|
[Contents.] [Subject Index] [Index of Names] [List of Illustrations] (In certain versions of this etext [in certain browsers] clicking on the image, will bring up a larger version.) (etext transcriber's note) |
OXFORD UNIVERSITY PRESS
LONDON EDINBURGH GLASGOW NEW YORK
TORONTO MELBOURNE BOMBAY
HUMPHREY MILFORD M.A.
PUBLISHER TO THE UNIVERSITY
PALACE AND MOSQUE
AT
UKHAIḌIR
A STUDY IN
EARLY MOHAMMADAN ARCHITECTURE
BY
GERTRUDE LOWTHIAN BELL
O X F O R D
AT THE CLARENDON PRESS
1914
TO MY FRIEND
DR. WALTHER ANDRAE
IN GRATEFUL RECOLLECTION OF HAPPY AND PROFITABLE
DAYS SPENT IN THE FIRST CAPITAL OF ASSYRIA
WHICH HAS BEEN REVEALED BY HIS
LABOUR AND RECREATED BY
HIS LEARNING
PREFACE
I have attempted in this book to bring together the materials, so far as they are known, which bear upon the earliest phases of Mohammadan architecture, to consider the circumstances under which it arose and the roots from which it sprang. No development of civilization, or of the arts which serve and adorn civilization, has burst full-fledged from the forehead of the god; and architecture, which is the first and most permanent of the arts, reflects with singular fidelity the history of its creators. Not only does their culture stand revealed in the crumbling walls which sheltered them and in the monuments raised for perpetual remembrance over their bones, but the links which bound them to that which had gone before are therein confessed, as well as their own contribution to the achievements of their predecessors, to mechanical skilfulness, to utility, and to beauty. It is the nature and the extent of this contribution which is of vital importance to the student, and it is this which lends to architecture its keenest significance. What, then, was the contribution of the first builders of Islâm?
It must be confessed that the question admits of no very striking rejoinder. The Mohammadan invaders were essentially nomadic; their dwelling was the black tent, their grave the desert sands. The inhabitants of the rare oases of western and central Arabia were content, as they are to-day, with a rude architecture of sun-dried brick and palm-trunks, unadorned by any intricate device of the imagination, and unsuited to any but the simplest needs. Even the great national shrine at Mekkah, the sacred house of the Ka’bah, was innocent of subsidiary constructions. It is true that on the northern trade-route the rock-cut tombs of Madâin Ṣâliḥ and of Petra bear witness to a higher order of artistic impulse, but it was an impulse which borrowed its power from without, from Hellenized Egypt and from Hellenized Syria. If there were an indigenous Arabian architecture worthy of the name, it can only have existed in the southern limits of the peninsula, where as yet exploration has been too imperfect to afford data for argument, nor is there evidence to show that in the seventh century of our era it can have played a part in the development of the northern tribes. Upon the northern frontiers the influence of the Byzantine and of the Sasanian empires would seem to have been predominant, and when the invaders established themselves in provinces which had been ruled from Constantinople or from Ctesiphon, they employed Greek and Persian artificers to fulfil their newly developed requirements and to satisfy their newly developed taste for architectural magnificence. The palaces of the conquerors were planned, constructed, and adorned by those whom they had conquered; their learning and their civilization were borrowed from them; even the ritual of their faith was shaped by contact with older forms of worship. No more significant example of the debt which Islâm owes to alien races can be cited than that which is afforded by the history of the mosque. Out of the mud-built courtyard of the Arab house, the open space for domestic and tribal assembly, Greek and Persian builders created an architectural type which governed the whole Mohammadan world. And the only contribution of the masters for whom they worked was the demand for just such large and open spaces, easily accessible, oriented in a certain manner, and partially shaded from the rays of the sun.
It is therefore scarcely possible to say that a specifically Mohammadan art existed during the first century after the Flight, though its germs were latent in the welding together of Hellenized with un-Hellenized, or barely Hellenized, regions under a single hand. The architecture of the first century gives evidence of the formative character of this process of compression; before the third century had ended it may be said to have been completed. If the monuments of the first century are still a faithful reflection of earlier and foreign creations, they hold the promise of further and more definitely characterized growth. But in an age and in lands where change was slow-footed, older conceptions continued to hold the field long after the political conditions under which they had arisen had vanished or had been baptized with other names. As we now know, the Mesopotamian palace builders of the ninth century of our era were guided by schemes which their Sasanian forerunners had inherited from remoter times; while the mosque builders had advanced little beyond the plan laid down in the camp-cities of the conquest. But the interchange of workmen between East and West was continuous, the intercourse unbroken; and from that intercourse, coupled with the needs of the age and the prejudices of the Faith, the arts of Islâm were born.
In the present study my eyes have been turned chiefly, and necessarily, backwards. I have not been so much concerned with the offspring as with the parentage of the buildings which I have passed under review. Of these buildings the most important is the great palace of Ukhaiḍir on the eastern side of the Syrian desert. I have given, also, the first plans and photographs of three small ruins in its vicinity, Qṣair, Mudjḍah, and ‘Aṭshân. If they do not belong to the same period as the palace, they cannot be far removed from it in date. The problems presented by Ukhaiḍir led me back to Sasanian architecture, and I publish here new plans and photographs of two vast constructions at Qaṣr-i-Shîrîn. I have, further, taken this occasion to publish the plans of two mosques, the one at Diyârbekr, the other at Mayâfârqîn, both of which belong to a later period. The first of these has been known to us only through a sketch made by Texier, which I found to be inaccurate in many significant points, as it is also incomplete. The second has not previously been studied.
The palace of Ukhaiḍir was practically unknown until the winter of 1908-9, although it had been seen by European travellers as early as the seventeenth century. Della Valle passed by it in June 1625 on his way from Baṣrah to Aleppo, and described it as ‘a great ancient fabric, perfectly square, with thirteen pilasters or round columns on each side without, and other compartments of arches; within which were many chambers, with a court of no great bigness and uncovered. The Arabians call this fabric Casr Chaider. I could not conjecture whether it had been a palace or temple or castle; but I incline to believe it a palace rather than anything else.’[1] Pedro Teixeira’s account is doubtful. He says:[2] ‘At eleven in the morning we came to a dry channel which in winter they say has much water, and I thought it likely by the nature of its situation and capaciousness. Over it, on a rising ground, is still an ancient square fort, with twelve bastions, three on each side, made of burnt brick and lime, strong and well built. Without it, at about sixty paces distance, is a small Alcoran, or Tower, ten cubits high, tho’ it appears to have been higher, of the same structure, all decay’d with age; yet it appears to be a royal fabrick by its goodness and the place it stands in, where it could not be raised without mighty cost and much labour, and difficulty. It was done by an Arabian king, grandfather to Xeque Mahamed Eben Raxet, whom I said before I was carried to see, to secure the caravans going that way before the Turks possess’d themselves of Bagdat and Bazora. The Arabs call it Alcayzar or Kayzar, which signifies a palace or Cesar’s House, for so they call all that belong to kings and princes. This they reckon the half-way from Bazora to Mexat Aly, whither we were going. We found some small wells in this channel, the water of them clear and fresh, but of an intolerable ill scent, yet necessity prevail’d.’ The only item in this description which connects Teixeira’s palace with Ukhaiḍir is the name. Teixeira reached Meshhed ‘Ali (Nedjef) six days after he had passed by Alcayzar and he gives the situation of the palace as half-way between Baṣrah and Nedjef, whereas Ukhaiḍir lies to the north-west of Nedjef. There is no ‘Alcoran’, i.e. minaret, at Ukhaiḍir, neither could the building be described, even by the least careful observer, as a square fort with three bastions on each side. I am therefore inclined to suppose that there is another ruin called Ukhaiḍir further to the south. We need not linger over the derivation which he assigns to the name.
Scarcely more correct as to architectural features is Tavernier’s allusion to Ukhaiḍir. There can, however, be no doubt that it is to Ukhaiḍir that he refers, by reason of the geographical position of his ‘grand Palais’. Coming from Aleppo, he turned off at ‘Ânah into the desert and after some twenty days of journeying he observes:[3] ‘Cinq jours aprés que nous eûmes quitté ces deux familles Arabes, nous découvrîmes un grand Palais tout de brique cuite au feu; et il y a de l’apparence que le pays a esté autrefois semé, et que les fourneaux où on a cuit cette brique ont esté chauffez avec du chaume: car à quinze ou vingt lieües à la ronde il n’y a pas une brossaille ni un brin de bois. Chaque brique est d’un demi-pied en quarré et épaisse de six pouces. Il y a dans ce Palais trois grandes cours, et dans chacune de beaux bastimens avec deux rangs d’arcades qui sont l’un sur l’autre. Quoy que ce grand Palais soit encore entier, il est toutefois inhabité, et les Arabes fort ignorans de l’antiquité ne me sceurent apprendre pour qui il a esté basti, ny d’autres singularitez dont je m’informay, et dont j’aurois bien voulu qu’ils m’eussent instruit. Devant la porte de ce Palais il y a un étang accompagné d’un canal qui est à sec. Le fond du canal est de brique, de mesme que la voûte qui est à fleur de terre, et les Arabes croyent que c’a esté un conduit par lequel on faisoit passer l’eau de l’Euphrate. Pour moy je ne sçaurois qu’en juger, et ne puis comprendre comme on pouvoit faire venir de l’eau de si loin, l’Euphrate estant éloigné de ce lieü-là de plus de vingt lieües. De ce Palais nous tirâmes au nord est et après une marche de quatre jours nous arrivâmes à un méchant bourg, autrefois nommé Cufa et à present Meched-Ali.’[4]
The least inaccurate description of Ukhaiḍir is furnished by an anonymous Englishman, quoted by Niebuhr.[5] ‘Ich habe’, says he, ‘in dem Tagebuch eines Engländers, der von Haleb nach Basra gereiset war, gefunden, dass er 44 Stunden nach Osten von Hêt eine ganz verlassene Stadt in der Wüste angetroffen habe, wovon die Mauer 50 Fuss hoch und 40 Fuss dick war. Jede der vier Seiten hatte 700 Fuss, und in der Mauer waren Thürme. In dieser Stadt, oder grossem Castell, findet man noch ein kleines Castell. Von eben dieser verlassenen Stadt hörte ich nachher, dass sie von den Arabern el Khader genannt werde und um 10 bis 12 Stunden von Meshed Ali entfernt sei. Sie ist ohne Zweifel gleichfalls wegen Mangel an Wasser verlassen worden: und da man hier gar keine Städte oder Dörfer in der Nähe findet, so ist dies wohl die Ursache, dass man davon nicht alle brauchbare Steine weggebracht hat, wie von Kufa und Basra, wo fast nichts mehr übrig ist.’ In the same volume (p. 236) Niebuhr gives the route from Baṣrah to Aleppo through the desert and mentions therein Ukhaiḍir under the name of el Chäder, remarking that it is the castle to which the Englishman referred. This Englishman I conjecture to have been Mr. Carmichael, whose route is shown in a map published by Ives,[6] and there called ‘the common route of the caravans from Aleppo to Bassora over the great desert of Arabia, as described in a journal kept by Mr. Carmichael in the year 1751’. Ukhaiḍir appears upon it as ‘Alkader, the ruins of a most magnificent building’.
Major John Taylor saw it in June 1790 and dismissed it with short shrift.[7] He too was following the desert road from Aleppo to Baṣrah. On leaving Shethâthâ he says: ‘The camels being loaded at half past 6 this morning, we set forward over a barren flat desert. We crossed the bed of a river and at 11 a.m. we passed to our left the ruins of a small square fort, distant about half a mile, which the Arabs call Ula Kayder.’
Ritter[8] gives a summary of all these notices by early travellers, including that of Teixeira, which he accepts unquestioned, in spite of the fact that Teixeira’s palace lies, according to his own account, at least seven days’ journey to the south of the site of Ukhaiḍir.
M. Massignon was, however, the first to make any record of Ukhaiḍir. His preliminary notes, together with a plan and some photographs, were published in the Bulletin de l’Académie des Inscriptions et Belles-Lettres of March 1909, and in the Gazette des Beaux-Arts of April 1909. The next visitor to the palace was myself. I left Aleppo in February 1909 and reached Ukhaiḍir on March 25, travelling by the east bank of the Euphrates and across the desert from Hît via Kubaisah and Shethâthâ. I had no knowledge of M. Massignon’s journey, neither did the Arabs, who were at that time inhabiting the place, give me any information concerning him. I did not hear of his discovery until I reached Constantinople in the following July. M. Massignon followed up his observations with the first volume of his Mission en Mésopotamie (published in 1910), which was concerned chiefly with Ukhaiḍir. I, in the meantime, had published a paper on the vaulting system of the palace in the Journal of Hellenic Studies for 1910 (p. 69), and I gave a more detailed account of the building in the following year (Amurath to Amurath, p. 140). I returned to the site in March 1911, in order to correct my plans and to take measurements for elevations and sections. Going thence to Babylon, I found that some of the members of the Deutsche Orient-Gesellschaft who were engaged upon the excavations there had been to Ukhaiḍir during the two years of my absence and were preparing a book upon it. They were so kind as to show me their drawings while I was at Babylon, and I had the advantage of discussing with them my conjectures and difficulties, and the satisfaction of finding that we were in agreement on all important points. Their book appeared in 1912 (Dr. Reuther, Ocheïdir, published by the Deutsche Orient-Gesellschaft), and is referred to frequently in this volume. For their generosity in allowing me to use some of their architectural drawings, I tender my grateful thanks, together with my respectful admiration for their masterly production.
I feel, indeed, that I must apologize for venturing to offer a second version of the features of a building which has been excellently described and portrayed already. But my excuse must be that my work, which was almost completed when the German volume came out, covers not only the ground traversed by my learned friends in Babylon, but also ground which they had neither leisure nor opportunity to explore; and, further, that I believe the time has come for a comparative study of the data collected by myself and others, such as is contained in this book.
I must also thank M. Dieulafoy, M. de Morgan, Professor Strzygowski, Professor Sarre, Dr. Herzfeld, Professor Brünnow, Professor Haverfield, M. Velazquez Bosco, the Director of the Imperial Museums in Berlin, the Council of the K. Akademie of Vienna, the Deutsche Orient-Gesellschaft, and Messrs. Holman, Macmillan, Gebhardt and Bruckmann, for permitting me to reproduce plans, drawings, and photographs prepared or published by them. I have in every case acknowledged my indebtedness in the text of this book. Dr. Moritz and Professor Littmann have been so kind as to give me their views on the graffito in the palace, and their suggestions as to its deciphering. Finally I should like to thank the Clarendon Press for the care which has been expended upon the publication of my work, and Sir Charles Lyall for the help which he has given me in revising the proofs.
