An ʿAbbāsid Fishpond Villa Near Makkah

by Terry Allen

published by Solipsist Press, Occidental, California, 2009

Copyright © 2009 by Terry Allen

An electronic publication     ISBN 0-944940-10-2

Table of Contents

The Villa at ʿAlwīyah
Formal Parallels for Pavilions in Pools
Roman Fishponds
Fishponds in the Islamic World and Norman Sicily
ʿAlwīyah's Reservoir as a Fishpond
ʿAlwīyah and ʿAbbāsid Residential Plans

Figure 1. ʿAlwīyah, reservoir, to south; stucco door frame.

In 1977 an archaeological survey team directed by James Knudstad, on which I served, investigated the ruins of a villa at a site called ʿAlwīyah approximately forty kilometers east-northeast of Makkah. It can be dated to the late second/early ninth century by the remains of carved stucco door frames found in one of its vestibules, which resemble stucco found at Raqqah.1 The villa does not share the plan of the lodges built under Hārūn al-Rashīd along the Darb Zubaydah, the pilgrim road from Iraq to Madīnah and Makkah (one such lodge, at Umm al-Ḍamīrān, is nearby).

The Villa at ʿAlwīyah

ʿAlwīyah has stuck in my mind partly for its stucco and partly because of the unusual reservoir that occupies much of its area. I think it reasonable to call it a villa, as “palace” implies a grander structure and its rustic setting and reservoir justify the use of the ancient term. Only recently have I found an explanation for its form.


Figure 2. ʿAlwīyah, plan (after Atlal; surveyed by James Knudstad; inked drawing by Fatḥī Fedā).

The main body of the villa, which is built of plastered rubble. measures about eighty-five meters north to south by forty meters east to west. It sits on the flank of the hills along the east side of the Wādī Shāmīyah overlooking the Darb Zubaydah in the wādī bed below. The villa has two formal entries, both on the wādī side and both emphasized by nonfunctional quarter-round buttresses. The northern entry opens onto a vestibule that gives access to a corner of a colonnaded courtyard with rooms ranged around it. The southern entry opens onto another vestibule, where the stucco door frames were found. This vestibule provides a similar bent entry into a corner of the northern courtyard through an intermediary room and a direct entry into a corner of a colonnaded courtyard that is mostly occupied by the reservoir. The northern courtyard and the larger southern, reservoir courtyard are separated by a formal wing. In the center of this wing is a small colonnaded courtyard, wider east to west than north to south, which is accessible through arcades on piers to the northern and southern courtyards and through doors to two flanking rooms. These flanking rooms are rectangular, the long sides of which run (roughly) north and south, also have doors in their short sides, opening onto both courtyards. A similar room lies to the east, communicating with yet another room too ruined to determine how it fitted into anything farther east.

Ruins of various subsidiary or secondary structures lie attached or adjacent to the villa. A floor of baked bricks, fragments of opus sectile, and fragments of glass molded in a honeycomb design were found in the main body of the villa.


Figure 3. ʿAlwīyah, reservoir to southeast and to west.

The reservoir itself is a trapezoid, about twenty-eight meters east to west by twenty-four meters north to south. It has interior buttresses like public reservoirs of the period along the Darb Zubaydah, suggesting it has considerable depth, but it was not excavated to determine how deep it actually is. Within it is a causeway about two meters wide on transverse barrel vaults extending south from the esplanade between the northern colonnade and the reservoir itself to a double octagonal structure in the center of the reservoir. Both the causeway and the octagonal structure rise to the level of the top of the reservoir. The octagonal structure, which is about ten meters across, consists of two twenty-centimeter-thick octagonal walls with arched openings in their faces,2 separated by about two meters.

The causeway and double octagon in the center of the reservoir suggest that a pavilion may have been built there, although how it might have been constructed (of wood? as a tent?) or why its foundations should be doubled are questions that were not investigated in 1977.

Formal Parallels for Pavilions in Pools

While a pavilion centrally located in an architecturally defined space is not unusual in Islamic architecture few early examples of a pavilion (or any structure) in a reservoir survive in fact or in textual descriptions. A notable and very odd example is the pavilion shoehorned into a shallow basin in the forecourt of Khirbat al-Mafjar.3 More conventional is the arrangement at Qayrawān, where ʿAbd Allāh b. ʿAbd al-ʿAzīz al-Bakrī (d. 487/1094) described an Aghlabid (mid-third/ninth century) cistern:

[It] is round and extremely large. In the middle of it is an octagonal tower (ṣaumaʿa) on the top of which is a qaṣaba (? pavilion [thus K.A.C. Creswell]) for observation opening with four doors (bāb).4

The pavilion was reached by boat.

