, 9. Henri Viollet, "Description du palais de al-Mouta- and the head of the project in Samarra, Sd. Hafiz al- sim fils d'Haroun-al-Raschid i Samara et quelques Hayani, for their kind reception and sharing of infor- monuments arabes peu connus de la Mesopota- mation on a site on which I did not myselfwork. I would mie," Mimoires presentis d lAcadimie des Inscriptions like to thank the Arthur M. Sackler Gallery, Washing- et desBelles-Lettres1 2 (1909): 567-94; H. Viollet, "Le ton, D.C., for access to the Herzfeld Archive, and the palais de al-Moutasim ils d'Haroun-al-Raschid a Islamisches Museum, Staatliche Museen, Berlin, for Samara et quelques monuments arabes peu con- access to materials of the Samarra Expedition. The nus de la Mesopotamie," Comptes Rendus de archaeological analysis depends upon publication work l'Acadimie des Inscriptions et des Belles-Lettres (1909): for the archaeological survey of Samarra carried out 370-75; H. Viollet, "Fouilles i Samara en Mesopot- with the support of the Fondation Max van Berchem, amie: Un palais musulman du IXe siecle," Mimoires the British School of Archaeology in Iraq, and the presentis l'Acadimie des Inscriptions et des Belles- British Academy. The research for the textual part of Lettres 12 (1911): 685-717; H. Viollet, "Fouilles a this studywas carried out during the tenure of a fellow- Samara: Ruines du palais d'Al Moutasim," Comptes ship of the Alexander von Humboldt Stiftung at Uni- Rendus de l'Acadimie des Inscriptions et des Belles- versitiit Tfibingen in 1989-91, and I am grateful to Lettres (1911): 275-86. Professor Heinz Gaube and the students of the Orien- talisches Seminar for their stimulating reaction to the 10. Ernst Herzfeld, Samarra:Aufnahmen und Unter- first presentation of this subject. suchungen zur islamischen Archdologie (Berlin, 1907). I. K. A. C. Creswell, Early Muslim Architecture, 1st ed., 2 vols. (hereafterEMA) (Oxford, 1940), 2:232-45; 11. The origins and events of the German Samarra Ernst Herzfeld, "Mitteilung fiber die Arbeiten der Expedition are briefly recounted in Alastair zweiten Kampagne von Samarra," Der Islam 5 Northedge, "Creswell, Herzfeld and Samarra," (1914): 196-204. Muqarnas8 (1991): 74-93.

2. This has happened inJames Allan's new edition of 12. "The excavations in the palaces have been com- Creswell's Short Account ofEarly Muslim Architecture pletely robbed of bricks; one sees only the trench- (Aldershot, 1989), fig. 210. In Creswell's original es, instead of walls. No pavements. The other edition of the Short Account, the plan of the palace excavations are blown away [i.e., silted] and vege- was not included at all, no doubt for reasons of tation begins to grow over them" (Journal N-83, p. space. 16, in the Herzfeld Archive, Arthur M. Sackler Gallery, Washington, D.C.). 3. Herzfeld, "Mitteilung fiber die Arbeiten," 196; Herzfeld, Ausgrabungen von Samarra,'vol. 6, Ge- 13. Herzfeld, "Mitteilung fiber die Arbeiten." schichtederStadt Samarra (Hamburg, 1948), passim. 14. The original field notes still exist in the Herzfeld 4. E.g., E. Esin, "The Turk al-c'Aam of Samarra and Archive. the Paintings Attributable to Them in the 6awsaq al-Haqani," Kunst des Orients 9 (1973-74): 47-88. 15. For the purposes of my survey and catalogue of sites at Samarra, the various site areas have been 5. There was, of course, no question of rediscovery of allotted letter designations from A to Z to identify the site, whose location had always been known. them. For individual buildings and archaeological sites, numbers are added to the letters in the series. 6. John Ross, "AJourney from Baghdad to the Ruins For example, P7 is a group of three brick kilns in of Opis and the Median Wall in 1834,"Journalof the site P. Royal Geographic Society 11 (1841): 121-36. 16. Ernst Herzfeld, Die Ausgrabungen von Samarra,vol. 7. J. F.Jones, 'Journal of a Steam Voyage to the North 1, Der Wandschmuck derBauten von Samarraund seine of Baghdad in April 1846," Journal of the Royal Ornamentik (Berlin, 1923); vol. 3, Die Malereien von Geographical Society 18 (1848): 1-19. Samarra (Berlin, 1927); vol. 6, Geschichte der Stadt Samarra (Hamburg, 1948). 8. General L. de Beyli6, Prome et Samarra: Voyage archiologique en Birmanie et en Misopotamie (Paris, 17. Khalid KhalTl Hammildi, "Qasr al-khalifa al- 1907); de Beyli6, "L'architecture des Abbassides Mu'tasim fi Samarra," Sumer 38 (1982): 168-205; au IX e siecle. Voyage archeologique a Samara, on 168, he remarks that work on the palace was 156 ALASTAIR NORTHEDGE