With this I must take leave of a field of study which formed for four years my principal occupation, as well as my chief delight. A subject so enchanting and so suggestive as the palace of Ukhaiḍir is not likely to present itself more than once in a lifetime, and as I bring this page to a close I call to mind the amazement with which I first gazed upon its formidable walls; the romance of my first sojourn within its precincts; the pleasure, undiminished by familiarity, of my return; and the regret with which I sent back across the sun-drenched plain a last greeting to its distant presence. The unknown prince at whose bidding its solitary magnificence rose out of the desert, the unknown lords who dwelt in its courts, cannot at the time of its full splendour have gloried and rejoiced in their handiwork and their inheritance more than I who have known it only in decay; and, in the spirit, I part from it now with as much unwillingness as that which I experienced when I withdrew, further and further, from its actual protection.
GERTRUDE LOWTHIAN BELL.
CONTENTS
| CHAPTER | PAGE | |
| [I.] | UKHAIḌIR | [1] |
| [II.] | QṢAIR, MUDJḌAH, AND ‘AṬSHÂN | [38] |
| [III.] | QAṢR-I-SHÎRÎN | [44] |
| [IV.] | GENESIS OF THE EARLY MOHAMMADAN PALACE | [55] |
| [V.] | THE FAÇADE | [122] |
| [VI.] | THE MOSQUE | [145] |
| [VII.] | THE DATE OF UKHAIḌIR | [161] |
| [SUBJECT INDEX] | [169] | |
| [INDEX OF NAMES] | [173] | |
LIST OF ILLUSTRATIONS
| FIGURES IN THE TEXT | ||
|---|---|---|
| FIG. | PAGE | |
| [1.] | Ukhaiḍir, north wall of palace, showing original scheme | [11] |
| [2.] | Ukhaiḍir, arch construction | [12] |
| [3.] | Ukhaiḍir, arch construction | [15] |
| [4.] | Ukhaiḍir, south side of court B | [31] |
| [5.] | Zindjirli | [61] |
| [6.] | Pasargadae | [62] |
| [7.] | Persepolis, Apadana of Xerxes | [63] |
| [8.] | Persepolis, palace of Darius | [64] |
| [9.] | Parthian palace at Niffer | [66] |
| [10.] | Hatra palace | [67] |
| [11.] | Relief from Quyundjik | [77] |
| [12.] | Modern Ṭarmah houses | [83] |
| [13.] | Balkuwârâ | [85] |
| [14.] | Scheme of Pompeiian house | [87] |
| [15.] | Priene, house 33 | [88] |
| [16.] | Priene, house 24 | [88] |
| [17.] | Palace at Pergamon | [89] |
| [18.] | Small palace at Hatra | [91] |
| [19.] | Ctesiphon | [95] |
| [20.] | Karkh | [95] |
| [21.] | Roman fort at Housesteads | [99] |
| [22.] | Odhruḥ | [100] |
| [23.] | Ledjdjûn | [101] |
| [24.] | Da’djaniyyeh | [102] |
| [25.] | Bshair | [104] |
| [26.] | Qasṭal | [105] |
| [27.] | Lagash | [107] |
| [28.] | Ṭûbah | [113] |
| [29.] | Kharâneh | [114] |
| [30.] | Petra, the storied tomb | [129] |
| [31.] | Hatra, façade of palace reconstructed | [138] |
| [32.] | Mosque at Raqqah | [154] |
| [33.] | Mosque of Abû Dulaf | [155] |
| [34.] | Assyrian fortress | [157] |
| [35.] | Ukhaiḍir, graffito in room 44 | [163] |
| MAPS | ||
| [1.] | Syria and Mesopotamia | —at end |
| [2.] | Ukhaiḍir, map of site | |
| PLATES (at end) | |
|---|---|
| 1. | Ukhaiḍir, ground-plan. |
| 2. | Ukhaiḍir, ground-plan of interior buildings. |
| [3.] | Fig. 1. Ukhaiḍir, first floor of palace. Fig. 2. Ukhaiḍir, second floor of palace. |
| 4. | Fig. 1. Ukhaiḍir, Section a-b. Fig. 2. Ukhaiḍir, Section c-d. |
| 5. | Fig. 1. Ukhaiḍir, Section e-f. Fig. 2. Ukhaiḍir, Section g-h. Fig. 3. Ukhaiḍir,The Ḥammâm. Fig. 4. Qṣair. |
| [6.] | Fig. 1. Ukhaiḍir from north-east. Fig. 2. Ukhaiḍir, central court, from south. |
| [7.] | Fig. 1. Ukhaiḍir, south-east angle of palace yard. Fig. 2. Ukhaiḍir, north-east corner. |
| [8.] | Fig. 1. Ukhaiḍir, south-west corner. Fig. 2. Ukhaiḍir, detail of tower chamber.Fig. 3. Ukhaiḍir, decoration on north wall. |
| [9.] | Fig. 1. Ukhaiḍir, south gate, interior. Fig. 2. Ukhaiḍir, south gate, exterior. |
| [10.] | Fig. 1. Ukhaiḍir, chemin de ronde of east wall, looking north. Fig. 2. Ukhaiḍir, northfaçade, showing loopholes of chemin de ronde. |
| [11.] | Fig. 1. Ukhaiḍir, north façade. Fig. 2. Ukhaiḍir, north gate. |
| [12.] | Fig. 1. Ukhaiḍir, room 1, looking north. Fig. 2. Ukhaiḍir, room 88, south-west endof vault. |
| [13.] | Fig. 1. Ukhaiḍir, room 4, north-east portion of dome. Fig. 2. Ukhaiḍir, room 4, south-westportion of dome. |
| [14.] | Fig. 1. Ukhaiḍir, great hall, looking south. Fig. 2. Ukhaiḍir, vault of great hall,looking south. |
| [15.] | Fig. 1. Ukhaiḍir, great hall, west side. Fig. 2. Ukhaiḍir, great hall, door of south-weststair. |
| [16.] | Fig. 1. Ukhaiḍir, great hall, looking north. Fig. 2. Ukhaiḍir, vault of south-west stairout of great hall. |
| [17.] | Fig. 1. Ukhaiḍir, corridor 5, looking west. Fig. 2. Ukhaiḍir, north end of corridor 20. |
| [18.] | Fig. 1. Ukhaiḍir, south wall of mosque. Fig. 2. Ukhaiḍir, miḥrâb. |
| [19.] | Fig. 1. Ukhaiḍir, east side of mosque. Fig. 2. Ukhaiḍir, east side of mosque, north end. |
| [20.] | Fig. 1. Ukhaiḍir, south-east angle of mosque. Fig. 2. Ukhaiḍir, south-west angle ofmosque. |
| [21.] | Fig. 1. Ukhaiḍir, door of mosque. Fig. 2. Ukhaiḍir, north end of gallery 108. |
| [22.] | Fig. 1. Ukhaiḍir, north-east angle of court A. Fig. 2. Ukhaiḍir, corridors 28 and 102from corridor 100. |
| [23.] | Fig. 1. Ukhaiḍir, court H, north side, and north wall of mosque. Fig. 2. Ukhaiḍir,second story, rooms 119, 120, and 121, from east. |
| [24.] | Fig. 1. Ukhaiḍir, second story, rooms to south and east of court. Fig. 2. Ukhaiḍir,second story, showing doors of 132, 137, and 117. |
| [25.] | Fig. 1. Ukhaiḍir, gallery 134. Fig. 2. Ukhaiḍir, squinch in north-west angle ofgallery 134. |
| [26.] | Fig. 1. Ukhaiḍir, north-west angle of central court. Fig. 2. Ukhaiḍir, east door andsouth-east end of central court. |
| [27.] | Ukhaiḍir, central court, east side of north façade. |
| [28.] | Fig. 1. Ukhaiḍir, south-east angle of central court. Fig. 2. Ukhaiḍir, fluted semi-dome,south-east angle of central court. |
| [29.] | Fig. 1. Ukhaiḍir, room 29 and south side of central court. Fig. 2. Ukhaiḍir, southside of central court, showing door of room 31. Fig. 3. Ukhaiḍir, south side ofcentral court, door into room 42. |
| [30.] | Fig. 1. Ukhaiḍir, vault of room 31. Fig. 2. Ukhaiḍir, room 31, showing decoration intop of vault. |
| [31.] | Fig. 1. Ukhaiḍir, south wall, east end, of room 32. Fig. 2. Ukhaiḍir, room 40 fromroom 30. Fig. 3. Ukhaiḍir, south-west angle of passage 36. |
| [32.] | Fig. 1. Ukhaiḍir, room 33, north-west column. Fig. 2. Ukhaiḍir, groin in north-eastangle of corridor 28. |
| [33.] | Fig. 1. Ukhaiḍir, court B, north-west angle. Fig. 2. Ukhaiḍir, court B, eastern halfof north façade. Fig. 3. Ukhaiḍir, court C, north-west angle. Fig. 4. Ukhaiḍir,court C, eastern half of north façade. |
| [34.] | Fig. 1. Ukhaiḍir, south door of room 44. Fig. 2. Ukhaiḍir, south doors of room 45. |
| [35.] | Fig. 1. Ukhaiḍir, south side of court B. Fig. 2. Ukhaiḍir, south-west angle of court H.Fig. 3. Ukhaiḍir, west end of No. 78. |
| [36.] | Fig. 1. Ukhaiḍir, door between rooms 44 and 45 from room 44. Fig. 2. Ukhaiḍir,court C, south door of room 55. |
| [37.] | Fig. 1. Ukhaiḍir, door from court C into palace yard. Fig. 2. Ukhaiḍir, south-westcorner of court E. Fig. 3. Ukhaiḍir, south side of court H. |
| [38.] | Fig. 1. Ukhaiḍir, from south-east corner of chemin de ronde. Fig. 2. Ukhaiḍir, fromeast gate. |
| [39.] | Fig. 1. Ukhaiḍir, south-west angle of court G. Fig. 2. Ukhaiḍir, east annex, north-eastend. Fig. 3. Ukhaiḍir, east annex, from north. |
| [40.] | Fig. 1. Ukhaiḍir, remains of stair. Fig. 2. Ukhaiḍir, room 140. |
| [41.] | Fig. 1. Ukhaiḍir, room 141, north-west corner of groin. Fig. 2. Ukhaiḍir, east annex,from south. |
| [42.] | Fig. 1. Ukhaiḍir, east annex from south, showing door of room 141. Fig. 2. Ukhaiḍir,north annex, showing roof. Fig. 3. Ukhaiḍir, north annex, detail of roof. |
| [43.] | Fig. 1. Ukhaiḍir, north annex, from north gate. Fig. 2. Ukhaiḍir, from north. |
| [44.] | Fig. 1. Ukhaiḍir, north annex, from west. Fig. 2. Ukhaiḍir, from north-west. |
| [45.] | Fig. 1. Qṣair, interior, showing apse. Fig. 2. Qṣair, detail of apse. Fig. 3. Qṣair,exterior from south. |
| [46.] | Fig. 1. Mudjḍah. Fig. 2. ‘Aṭshân. |
| [47.] | Fig. 1. Mudjḍah. Fig. 2. Mudjḍah. Fig. 3. Mudjḍah, detail of lower niches. |
| [48.] | Fig. 1. Ṭâûq, minaret. Fig. 2. ‘Aṭshân, from north-east. |
| [49.] | Fig. 1. ‘Aṭshân, north gate, exterior. Fig. 2. ‘Aṭshân, north gate, interior. |
| [50.] | Fig. 1. ‘Aṭshân, rooms 2, 3, and 5, from north. Fig. 2. ‘Aṭshân, rooms 5 and 8, fromnorth. |
| [51.] | Fig. 1. Palace of Khusrau, corridor 103, east side. Fig. 2. ‘Aṭshân, west door ofroom 6, from west. |
| [52.] | Fig. 1. ‘Aṭshân, room 8, from west. Fig. 2. Palace of Khusrau, vault of room 71. |
| 53. | Qaṣr-i-Shîrîn, palace of Khusrau, upper level. |
| 54. | Qaṣr-i-Shîrîn, palace of Khusrau, lower level. |
| [55.] | Fig. 1. Palace of Khusrau, east end of hall 3. Fig. 2. Palace of Khusrau, west end ofhall 3. |
| [56.] | Fig. 1. Palace of Khusrau, vaulted ramp in corridor 12. Fig. 2. Palace of Khusrau,court M, south antechamber, showing door leading into corridor 42. |
| [57.] | Fig. 1. Palace of Khusrau, south-west corner of court M, showing corridor 42. Fig. 2.Palace of Khusrau, east side of courts O and Q. |
| [58.] | Fig. 1. Palace of Khusrau, west side of courts Q and S. Fig. 2. Palace of Khusrau,south-west corner of court S. |
| [59.] | Fig. 1. Palace of Khusrau, vault of room 73. Fig. 2. Palace of Khusrau, corridor 43,looking west. |
| [60.] | Fig. 1. Palace of Khusrau, court V, looking west. Fig. 2. Palace of Khusrau,gateway between courts U and V, west arch. |
| [61.] | Palace of Khusrau, gateway between courts U and V, south-east angle of room 82. |
| [62.] | Palace of Khusrau, court W, with rooms 97 and 98. |
| [63.] | Fig. 1. Palace of Khusrau, eastern double ramp. Fig. 2. Palace of Khusrau, northbuildings. |
| [64.] | Qaṣr-i-Shîrîn, Chehâr Qapû. |
| [65.] | Fig. 1. Chehâr Qapû, interior of east gate. Fig. 2. Chehâr Qapû, niche in room 8.Fig. 3. Chehâr Qapû, squinch in room 6. |
| [66.] | Fig. 1. Chehâr Qapû, niche in room 6. Fig. 2. Chehâr Qapû, squinch in room 14. |
| [67.] | Chehâr Qapû, court D and hall 54, from east. |
| [68.] | Fig. 1. Chehâr Qapû, vault of room 31. Fig. 2. Chehâr Qapû, squinch in room 39. |
| [69.] | Fig. 1. Chehâr Qapû, hall 54, south-east corner. Fig. 2. Chehâr Qapû, hall 54,squinch in south-west corner. |
| [70.] | Fig. 1. Chehâr Qapû, hall 54, exterior of south door. Fig. 2. Chehâr Qapû, hall54, interior of south door. |
| [71.] | Chehâr Qapû, hall 54, from south. |
| [72.] | Chehâr Qapû, hall 54, from west. |
| [73.] | Fig. 1. Qaṣr-i-Shîrîn, Qal’a-i-Khusrau. Fig. 2. Firûzâbâd. |
| [74.] | Fig. 1. Sarvistân, small domed chamber. Fig. 2. Hatra, oversailing vault in mainpalace. |
| [75.] | Fig. 1. Kerkûk, Mâr Ṭahmâsgerd. Fig. 2. Hatra, vaulted passage in so-calledtemple. |
| [76.] | Sarvistân. |
| [77.] | Sargon’s Palace at Khorsâbâd. |
| [78.] | Fig. 1. Gate at Khorsâbâd. Fig. 2. Ḍumair. |
| [79.] | Fig. 1. Kharâneh. Fig. 2. Kharâneh, gateway. |
| [80.] | Fig. 1. Kharâneh, interior of court. Fig. 2. Kharâneh, interior of audience hall. |
| [81.] | Mshattâ. |
| [82.] | Fig. 1. Petra, Corinthian tomb. Fig. 2. Petra, al-Dair. |
| [83.] | Ctesiphon. |
| [84.] | Fig. 1. Doorway of mosque, Ḥasan Kaif. Fig. 2. Gateway of mosque, Ḥarrân.Fig. 3. Mayâfârqîn, north façade of mosque. |
| [85.] | Ukhaiḍir, reconstructed north façade of central court. |
| [86.] | Fig. 1. Parthian decoration, Assur. Fig. 2. Sasanian silver dish (Hermitage,St. Petersburg, No. 2969). |
| [87.] | Details of decoration from Medînat al-Zahrâ. |
| [88.] | Fig. 1. Djebel Sindjâr, khân. Fig. 2. Ḥasan Kaif, mosque. |
| [89.] | Fig. 1. Cairo, mosque of Ibn Ṭulûn. Fig. 2. Mosque of Abû Dulaf. |
| [90.] | Diyârbekr, Ulu Djâmi’. |
| [91.] | Fig. 1. Cairo, mosque of Ibn Ṭulûn. Fig. 2. Sâmarrâ, mosque. |
| [92.] | Mosque of Ṣalaḥ al-Dîn, Mayâfârqîn. |
| [93.] | Fig. 1. Diyârbekr, mosque, fragment of old wall. Fig. 2. Mayâfârqîn, mosque. |
CHAPTER I
UKHAIḌIR