A description of a pavilion in a reservoir (buḥayrah) at a fifth/eleventh-century palace at Toledo is preserved by Shihāb al-Dīn Abu'l-ʿAbbās Aḥmad al-Maqqarī (ca. 986–1041/1577–1632). It was supposedly made of glass and water poured down its outside from an outlet at the top.5

Later platforms or pavilions in reservoirs are not uncommon in India.6

Roman Fishponds

I believe an explanation for the features of the ʿAlwīyah reservoir may be found in Roman architecture. James Higginbotham has published a survey of Roman fishponds built in Italy dating from the late Republic and the first century A.D., when he believes they were most numerous. They are divided between fishponds built at the edge of the sea, generally enclosing part of the sea itself, and those constructed inland.7 Some Roman fishponds possessed central features.

A Roman author, Lucius Junius Columella (4–70 A.D.), in a treatise on pisciculture,8 advocated the construction of recesses in the walls of the fishpond to provide shade. Shade was also provided by

structures attached to the sides of fishponds so that they project out into and cover part of the enclosure [as] can be seen in the piscinae at the so-called Villa of Quintilius Varus near Tivoli, the [seaside] Piscina di Lucullo near Circeo, and the pond located in the domestic quarter at Paestum. At Tivoli and Circeo platforms are supported by small concrete barrel vaults, which project out toward the center of the enclosure. The submerged tunnels, created by the vaults, provided shade for the fish as well as supporting a platform for viewing the pond and its inhabitants.9

These formed rectangular platforms.10

Closely placed parallel walls in the fishponds at Torre Valdaliga and La Mattonara may have supported platforms that stood over portions of the ponds. These broad platforms could have afforded better views of the ponds while providing shade for fish confined in the enclosure.

On a smaller scale, the ponds at the so-called Villa of Horace near Licenza, the Villa of Quintilius Varus at Tivoli, and the Piscina di Lucullo were equipped with small platforms that projected toward the middle of the enclosures. From these platforms, visitors could enjoy the view from a slightly better point of vantage. While providing similar protection to the fish these features also serve to emulate the larger and more prestigious seaside piscinae.11

These seaside fishponds could be quite grand and quite complex (in part in order to accomodate management of supplies of water both fresh and salt). At Torre Astura

a large maritime villa is located on the point [Punta di Astura] and connected by a bridge to an artificial island offshore. … The residential buildings on the shore were connected to an artificial island fashioned from hydraulic concrete. The island, to the south, was reached by way of a concrete bridge, approximately 130 meters long, which carried a roadway (4 meters wide) and the channel of a freshwater aqueduct. The bridge led to a roughly rectangular terrace (82 meters square) supported on concrete vaults. This part of the complex was positioned in the center of the north side of a large rectangular enclosure given over to the raising of fish.12

This rectangular enclosure had at least nineteen tanks and was about 170 by 125 meters.

The most prominent Roman fishpond, easily visited today, is in the Domus Augustana on the Palatine Hill in Rome:

The piscina in the upper peristyle is the largest structure in the palace constructed to hold water. The pond is basically rectangular in plan measuring approximately 32 by 24 meters. The two corners of the side away from the private quarters of the palace (the northeast) are rounded while the other two are square. The enclosure was constructed in brick-faced concrete lined with cocciopesto [mortar with crushed ceramic inclusions] and then revetted with marble. The remains of marble flooring near the walls indicate a depth of just over 1 meter. The sides of the pond were provided with a continuous series of recesses that consisted of a repeating pattern of semicircular, triangular, and rectangular niches. The entire arrangement consisted of at least eighty recesses around the interior of the pond. This design, while aesthetically appealing, provided retreats for the fish. The pond, in the center of the upper peristyle, was constructed as part of [the architect] Rabirius's Flavian palace, which was inaugurated in A.D. 92.

A second phase of construction saw the addition of an island, which was connected to the northeast side of the pond by a bridge. The island and the connected bridge were positioned slightly off center lying closer to the southeast side. Both structures were built of concrete faced with courses of brick. The bridge is 7.4 meters long, 1.05 meters high, 2.4 meters wide, and supported by seven arches. The piers of the bridge sit right on the pond floor with no foundation. The solid concrete island measures 10.1 by 9 meters and is 1.45 meters high. Foundations on the island appear to have supported a small temple or aedicula. This addition is dated by a brick stamp, which provides a terminus post quem of the late third or early fourth century A.D.13

Internal divisions were used to control the movement of fish and to segregate populations; these included “vertical slots [that] would have secured movable grates (cancelli) [that] could be slid up and down to allow fish to transit”.14

Higgenbotham emphasizes that fishponds (at least those sufficiently preserved that he could identify them as such) were commonly built in association with villas, especially at the seaside:

The location[s] of these residences were often chosen to exploit the panoramic vistas or to take advantage of the natural coves and inlets along the shore. The artificial fishpond, with its perimeter moles reaching out and enclosing large areas of seascape, reflects the desire to control nature and to include plants, animals, and fish within the confines of the villa property.15

There are examples of fishponds located close to dining areas,16 surely to reinforce the thought that the wealthy owner himself provided the fish the diners were enjoying.