carried out from 1936 onwards, although it is not Iraq Museum [Baghdad, 1976]). evident what this work was. 35. Herzfeld, Die Ausgrabungen, vol. 3. 18. Hammfid, "Qasr al-khalifa al-Muctasim." 36. Surface fragments found at al-Haruni in 1989. 19. The publication is HAfiz Husayn HIayan!, "al-H ir," Sumer44 (1985-86): 139-57 (Ar. sect.). The struc- 37. Creswell, EMA, 243, fig. 193. ture is identified in this article and on the informa- tion signs at the site as "the palace of al-Hayr," a 38. Herzfeld, "Mitteilung fiber die Arbeiten." building whose location is not given in the texts which refer to it (Walid b. 'Ubayd al-Buhturi, 39. H.ammfid, "Qasr al-khalifa al-Mu'tasim," 187. Diwan al-Buhtuni, ed. Hasan Kamil al-Sayrafi [Cairo, 1963-64], 1.44; Yaqfit b. 'Abdallah al-Hamawl al- 40. H.ammfldi, "Qasr al-khalifa al-Muctasim," 201,204. Rfimi al-Baghdadi, KitabMujam al-Buldan, 6vols., ed. Wuistenfeld [Leipzig, 1866-73], s.v. al-.Hayr). 41. Information from Sd. Hafiz Hayani. Hayani does not explain the identification or tell us who first proposed it. 42. Herzfeld, "Mitteilung fiber die Arbeiten."

20. Creswell, EMA, 2:265-70; Creswell-Allan, Short Ac- 43. Ahmad b. Abi Ya'qfib b. Wd.ih al-Yac'qibi, Kitab al- count, 365-67. Buldan, ed. M. J. de Goeje, Bibliotheca Geogra- phorum Arabicorum 7 (Leiden, 1892), 261; Tabari, 21. Alastair Northedge and R. Falkner, "The 1986 Tanikh al-Rusul wa'l-Muliik, 3:1807. SurveySeason atSamarra,"Iraq49 (1987): 143-73. 44. The name Surra Man Ra'a, which appears to have 22. Creswell, EMA, 2:50-98; Creswell-Allan, Short Ac- been the formal name of the city of al-Muctasim, is count, 248-63. used here to refer to the central city, in contrast to the outlying cantonments of al-Matira and al-Karkh, 23. Creswell-Allan, Short Account, 10-15. and the city of al-Mutawakkiliyya.