The fortified palace of Ukhaiḍir stands in the desert about three hours’ journey to the south-east of the oasis of Shethâthâ and some seven hours’ south-west of Kerbelâ. Its exact site has been fixed by Sir William Willcocks’s survey and it is upon his map that mine is based ([Map 1]). Ukhaiḍir is not far from the south-west end of the low ground which Sir William Willcocks has called the Ḥabbâniyyeh depression. The southern part of this depression covers an area of 146 square kilometres at a level of 46 metres above the Persian Gulf;[9] at its lower end it still contains a lake of brackish water, the lake of Abû Dibs, the water-level of which is 19 metres above the Persian Gulf. The northern part is occupied by the Ḥabbâniyyeh Lake. That the whole area was once filled with escape water from the Euphrates is shown by the fact that it is covered at a level of 25 metres above the Persian Gulf by a thick belt of Euphrates shells; at this level it extends over an area of 1,200 square kilometres. The oases of Raḥḥâliyyeh and Shethâthâ are situated upon the edge of this ancient reservoir. Between Shethâthâ and Ukhaiḍir a shallow valley, the Wâdi al-Ubaiḍ, makes its way up from the south-west to the lake of Abû Dibs. I have been told that after heavy winter rain a stream has been known to flow down the ghadîr, the water-course, which winds through the sand and stones of the valley bed. Whether this be true or no, a well of good sweet water exists in the Wâdi al-Ubaiḍ, fed, in all probability, by a spring, like the famous water of Muḥaiwir in the Wâdi Ḥaurân, or the wells of ‘Asîleh in the Wâdi Burdân. At no other point in the immediate vicinity of Ukhaiḍir is fresh water to be obtained; whether you dig within the palace walls, or without, the water, if water there be, is brackish and unfit to drink. To the north of the Wâdi al-Ubaiḍ the ground opposite Ukhaiḍir, sloping gradually down to the Ḥabbâniyyeh depression, is intersected by gulleys, narrow and steep, cutting through hillocks of gypsum, and among these hillocks is the small ruin which the Arabs call Qṣair. Here, I take it, the gypsum was obtained for the mortar which binds the masonry of the palace, and its good qualities are attested by the excellent preservation of wall and vault until this day. I have not visited the quarries, but the Arabs told me that the stone had been brought from a distance of about an hour to the south of Ukhaiḍir, where there are traces of working ‘taḥt al-arḍ’, below the ground—not in a hill-side. Near the quarries there is said to be a well of good but not abundant water; Shakhârîz is the name of the well. It is built of stone. Behind it, some three hours’ journey from Ukhaiḍir, there is a low line of hills, the Djebel Ḍaba’. From the castle walls the long levels of the desert spread out invitingly to the hills, and I would gladly have gone thither, but I had not time to spare during either of my visits. Ukhaiḍir does not reckon security among its many charms. The plentiful sweet water of the well in the Wâdi al-Ubaiḍ makes it a trysting-place for raiding parties, and after four or five days’ sojourn it is best to be gone, lest the news that a foreigner is lodged within the palace walls should run too temptingly among the tribes. In 1911, the date of my last visit, I came to Ukhaiḍir from Shethâthâ, having ridden straight across the desert from Ramâdi, skirting the Ḥabbâniyyeh Lake and the east side of the Ḥabbâniyyeh depression. When I left I did not follow the usual way, by Abû Dibs to Kerbelâ, but rode almost due east, to the foot of a cliff of sand and rock, which is the western limit of a flat desert plateau that stretches eastward to the Hindiyyeh. An abrupt rise of this nature is called in colloquial Arabic a ṭâr.[10] From Ukhaiḍir the ground dropped gradually. After two hours’ riding (about six miles) we reached the khabrâ of Wizikh. A khabrâ is a hollow bottom where rain water lies and stagnates till it evaporates. The khabrâ of Wizikh, which was dry and sandy, appeared to stretch along the foot of the ṭâr, northward to Abû Dibs, and also southwards. My Arab guide, a sheikh of the Zaqârît, which is a sub-tribe of the Shammar, informed me that there were wells of brackish water in the khabrâ further to the south, the Biyâr Slâm. The khabrâ was about a fifth of a mile wide. At the further side we rode up the sandy gulleys of the ṭâr and in ten minutes reached a well, the Bir Sbai’i, the water of which was brackish but drinkable. From here to the Hindiyyeh there is no water of any kind. Another ten minutes brought us to the summit of the ṭâr, whence we could see Ukhaiḍir on the one hand and the tower of Mudjḍah on the other. The bearings here were as follows: Ukhaiḍir (south-east angle of the castle) 300°, Mudjḍah 97°, central point of the Djebel Ḍaba’ 244°. Mudjḍah is a solitary tower without any provision for the storage of water, or any ruins round it. I think it can have served no other purpose than that of a landmark on the line of the caravan track, which must have passed this way from the great city of Kûfah to the oasis of Shethâthâ, or ‘Ain al-Tamr, to give it its earlier name. From the top of the ṭâr to the modern Kerbelâ-Nedjef road the desert is absolutely flat and featureless, and we ourselves came near to losing our way across it. The existence of a former caravan track across this waste is assured by the ruined khân of ‘Aṭshân, half-way between Mudjḍah and the modern Khân Ḥamâd.
Such are the characteristics of the country round Ukhaiḍir. The ṭâr, standing over the low ground of the khabrâ, bounds the view to the east; to the north-east, across the Wâdi al-Ubaiḍ, the gypsum hillocks lead down to the Ḥabbâniyyeh depression; to the north-west a few shallow desert wâdis cross the path to Shethâthâ; to south and west stretches the immense expanse of the Syrian desert, broken only by the small group of the Djebel Ḍaba’. It is, however, by no means certain that in the seventh and eighth centuries, that is to say, at the period during which it is probable that the palace was built, the local conditions were the same as they are at present. It is indeed likely that the Ḥabbâniyyeh depression contained at that time more water than it does now, that the lake of Abû Dibs stretched across a considerable part of it, and that its margin approached nearer to Ukhaiḍir. The scrub and reed round the edge of the lake would have given cover for water fowl, for boar and other wild animals, and the lords of Ukhaiḍir, when they went out to the chase, would have had an ample supply of game. Moreover the oasis of Shethâthâ was certainly a more important place then than it is at present, for all its 160,000 palm-trees.[11] There can be no doubt that it occupies the site of ‘Ain al-Tamr, famous in the days of the Persian kings[12]—that same oasis which Khâlid ibn al-Walîd took and sacked in the year A.H. 12. It is my belief that the Mohammadan invasion did not diminish its importance, and in proof I would adduce the evidence afforded by the khân of ‘Aṭshân and the landmark tower of Mudjḍah, showing that from Kûfah to ‘Ain al-Tamr there must have been a direct caravan road across the desert. Muqaddasi, writing in the year A.D. 985, describes ‘Ain al-Tamr as a little castle;[13] Yâqût, who mentions the name Shefâthâ as part of ‘Ain al-Tamr, praises its dry dates above those of other towns,[14] and to this day they maintain that honourable pre-eminence. Ukhaiḍir, then, with the marshy haunts of game a mile or two from its gates, and a much-frequented oasis three hours to the north, presented in the eighth century advantages which it no longer enjoys now that the waters have retreated to the confines of the modern Abû Dibs, and the traffic of Shethâthâ has shrunk to an occasional small caravan of merchant and citizen passing along the Kerbelâ track, or the visit of a ragged crew of Beduin date-buyers. Yet it is difficult to conjure up any picture but that of isolation when, after a weary struggle through sand or marsh, according to the season, the gaunt walls and towers of the palace rear themselves out of the solitudes of the desert—in all that barren waste sole vestige of mortal energy, of the fleeting splendour of mankind. ([Plate 6], Fig. 1).
The palace consists of a quadrangular area bounded by a wall which measures 163·60 metres from east to west, and 175·80 metres from north to south ([Map 2]). It is almost exactly oriented. The wall is provided with round towers, projecting 2·70 metres from its face, and with a gate in the centre of each side. At the north-west angle, at a distance of 13·25 metres from the palace wall, a building consisting of fifteen vaulted rooms runs out due north. It has a length of 81·20 metres and a width of 11·45 metres. To the west of the six southerly chambers lies a rectangular court, 35·20 metres from north to south and 25·80 metres from east to west, with round towers like those of the main palace, projecting 2·75 metres. North-east of the palace there is a small irregularly-shaped building, known to the Arabs as the Ḥammâm, the bath. Its greatest length is 12·90 metres and its greatest width, including the rectangular buttresses, 8·65 metres. With the exception of the Ḥammâm, these edifices have been enclosed by a second stone wall, but this wall cannot have been a considerable structure, for at the only point where its width can be determined, north of the palace, it is but 1 metre thick. Its present aspect is that of a low mound of sand, and in places even this mound is by no means clearly to be traced. Owing to the very fragmentary character of the northern line of the outer wall, it is not possible to fix the position of the north gate, though there can be little doubt that a gate existed opposite the north gate of the palace, at a distance of about seventy paces from it. South of the Ḥammâm the wall is easier to make out. It runs parallel to the east wall of the palace, and is broken by a gateway opposite the eastern palace gate. At intervals large heaps of stones seem to indicate the presence of towers. Two hundred and thirty paces to the south of the palace, this outer towered wall turns to the west and runs parallel to the south wall of the palace. Traces of a gate can be seen opposite the south gate of the palace. From the south-west angle of the palace wall a second low sandy mound runs down to join the outer wall, and immediately to the west of this division wall there had been another gate in the outer wall, which then ran on westward for two hundred paces. The west wall is not exactly parallel to the palace; it was broken by a gate opposite the west gate of the palace. The north-west angle of the outer wall is very nearly obliterated. It turns off eastward almost at right angles and joins an inner dividing wall which comes up from a point about twenty paces west of the north-west tower of the palace, and seems to have been connected with that tower by a cross-wall. At the point of junction between this dividing wall and the outer wall, a mound runs out north-west for a great distance into the desert. I did not follow it, but from the top of the palace its course can be traced for more than a mile. The northern outer wall then turns slightly to the south of east and passes close to the south-east corner of the detached northern building, beyond which point it is almost obliterated. Between the Ḥammâm and the north-east angle of the outer wall there are some low sandy mounds wherein the Arabs say that they have dug and found brackish water.
When I first visited Ukhaiḍir in March 1909 it was occupied by Arabs from Djôf in Nedjd who were anxious to establish themselves there permanently. To this end they wished to receive official recognition from the Government, and they proposed to earn a livelihood by supplying Baghdâd with camels bought from the tribes of the Syrian and Arabian deserts. When I returned in 1911 they were gone, and Sheikh Ṣukhail, of the Zaqârît, who was camped under the walls, could give me no account of their departure, except that it had taken place some months previously. Possibly they found Ukhaiḍir an unsatisfactory centre for commercial enterprise, and there can be no question but that their project would have been ill looked upon by the Beduin, who regard the sweet waters of the Wâdi al-Ubaiḍ as their peculiar property. Whatever may have driven them forth, the Djôfîyîn had left no memorial of their residence save heaps of filth and refuse in the halls and courts of the palace, new stonework round the well in the Wâdi al-Ubaiḍ, a meagre plantation of half-withered palm-shoots close by it, and evidences of an equally unsuccessful attempt to establish a few palm-trees within the palace walls near the west gate, where there is a small deep well of brackish water. And we, finding Ukhaiḍir untenanted, took possession of it and pitched our tents in the central court.
The towered wall of the palace encloses a yard and a quadrangular block of building which covers an area measuring 111·40 metres from north to south and 68·50 metres from east to west (Plate 1). On three sides of this block, rounded towers project 1·75 metres from the face of the wall, while the north side is connected with the main wall. The northern part of the building is three stories high, the upper story being on a level with the chemin de ronde which runs round the main wall. The rest of the building, 73·95 metres from north to south, is one story high. The palace yard runs round three sides of the building. To the west and south it is unoccupied by any structure; north of the west gate lies a well of brackish water, and it was there that the Djôfîyîn had planted their palm-shoots. This well I believe to be modern; it bears no mark of antiquity. To the east, north of the east gate, the yard is blocked by an edifice, a single story high, the chambers of which are numbered on the plan 140-152. It is a later addition, as will be seen, to the original scheme of the palace.