It is in their close association with dining and display that inland fishponds most resemble and imitate the larger seaside piscinae. The public and private vantage in both settings served to emphasize the social station of the fishpond owner.17

Higgenbotham believes that fishponds were particularly popular in the villas of the wealthy during the period he studied (the late Republic and the first century A.D.):

These two centuries encompassed a time in Roman history when the open competition among the elite of the Republic led to civil wars and eventually gave way to the autocratic rule of the Empire. During these years the possession and operation of an artificial fishpond were indicative of great wealth and symbolic of almost royal status. The gradual decline in the popularity of the large seaside fishpond in the first century A.D. parallels the consolidation of power under the emperors, who discouraged the conspicuous display of wealth and status common in the Republic. Smaller freshwater ponds, which become common during the first century A.D., belong largely to those of lower station seeking to emulate the image of success created by the piscinarii of the late Republic.18

The resources expended on the construction and operation of Italian piscinae were beyond the grasp of most individuals and, therefore, the possession of an artificial fishpond was a mark of singular status. Consequently, the piscinarii of the late Republic drew considerable interest and excited the contemporary imagination.19

From contemporary sources Higgenbotham argues that while fishponds could have been practical sources of fish for their owners, seaside fishponds connected with villas were too expensive to be explained on practical grounds alone. He is unconvinced that such fishponds produced for the market, but it appears from his evidence that such may have been the case, as there was high demand for fish among the wealthy.20 Nevertheless, aesthetic pleasure and social status must have been important reasons for building and maintaining a fishpond:

Most of the fishponds in Italy appear to have been designed so that effective fish raising could be coupled with the pleasure of owning a private body of water replete with aquatic life21

including even pet fish. Higgenbotham sees inland fishponds as smaller imitations of the seaside variety, but there seems to be no reason to give precedence to the latter.

Roman fishponds could be accessible to both the more public and the more private areas of a residence, serving to impress visitors.22

Fishponds in the Islamic World and Norman Sicily

There must be many references in early Arabic literature to fish in water and ponds. The Birkah al-Jaʿfarīyah at Samarra was eulogized by the poet al-Walīd b. ʿUbayd al-Buḥturī, who mentioned fish swimming in it.23 Aḥmad Sūsah locates the Birkah al-Jaʿfarīyah at the palace known as al-Musharraḥāt, according to Alastair Northedge, who finds “no particular reason to locate” it there.24 There is a reservoir about two hundred meters square at al-Musharraḥāt, which Northedge plausibly characterizes as a hunting palace because of the adjoining game reserve in which the reservoir itself is located; while Northedge thinks it “doubtful that the basin was a necessary requirement of such a hunting park” the rationale for it he suggests is that “perhaps the caliph thought that it would be possible to watch the animals coming to drink”.25 Wherever it was, according to Stefan Sperl's translation of al-Buḥturī the Birkah al-Jaʿfarīyah had “a wide basin in its lower regions … and ample space in its upper reaches”, a statue of a dolphin, and dakkatayn facing each other, which Sperl translates as “two esplanades”. These two dikkahs (“benches”) might be an attempt to describe platforms projecting into the birkah.

Al-Maqqarī mentions fishponds (buḥayrāt) containing fish (ḥītān) at Madīnah al-Zahrā' and that the fish were fed bread and chickpeas, but does not describe the shape or features of the ponds.26

A reservoir that very likely was a fishpond was excavated near Cordoba in 1910 by Ricardo Velázquez Bosco, who published it in 1912 as being from the Umayyad estate known from texts as the Munyah al-ʿĀmirīyah. This identification was not generally accepted, and in a 1984 article Manuel Ocaña Jiménez proposed to identify the site as the fourth/tenth-century Munyah al-Rumānīyah.27

This reservoir, built of ashlar, is trapezoidal, nearly fifty meters long, an average of twenty-eight meters wide, and four meters deep; it could be filled to a depth of 3.85 meters It might well have been filled to a somewhat shallower level. The reservoir's remarkable feature is a walkway all around its circumference, corbelled out over the body of the reservoir on a series of shallow barrel vaults.

Velázquez Bosco thought this walkway implied recreational use and likened the reservoir to others in Persian palaces, in the Alhambra (the Court of the Myrtles and the Partal), the fishponds at Madīnah al-Zahrā', and that in the Dār al-Baḥr in the Qalʿah of the Banū Ḥammād. Ruggles thinks that it could be a fishpond on account of its depth. While its depth is indeed favorable for pisciculture the reservoir's corbelled walkway in particular was a pointless exercise unless it was meant to provide shade for fish or inhibit them from jumping out of the pond.28

In Palermo the twelfth-century Norman kings built suburban garden residences tightly integrated with reservoirs, among them the Cuba, which projected into a reservoir. This reservoir has been identified as the fishpond a sixteenth-century literary source states was at the the Cuba. The residence block stood bordered by water on three sides, with a small pavilion, the Cubula, standing as an island within the reservoir. In front of the Zisa lay a reservoir with a rectangular pavilion reached by a short bridge; a fishpond at the Zisa is mentioned by a contemporary source and this reservoir, too, was identified as a fishpond by a sixteenth-century source.29 The Dār al-Baḥr at the Qalʿah of the Banū Ḥammād in Algeria is often cited in connection with the Norman residences as it had an enormous basin, but I know of no evidence that it was used as a fishpond.