24. HaydnT, "al-Hir." 45. The full argumentation of the subject of this ave- nue and its relationship to the layout of the origi- 25. Alastair Northedge, '"The Racecourses at Samar- nal Surra Man Ra'a must be reserved for another ra," Bulletin of the School of Oriental and African publication. Studies 53 (1990): 31-56. 46. Tabari, Trniikh al-Rusul wa'l-Muliik, 3:1820-22. 26. Muhammad b.Jarir al-Tabari, Taiikhal-Rusulwa '- Mulik, ed. de Goeje et al. (Leiden, 1879-1901), 47. Al-Ya'qibi, Buldan, 262. At this point there is a 3:2040. grand palace on the east side of the avenue, and another on the west. One should be identified with 27. Tabari, Tarikh al-Rusul wa'l-Mulik, 3:1394-1403. the house of Itakh, and the other probably with al- 'Umari (al-Yac'qibl, Buldin, 258; Tabari, Tafikh al- 28. A type of boat. Rusul wa'l-Mulhk, 3:1300-1, 1503, 1504; Abfi al- HasancAli b. al-Husayn al-MasCidi, Muruijal-Dhahab 29. Abi al-Faraj 'Al b. Husayn al-Isfahani, Kitab al- wa Maaidinal-Jawahir, ed. and trans. C. Barbier de Aghani, 21 vols. (Bulaq, 1867), 9:32. Reinard and P. de Courteille, 9 vols. [Paris, 1861- 77], 7:122). 30. Tabarl, Tanikh al-Rusul wa'L-Mulfk, 3:1696. 48. It should be remembered that the written sources 31. J. Lassner, The Topography of Baghdad in the Early are of varied origins. Consistency in naming struc- Middle Ages (Detroit, 1970), 54. tures cannot be expected.

32. E.g., Qasr Kharana (Creswell-Allan, Short Account, 49. Herzfeld excavated the entrance to this building 96-105); orJabal Sais (pp. 118-22). from the great courtyard. However, his reading of the plan differs somewhat from the impression 33. Not so far published. given by the aerial photographs.

34. In 1977 this basin was in the courtyard of the 50. Creswell, EMA, fig. 64. Abbasid Palace in Baghdad. Previously it had been in the Khan Mirjan (F. Basmachi, Treasures of the 51. The plan of the palace at al-Musharrahat can be AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 157

consulted in Northedge, "Creswell, Herzfeld and 66. H.aynT, "al-H.ir," 139-57. Samarra," fig. 12, and Northedge, 'The Palace of al-Istabulat at Samarra," Archiologie islamique 3 67. Herzfeld, "Mitteilung fiber die Arbeiten." (1993), fig. 8; a publication is being prepared by Petersen and Northedge. 68. Herzfeld, "Mitteilung fiber die Arbeiten."