The main wall consists of a core of masonry 2·60 metres thick, rising about 10 metres above the present level of the ground (section e-f, Plate 5, Fig. 1). It is difficult to get absolutely accurate measurements of height as the surface-level varies slightly according to the depth of ruin strewn over it. Blind arcades on the interior and on the exterior carry the chemin de ronde. On the interior, pilasters 1 metre deep are united by arches very slightly pointed ([Plate 7], Fig. 1). The pilasters are without capital or impost, the arches springing directly from them. The arches rise to a height of 8·50 metres, and their span averages on the east wall a little under 3·85 metres, while the width of the pilasters averages 1·55 metres. The arches are composed of two rings of stone voussoirs, the inner ring laid vertically; i.e. with the broadside showing, the outer ring laid horizontally, with the narrow end showing. Dr. Reuther notices that in some instances the horizontal outer ring is lacking. The walls and pilasters, like all the walls of Ukhaiḍir, are built of thin irregular slabs of stone, very roughly coursed, with a binding course laid through them at intervals. In or above the binding courses are holes for wooden beams. There are four such holes in each pilaster and one in the spandrel between the arches. In the back wall of each arcade there are three holes up the centre, and two level with the springing of the arch. Similar holes for beams occur in all the walls of Ukhaiḍir. At a height of 1·50 metres above the level of the arches, the wall is set back ·40 metre and broken by windows, 11·80 metres above the ground, and 1·80 metres above the floor of the chemin de ronde. As the authors of Ocheïdir have observed, these windows cannot have served any purpose of defence, since they are so high above the floor. There was thus no means of attacking from the wall a foe who had penetrated into the palace yard. Between each pair of windows, shallow pilasters, corresponding to the pilasters below, are carried up to the top of the wall. There are holes for beams between the window arches on wall and pilaster, and also directly above, along the top of the wall. On the exterior there is again a blind arcade 1 metre deep, consisting of two round arches between each tower ([Plate 7], Fig. 2). The towers have a projection of 2·75 metres beyond the face of the arcade. The exterior arches bear no relation to the arches of the interior arcade. Two arches, with an average span of 3·85 metres, separated by a pilaster 1·60 metres wide, stand between each of the piers, 4·10 metres wide, against which the three-quarter round towers are placed. There are five of these towers between gateway and angle tower. They have a diameter of 3·30 metres, whereas the angle towers have a diameter of 5·10 metres. The holes for beams appear as on the inner side of the wall, but they do not correspond with the interior holes. As in the interior arcade, the outer arches are slightly pointed and spring directly from the pilasters. The top of the exterior arches is ·30 metre above the level of the floor of the chemin de ronde. The chemin de ronde does not occupy the whole width of the core of the wall ([Plate 3], Fig. 2). The passage is 1·90 metres wide. On the inner side, the wall is 1 metre thick and broken by the above-mentioned windows looking into the yard; on the outer side there is a series of recesses covered by ovoid arches. Each recess, 1·45 metres wide and ·40 metre deep, contains either a loophole window or a door. The loopholes, of which there are four between each tower, open on to the exterior of the palace and command a wide view of the desert. They are ·65 metre wide on the inside and narrow outwards to ·20 metre. On the inside they are covered by a lintel with an arched niche above it, on the outside they have a triangular head with a single upright stone placed within it, supporting the side stones of the triangle, and a small inverted triangular aperture above ([Plate 8], Fig. 3 and Plate 10, Fig. 2). Each window recess is machicolated, there being an interval of ·20 metre between the outer edge of the floor of the recess (which corresponds with the outer face of the core of the wall) and the inner side of the arches of the exterior arcade. Through this gap an enemy standing at the foot of the wall could be attacked. Every fifth recess contains a door, ·75 metre wide, which gave access to a small round chamber hollowed out of the thickness of the tower. In the whole circuit of the wall not one of these tower chambers is intact, but enough remains to determine their construction ([Plate 8], Fig. 2). Each chamber was covered by an ovoid dome, in the masonry of which there are traces of flat ribs. There was a loophole in the walls on either side, from which the defenders could cover the curtain wall between tower and tower, and it is reasonable to suppose that there must have been a third loophole fronting the desert. The loopholes were constructed in the manner already described. It seems probable that the towers exceeded the curtain walls in height; many of the towers show fragments of masonry higher than the present summit of the walls. The angle towers rose a story above the chemin de ronde and contained a second round chamber above the chamber on the level of the chemin de ronde. Traces of this second chamber remain in the north-east and in the south-west towers ([Plate 8], Fig. 1). A stair was placed in each of the four angles of the castle yard ([Plate 7], Fig. 1). The stairs, which were vaulted in a manner which will be described later (below p. 16), wound twice round the newel post before they reached the gallery of the chemin de ronde, and thereafter rose one story higher in order to reach the summit of the wall, and the upper chamber of the angle towers. It is probable that the summit of the wall was given a crenellated parapet in order to protect those who walked along it. Nor was it only from the angles of the yard that the chemin de ronde could be approached. It was accessible from the top story of the palace and also by means of stairs which were situated on either side of the east, south and west gates. None of these gates are well preserved and in no case have the stairs escaped ruin, but the mark of the stair can be seen clearly on the inner face of the wall ([Plate 9], Fig. 1). The three gateways are all alike (section g-h, Plate 5, Fig. 2). They are flanked on the exterior by segments of towers ([Plate 9], Fig. 2). The outer archway, which contained the door, has in every case been blocked up by the Beduin; it is therefore impossible to tell its exact depth, though its width, 2·10 metres, can be determined. I omitted to note the portcullis of which the authors of Ocheïdir found traces outside the door.[15] An inner arched niche, 1·45 metres long by 2·50 metres wide, is visible from the interior, together with a portion of the chamber into which it led. This chamber was 6·30 metres long by 3·10 metres wide, and was covered by a pointed barrel vault oversailing the face of the walls. Over the doorway on the inside, there is an arched niche which communicated with the arch of the outer gate by a rectangular funnel. It is impossible to imagine what can have been the purpose of this funnel, which connected the bottom of the niche with the top of the arch, unless it were meant to receive the bolt of the door, but I do not think that even this explanation will hold. The authors of Ocheïdir observed a similar communication between every niche placed over a doorway and the arch below it. The construction is made clear in their admirable drawing (Ocheïdir, Fig. 19). They offer no conclusion as to its purpose, but since it occurs in archways which show no sign of having contained a door, the idea that it was meant to provide space for a bolt cannot be maintained. The inner wall of the gate-house, which has in every case fallen, projected into the palace yard 3·50 metres from the face of the inner pilasters of the enclosing wall. Besides the vaulted passage or chamber in the centre, it comprised the above-mentioned staircases. I detected traces of a door between the gate-room and the staircase on either side. The stair wound once round the rectangular newel post and reached a chamber on the first floor, above the gate-room. The doors of communication between the stair and this chamber are not preserved. The chamber is unusually low, 3·30 metres from the floor to the top of the vault. It is provided with a large window, 2·50 metres high, in the outer wall, opening on to the desert. The stair turned once more round the newel post and led into the chemin de ronde, with which the upper chamber of the gate-house communicated by doorways. The vaulting construction of the south gateway, which is the best preserved ([Plate 9], Fig. 1), shows that the vault of the upper story must have cut across the vaults of the passage, from which it was separated by transverse arches. A big window in the outer wall opens down to the floor of the chamber and the learned authors of Ocheïdir place here, no doubt correctly, a hourd projecting from the wall over the doorway below. There are small rectangular domed chambers in the towers on either side of the gate, the domes being set over the angles of the square on horizontal brackets. The gate-house was probably carried up, like the angle towers, a story higher, and the stairs must have communicated with the upper story, to judge by the evidence afforded by the south gate-house. On the north façade, and there only, the summit of the wall was given a decoration consisting of a row of arched niches carried by small engaged columns ([Plate 8], Fig. 3). The authors of Ocheïdir describe these arches as horse-shoed; they seemed to me to be merely slightly stilted and adorned with a double fillet. Below the niches runs a band of lozenges. Between each niche is set a larger engaged column, and these columns appear to have been carried up higher than the arches and in all probability bore an architrave, thus forming a rectangular frame to each niche, but the exact nature of the decoration here is uncertain, since the wall has broken away. The chemin de ronde was covered by a pointed stone vault, most of which has fallen in ([Plate 10], Fig. 1). Like all the vaults of Ukhaiḍir it oversails the face of the wall. The lower part is built of horizontal courses, while in the upper part the stone slabs are laid in vertical rings so as to dispense with centering, and this is the construction in all the vaults of the palace. At the springing of the vault a wooden beam crossed the passage from wall to wall. The holes for these beams are visible, and in some places a splintered fragment of wood projects from the masonry. At the angles of the passage the vaults from either side come together in a simple diagonal section, i.e. there was no intersection of the vaults.
The principal entrance of the palace is the north gate ([Plate 11], Fig. 1). Before the door there is an artificial platform thirty-two paces from north to south by eighty-seven paces from east to west. The door is placed in a rectangular tower, 15·70 metres wide, which projects 5·10 metres from the face of the wall, 2·40 metres from the face of the towers. Between the west side of the gate-tower and the first of the western round towers is stretched a vault 2·50 metres in depth ([Plate 11], Fig. 2). Upon this vault rests a small platform, immediately below the loopholes of the chemin de ronde, at the level of the second story. On the east side of the gate-tower there are traces of a similar vault, but this must have fallen at a period when the palace was still inhabited, since the place which it occupied upon the wall has been carefully plastered over. The pointed arch over the north door is a later reconstruction. The door leads into a narrow room, No. 1, 5·95 metres by 3 metres, from which there is access to rooms 2 and 3. These rooms are irregular in shape, unlighted, and built over vaults which are now filled with débris. The authors of Ocheïdir suggest that they may have gone down to the water-level. I doubt it. The water-level in the palace yard is considerably deeper than these vaults are likely to have been, and the water there is too brackish to drink. It is more likely that these subterranean chambers were dungeons. The vault over room 1 is not continuous. It is composed of a series of seven transverse arches, ·65 metre wide, separated by spaces ·20 metre wide ([Plate 12], Fig. 1). These apertures enabled the occupants of room 88, on the first floor, to pour boiling liquids on any foe who had passed through the door. Room 1 is bounded to the south by an arched doorway, oversailing the wall, as is the case with all wide arched openings at Ukhaiḍir, beyond which lies the smaller chamber No. 4, 4·15 metres long by 3·10 wide. A transverse arch cuts off 1·05 metres of this space, leaving a square of 3·10 metres to be covered by a fluted dome ([Plate 13], Fig. 1).[16] The remaining three sides of the chamber are broken by pointed archways which give access to the great hall (No. 7), and to the passages Nos. 5 and 6. The fluted circle of the dome is set upon a fillet which has a projection of about 1 centimetre from the face of the wall below ([Plate 13], Fig. 2). The circle is accommodated to the square by a course of stones forming at each corner a flat triangular bracket, rounded upon the inner side. The upper part of the dome is much ruined. The curve must have been ovoid and it is probable that an aperture was left at the summit, since the dome, if closed, would have projected considerably above the floor level of room 88. The hole in the upper floor, like the slits in the roof of room 1, would have served for purposes of attack when the enemy had forced an entrance.
The authors of Ocheïdir have pointed out that the original scheme of the castle did not include the present north door, nor yet the massive enclosing wall with its towers and gates. As it was first planned, the north door stood well within the existing entrance, between two segments of towers. A part of these towers is visible in rooms 2 and 3. But when the walls had been raised about 2·80 metres from the ground, the plan was altered and the outer wall and north door added to it. The north palace wall, with its round towers and gateway, was then incorporated in the larger outer wall. A glance at Dr. Reuther’s plan will show how this was effected ([Fig. 1]). Although the alteration took place while building was in progress and does not denote a later period of construction, it is yet of importance, as I shall have occasion to show later.
On the first floor the gate-tower is occupied by three vaulted chambers, 88, 89, and 90. The central room, 88, is 4·50 metres wide and therefore wider by 1·50 metres than the passage room, 1, below it. Consequently the slits between the transverse arches of 1 do not take up the whole width of 88, but leave a passage along the wall on either side. The chamber is low, measuring only 3·55 metres to the top of the vault. The vault oversails the wall; the lower part is composed of stones laid horizontally, the upper part of stones laid in vertical rings, with an inclination backwards against the north wall. At the southern end a space between the vertical rings and the south wall is filled in with horizontal courses ([Plate 12], Fig. 2). The arches of the side doors break into the vault. In the north wall there is a large window, the upper part of which has fallen away, though some of the lower part remains. It is slightly recessed on the exterior ([Plate 11], Fig. 2), and Dr. Reuther gives the explanation of this recess. It contained the groove of the portcullis, which has been obliterated below owing to the rebuilding of the north door at a later period. In the south wall of room 88 there are three arched windows opening into the great hall. The central window is the largest; in all three the arch is surmounted by a shallow arched niche. The narrow vaulted rooms 89 and 90 are approached by round-arched doorways and lighted only by very small windows high up in the north wall. In room 89 there is a staircase leading up to the second floor. Rooms 89 and 90 open into long corridors corresponding in width with the corridors 5 and 6.
Fig. 1. North wall of palace, showing original scheme. (From Ocheïdir, by kind permission of Dr. Reuther.)
Fig. 2. Arch construction. (From Ocheïdir, by kind permission of Dr. Reuther.)