ʿAlwīyah's Reservoir as a Fishpond

Here, in Roman and Islamic fishponds, is an explanation for the causeway built on arches and the octagonal central feature in the reservoir at ʿAlwīyah. Although the causeway might well have been built on arches instead of without them to allow for better circulation of water even if it had not been intended to sit within a fishpond, the central feature seems particularly well designed to provide fish with shade. There need not have been a pavilion built above these concentric octagons—although some relief from shade for humans admiring fish would have been most desireable—but as the octagons are too narrow to walk on there must at least have been some flooring connecting them, perhaps in the form of wooden planks; the center could have been left open.

There was no lack of water at the site. Remains of substantial hydraulic infrastructure were recorded in 1977 along the Wādī Shāmīyah, including dams, reservoirs to serve pilgrims, and aqueducts along the side of the wādī. Reservoirs built to support the Darb Zubaydah in more open areas are supplied with water only by flash floods which wash into them all recent surface deposits from a wide area; as a result their water is permanently turbid, but ʿAlwīyah seems to have had purer water from better sources.

The point of ʿAlwīyah, which must have belonged to someone wealthy and therefore presumably important, seems to have been the southern, reservoir courtyard, which could be entered from the northern courtyard through the formal wing or more directly from the south entrance to the villa. While my illustrations show that today (or at least in 1977) one could look down into the wādī from the reservoir, of course originally the courtyard, which was walled, would have had no such view. The point of the octagonal structure must have been to provide a viewpoint for enjoying the reservoir, not the landscape outside. It would also have been a place where fish might congregate and be caught. The social value of possessing such a fishpond would have been just as it had been in Roman Italy: “reflect[ing] the desire to control nature and to include [it in the form of fish] within the confines of the villa”. The ability to produce fresh fish so far from fresh water would have both impressed the owner's guests and perhaps catered to his own tastes.

Establishing and operating a fishpond at ʿAlwīyah would have presented certain practical problems, although provision of fresh water is not among them in this location. There are fresh-water fish in Arabia,30 and they could have been transported to the site in water bags. As for feeding the fish, according to Higgenbotham among the foodstuffs Columella recommended for fish was dried figs;31 dates are plentiful in the Wādī Shāmīyah.

Keeping the water fresh would not have been a problem, as the reservoir has inlets and outlets. Protection from too much sun, even with the central octagonal feature, might have been the greatest problem. It would be interesting to discover the depth of the reservoir as built (on the shallow side the Roman fishpond at Grottarossa, which has been restored, is only 1.2 meters deep32) and whether there is any trace of other devices for providing shade, such as recesses built into the sides or even provision for an awning. Once established the fish could be raised as a self-sustaining population. Of course there is no telling whether such a scheme would have succeeded, but it might well have been planned and built nonetheless.

ʿAlwīyah and ʿAbbāsid Residential Plans

My primary interest in writing up the connection I see between ʿAlwīyah and Roman fishponds has been to draw attention to an unusual building that caught my fancy, but in looking for parallels in ʿAbbāsid architecture I discovered that it may be possible to use ʿAlwīyah to prise a bit more information out of other ʿAbbāsid residential plans, although in an indirect way.

To deal first with one misleading similarity, the arcaded southern wing of the building surrounding the reservoir at al-Rumānīyah might be adduced as a parallel to the formal wing at ʿAlwīyah, especially as (from their plans, anyway) both can be thought of as screens between two areas, but instead of separating two enclosed courtyards the southern wing at al-Rumānīyah separated the reservoir courtyard from the upper terrace of the garden and afforded a view as well as a pleasant breeze, unlike the formal wing at ʿAlwīyah.

The key to placing ʿAlwīyah within ʿAbbāsid residential architecture is not its reservoir but the overall form of the building, with its two courtyards, the northern one surrounded by rooms and the other containing the reservoir, separated by a distinctively planned formal wing. This plan exists in contemporary palaces at Raqqah and several generations later at Samarra. The remains at Raqqah have been analyzed recently by Ulrike Siegel, who classifies the various tracts of the buildings quite reasonably as reception, living or service, and garden or courtyard areas (Empfangsbereiche, Wohn-/Nebenbereiche, and Garten-/Hofbereiche; I interpret Nebenbereiche freely).33

The Samarran parallels to the formal wing between the courtyards at ʿAlwīyah are somewhat more obvious, so I shall discuss them first. They are found in the large- to medium-size houses, which have recently been published or republished handily by Thomas Leisten and Alastair Northedge: House 4, between courtyards 41 and 45,34 House 2 in H109, and House 11.35 In all cases three rectangular chambers, joined along their long sides, stand between two courtyards. In front of these groups, on both sides, stand arcades on piers. The thickness of the walls of these groups is the same as that of the rest of the construction. In House 2 of H109 one of the courtyards is surrounded by rooms but without a large reception hall and the other has rooms only at the far end. In House 11 neither courtyard gives onto any rooms except passageways. Northedge identifies the grouping of rooms between the courtyards as a “reception block”, which is no doubt correct, as there must be reception chambers and there are no other candidates for them.