52. The plan of Sur 'Isa regrettably remains unpub- 69. Modification of building plans to correct prob- lished. See drawing D-1084 in the Herzfeld Ar- lems evident in earlier versions is known elsewhere chive, Arthur M. Sackler Gallery, Washington, D.C. at Samarra: one can argue that the plan of the cloverleaf racecourse was dictated by the obvious 53. Al-Ya'qiibi, Buldan, 240. difficulty of the traditional out-and-back plan (race- courses 1 and 2), that the caliph could not see the 54. Tabarl, Tdaikh al-Rusul wa'l-Muliik, 3:1503, 1504. whole race (Northedge, "Racecourses at Samar- ra"). The cloverleaf racecourse also appears to 55. Tabari, Tahrkh al-Rusul wa'l-Muliik, 3:1504. date to the reign of al-Mutwakkil. One suspects that these modifications were the result of the 56. Tabarl, Tahikhal-Rusulwa'-Muluk, 3:1815,1823-24. personal intervention of al-Mutawakkil, who was a great lover of architecture, and personally super- 57. Tabarl, Tiinkh al-Rusul wa'l-Muliik, 3:1680. Trans- vised the construction of al-Mutawakkiliyya from lation of G. Saliba. the palaces of al-Muhammadiyya on the raised canal levee of the Qatul al-Kisrawi to the east of the 58. Herzfeld, "Mitteilung fiber die Arbeiten," 201. site of the Abu Dulaf mosque (Tabari, Tiikh al- Rusul wa 'l-Muluk, 3:1438). 59. J. Oates, Babylon (London, 1979), 151, ill. 101; R. Koldewey, Das Wiedererstehende Babylon (Leipzig, 70. Al-Yac'qibl, Buldan, 260. 1925). Koldewey in fact proposed that the vaulted building was the site of the Hanging Gardens of 71. Viollet, Fouilles d Samarra,pl. III; Herzfeld Archive, Babylon, although he recognized that the hypoth- drawings D-1031 and D-1124. esis "bristled with difficulties." The building con- tained an archive of tablets detailing lists of rations 72. Herzfeld Archive, drawing D-1044. for foreign exiles in the city and an unusual type of triple well. A series of wells is also to be found in the 73. Herzfeld, "Mitteilung fiber die Arbeiten." similar building A12 at al-Jacfari. 74. Arthur Upham Pope and PhyllisAckerman, eds., A 60. By contrast no examples of the cross-hatch style, Survey of Persian Art from Prehistoric Times to the style B, or the beveled style, style C, are yet known Present, 4 vols. (Oxford, 1938-39), 1407, fig. 512. dating from before the foundation of Samarra. R. D. McChesney, "Four Sources on the Building of Isfahan," Muqarnas5(1988): 103-35, quotes the 61. It is of course possible that the decoration of the Nuqawit al-Athir of Mahmfid b. Hidayat Allah Bab al-'Amma is later than the building itself. In Natanzi on the purpose of the Maydan-i Shah: fact, one would expect the building to have been 'T"For polo (chtigan-bz1) and horse racing (asb-tzi) redecorated during its occupation. the maidan was leveled." In a second phase, how- ever, the purpose of the maydan was changed. 62. Directorate General of Antiquities, Hafriyyat Samarr, 1936-39, 2 vols. (Baghdad, 1940). 75. Ibn Qutayba, 'Uyin al-Akhbar, 2 vols., ed. Brockel- mann (Berlin, 1900), 1:166; A. Mez, Die Renaissance 63. Herzfeld Archive, drawings D-1049 and D-1125. deslslams (Heidelberg, 1922),385. Mez incorrectly noted the width as 60 meters. The caliph's maydan 64. It is assumed that a service department of the was no doubt larger than those of ordinary nota- palace would require a majlis for the chief which bles. would be impressive in proportion to the individ- ual's importance, and that the work areaswould lie 76. For a detailed discussion of the sport of horse behind and perhaps out of sight. An alternative racing, see Northedge, "Racecourses at Samarra." explanation might be that this complex is the residence of someone close to the caliph, although 77. In Northedge, "Racecourses at Samarra," the fig- this building does not have any apartments for the ure of 10,500 m was quoted as the length of accommodation of dependents. racecourse 3. The new figure of 10,420 m is based upon electronic measurement of the same map 65. Viollet, Fouilles Samara, pl. III. source dated 1917, cf. Northedge, "Racecourses at 158 ALASAIR NORTHEDGE

Samarra," n. 87. Nevertheless the new figure seems possibility that the usage of Dar al-Khilafa at to be more correct, for it matches a corrected Samarra is a back projection from later times measurement for racecourse 1of 10,417 m. If these seems to be excluded by its use by a contemporary distances represented 20,000 cubits, then the cu- author such as al-Yacqfibi. bit was 0.52 m. 97. Al-Ya'qfbil, Buldan, 261. 78. Y6 is one of a group of five identical structures in the steppe which are oriented to the qibla and 98. Al-Yacqlbi, Buldan, 261. would appear to be musallas. 99. Cf. Tabari, Tnikh al-Rusulwa 'I-Muliik, 3:1350, for 79. Northedge, "Racecourses at Samarra." an exception.