The great hall, to the south of room 4, is the largest chamber in the palace. It is 15·50 metres long by 7 metres wide, but its width is increased on either side by arched recesses 1·40 metres deep and from 2·20 metres to 2·30 metres wide ([Plate 14], Fig. 1). These recesses, five on either side, are separated from one another by squat engaged columns set against piers which are ·80 metre deep. The columns carry rectangular impost-capitals from which spring the shallow slightly pointed semi-domes, or calottes, which cover the recesses. The capitals are very roughly constructed of small stones. There are traces of a shallow abacus, while a cavetto moulded in plaster seems to have been interposed between capital and shaft. At the corners a triangular stone adjusted the circle of the column to the square of the abacus, and the whole was no doubt covered with plaster. The abacus projection runs back along the walls of the niche and above it the calotte springs from another small projection ([Plate 15], Figs. 1 and 2). The calottes are bracketed over the angles, the construction being the same as that described in the dome of room 4. All the niches of Ukhaiḍir are treated in like fashion. The method employed in constructing the archivolts is admirably described by Dr. Reuther.[17] The face of the arch is formed by a permanent centering composed of gypsum and reeds. The vaulting takes place, not above the centering but between the two centering arches, the vault being built in vertical rings ([Fig. 2]). When the arches are of wide span an outer ring of horizontal voussoirs is added to the inner arch. This system is common in Mesopotamia to the present day, and is found frequently at Ukhaiḍir. In the great hall there are holes for wooden beams below the abacus of the capitals and in the spandrels of the arches. The northern recess on the east side is open and gives access to a ramp which leads to the first floor. The second, third, and fifth recesses contain low doors covered by a segmental arch. On the west side similar doors are set in the first, third, fourth, and fifth recesses, the last named giving access to a stair ([Plate 15], Fig. 2). The calotte archivolts at their highest point are 3·50 metres above the present level of the floor. The wall is carried up for another 1·25 metres, where there is a double outset from its face. Above this outset the stone vault runs up perpendicularly for about ·80 metre and the remainder of the vault is of brick ([Plate 14], Fig. 2). For a height of about 1·50 metres the brick tiles are laid horizontally, but when the curve of the vault increases the bricks are set upright in vertical rings. The vault thus formed is built without centering; it has a slightly pointed, ovoid shape and is much stilted. The north wall remains intact and its scheme of decoration is instructive ([Plate 16], Fig. 1). The arched door, 3·50 metres high, is set back within a niche 1 metre deep. About ·90 metre above the arch of the door stands a very shallow calotte covering the niche. The face of the calotte is recessed, which enhances its decorative value by giving it a double outline. As Dr. Reuther has observed,[18] the calotte is not ‘the segment of a pointed dome, but its curve in horizontal section springs sharply back from the face of the archivolt and flattens rapidly behind. Thereby the effect of the shadow is strongly felt at the edge, and the calotte seems to be deeper and more markedly vaulted than it is in reality’. At the base of the calotte there is a small niche which has been broken through owing to the partial ruin of the dome behind it.[19] In the wall on either side of the calotte there is a shallow arched niche. The arch is carried on pairs of engaged columns and is enclosed in a rectangular label. Above the calotte are the three windows of the first floor room, 88, covered by segmental arches. The windows are framed by engaged columns which carry stilted round-arched calottes. The south wall of the great hall is partly ruined. The doorway seems to have been of the same proportions as the door in the north wall, but it was not set back within a niche. The small decorative niches reappear on either side, and there were probably three windows opening into room 101 in the upper story, indeed on the west side the window jamb can still be seen. Even with these windows the great hall must have been most insufficiently lighted, since neither its doors nor its windows open directly on to the exterior of the building. To the south lay the small rectangular chamber, No. 27, which was probably, as Dr. Reuther suggests, covered by a dome similar to the dome of No. 4. It opens to east and west into the vaulted corridor 28, and on the south into the central court.
Holes for wooden beams can be seen on the north wall of the great hall, two on either side of the portal niche, one on either side of the shallow decorative niches, and one on either side of the group of windows. On the south wall they have been somewhat differently disposed, one on either side of the door at the level of the arch, one almost immediately above, higher than the top of the arch, and three higher up still, following the curve of the vault ([Plate 14], Fig. 2).
The masses of masonry on either side of the vault are lightened by the tubes which are characteristic of the vaulting system of Ukhaiḍir (section a-b, Plate 4, Fig. 1). One of these tubes pierces the wall on either side, partly above the calottes of the recesses. On the east side the opening of this tube can be seen high up in the wall of the corridor 28; on the west side the tube is not visible owing to the interposition of a stair behind the corridor, but there can be no doubt that it exists. Again towards the top of the vault there is another pair of tubes. The western of these two can be seen through a breach in the wall of the stair which leads from room 89 to the second floor; I infer its eastern counterpart. The vault of the great hall is buttressed by the vaults of the chambers of the ground floor and of the first floor which lie at right angles to it.
The wings of the three-storied block, of which the great hall forms the centre, are bounded to the north by the two vaulted corridors 5 and 6 ([Plate 17], Fig. 1), the western corridor, 5, being 34 metres long, and the eastern, 6, 34·90 metres long. The vaults are constructed in the usual fashion, oversailing the wall and built of thin slabs of stone, laid vertically in concentric, slightly pointed rings. The corridors lead into the palace yard. The door of the west corridor is much ruined. The door of the east corridor is set in a niche surmounted by a shallow calotte, of which the archivolt is slightly pointed. Below the calotte, between it and the arch of the door, is a second small arched niche, connected by the usual funnel with the top of the door arch. The calotte is outlined by a singular decoration composed of a crenellated motive.[20] The crenellated motive is common in the ornament of Ukhaiḍir and elsewhere, but I am not acquainted with any other example of its application to the archivolt.
To the south of the east corridor runs a vaulted ramp, a sloping passage from the great hall to the first floor. To the south of the ramp lie two groups of three vaulted chambers. In the inner group, Nos. 12, 13, and 14, the rooms are 7 metres long with an average width of 3·50 metres. They are separated from each other by walls 1 metre thick, and communicate with each other by doors covered by ovoid arches set back from the jambs. Each room possesses a door into the great hall, but since the position of these doors is determined by that of the recesses in the hall, which do not correspond with the rooms behind them, the doors are never in the centre of the rooms, and in one case, No. 13, the side wall is narrowed to allow space for the door. The wall which separates the rooms from the recesses of the great hall is 1·50 metres thick. A door at the east end of each room leads into the corresponding room of the second group. In this group the rooms 15, 16, and 17, while they have the same width as those of the first group, are considerably shorter, measuring only 4·80 metres. They communicate with each other and with the vaulted passage, 20. Room 17 has further a door in the north wall, which leads into the small vaulted room, No. 18, and this in turn is connected with a still smaller room, No. 19. Nos. 18 and 19 lie under the ramp, and No. 19 is in consequence extremely low. None of the chambers above described are provided with windows; what light they possess filters in through the doors. Nos. 12, 13, and 14 are therefore exceedingly dark, and must have been darker still when the south wall of the great hall was intact. Nos. 18 and 19 are totally unillumined, and for this reason, and on account of the inconvenience of their low vaults, it may be presumed that they were not used for dwelling purposes.
Fig. 3. Arch construction. (From Ocheïdir, by kind permission of Dr. Reuther.)
The arches of the doorways in these rooms, and in all other small doorways in the palace, are constructed in a manner different from that which has been detailed above. Again I borrow the description from Dr. Reuther. A wooden centering has been placed upon the jambs; over this centering was laid a band of gypsum mortar and small stones, irregularly bedded, which, when it hardened, formed an inner arch of concrete (Fig. 3). When the span was narrow no other arch was considered necessary. When it was wider an outer arch of voussoirs laid horizontally encompassed the inner concrete arch. Not infrequently, besides the wooden centering, a permanent centering of mortar and reed was placed on either face of the concrete arch. When the wooden centering was removed the concrete arch remained, set back from the jambs, whereas in all the wide archways, such as those of room 4, the arch follows the principle of the vault and oversails the wall.
The passage, No. 20, which is 12·25 metres long by 2·80 metres wide, communicates by a door at its northern end with the small unlighted room, No. 21. The construction here is of interest ([Plate 17], Fig. 2). The passage is finished by a shallow pointed calotte, standing out from the face of the wall and spanning the angles in the usual fashion with a horizontal masonry bracket. Below it, but not in the centre of the passage, is the small doorway, which is covered by a masonry lintel. The passage opens on to court A through an arcade of two pointed arches. The arches spring from engaged columns and from a squat masonry column placed between them. The rough capital and engaged capitals, from which the stucco has disappeared, are constructed in the same way as the engaged capitals in the great hall. On the opposite side of the court there was once a similar arcade of two arches which has now fallen; indeed, the arcade of No. 20 is the only free-standing arcade which remains intact in the whole palace, with the exception of those in rooms 33 and 40. Court A, 10·70 metres by 6·25 metres, communicates with corridor 6 by a vaulted passage, 1·90 metres wide and 4·25 metres high, leading to an arched doorway 1·60 metres wide and 2·55 metres high. East of this passage lies a vaulted room, No. 26, the door of which stands in the ruined cloister, No. 25. Room 26 is lighted by two small windows in the south wall, opening on to the court, and by a window-slit in the east wall, opening on to the palace yard. To the south of court A lie three chambers, Nos. 22, 23, and 24, which have a width varying from 4·05 metres to 3·85 metres and a length of 5 metres. They communicate with each other and with the court, added to which No. 22 possesses a third door leading into No. 20, and No. 24 a third door leading into No. 25. For the door leading from No. 24 into court A space has been provided by removing a section of the dividing wall between Nos. 23 and 24.
The arrangement of the west wing of the three-storied block is dissimilar from that of the east wing. Three chambers, 8, 9, and 10, lie to the west of the great hall. They have an average width of 3·70 metres, but in length they are only 5·75 metres. They are lighted by small windows high up in the west wall. They communicate with one another by doors covered with ovoid arches set back from the jambs, and with the great hall by small doors in the recesses. The vaults are pointed and oversail the walls. South of No. 10, a stair leads up from the southernmost doorway in the great hall to the first floor. The vault over this stair, of which I give a photograph ([Plate 16], Fig. 2), will serve to illustrate the construction of all vaults at Ukhaiḍir over an inclined plane. They are built in horizontal sections, which form inverted steps; an unbroken rising vault is not to be found in the palace. To the east of this group of rooms with its stair is the cloistered court which I suggested, after my first visit, might be a mosque.[21] The suggestion has been borne out by the discovery of an arched niche in the south wall, which I believe to be the miḥrâb.[22]
The mosque (since I may now give it this title without hesitation) is approached by two doorways from the west corridor, 5. These doorways lead into an open rectangular court, the ṣaḥn, 10·30 metres from north to south by 16 metres from east to west. To east, south, and west of the court ran porticoes, or riwâqs, to use their Arabic name, which have now fallen ([Plate 18], Fig. 1). The engaged columns on the north side and the south-east angle pier are, however, standing, and they determine the width of the riwâqs. The southern riwâq was the widest (4·05 metres), and this is the portion of the mosque which is known as the ḥaram. The east and west riwâqs are alike 3 metres wide. The arcades, which separate the riwâqs from the ṣaḥn, occupy a space 1 metre thick. On the west side the arcade is entirely ruined, but on the east side part of the arches at either end are still to be seen ([Plate 19], Figs. 1 and 2). From these fragments it is apparent that there must have been three arches on the east and west sides, while approximately similar proportions would allow five arches on the south side. (The span of the south arches must have been about ·30 metre less than the span of the east and west arches.) The north end of the east and west vaults rested against the north wall, the south end against a transverse arch, in order to avoid intersection with the vault of the ḥaram. The east vault, which is best preserved, is a slightly pointed ovoid and oversails the east wall. Below the spring of the vault can be seen the windows of rooms 8 and 9; the window of room 10 opens into the ḥaram. Immediately above the springing of the vault there are three holes for cross beams, the decay of which has entailed the ruin of the vault. The fallen masses of masonry columns and vault form heaps of débris on all three sides of the court. At the eastern end of the ḥaram there is a low door, almost blocked by ruin heaps, which gives access to a narrow blind passage situated under the stair. The vault of the ḥaram has received an elaborate decoration in stucco. It was divided into sections by nine transverse arches, 1 metre wide. They cannot have had any correspondence with the columns and arches of the arcade, nor was this necessary, for they sprang from above the line of the vault and therefore from above the summit of the arches of the arcade. The transverse arches were decorated with lozenges (wards as they would be called in modern Arabic) having a zigzag outline ([Plate 18], Fig. 1). In the centre of each lozenge there was a round hole, or rosette, recessed back in concentric circles. Between the transverse arches the vault was worked in parallel bars of stucco, the one oversailing the other. The bars begin at a distance of about ·80 metre above the spring of the vault. It is evident that this vault must have been constructed over a light centering, and Dr. Reuther is of opinion that the singular ridged decoration was suggested by the impression left by the centering boards upon the plaster.[23] The top of the vault was probably treated as in room 31, where a decoration similar to that of the ḥaram is more fully preserved. Holes for cross-beams break the fourth and fifth stucco ridge between each transverse arch. Between the terminal transverse arches and the wall at either end of the ḥaram there is a space 1·60 metres long. It is divided into two quarter-domes by a transverse arch which springs from the back wall, at right angles to the transverse arches of the vault. This arch is decorated in exactly the same manner as the others and must have joined the first transverse arch at either end, at the summit of the vault. The quarter-domes are covered with stucco ornament. At the east end ([Plate 20], Fig. 1) a fluted squinch occupies the two angles; on either side of it are two shallow calottes. Three concentrically recessed rosettes are set above each of the calottes, and there is a like motive in the apex of the calotte. Above the squinch and calottes there is a band of four isolated crenellations, the same motive which appears on the archivolt over the doors of corridors 5 and 6. Above the crenellations are vestiges of a decorated band, and above the band the apex of the quarter-dome is fluted. At the west end there is a slight variation in the proportions and in the motives of the lower register of the quarter-domes ([Plate 20], Fig. 2). The squinch, instead of being fluted, is decorated with three concentric bands, sunk one within the other. At its base lies one of the usual concentric rosettes. The same rosette is placed on either side of each calotte and within the calotte, the rosette above the calotte being omitted. The crenellated motive of the east end is repeated at the west end, but the band between the crenellations and the flutes of the quarter-domes is omitted.
The miḥrâb niche is not placed exactly in the centre of the south wall, but a few centimetres to the east ([Plate 18], Fig. 2). If there was any stucco ornament upon it, it was all carried away by the fall of the vault. The semi-dome which covers it is set over the rectangular niche on horizontal brackets of masonry, like all other semi-domes and calottes in the palace. The archivolt is constructed of a double ring of voussoirs, the inner ring laid vertically, the outer horizontally. There is no reason to doubt that the miḥrâb is contemporary with the wall. The plaster which remains upon the interior of the semi-dome shows no sign of decoration. Below the semi-dome the face of the walls of the niche is much injured by the heavy masses of fallen masonry.
The angle pier which took the corner arches of the ḥaram and the east arcade shows, on the sides facing the arcades, returns in the shape of engaged columns. A third return is rectangular and corresponds with a return on the east wall, the two carrying the transverse arch which terminates the eastern vault. In the fragment of this vault which is standing the principles of construction can be discerned unusually well ([Plate 19]). The vault is built of thin slabs of stone, laid in rings, with a marked inclination against the northern head wall. At the southern end these rings fan out so as to meet the transverse arch.
One more detail remains to be noticed. The two doors from the west corridor, 5, stand in recesses 1 metre deep. The recesses are covered by a calotte, and round the archivolt is placed a stucco decoration consisting of seven cusps ([Plate 21], Fig. 1).
The first floor of the north gate tower has already been described. The east door of room 90 communicates with the vaulted and unlighted room, 93. A thin dividing wall separates room 93 from room 94 (there is a small aperture like a window in this wall). Beyond another thin dividing wall lies room 95, with a window at its eastern end looking into the palace yard. These three rooms, 93, 94, and 95, occupy the space above the east corridor, 6. Room 107 is on a lower level; it is approached from 93 by a doorway with steps and is wholly unlighted. The group of rooms Nos. 103, 104, and 105 are on the same level as 107. They are 14·75 metres long and correspond in width with the rooms below them. At their western end they are provided with a masonry divan, 1·20 metres wide and raised ·55 metre above the level of the floor. The meaning of this divan is apparent in the section (section a-b, Plate 4, Fig. 1); it was needed in order to lift the floor of the three rooms above the vaulted tube which lies parallel to the vault of the great hall. The height of these rooms from the floor to the top of the vault is 4·20 metres. They communicate with each other and with the vaulted passage 108, and room 103 possesses further a door in the south wall leading into room 102. The latter returns to the level of rooms 93, 94, and 95, and consequently steps are placed in the doorway of 103.