Grander houses and palaces at Samarra have courtyards separated by formal wings in the same fashion as those just discussed, but with more familiar reception chambers. Here I can discuss only selected examples. In Houses 1 and 3 the wings between the courtyards nearer the entrance and the farther courtyards have an additional, internal arcade in front of the central chamber, which is open on the side toward the entry. The combination is common at Samarra; Northedge calls it a “T-īwān”.36 An additional chamber has been added at each side of the whole arrangement.37 These reception chambers point clearly to the courtyard nearer the entry as being the more public and formal one, to which formal display was more necessary.

With these examples in mind, it is easier to see that the same pattern of division of space occurs at Raqqah, although at Raqqah the form of the reception areas is more varied and nearly all the plans are composed without regard for the symmetry of the complete building.

In Palace B, beyond the forecourt, the plan is split in two, with living quarters and courtyards on the east and a reception suite between a more formal courtyard (decorated by applied pilasters) and a rear courtyard that contains no rooms and does not communicate with the adjacent rear courtyard of the living quarters.

Palace C is divided in thirds north-to-south, with living quarters in the middle third and a poorly preserved reception suite near the middle of the western third, between two courtyards. The eastern third is open.

The plan of Palace D is obviously incomplete (lacking the complete exterior wall) but can be restored in the imagination to resemble the Samarra palaces: a reception suite of three chambers divided by narrow halls and backed by a large portico stands between a known courtyard and another that must be inferred from the openness of the portico, which must have faced an enclosed space.

The plan of the West Palace is a bit more complex than those of the previous examples and is also incomplete on the south end, but the eastern half of the main body of the palace exhibits the same arrangement of reception suite fronted by a courtyard, with another very likely behind it to the south (cf. Palace B). On a much reduced scale this arrangement occurs again in the East Palace.38

By analogy with palaces or large houses at both Raqqah and Samarra, what I have called the formal wing of ʿAlwīyah must have been used as a reception area by the owner of the villa. It could be so used for guests approaching from the northern courtyard or for those who entered the southern courtyard directly from the southern vestibule. For the comparable Samarra buildings no separate entries into the farther courtyards appear to exist (although there is something wrong with the plan of House 4, and it is difficult to see just how it was to be entered by guests of status).

The distinctive difference between the two courtyards at ʿAlwīyah, which is not apparent at Raqqah and Samarra, is that one of them was equipped with infrastructure for what can be characterized broadly as outdoor entertainment. So, too, it may have been at Raqqah and Samarra. I think these pairs of courtyards were meant to accomodate two different, although overlapping, classes of people. These classes can be characterized as visitors, who had to be dealt with as part of the villa owner's function as patron, and guests, who were invited to be entertained. I think it reasonable to conclude that the courtyards nearer the entry (at Raqqah and Samarra; at ʿAlwīyah, the northern courtyard) were used for the most public occasions, attended by the widest range of visitors, and the farther courtyards (at ʿAlwīyah the southern, reservoir courtyard) for more private affairs, attended by more rigorously selected guests.39 At Raqqah and Samarra these guests would necessarily have entered through the more public courtyards; at ʿAlwīyah there was a short-cut (the southern vestibule). The obvious infrastructure for outdoor entertainment in these farther courtyards at Raqqah and Samarra would be irrigated gardens.40

Returning to Samarra and the largest palaces there, the plan of the Dār al-Khilāfah has been inflated to the point of distortion but the sequence of entry–courtyard for visitors–formal reception chambers–farther courtyard is maintained. The farther courtyard (Creswell's “Great Esplanade”, Northedge's H302) is entered through a large portico and is divided by a canal into two sections. In the western section there are two axial features of some sort, and beyond the eastern section lies the “Small Serdab”, the maydān, and the racecourse.41 With due caution I suggest that one or the other, or both, sections of H302 is likely to have been planted as a garden. If so it might have been a forerunner of the Jauṣaq al-Muḥdith in al-Muqtadir's palace in Baghdad, to which Greek ambassadors were brought in 304–05/917:

Then the envoys passed to what was called the Jawṣaq al-Muḥdith (New Kiosk), which is a palace in the midst of gardens. In the center of it is a tank made of tin (raṣāṣ qalʿī), round which flows a stream in a conduit also of tin more lustrous than polished silver. This tank is thirty ells in length by twenty across, and round it are set four magnificent pavilions with gilded seats adorned with Dabīqī embroidery, and the pavilions are covered over with the gold work of Dabīq. All round this tank extends a garden with lawns where palm trees grow, and it is said that their number is four hundred, and the height of each is five ells. The entire height of these trees, from root to spathe, is enclosed in carved teak wood, encircled with gilded copper rings. And all these palms bear full-grown dates, which are ripe in almost all seasons and do not decay. Round the sides of the garden also are melons of the sort called Dastabūya, and also other species.