80. Viollet, Fouilles d Samara, 25, pls. XIV, XXI. 100. Tabari, Tiiikh al-Rusul wa '-Mulfik, 3:1503, 1788. The mazilim was the law court presided over by 81. The buttresses were presumably 20 cubits in diam- the caliph. eter (20 x .52 m = 10.4 m). 101. Al-Ya'qbil, Buldan, 255. 82. Demolition seems to be the most logical explana- tion for the disappearance of the east end which 102. TabarT, Trikh al-Rusul wa'l-Mulhk, 3:1350-51, appears on Herzfeld's plan and seems to be con- 1503. firmed by examination of the aerial photographs. Regrettably I did not address this question during 103. Dominique Sourdel ("Questions de c6r6monial a visit to this site in 1989. abbaside," Revue des itudes islamiques 38 [1960]: 121-48) must be right in saying that the Bab al- 83. This plan is paralleled in a building in al-Jacfari cAmma was not itself a reception room. (A188), but not in any of the excavated buildings. 104. Tabari, Tanikh al-Rusul wa'l-Mulihk, 3:1230. 84. Cf. above, description by Ross, 'Journey from Bagh- dad." 105. Tabarl, Taikh al-Rusul wa'l-Mulfik, 3:1433. The Bujja were from Nubia. Cf. also Ahmad b. Abi 85. E.g., Hammfdi, "Qasr al-khalifa al-Mutasim"; Yacqfib b. Wad.ib al-Yacqfibi, Thankh, ed. M. Hout- HayanT, "al-Hlir." sma, 2 vols. (Leiden, 1883), 2:480, for the arrival of the rebel Abu Harb al-Mubarqac from Pales- 86. P. Schwartz, Die Abbasiden-Residenz Samarra. Neue tine. historisch-geographische Untersuchungen (Leipzig, 1909). 106. Tabari, Tanikh al-Rusul wa'l-Mulfk, 3:1318. The bodies are said to have been crucified after execu- 87. A. A. N. al-CAni, "Istadrikat trikhiyya li-maw5qi' tion. athariyya III," Sumer39 (1983): 261-66 (Ar. sect.). 107. Tabari, Tarikh al-Rusul wa'l-Muluhk, 3:1811. 88. Al-YaCqfibi, Tanikh, 2:473. 108. Tabarl, Taiikh al-Rusul wa'l-Mulih, 3:1874. Cf. 89. Tabarl, Tinkh al-Rusul wa'l-Mulk, 3:1787-88, also Tabari, Tanikh al-Rusul wa'I-Muliik, 3:1419, 1823-24. 1434, 1522, 1722-23, 1859.

90. Al-Ya'qibi, Buldan, 261. 109. The sinister reputation of the Bab al-'Amma may also explain why it is that the fired bricks of the 91. Tabari, Tii kh al-Rusul wa'l-Mulik, 3:1350, 1788. triple iwan have not been robbed out along with the rest of the palace. 92. Tabari, Tiinkh al-Rusul wa'l-Muliik, 3:1706. 110. Tabarl, Taiikh al-Rusul wa'l-Mulfk, 3:1504. 93. Tabari, Tankh al-Rusul wa'l-Muluik, 3:1804, 1807. 111. Tabari, Tiinkh al-Rusul wa 'I-Muliih, 3:1503, 1505, 94. Tabari, Taiikh al-Rusul wa'l-Muluk, 3:1383. 1722-23, 1821, 1823-24.

95. Tabari, Tanikh al-Rusul wa'l-Muluk, 3:1787-88, 112. Al-Ya'qfib, Buldiin, 261. 1820-22, 1823-24. 113. Sourdel, "Questions de c6r6monial," 126-27. 96. Lassner, Topography of Baghdad, 85-91. The There is no sign of a monumental avenue on the AN INTERPRETATION OF THE PALACE OF THE CALIPH AT SAMARRA 159

east side of the palace, a possibility suggested by 'Izz al-Din ibn al-Athir, al-Kamilfi al-Taikh, ed. C. Sourdel. J. Tornberg, 14 vols. (Leiden, 1866-71), 6:108.

114. The principal arguments for proposing that the 127. Tabari, Tanikh al-Rusul wa'l-Mulfik, 3:1716. garden is original are (1) what did the Bab al- cAmma look out onto if the garden was not there? 128. Tabari, Taikh al-Rusul wa'l-Muliihk, 3:1791. (2) There is no evidence that it replaced other construction, and it is well adapted to the archi- 129. Tabari, Tainkh al-Rusul wa'l-Mulfik, 3:1307. tecture of the palace. (3) The palace of Balku- wara, which is a single-phase building, was con- 130. Tabari, Taikh al-Rusul wa '-Mulikt, 3:1507, 1545. structed with a garden facing onto the Tigris (Creswell, EMA, fig. 214). 131. Tabari, Tanikh al-Rusul wa'l-Mulfik, 3:1545, 1668.