At the north end of the passage 108 there is a door sunk below the level of the floor and covered by an arch oversailing the jambs ([Plate 21], Fig. 2). It communicates with the ramp which comes up from the great hall. East of this door there are the remains of an engaged column, and it is obvious that the passage must have been flanked here by an open arcade ([Plate 3], Fig. 1). Steps in the doorway at its southern end lead up to room 106, which is on the same level as 102. South of court A lie three rooms, 109, 110, and 111. They are not as deep as the rooms below them on the ground floor (4·40 metres as against 5 metres) since space has to be provided for a narrow ledge above court A. On to this ledge the north doors of the three rooms open. On the north side of court A the ramp, after passing the doorway of 108, is continued upwards (its windows can be seen in the wall of the court ([Plate 22], Fig. 1)). A wide doorway opens on to a stair, which will be described later, coming up from the palace yard. The ramp is then carried on along the east side of court A, and finally opens on to the roof of 111 and of the narrow passage to the east side of it. The last portion of the ramp is ruined, but traces of the vault which supported its floor can be seen in the east wall of court A, together with the spring of the vault with which it was roofed. Between the ramp and the vault of 25 there appears to have been a vaulted passage, very low at its northern end, and lighted by a rectangular window which overlooks the palace yard. It opened at the southern end, through a narrow vaulted way, on to the roof of No. 47.
The outer stair from the yard is a later addition ([Plate 40], Fig. 1). The round tower at the northern end of the wall has been cut away to receive it, and it was supported further by four rectangular piers, two on either side of the tower, which were built up against the wall. These piers were not bonded into the wall, and the northernmost has entirely fallen away, but it can still be traced on the face of the masonry. The communication with the first floor was effected, as has been mentioned, by means of a door at the north-east angle of the ramp.
Room 106 occupies the vaulted space at the west end of 47 and has a door to the south opening on to the roof of 45. To the west a door leads into corridor 102, which lies above the eastern wing of corridor 28 ([Plate 22], Fig. 2). It has a door to the south opening on to the roof, and is lighted by narrow windows in the south wall. West of 102 was the small room, 101, now ruined, and beyond it rooms 100 and 99 above the west wing of corridor 28. The height of these rooms on the first floor is only 3·55 metres to the top of the vault. No. 100 communicates by a door and steps with the stair leading up from the south-west corner of the great hall, and so with the first floor chambers of the west wing. These can be approached also from the west door of room 89, which opens into the passage room No. 92. In the south wall of 92 there is first a door and steps which lead down to No. 96, secondly a door giving access to the roof of the east riwâq of the mosque, and further west a narrow window which overlooks the ṣaḥn. There are two similar windows in the south wall of 91 and a door on to the roof of the west riwâq of the mosque. (The windows and the door of the west riwâq can be seen in Plate 23, Fig. 1.) At the western end of 91 a window opens on to the palace yard. Rooms 96, 97, and 98 lie above 8, 9, and 10. They are lighted by narrow windows in the west wall, which can be seen in Plate 19, Fig. 1. They communicate with each other by doors covered by ovoid arches set back from the jambs and breaking into the curve of the vault, and each has access through an arched opening in the east wall to a small room ·85 metre wide, lying at a higher level. The northernmost of these three small rooms lies under the stair leading from No. 89 to the second floor, and its vault slopes down at the northern end in order to leave space for the stair. No. 98 opens by a door on to the staircase from the great hall. At the west end of the staircase there is a door leading out on to the roof of the ḥaram, and above it is placed a window. Both door and window can be seen in Plate 19, Fig. 1. Opposite to this door and window there is a large opening in the west wall of the great hall, doubtless in order to secure a little additional light in that dark edifice.
The stair and the ramp from the great hall were therefore the sole means of approaching the first floor until the outer stair from the yard was added. The second floor could be approached in a circuitous manner by the upper part of the ramp and over the roof of rooms 111, 110, and 109, or more directly by the stair leading out of room 89. But this stair could only be reached either by the ramp and through rooms 105, 107, 93, 90, 88, and 89, or by the stair out of the great hall and through rooms 98, 97, 96, 92, and 89. The second floor could also be reached from the yard, by the stairs in the north-east and north-west angles and thence along the chemin de ronde.
The rooms on the second floor do not correspond regularly with those of the floors below ([Plate 3], Fig. 2). The second floor of the gate-tower is much ruined. It is possible that, as the authors of Ocheïdir suggest, it was originally divided into three chambers lying north and south. Parts of the south wall remain, and there is clear evidence of a door jamb near its eastern end. On the east side the doorways leading into 117 and into the chemin de ronde are standing, together with the south jamb of a doorway which undoubtedly gave access to the roof of the vault between the gate-tower and the first round tower. The door into the corresponding balcony on the west side is gone, the door of the western wing of the chemin de ronde is much ruined, but the door into No. 116 is still perfect. Neither of these walls, to east and to west, shows any trace of a vault; the vault, if vault there were, covering the gate-tower chambers must therefore have sprung much higher than the vaults of the adjoining chambers.[24]
To the west of 116 is a small room, 115, with a door into the chemin de ronde and a door into the open court, 114. A window in the south wall of this court overlooks the ṣaḥn of the mosque ([Plate 23], Fig. 1). Still further west is a vaulted room, 113, presumably with a window looking out into the yard, but the west wall is much ruined. On the opposite side of the gate-tower, No. 117 opens into a small rectangular area, 118, where there is no sign of a roof; to the east of it lies an open space embracing the roofs of Nos. 94 and 95 together with a part of 93. Here, too, there is no trace of a vault in the north wall, nor of any party walls. The series of rooms on either side of the gate-tower, occupying the area over the corridors on the ground floor and of the corresponding rooms on the first floor, are designated by Dr. Reuther casemates because they were connected with the chemin de ronde and probably played some part in the defence of the palace. In all of them the vaults, which oversail the walls in the usual fashion, are slightly flattened at the top.
A door in the south wall of No. 117 leads into an open court, 16·95 metres from east to west by 12·60 metres from north to south. It does not lie in the centre of the three-storied block, but extends considerably to the east of the central axis. The stair from the first floor reaches the second floor at the north-west angle of this court. The door into 119 opens awkwardly over the stair. On the east, south, and west sides of the court stand groups of three chambers, the central chamber opening into the court by a wide archway springing from engaged columns, the side chambers by doors covered by ovoid arches set back from the jambs ([Plate 23], Fig. 2); and here we have an architectural group which dominates all the courts upon the ground floor of the palace that are yet to be described. The central chamber with its wide archway is the lîwân or reception-room,[25] the side chambers are, in one form or another, its invariable or almost invariable complement. I shall henceforward speak of the whole as a lîwân group. As Dr. Reuther has pointed out, the occupants of an oriental room seat themselves upon cushions or dîwâns against the wall, the dîwân, cushion or carpet, which is placed against the back wall, being the place of honour. In order not to break up the company, the side doors of every room are situated as far as possible from the back wall, and it will be noticed that this rule holds good in every living-room of the palace. At Ukhaiḍir (though this is not always the case) in every lîwân group the rooms communicate with each other. It is common in oriental houses to build lîwâns facing different points of the compass so as to secure a comfortable shade at different hours of the day, and warmth or coolness at different seasons of the year. The lîwân group, if such it were, over the gate-tower would have served the purpose of a winter reception-room, for it faced south; the group facing north would be used in summer.
In the lîwân group on the west side of the court the rooms are 5·95 metres long with an average width of 4 metres. The vaults here are all standing, and the rooms are considerably higher than those on the first floor, measuring 5·25 metres to the top of the vault. (It is difficult to get exact measurements for the height of the rooms on the ground floor owing to irregularities in the level of the ground, but I think that a height of 5 metres to the top of the vault is not far wrong.) Between the parallel barrel vaults are masonry tubes, which are visible upon the façade in the form of small openings like windows between the arches of the central and of the side rooms. To the south of No. 121 there is a small open court, 123, which is approached by a narrow passage from the main court. A door from it leads into No. 122, which is completely ruined. On the north side of the court, 123, there was a stair which gave access to the flat roof of Nos. 121, 120, and 119. On the north side of 119 a fragment of wall rises above the level of the roof; it was probably connected with the high vault of the gate-house chambers. In the lîwân group on the south side of the court, the rooms, 124, 125, and 126, are 7 metres long, but their exact width is difficult to determine since the party walls have fallen ([Plate 24], Fig. 1). It must, however, have averaged about 4 metres like the width of the rooms on the west side. On the east side of the court a vaulted passage runs parallel to 137; the door into the court is standing and its arch oversails the jambs, whereas the arches of all the other doors are set back ([Plate 24]). Above the door there is a narrow window. A lîwân group follows to the south of the passage ([Plate 24], Figs. 1 and 2). The rooms are 7·45 metres long; their width varies, as far as I could ascertain in their ruined condition. According to my estimates No. 132 is 2·85 metres wide, No. 131 is 3·95 metres wide, and No. 130 is 4 metres wide. Still further south there is a small open court, No. 127, corresponding to No. 123. A door in the south wall opens on to a narrow parapet or balcony which crowns the façade of the first floor. To the east lies an irregular chamber, 128, which is totally ruined.
The passage, 137, leads into a gallery, No. 134, which was finished on the east side by an open arcade ([Plate 25], Fig. 1). Traces of an engaged column remain at the north end of the arcade, and the vault was constructed with transverse arches in the same manner as the vaults round the ṣaḥn of the mosque. There was, however, no stucco decoration in this upper gallery. At the angles stood quarter-domes over unadorned squinch arches ([Plate 25], Fig. 2). The gallery opens at its south-eastern end on to the roof of No. 109. To the south of the gallery there are two narrow chambers, one with a door into the gallery, the other with a door on to the roof of 109. They are almost completely ruined. Dr. Reuther places in them a stair leading by a double flight on to the roof.
The main part of the palace, one story high, lies to the south of the three-storied block. Except for a group of rooms in the east side of the yard, which is a later addition, it is symmetrically arranged round a central court. It falls into three divisions: two courts, B and C, with their living-rooms on the east side; two exactly similar courts, G and H, on the west side; a central court with a group of chambers to the south of it, and further south a small court, E, with rooms on three sides of it, and a subsidiary court, D, further east. The long vaulted corridor, 28, which runs from east to west between the great hall and the central court, turns at right angles and runs from north to south between the central court with its chambers and the side wings. It is then carried round to the south of the chambers dependent on the central court, and runs from east to west between them and court E with its chambers.
The central court is 32·70 metres from north to south and 27 metres from east to west. It is surrounded to east, north, and west by a blind arcade which forms part, on the north side, of the façade of the three-storied block ([Plate 6], Fig. 2). The arcade is 1 metre deep. Engaged half-columns set against rectangular piers carry shallow calottes, the archivolt of which is slightly horse-shoed ([Plate 26], Fig. 1). The intercolumniation varies from 2·35 to 2·55 metres. All the details were of stucco, which has now broken away. The columns, piers, and walls are of stone masonry; the capitals, calottes, and archivolts, together with the wall above them, are of brick. The capitals, which are much damaged, are cubes formed of three courses of bricks; the calottes are of brick laid in horizontal courses and carried over the angles of the niches by horizontal brackets; the horse-shoed archivolts are composed of an inner ring of brick tiles laid horizontally, and an outer ring laid vertically. Of the outer ring only fragments remain. In one case (the calotte immediately to the south of the east door) the tiles are laid in rings, and the curve of the archivolt is not horse-shoed ([Plate 26], Fig. 2). The corresponding calotte on the west side has fallen. In the centre of each calotte, and impinging upon the stonework below, there is an oblong window which lights corridor 28. On the north side of the court only two of the niches and calottes remain intact to the east of the central door, and only one to the west of the central door. In the centre the whole face of the wall has fallen, carrying with it parts of the corridors on the first floor and part of the south wall of the great hall. The small chamber, 27, which was probably covered with a dome, is entirely ruined, together with room 101 above it. It is therefore impossible to determine the exact form of the doorway which led from 27 into the central court, but there is no reason to suppose that it differed materially from the door on the east side of the court. The nature of the horizontal decorations which govern the façade preclude all idea of a large central door. The blind arcade of the first floor is not so high as the arcade below it (Plates 27 and 85). Instead of the half-columns and piers of the ground floor, the archivolts of the first floor spring from a cluster of four small engaged columns which must have been finished in stucco. Nothing remains of the capitals. In the spandrels are placed oblong windows lighting the upper corridors, 100 and 102. On the face of the pointed arches of the arcade it is still possible to trace a scolloped ornament in plaster, like that which exists over the doors of the mosque. Within the large arches there is a system of small blind arched niches flanked by slender engaged colonnettes of which little trace remains. There are five of these niches within each of the large niches, two below and three above, the central niche in the group of three being the largest. There is a slight error here in Dr. Reuther’s reconstruction, an error to which he himself called my attention. He has placed only one small niche in the upper register instead of three. The side niches can be seen in Plate 27. He suggests that in the middle of the façade one or more of these small niches must have contained windows in order to give additional light to room 101, since it was from room 101 that most of the light in the great hall was derived. Beyond the arcading on either side of the façade the wall was finished by a solid pier, the surface of which was broken by three projecting horizontal bars. The cornices are not preserved, but, as I shall show later, they cannot have been very important. The decoration of the façade ends on the level of the second floor and forms a narrow balcony a little over 1 metre wide which runs along the face of the building. The wall of the second floor is recessed a few centimetres to give additional width to this balcony. On to it open the doors of Nos. 123 and 127. These doors are not placed symmetrically with respect to the façade; the west door is nearer the centre than is the east door. The plain wall is carried up to the top of the door arches; above that level there is a band of shallow arched niches which appear to have been divided from one another by engaged columns, probably carrying an architrave, like the niches on the summit of the outer north wall of the palace.
To return to the central court. On the east side there is a doorway in the third intercolumniation from the south end ([Plate 26], Fig. 2). It leads into corridor 28. The arch of this door is set back from the jambs, but the upper part is ruined. The corresponding door on the west side has disappeared, together with most of the south-west end of the wall. On the east side the arcading is not carried into the angle of the court. The southernmost archivolt ends against a quarter-column, beyond which space is provided for the entrance of a stair which leads down to a vaulted chamber below the level of the ground ([Plate 28], Fig. 1). Above this entrance there is a fluted semi-dome finished by a fillet ([Plate 28], Fig. 2). The semi-dome is set horizontally over the angles of the niche in the accustomed manner. The actual entrance to the stair is covered not by an arch but by a masonry lintel (compare the door between 20 and 21).