The ambassadors passed out of this palace, and next came to the Palace of the Tree, where [as has already been said] is a tree, which stands in the midst of a great circular tank filled with clear water. The tree has eighteen branches, every branch having numerous twigs, on which sit all sorts of gold and silver birds, both large and small. Most of the branches of this tree are of silver, but some are of gold, and they spread into the air carrying leaves of divers colors. The leaves of the tree move as the wind blows, while the birds pipe and sing. On one side of this palace, to the right of the tank, are the figures of fifteen horsemen mounted upon their mares, both men and steeds being clothed and caparisoned in brocade. In their hands the horsemen carry long-poled javelins, and those on the right are all pointed in one direction, as though each were attacking his adversary. …42

Evidently neither of these reservoirs was a fishpond, as no fish are mentioned even though it would have been appropriate to do so had they existed, and either the clearness of the water or the reflectiveness of the tin lining is emphasized. On the bank of the Tigris no fishpond was needed.

At ʿAlwīyah one element of outdoor entertainment was selected—perhaps the most impressive one in a dry land—and made the focus of the building. The gardens of ʿAlwīyah's owner might have been found in the nearby plantations he no doubt also owned.43 He had relatively little need for formal display here (the northern courtyard may even have been residential, though it does not appear organized for the purpose), and the result is a somewhat abbreviated plan with its single entertainment element disproportionately prominent. But that plan appears to have been derived from a series also represented in Raqqah and Samarra, and probably with the same functional identity of its parts.

Certainly not every courtyard behind every reception suite need have been planted as a garden, although when one considers the possibilities, what could have been a more attractive use of such spaces? The lodges along the Darb Zubaydah in the Hijaz, apparently contemporary with ʿAlwīyah, generally share a distinctive plan somewhat at variance with its two-courtyard plan. At Barūd, Umm al-Ḍamīrān, al-ʿAqīq, and possibly Kuraʿ and Maʿdan Bani Sulaym, a square enclosure with a small mosque outside its main entry (as at Raqqah, in Palaces C and D) is usually divided more or less symmetrically along the axis of the main entry into a set of courtyards with multiple entries and an īwān-hall in the central courtyard, beyond which are other courtyards with apparently residential suites. Given the known function of these buildings—to accomodate high-status travellers for short periods—it seems reasonable to interpret their farther courtyards as domestic spaces.44

Whether the reservoir at ʿAlwīyah was a fishpond or not—and again, what would its attraction have been otherwise?—and whatever the links between it and the Roman domestic fishponds it resembles, the real value of the fishpond villa at ʿAlwīyah is that it helps define the two-courtyard residence plan found at Raqqah and Samarra by clarifying its probable social functions and resolving the question of where the gardens of these residences were located.

1. The villa was published in Salah al-Helwa et al., “Preliminary Report on the Second Phase of the Darb Zubayda Reconnaissance 1397/1977”, Atlal, v. 2, 1978, pp. 51–64, pp. 53–55, as “al-ʿUlwiyya”; “ʿAlwīyah” was recorded in the field (possibly neither is a correct toponym for the ruins). Plans can be found on pl. 49 and 60. Working from the published coördinates (21° 39' N and 40° 6' E, p. 53) and my memory of the site I can find the general location of the ruins in satellite images. I believe that they are at about the location shown in the center of (decimal: latitude 21.6588 and longitude 40.1137). For the Raqqah stucco and comparison with ʿAlwīyah see Michael Meinecke, “Early Abbasid Stucco Decoration in Bilād al-Shām”, Bilād al-Shām During the Abbasid Period (132/A.H./750 A.D.—451 A.H./1059 A.D), Proceedings of the Fifth International Conference on the History of Bilād al-Shām, ed. Robert Schick and Muhammad Adnan al-Bakhit, Amman, 1991, pp. 226–67, p. 229–30).

2. The description of these features in the published report is garbled: I believe there are openings in each segment of both the inner and outer octagons.

3. Robert Hamilton, Khirbat Al Mafjar: An Arabian Mansion in the Jordan Valley, Oxford, 1959, pp. 110–21; idem, Walid and his Friends: An Umayyad Tragedy, Oxford, 1988, fig. 27, which leaves out the bridge Hamilton thought connected it at gallery level to the palace, p. 56.

4. Trans. Creswell, Early Muslim Architecture, Oxford, 1940, v. 2, pp. 289–90. For the text see the section of Kitāb al-masālik wa'l-mamālik ed. MacGuckin de Slane as Description de l'Afrique septentrionale, Algiers, 1857, p. 26.

5. Encyclopaedia of Islam, 2nd ed., s.v. “al-Maḳḳarī”; Shihāb al-Dīn Abu'l-ʿAbbās Aḥmad al-Maqqarī, Nafḥ alṭīb min ghuṣn al-andalus al-raṭīb, ed. Reinhart Dozy et al., Leiden, 1855–61, v. 1, p. 347; D. Fairchild Ruggles, Gardens, Landscape, and Vision in the Palaces of Islamic Spain, University Park, Pa., 2000, pp. 147–48.