115. Al-Yacqfibl, Buldan, 260. 132. Tabarl, Tainkh al-Rusul wa'l-Mulfk, 3:1831.

116. Tabari, Tafnkh al-Rusul wa'l-Mulfik, 3:1503. 133. Cf. also "al-Jawsaq and other maqasir" (Tabari, Taikh al-Rusul wa'-Muliik, 3:1820); "al-Jawsaq 117. Al-Yacqfibi, Buldan, 258. There is also a mention and the palaces of the Caliphate" (al-Yacqfibi, of "al-Jawsaq al-Ibrahimi," which cost 2 million Buldan, 267). dirhams (Abfi al-Faraj Ali b. H.usayn al-Isfahani, Kitab adab al-ghurabi', ed. S. Munajjid [Beirut, 134. The hunting reserve east of the city. The area 1972], 47-50), and "al-Jawsaq fi Maydan al-Sahn" seems later to have lost its specific function, but or "al-Sakhr," which cost 500,000 dirhams (Al- the name remained in use. Isfahani, Kitab adabal-ghurab&; Yaqit, Mu'jam al- Buldiin, s.v. Samarra). Both constructions are at- 135. Tabari, Tarikh al-Rusul wa'l-Mulfik, 3:1787-88. tributed to al-Mutawakkil. 136. Tabarl, Tiifikh al-Rusul wa'l-Mulhk, 3:2040; al- 118. Tabari, Tafikh al-Rusul wa'l-Mulik, 3:1881. YaqibT1, Tainkh, 2:473.

119. Tabari, Tarnkh al-Rusul wa'l-Mulfk, 3:1307. 137. The evidence of relationship with racecourse 1 and Tell al-'Alij shows that site X must be early in 120. Al-Yacqfibl, Tankh, 2:478. date (Northedge, "Racecourses at Samarra").

121. Al-Yaqfibi, Buldan, 265; Tabari, Tahikh al-Rusul 138. Al-Isfahani, Adab, 47-50; Abu al-Hasan Al b. wa'-Mulfk, 3:1446. Muhammad al-Shabushti, Kitab al-Diyarat, ed. Gurgis Awwad, 2nd ed. (Baghdad, 1966), 170. 122. Abf Muhammad Ahmad ibn Actham al-Kffi, Kitab al-Futuih, 8 vols. (Haydarabad, 1975), 8:354; al- 139. Tabarl, Tanikh alRusul wa 'l-Mulfik, 3:1788, 1791, Khatib al-Baghdadi, Tanikh Baghdad, Cairo, ed. 1823-24. 2:121. Already doubted by Oleg Grabar ('The Earliest Islamic Commemorative Structures: Notes 140. Tabari, Taiikh al-Rusul wa'-Mulik, 3:1787-88, and Documents," Ars Orientalis 6 [1966]: 7-46) 1820-24. and Sheila Blair ('The Octagonal Pavilion at Na- tanz," Muqarnas1 [1983]: 69-94), Herzfeld's view 141. Tabarl, Taiinh al-Rusul wa'l-Mulh, 3:1479. that the Qubbat al-Sulaybiyya was the mausoleum of al-Muntasir now seems definitely incorrect 142. Al-Mutawakkil was said to have neglected his pub- (Friedrich Sarre and Ernst Herzfeld, Archiologische lic duties in the weeks before his murder (Tabari, Reise im Euphrat-und Tigris-gebiet, 4 vols. [Berlin: Taiikh al-Rusul wa'l-Mulfik, 3:1453-54). D. Reimer, 1911-20], 2:86; Creswell, EMA, 285). 143. Regrettably Sur Ashnas is not published. For the 123. Al-Yacqfibl, Buldiin, 267. Turkish cantonments of al-Karkh, see A. North- edge, "Archaeology and New Urban Settlement 124. Tabarf, Tdarfkh al-Rusul wa 'l-Muluk, 3:1711, in Early Islamic Syria and Iraq," in Studies in Late 1823. Antiquity andEarly Islam, ed. G. R. D. King and A. Cameron, vol. 2, Settlement Patternsin the Byzantine 125. Al-Yacqfibi, Buldan, 267; Tabari, Taiikh al-Rusul and Early Islamic NearEast (forthcoming). wa 'l-Mulik, 3:2040. 144. Sourdel, "Questions de ceremonial," 126; Tabard, 126. Tabari, Tikh al-Rusul wa'l-Mulk, 3:2223-24; Taiikh al-Rusul wa '-Muliik, 3:1496.