The south side of the court is also arcaded, but not in the same fashion. The arcades are much shallower (·40 metre deep) and they are differently grouped. In the centre of the south wall there was a wide archway (4·20 metres wide) leading into room 29. This arch rose above the level of the arcade on either side of it and the chambers behind it were higher than the adjoining chambers ([Plate 29], Fig. 1). On either side of the entrance there is an unusually large engaged column; beyond these columns there is a flat pier and an engaged quarter-column, followed by a niche ·80 metre wide covered by a shallow calotte ([Plate 29], Fig. 2). Three more recesses, measuring in width 1·95 metres, 2·10 metres, and 2·50 metres, and separated from each other by engaged columns of about ·70 metre diameter, occupy the remainder of the façade. In no case is the capital preserved, but it is noticeable that all the columns swell outwards towards the top. The archivolts are ovoid, not horse-shoed. The first niche on either side of the small niches contains a door leading on the west side into No. 31 and on the east side into No. 42. The third big niche on the east side contains another and a smaller door which gives access to a stair leading to the roof ([Plate 28], Fig. 1). The doors of Nos. 31 and 42 offer good examples of arch construction ([Plate 29], Fig. 3). The arch is set back from the jambs and formed of an inner ring of concrete and an outer ring of stone voussoirs laid horizontally. The calottes covering the niches are of brick, but unlike the calottes on the other three sides of the court, the bricks are set horizontally and vertically and used in half and quarter lengths so as to form intricate designs which Dr. Reuther compares very aptly to the Hazârbâf motives so common in oriental woodwork ([Plate 29], Fig. 2).
South of the central court lies a group of rooms of a ceremonial character. In the centre of this group is the lîwân No. 29, 6 x 10·70 metres. It was covered by a barrel vault of brick, which has now fallen in. The vault oversailed the wall and its point of springing is 4·30 metres above the level of the ground, instead of the 3·40 metres above ground-level at which the vaults spring in the adjoining chambers to east and west. It is therefore clear that the vault of 29 must considerably have overtopped the other vaults, and as I shall show later, it is usual to find the ceremonial lîwân higher and more important than the remaining chambers of the group. I have followed Dr. Reuther in giving it a rectangular frame upon the façade of the court (section e-f, Plate 5, Fig. 1). Two large doors, 1·50 metres wide and 3·64 high to the top of the arch, open on either side of the lîwân, on the east into rooms 41 and 42, and on the west into rooms 31 and 32, which lie at right angles to the lîwân. At the south end a similar door leads into No. 30, a chamber 6 metres square, which has been covered by a barrel vault of brick running north and south, and doubtless the same height as the vault of the lîwân. Doors of the same character, with ovoid arches set back from the jambs, are placed in the middle of the east, south, and west walls of No. 30. The fact that the high vaults of Nos. 29 and 30 were not sufficiently buttressed by the lower vaults on either side accounts for their fall.
Rooms 31 and 32 are distinguished by a plaster decoration more elaborate than any which is to be found elsewhere in the palace, with the sole exception of the mosque. The vault of No. 31 resembles the vault of the ḥaram, and like the ḥaram vault it must have been built over a centering. It is divided into two compartments by three transverse arches, one spanning the centre of the chamber, the other two placed respectively against the east and west walls ([Plate 30], Fig. 1). These transverse arches, which are ·95 metre wide, spring from a double outset at a height of 2·80 metres from the ground. The vault between the arches springs at a point ·25 metre higher. It is composed, like the ḥaram vault, of narrow oversailing ridges worked in stucco. Along the top of the vault are placed between each pair of transverse arches four square stucco motives, some of which remain intact. They differ slightly from each other, but all are variants of the same theme ([Plate 30], Fig. 2). The first from the east end consists of four squares within one another, like a Chinese box, each sunk behind the other. In the centre there is a circular rosette, doubly recessed. In the second a single recessed square contains a saucer-shaped motive, the surface of the saucer being covered with rings of small plaster excrescences. In the third the usual recessed square is filled with a triply sunk diamond, with a recessed rosette in the centre. In the fourth the recessed square frame is filled with a recessed diamond, within the diamond is a recessed square, within the square a second recessed diamond, in the centre of which is a rosette. In the western compartment two of the motives consist of squares sunk within one another, a third of a doubly sunk square containing a triply sunk rosette, while the fourth is obliterated. Finally high up in the east and west walls under the vault is placed a small niche whereof the arch springs from engaged colonnettes.
No. 31 is connected with No. 32 by a door opposite to the door in the central court. The construction of the roof in No. 32 is different from any other example of roofing in the palace. It is divided into three compartments by four heavy transverse arches which spring at a height of 2·85 metres from the floor, level and are set forward twice from the face of the wall ([Plate 31], Fig. 1). Between the arches small barrel vaults are stretched across the chamber from north to south. In the eastern compartment the north and south head walls are carried up to the height of the vault. Immediately below the spring of the vault there is a sunk band in the head walls decorated with three recessed circles or rosettes. In the central and western compartment the vault terminates against a semi-dome, set over the angles in one case horizontally, in the other (the western compartment) by means of small recessed squinches (compare the west end of the ḥaram). Below the semi-domes there are a couple of narrow fillets, and below the sunk band of the eastern compartment a single wide fillet. Below these, at the same level in all the compartments, the head wall is decorated with pairs of arched niches, the arches being supported by engaged colonnettes. The colonnettes have no bases; a narrow impost serves them as capital. The face of the arches is decorated in two of the compartments by fillets and in the third (the western) by a zigzag motive. Within each niche there is a spear-shaped ornament sunk in the wall. In the spandrel between the arches there lies a recessed rosette. At a height of ·35 metre above the springing point of the transverse arches the head wall is set very slightly forward, in imitation of the outset of an oversailing vault. The arches of the doors rise higher than the level of this outset, which is lifted in a rectangular label over them. The barrel vaults between the transverse arches are variously treated. The eastern vault is divided into sections by three short transverse arches, each of which is decorated by a square sunk motive. The central vault has the same number of short transverse arches, but these are undecorated. The western vault is provided with a transverse arch against the semi-dome at either end, while the remainder of its length is decorated with stucco ridges. A pair of niches, smaller than those upon the side walls, is placed in the east and in the west wall under the transverse arches, but the spear-shaped ornament and the recessed rosette of the side niches is omitted.
Rooms 31 and 32 are 10·05 metres from east to west and 4·90 metres from north to south. Room 41, lying opposite to room 32, has an equal length and the same system of doors, but no decoration. Room 42, which corresponds with room 31, is only 7·25 metres from east to west, since space had to be allowed for the two stairs leading out of the central court, one to the roof and one to the underground chamber. In the south-east corner of No. 42 there is a small door giving access to a narrow passage behind the block of masonry which contains the upper stair. It turns at right angles into a short passage lying above the lower stair. The vaulted underground chamber corresponds in length and width with No. 42 (section e-f, Plate 5, Fig. 1). It is lighted by three small windows which are splayed upwards to the ground-level—one of these can be seen in Fig. 3 of Plate 29. The room was filled with débris, so that I cannot be certain of its height. In the west wall there is an arched niche or ṭâqchah. In the intense heat of southern Mesopotamia it is customary to provide all houses with underground chambers, wherein the inhabitants spend the greater part of their day in summer. They are known as serdâbs. To the authors of Ocheïdir I am indebted for an interesting observation with regard to the vault of No. 41.[26] It was built in sections over a movable centering which has left its mark upon the concrete of which the vault was formed.
Rooms 32 and 41 communicated by doors in the south wall with the columned chambers 33 and 40 ([Plate 31], Fig. 2), which are exactly alike in every respect, except that No. 40 is connected by a door with the room to the south, No. 39, whereas there is no south door in No. 32. Both 33 and 40 have doors, covered with ovoid arches set back from the jambs, leading into the corridor 28, and both are divided into three aisles by two arcades of three arches carried on two masonry columns. The aisles run north and south. The innermost aisle in either case forms part of the vaulted corridor, 36, which runs round three sides of No. 30. This aisle is only 2·50 metres wide, as compared with the 2·85 metres of the other two aisles. All the aisles are roofed with barrel vaults. Though the columns are of stone masonry, the capitals, together with the arches and walls they carry, and the segmental vaults, are of brick. The columns are separated from one another from north to south by a distance of 2·50 metres, but the distance between each column and the wall behind it is only ·90 metre; hence the wide central arches rise almost to the spring of the vault, whereas the side arches are from their narrow span necessarily much lower ([Plate 32], Fig. 1). The curve of all the arches is a pointed ovoid, and the narrow arches are considerably stilted. These last are built of concentric rings of small brick tiles, the inner band laid vertically, the outer horizontally. The large arches are composed of two concentric rings of voussoirs, both laid vertically, the inner ring being of large tiles used in their full size, the outer ring of half of the same tiles. The capitals are better preserved than any in the palace, and from one of the capitals of No. 33 in particular, an excellent idea of the form of the impost-capital commonly used at Ukhaiḍir can be obtained. (It is the capital seen in Plate 32, Fig. 1.) The cube of the capital is adapted to the circle of the column by placing an angle of brick under each corner. The capital is composed of a shallow ovolo in moulded plaster surmounted by an abacus which consists of a single course of bricks and carries an impost formed of three courses of brick. Within the arches the impost slightly oversails the abacus.
On the south side of corridor 36 the vault has fallen, together with the columns between the engaged piers which must have supported the arcade ([Plate 31], Fig. 3). The spring of the arches can be seen against the piers. From the fragments that exist, the barrel vaults do not seem to have intersected one another but to have met diagonally at the angles. At the east and west ends of No. 36 a door opens into rooms 39 and 34. No. 34 communicates with a parallel chamber, No. 35, which opens independently upon the narrow open court, F, between 36 and the corridor 28. The eastern side of this court was much ruined. In the south-east corner was a stair which led up to the roof. To the north, and partly under the stair, lies a small room, 38, communicating with another narrow room, 37, which was not entirely vaulted over. That it was intended to contain a fire is clear from the fact that the vault is pierced by two terra-cotta pipes, the one 29 centimetres in diameter, the other 12 centimetres, which must have served as chimneys. Similar pipes occur elsewhere and will be mentioned later.
The long corridor, 28, which lies to east and west of the central court and its group of chambers, turns at right angles and encloses the whole central block. The corridor is covered by a semicircular stone vault, oversailing the walls; at four points, however, it is left unroofed in order to admit light and air. These openings are flanked by transverse arches, springing a few centimetres lower than the spring of the vault. The angles of the corridor are roofed with groined vaults, and groined vaults occur in two places, towards the middle of each of the long sides of the corridor. Moreover, a small extension of the east arm of the corridor, No. 61, is also roofed with a groin. This last is the example given by Dr. Reuther on Plate 13 of Ocheïdir; it is the only groin in the palace which is built of brick. Where the groins do not rest on the head wall, they are laid against transverse arches, springing from a point lower than the springing of the vault. The lower parts of the groin are built of stones laid horizontally and forming a bracket from which spring the intersecting vaults ([Plate 32], Fig. 2). The vaults are also built of thin slabs of stone, cut in the shape of bricks, and laid with a slight inclination backwards against the head wall or the transverse arch. This construction demanded little or no centering. In the north-east angle of the corridor there is a small door in the east wall which gave access to a stair or passage running under the wall. It was so much blocked by ruins that I could not penetrate into it.
From the corridor a door opens into each of the five courts, B and C on the east side, forming the eastern wing of the palace, H and G on the west side, forming the western wing, and E to the south. The courts have no direct communication with each other. The chambers on the north and south sides of these courts are all arranged in lîwân groups, but there are differences in detail between courts B and H on the one hand, and courts C and G on the other, while the position and size of court E has led to further modifications. Court B ([Plate 33], Figs. 1 and 2) measures 15·20 metres from north to south, and 17·60 metres from east to west, but on the west side ·40 metre is occupied by a shallow blind arcade, and on the east side 3 metres was taken up by an arcaded passage which is now ruined. The blind arcade is composed of five arches carried by engaged piers which have an average width of ·70 metre. The arches are round and spring directly from the piers without the interposition of impost or capital. In the central of the five intercolumniations is placed the door from the corridor. To the north and to the south of the court lies a lîwân group of three vaulted chambers. The lîwân opens on to the court through an archway 2·60 metres wide flanked by engaged columns and piers ([Plate 34], Fig. 1). The side chambers communicate by means of arched doorways with small antechambers, which in turn open into the court through arched doorways 2·05 metres wide, flanked by engaged columns ([Plate 34], Fig. 2. The mass of brickwork which partly blocks the doorway is a later addition). The antechambers are roofed with barrel vaults running east and west, which are separated from the outer end of the lîwân vault by transverse arches; thus the vault of the lîwân is enabled to run through to the wall of the court ([Plate 35], Fig. 1). Structurally, the antechambers are therefore distinct from the outer end of the lîwân; practically the antechambers and the outer end of the lîwân form a kind of narthex, the outer end of the lîwân being part of the narthex and not an integral part of the reception-room. This fact is accentuated by the position of the side doors in the lîwân. The sitting space along the walls ends with these doors, and for practical purposes the lîwân is no longer than the side chambers. The capitals of the engaged columns are rectangular impost blocks of stone masonry. Between the parallel barrel vaults there is the usual system of tubes ([Fig. 4]). The tubes running north and south are carried over the transverse arches of the antechambers, and their openings appear on the façade of the lîwân groups. Where the façade has fallen, as, for example, on the south side of court B, the construction can be clearly traced, and it is also possible to observe that tubes ran from east to west between the wall of the façade and the barrel vaults of the antechambers, as well as on the inner side of the same barrel vaults. Perhaps these tubes were connected with a tube running north and south parallel with the vault of the corridor. The vaults are ovoid and are constructed of a single course of stones laid vertically supporting a mass of stone and concrete. In all the interior doors the arches are set back from the jambs ([Plate 36], Fig. 1) and constructed in the manner described on p. 15. Upon the plaster of the west wall of No. 44, south of the door leading into No. 45, there is a graffito inscription in Arabic (see below, [p. 161]).[27]
Fig. 4. South side of court B. (From Ocheïdir, by kind permission of Dr. Reuther.)