6. See Attilio Petruccioli, “Der Garten als Antizipation der Stadt: Parallele Entwicklungen”, Der islamische Garten, ed. idem, pp. 85–108, lower fig. on p. 101 (Fatehpur Sikri) and on right on p. 103 (Delhi); Elizabeth B. MacDougall, et al., The Islamic Garden, Washington, D.C., 1976, pl. 3 (Rajput Palace, Amber); Yves Porter, “Jardins pré-moghols”, Res Orientales, v. 3, 1991, pp. 37–53, fig. 3 (Mandu), and fig. 13 for an octagonal platform reached by a causeway at Kalyadeh.

7. Piscinae: Artificial Fishponds in Roman Italy (an admirably informative title), Chapel Hill, N.C., 1997. For the division between seaside and inland ponds see pp. 18–22.

8. De re rustica, volume 8, cited in footnotes without reference to edition and omitted from Higgenbotham's bibliography.

9. Higgenbotham, op. cit., pp. 25–26.

10. Ibid., fig. 62 and 63 show different reconstructions of the Piscina di Lucullo, in one of which the rectangular platform is reached by a causeway.

11. Ibid., p. 31; for Licenza pp. 128–31; for Tivoli pp. 122–25; for the Piscina di Lucullo pp. 152–57.

12. Ibid., pp. 143–44.

13. Ibid., pp. 120–22; cf. the so-called Grotto of Tiberius at Sperlonga, pp. 68, 159–63.

14. Ibid., p. 23.

15. Ibid., p. 31.

16. Ibid., pp. 31–33.

17. Ibid., p. 33.

18. Ibid., p. 40.

19. Ibid., p. 55.

20. Although there were also industrial-scale operations, as at Cosa, ibid., pp. 80–84.

21. Ibid., p. 57.

22. Ibid., pp. 33, 64.

23. Julie Scott-Meisami, “The Palace-Complex as Emblem: Some Samarran Qaṣīdas”, A Medieval Islamic City Reconsidered: An Interdisciplinary Approach to Samarra (Oxford Studies in Islamic Art, v. 14), ed. Chase F. Robinson, Oxford, 2001, pp. 69–78, pp. 70–71. The poem is transcribed and translated, with commentary, in Stefan Sperl, Mannerism in Arabic poetry, Cambridge, 1989, pp. 182–85 and 196–99. Amazingly, in the paperback edition of 2004, which I consulted, the Arabic text begins on p. 183, continues on p. 182, and concludes on p. 185, which order I assume reflects an error in the original printing that Cambridge University Press did not care to correct in the paperback version.

24. “The Palaces of the Abbasids at Samarra”, A Medieval Islamic City Reconsidered, pp. 29–67, pp. 55.

25. Ibid., p. 62. The plan, fig. 13, also appears (with scale) in Northedge, The Historical Topography of Samarra, London, 2005, fig. 90; the reservoir has not been excavated and only its outline is evident. The index entries for birka and specific birkas in this latter work point to discussions of other reservoirs at Samarra.

26. Al-Maqqarī, ed. Dozy, pp. 373–74.

27. Velázquez Bosco, Medina Azzahra y Alamiriya, Madrid, 1912, pp. 19–33; Ocaña Jiménez, “Las ruinas de ‘Alamiría’ un yacimiento arqueológico erróneamente denominado”, Al-Qanṭara, v. 5, 1984, pp. 367–81, with the identification pp. 376–79; the identification and Velázquez Bosco's find are reviewed (with illustrations from Velázquez Bosco) by Ruggles, op. cit., pp. 111–18. The absolutely latest discussion of al-Rumānīyah, which arrived as I was completing this article, is Felix Arnold et al., “Das islamische Landgut ar-Rumanīya bei Córdoba: Vorbericht einer Bauaufnahme”, Madrider Mitteilungen, v. 50, 2009, pp. 503–23.

28. For the latter purpose see Higgenbotham, op. cit., p. 24.

29. In no particular order: Hans-Rudolf Meier, Die normannischen Königspaläste in Palermo: Studien zur hochmittelalterlichen Residenzbaukunst, Worms, 1994, pp. 54–90, 148–49; Giuseppe Bellafiore, La Zisa di Palermo, Palermo, 1978, pp. 9, 16, 47–50, fig. 67; Paola Caselli, “Die Conca d'Oro und der Garten der Zisa in Palermo”, Der islamische Garten, ed. Petruccioli, Stuttgart, 1995, pp. 185–200.

30. Indeed, during the 1977 season we were told that a member of the Chinese Embassy staff had caught a fish in the mountains of the Hijaz.

31. Higgenbotham, op. cit., p. 34.

32. Ibid., p. 115.

33. “Frühabbasidische Residenzbauten des Kalifen Hārūn ar-Rašīd in ar-Raqqa/ar-Rāfiqa (Syrien)”, Madrider Mitteilungen, v. 50, 2009, pp. 483–502; fig. 10 shows her classification of tracts.

34. Leisten, Excavation of Samarra, Volume I: Architecture. Final Report of the First Campaign 1910–1912 (Baghdader Forschungen, v. 20), 2003, p. 144, fig. 91.

35. Northedge, Historical Topography, p. 126, fig. 51; p. 221, fig. 98; and cf. House 4, p. 129, fig. 52, and M18 and M19, p. 202, fig. 87, the latter two shown in sketch plan only.

36. Ibid., p. 127.

37. For Houses 1 and 3, ibid., p. 221, fig 98. The plan of House 1 must be incorrect in showing a mostly solid wall instead of an arcade on the side of the farther courtyard.

38. Siegel, op. cit., fig. 3, 4, 5, 7, and 8.

39. Herzfeld's characterization of houses with two courtyards as divided between Sarai and Ḥarīm ( Erster vorläufiger Bericht über die Ausgrabungen von Samarra, Berlin, 1912, p. 14, trans in Creswell, op. cit., p. 282), is belied by the absence of rooms adjoining some of the farther courtyards. See Siegel, op. cit., for further discussion of traditional typologies of palace plans.

40. Siegel, op. cit., classifies all of the courtyard–reception suite–courtyard sequences as reception areas; I think the farther courtyards crosscut her classification.

41. Northedge, Historical Topography, ch. 6; Creswell, op. cit., pp. 241–43 and plan between pp. 242 and 243 for the canal. Northedge writes “there is some evidence that the grand Esplanade was called al-Ṣaḥn, and a qubba was located there”, citing Muḥammad al-Ṭabarī, Leiden ed., v. 3, pp. 1706–08. The term ṣaḥn suggests an open courtyard, but working from the translation published by the State University of New York Press alone (v. 35, pp. 161–63; the passage is evidently that on p. 163, ll. 3–4), I do not see that the courtyard and pavilion mentioned can be identified with H302 securely. Perhaps Northedge's judgement is based on additional material in the sources. For the general disposition cf. Sūr ʿĪsā, Northedge, op. cit., pp. 125–27 and Leisten, op. cit., pp. 212–14.

Seton Lloyd, “Jausaq al-Khaqani at Samarra: A New Reconstruction”, Iraq, v. 10, 1948, pp. 73–78, thought the portico an incoherent afterthought (p. 77) but it appears in other palaces cited here; his reconstruction of the “Great Esplanade”, as lined with trees but otherwise mostly empty (on the model of the Maydān-i Shāh in Isfahan?) is itself an argument for the space having been planted as a garden: what would have been the point or use of this space had it been left open, when both a polo maydān and a racecourse lay near at hand?

A similar analysis for Balkuwārā is complicated by the fact that the double courtyard on its inland side, away from the river entry, is the entry from the cantonment within which it sits (Northedge, op. cit., p. 190, fig. 84), so that it has a different, two-sided circulation pattern than the buildings discussed here. Nevertheless, Herzfeld noticed that good soil had been added to the courtyard on the river side (Leisten, op. cit., p. 104); Leisten's doubts that the inland courtyards could have been gardens despite their cross-axial layout (p. 88, n. 25) are based on negative evidence. I do not understand Leisten's assertion that “there was no direct access by boat” to Balkuwārā (p. 84) as the riverside end of the garden courtyard has been eroded away. However, open courtyards to the east and a strictly enclosed garden courtyard to the west (river side) would suit the pattern I sketch here.

42. Guy Le Strange “A Greek Embassy to Baghdad in 917”, Journal of the Royal Asiatic Society, 1897, pp. 35–45; pp. 37–44, translating Abū Bakr Aḥmad, al-Khaṭīb al-Baghdādī (392–463/1002–71); for the text see Ta'rīkh Baghdād, 14 v., Cairo, 1349/1931, v. 1.

43. A brief survey of earlier landholdings, many of them specified by their water sources, is suggestive: Saleh A. el-Ali, “Muslim Estates in Hidjaz in the First Century A.H.”, Journal of the Economic and Social History of the Orient, v. 2, 1959, pp. 247–61.

44. All published in Atlal. Barūd: James Knudstad, v. 1, 1997, “The Darb Zubayda Project: 1396/1976. Preliminary Report on the First Phase”, pp. 41–68, p. 49 (Structure no. 4). Umm al-Ḍamīrān: ibid., p. 64 (Structure no. 1), shown on pl. 61. Al-ʿAqīq: al-Helwa, op. cit., p. 62 (Structure no. 1) l. 51. Kuraʿ: Neil MacKenzie et al., v3, 1979, “Preliminary Report on the Third Season of Darb Zubaydah Survey 1978”, pp. 43–54, pp. 46–47 (Formal Building Complex), p. 32. Maʿdan Bani Sulaym, ibid., pp. 47–48 (Structure no. 2), pl. 33–34.