East of the lîwân group on the north side of court B there is a stair, and still further east a narrow passage within the outer wall. A small door in the north-east corner of the side chamber, 46, gives access to an unlighted blind passage under the stair. The stair runs up to a landing-place which is connected by a low doorway with a small chamber situated above the eastern passage. Another door leads into a gangway hollowed out of the thickness of the outer wall, and from this gangway a door leads into a tiny circular room in the outer towers. I did not determine whether the gangway in the wall runs on interruptedly from court to court. On the whole, as Dr. Reuther has observed, this would seem to be improbable since the strict isolation of the courts is in all other respects preserved. Almost exactly above the entrance to the stair (an awkward piece of construction) sprang the first arch of the arcade which flanked the court from north to south. In every court this arcade has fallen, but on the south side of court H a portion of the first arch remains, together with the vault behind it ([Plate 35], Fig. 2). I cleared away the ruins at the south end of this arcade and found the remains of the first column at a distance of 2·40 metres from the south wall. The arcade must therefore have been composed of four columns carrying five arches, corresponding with the blind arcade on the opposite wall. The massive stone vaulting of Ukhaiḍir was not suited to free standing arcades, and, as has been noticed in the mosque, when the wooden cross-beams perished, their collapse was inevitable.
To return to court B. The passage already mentioned, running parallel with the outer wall, leads into an oblong room, 47, 3·55 metres wide, which lies from east to west across the back of the lîwân group and the stair. This room is vaulted at either end but is left open near the centre ([Plate 35], Fig. 3). The same oblong room is found behind the southern lîwân group of court B, and behind each of the lîwân groups in courts C, G, and H. In every case the vault next to the outer wall is pierced by a pair of terra-cotta pipes similar to the pipes described in No. 37. It is probable, as I shall show later (p. 82) that these rooms were intended for kitchens. On the south side of court B there is no stair; above the vault of the passage which leads into the oblong room, 51, there is a blind corridor accessible from No. 50 by a door placed in the east wall, some 2 metres from the ground. This door must have been approached by a wooden ladder or steps, but I climbed up into it over a heap of ruins. On the west side the antechamber of No. 49 is provided with a door into corridor 28. Immediately to the south of this door a wall, broken by a doorway, has been built across the corridor. This wall is a later addition; it is not bonded into the walls of the corridor, and it does not occur in the corresponding west arm.
Court C differs from court B in the absence of antechambers to the lîwân groups ([Plate 33], Figs. 3 and 4). The lîwân opens into the court through a wide pointed arch carried on engaged columns; the side chambers are provided with doorways into the court, covered by ovoid arches set back from the jambs ([Plate 36], Fig. 2), and the façade thus formed corresponds exactly with the façades of the court on the top floor of the three-storied block. Near the south-east corner of court C there is an arched doorway leading into the palace yard ([Plate 37], Fig. 1). In the oblong chamber, 60, behind the southern lîwân group, the south wall is occupied by a blind arcade of four arches borne by piers 1·10 metres wide and 1·05 metres deep. A similar blind arcade occurs in the corresponding chamber of court G, and indeed, except for slight variations in the measurements, the only difference between courts C and G is that in the latter there is no door into the palace yard. In the same way court H re-echoes court B save that in court H there is no doorway between the southern antechamber, 82, and the corridor 28 ([Plate 37], Fig. 3).
The arrangement of the rooms in court E is not symmetrical. On the east side court E is curtailed by the small oblong room, 61, and an open court, D. No. 61 is a continuation of the east arm of the corridor 28. It measures 5·25 metres from north to south and 3·50 metres from east to west. The square for the brick groin with which it is roofed is obtained by laying a transverse arch to north and south. It opens by two arched doors, divided by a pier, into court D, which measures 10 metres from north to south and 9·20 metres from east to west. In the south wall there is an arched doorway into the palace yard. To the east of court E there is space for one chamber only (62) and a winding stair which leads to the roof. On the west side there are two chambers, 67 and 68, communicating with one another and with the court. To the south of 67 there is a narrow passage ([Plate 37], Fig. 2) which leads into an oblong room, 69, similar in all respects to the oblong rooms behind the lîwân groups in courts B, C, G, and H.[28] Between the barrel vaults of 67 and 68 and the south arm of corridor 28 are the usual tubes. The doorways of 67 and 68 are covered with ovoid arches set back from the jambs, but the opening into the narrow southern passage follows the line of the vault and oversails the wall. Above the vault of the passage there is an inaccessible passage or tube which exists for structural reasons only. To the south of court E lies a lîwân with its side chambers, the lîwân, 64, opening into the court by a wide archway, the side chambers by small doors, as in courts C and G. Finally, the space between 65 and 69 is filled up by a fourth room, 66, which communicates with 65 and with the narrow passage. Tubes are laid between all the barrel vaults of these rooms.[29]
The whole building above described is enclosed on three sides by a wall 1·60 metres thick, set with towers 2·40 metres in diameter which project 1·80 metres from the face of the wall ([Plate 38], Fig. 1).[30] Through the upper part of the wall runs the low, vaulted, and unlighted gangway which has already been mentioned ([Plate 39], Fig. 1). It is no more than a tube between the wall and the vaults that adjoin the wall, but it serves to give access to the round chambers hollowed out of the towers. Access to the roof can be obtained at three points, the stair at the south-east angle of the central court, the stair at the south-east angle of court F, and the stair at the south-east angle of court E. Further, the three doors out of the first floor rooms 99, 102, and 106 open on to the roof of the single-storied block. There are traces of a narrow parapet round the edge of the roof, and the different courts seem to have been divided from one another and from the corridor 28 by low walls on the roof ([Plate 38], Fig. 2).
One other building stands within the palace yard, the group of rooms 140-152 to the east of the main palace. It is a later addition, though it resembles the rest of the palace too closely to admit of its having been added after the lapse of any considerable period of time. The north façade is prolonged beyond the chambers at either side, and is joined at the east end to one of the pilasters of the outer wall and at the west end to one of the towers of the inner wall, but it is not bonded in to the pilaster or to the tower. The northern end of the palace yard is thus divided off into a large court, which bears the same relation to the east annex as does the central court to the ceremonial chambers to the south of it. The stair to the first floor of the main palace was placed in this court, and it was approached from the main entrance through corridor 6. At the south-east corner the east annex does not connect with the angle of the east gate staircase, but is divided from it by an interval of ·30 metre.
The group of rooms 140-152 (the east annex) resembles in its main lines the group 29-42, south of the central court, and must have been intended for the same purposes. The north façade is decorated with blind arcades projecting ·25 metre from the face of the wall ([Plate 39], Fig. 2). The ovoid arches, which contain very shallow calottes, are carried by engaged columns having a diameter of ·40 metre. A recessed polygon was placed in the spandrels. The arcade is best preserved at the west end, and it is there possible to see that a narrow cornice, consisting of a single course of stones, ran along the wall above the arches, and that above the cornice the top of the wall was adorned with small arched niches, borne on stumpy half-columns and separated from one another by larger engaged columns (compare the top of the outer north wall of the palace and the top of the north façade of the central court). At the west end of the façade, in the first intercolumniation of the blind arcade, there is a gateway 1·90 metres wide, covered by a pointed arch. A similar gateway seems to have existed in the second intercolumniation from the east end, but the façade here is much ruined. The north wall of rooms 140, 142, and 145 has fallen ([Plate 39], Fig. 3). There can be no doubt that access was obtained to the lîwân, 140, by a wide archway, as in the case of the corresponding lîwân, 29, south of the central court. I saw no trace of a north door into chambers 142 and 145, though in all probability it existed. The lîwân, 140, is 5·40 metres wide by 10·50 metres long. Like the lîwân 29, it has two doors on each side and a door in the south wall. It is, however, vaulted in stone, not in brick, and the vault does not rise above the level of those on either side. The door-jambs are enriched with shallow pilasters, ·18 metre wide and ·4 metre deep, worked in stucco ([Plate 40], Fig. 2). They do not carry an arch over the archivolt of the door. In the side doors the archivolt cuts into the line of the oversailing vault which is carried over them. Above the south door there is a high narrow arched window, giving additional light to room 141. On either side of the door is placed a shallow arched niche, 1 metre wide and ·5 metre deep. The arch is filled in with a calotte, the lower edge of which is sunk behind the face of the wall. To the west of 140 are two vaulted chambers, 142 and 143, communicating with one another and with a similar chamber, 144, lying further to the south. The vaults of 142, 143, and 144 are set at right angles to the vaults of 140 and 141, so as to form buttresses to them. On the east side the same arrangement is observed in rooms 145, 146, and 147. These six chambers correspond to the more elaborate chambers 31, 32, 33, and 40, 41, 42 of the main palace. No. 141 (which corresponds with No. 30) is provided with four doors, one in the middle of each side. It was covered, not by a barrel vault, but by a stone groined vault, which has now fallen ([Plate 41], Fig. 1). The chambers east and west of 141 (Nos. 144 and 147; compare the columned rooms of the main palace) communicate with the yard on either side and also with the vaulted passage or antechamber 148. Into this passage ([Plate 41], Fig. 2; compare No. 36 of the main palace) the south door of No. 141 opens. The vault of the passage has fallen. It was no doubt carried on the south side by columns and arches like No. 36. There are no chambers to east and west of the passage, but on either side of the open space to the south of it were two chambers, 149 and 150 to the west, 151 and 152 to the east. They communicated with one another and with the yard to the north, as well as with the corridor south of 141. Their vaults ran east and west. No. 150 has fallen almost completely and No. 152 is much ruined.[31] A doorway in No. 148 gives access to a stair which leads down into an underground room lying beneath Nos. 144, 143, and 142 (section e-f, Plate 5, Fig. 1). It is lighted by three splayed windows in the north wall; under the windows there is an arched niche or tâqchah. To the west of No. 142 there is a ruined chamber which contained a stair leading to the roof. Thus the analogy with the block of rooms Nos. 29-42 is complete even to the serdâb and the stair to the roof.
The vault construction in the east annex shows a variation from that of the main palace. Instead of the long tubes running parallel with the barrel vaults, the masonry between the parallel barrel vaults of the annex is lightened by short compartments set at right angles to the vaults. Plate 39, Fig. 3, shows this construction between the vaults of 143 and 146 and the ruined vaults of 142 and 145; Plate 42, Fig. 1, the same construction between the vaults 144, 141, and 147, and the ruined vault of the passage 148. This system is an improvement upon the tubular scheme, inasmuch as it fills in the space between the vaults more completely and gives greater solidity to the roof. Moreover, it has the advantage of leaving no long inaccessible tubes to serve as a home for birds and snakes. The decorative effect of the openings of the tubes is lost, but it was not needed in the blank east and west walls of the annex, nor yet in the arcaded north wall.
The fact that a similar system of small compartments is to be observed in the building outside the palace to the north (though they are here laid parallel to the barrel vaults) leads me to suspect that it must have been built at about the same period, and is therefore a later addition to the original plan. It is completely detached from the palace, but it stands in line with the west wall of the palace and parallel to the north wall ([Plate 43], Figs. 1 and 2). It is separated by a distance of 13·25 metres from the face of the arcades of the north wall. It was itself constructed at two different periods. The older portion lies to the south, nearest to the palace, and consists of a large open court, J, 33·20 metres from north to south and 24·80 metres from east to west, flanked on the east side by six vaulted rooms. The southernmost of these six rooms, 153, is 9·55 metres from north to south and 7·80 metres from east to west. It is separated from court J by a wall 1 metre thick, but on the east side its wall is 1·90 metres thick and shows upon the exterior traces of an outer stair, leading to the roof, which passed over the wide arched opening in the east wall. The vault, which must have stood two stories high, like the vault of the great hall, has fallen. The remaining rooms, 154-158, have doors in the east wall and small loopholed windows in the west wall ([Plate 42], Fig. 2). The rooms are divided across the centre by a transverse arch and vaulted in two compartments, the vaults running east and west. Court J had a cloister upon the west side; it has entirely disappeared, but the spring of its vault is visible on the inner side of the west wall. Probably the vault was carried on the east side by columns and arches. Four round towers project at irregular intervals from the exterior of the west wall ([Plate 44], Figs. 1 and 2); they have the same diameter as the towers in the outer palace wall. The southernmost is about 3·40 metres from the southern angle of the court—an exact measurement is difficult because the angle of the wall is ruined. The next tower lies 5·65 metres to the north of the first; an interval of 7·35 metres separates it from the third tower, and the third tower is 10·70 metres from the larger tower at the north-west angle of the court. The angle tower contains a winding stair. The three smaller towers seem to be a later addition to the wall; they bear no relation to the three doors, and they block some of the windows. The windows are placed in groups of three, two groups between the south-west angle and the first door, one group between the first and second, and the second and third doors, and two groups between the third door and the angle tower. There are traces of a similar group in the north wall immediately to the east of the angle tower, and the straight face at the east end of the north wall gives reason to believe that there was a group of windows here also. The north wall is much ruined, and the ruin heaps are covered with blown sand. The arches of the windows are carried by engaged columns.[32]
To the north of room 158, and in a line with it, lie nine vaulted chambers which were added at a later date ([Plate 44], Fig. 2). They are separated from No. 158 by a stair running up to the roof, with a doorway to the west. At the east end there is a small room under the top of the stair with a loophole window in the east wall. From this room, which is accessible from No. 159, a stair, now completely ruined, led down into a substructure. Nos. 159 and 160 are 4 metres broad; they are covered by barrel vaults and have a door at either end. No. 161 opens by two doors into No. 162. No. 162 is 4·80 metres broad and is divided across the centre by a transverse arch. East of the transverse arch only half the space is vaulted over. Besides the doors, there are two small windows high up in the north and south wall. In the east and west walls there is a wide archway instead of the usual doors. The five rooms 163-167 resemble in all respects Nos. 159-161. Except over No. 162, where the vault is higher than in the other chambers, the roof of rooms 154-167 is raised above small compartments lying over the barrel vaults ([Plate 42], Figs. 2 and 3), and the mass of masonry between the vaults was lightened in the same manner. Slit-like windows appear high up in the east wall between the vaults (not, however, in rooms 153-162), doubtless in connexion with these compartments.
At a considerable distance to the north-east of the palace stands the small building which is known as the Ḥammâm (Plate 5, Fig. 3). Unlike the rest of the palace, it is not oriented. It consists of a long chamber running slightly to the west of north (about 24°), 10·65 metres long by 5·30 wide. It was covered by a vault which has now fallen. The door is on the east side; in the north and south walls there is a deep rectangular niche. A door in the north-east corner leads into a smaller chamber, 4·10 × 3·30 metres. In this building the thrust of the vault over the larger chamber is taken by outer buttresses, the only instance of such construction at Ukhaiḍir. On the east side there is one buttress ·60 metre deep; on the west side three, 1·25 metres deep. A stair leading to the roof ran up over the western buttress.