Calculating the population of Samarra

The Historian of Islam at Work Essays in Honor of Hugh N. Kennedy Edited by Maaike van Berkel Letizia Osti leiden | boston For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  Contents List of Figures and Tables xi Notes on Contributors xv Tabula Gratulatoria xxi A Lifelong Passion for Islamic History xxiii Maaike van Berkel and Letizia Osti Bibliography of Hugh N. Kennedy xxvii part 1 Caliphate and Power 1 A Ḥimyarite Restorationist Prophecy 3 Michael Cook 2 Kinship, Dynasty, and the Umayyads 12 Andrew Marsham 3 He Reigned as Caliph; Then He Died: The Reigns of Caliphs Versified 46 Geert Jan van Gelder 4 Versifying History in Abbasid Iraq: The Universal History of ʿAlī b. al-Jahm 69 Harry Munt 5 How to found an Islamic state: The Idrisids as rivals to the Abbasid Caliphate in the Far Islamic West 91 Corisande Fenwick 6 Rethinking “the Mamlūk State” with Ibn Khaldūn: “Mamlukization,” ʿaṣabiyya, and Historiographical Imaginations of the Sultanate of Cairo (1200s–1500s) 117 Jo Van Steenbergen 7 Ibn Khaldūn and the Ḥafṣid Caliphate 140 Allen Fromherz For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  viii contents part 2 Economy and Society 8 A Three-Centered System: Aleppo, Damascus, and Cairo in the Age of the Ayyubids 157 R. Stephen Humphreys 9 Informal and Formal Trading Associations in Egypt and Ifrīqiya, 850–1150 171 Chris Wickham 10 Good Governance in Theory and Practice: Comparing Abū Yūsuf’s Kitāb al-Kharāj with Papyri 183 Petra M. Sijpesteijn 11 A Matter of Trust: On Some Principles of Governance in the Letters of Qurra b. Sharīk 201 Arietta Papaconstantinou 12 Calculating the Population of Samarra 210 Alastair Northedge 13 Flour for the Caliph: Watermills in the “Land behind Mosul” 234 Cristina Tonghini 14 Bedouin, Bandits, and Caliphal Disappearance: A Reappraisal of the Qarāmiṭa and Their Success in Arabia 254 Peter Webb 15 Zinā and muḥṣanāt in the Quran 283 Richard Kimber part 3 Abbasids 16 Muslim Nostalgia: Longing for the Abbasid Past in the Mamluk Era 299 Robert Irwin For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  contents ix 17 The al-Mustanṣiriyya madrasa in Baghdad and Its Founder, al-Mustanṣir 320 Carole Hillenbrand 18 Hārūn al-Rashīd in Premodern Arabic Literary Imaginary: Ideology of Monogamy, Harem Politics, and Court Intrigues 340 Wen-chin Ouyang 19 The Representation of the Barmakids in Bodleian Manuscript Ouseley 217 and Other Monographs 356 Arezou Azad and Pejman Firoozbakhsh 20 Eutychius of Alexandria Vindicated: Muslim Sources and Christian Arabic Historiography in the Early Islamic Empire 384 Robert Hoyland 21 Bureaucrats on the Move: Messengers in Fourth/Tenth-Century Iraq 405 Maaike van Berkel, Nadia Maria El Cheikh and Letizia Osti 22 Al-Ṭabarī’s Unacknowledged Debt to Ibn Abī Ṭāhir Ṭayfūr 432 Sarah Bowen Savant 23 Heraqleh: A New Interpretation 448 Andrew Petersen part 4 Frontiers and the Others 24 The Interface between Byzantium and the Ilkhanids in Fourteenth-Century Book Painting 475 Robert Hillenbrand 25 Exploring Europe through Medieval Islamic Folk Literature 503 Niall Christie 26 The Lordship and Bishopric of Banyas in the Latin Kingdom of Jerusalem (1126–1164) 521 Alan V. Murray For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  x contents 27 Fortresses and Frontiers: Castles and Northern Syria in the Sultanate of Cairo 538 Angus D. Stewart 28 The Sasanian Fort of Pānkān 560 Balázs Major 29 Negotiating the North: Armenian Perspectives on the Conquest Era 591 Tim Greenwood 30 New Palaeoenvironmental Evidence on the Possible Impact on Agriculture of Early Arab-Islamic Raiding Activity on Crete 614 John Haldon Index 635 For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  chapter 12 Calculating the Population of Samarra Alastair Northedge One of the most striking of Hugh Kennedy’s article titles has always been ‘The feeding of the five hundred thousand: Cities and agriculture in early Islamic Mesopotamia.’ Evidently, from the beginning one of his major interests has been the Abbasids and their cities, and in this case how to feed them from a distance. The question of how populations were fed, although the main object- ive of Kennedy’s study, is not the part of the question dealt with here, but rather, how many of them there were to feed. This is still a work in progress, and review of the variables may lead to a better approximation in the future. The urban population of Iraq was certainly very large under the early Ab- basids, outstripping capitals of antiquity such as Ctesiphon.1 Its expansion falls into two phases: first, the settlement of the Arab tribes in the amṣār of Kufa and Basra, where large numbers were settled as fighters (muqātila) in the years after the conquest; and second, the expansion of the Abbasid capitals of Baghdad and Samarra, which was partly a consequence of the settlement of the army in the capital and partly, in the case of Baghdad, a product of the economic expansion of the city, which lasted from the foundation in the 760s until the decline of Abbasid Iraq in the first half of the fourth/tenth cen- tury. In the case of Kufa and Basra, the numbers in around 660 were of the order of 60,000–80,000 fighters, according to the reports. In the governorship of Ziyād b. Abīhi, there were 80,000, with 120,000 dependents, in Basra. That period may have been brief, as many departed for the conquests and did not return. Although it is not possible to confirm these figures from archaeological evid- ence, it is true that the outlines of the two amṣār can be detected on the ground in corona satellite imagery from the 1960s.2 Basra was walled in 155/771–772 1 This was the view of Adams, Land behind Baghdad 62–63, 72. However, Ctesiphon may have been bigger than supposed, as there were unwalled suburbs that have not been taken into account. It is regrettable that no thorough modern publication of Sasanian Ctesiphon exists at present. 2 Declassified corona satellite imagery can be downloaded from the US Geological Survey (https://earthexplorer.usgs.gov/) for free if the image has already been scanned, the ori- ginal imagery being taken on film. Most of the useful imagery was taken between 1966 and 1972. © Alastair Northedge, 2022 | doi:10.1163/9789004525245_013 For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 211 figure 12.1 The site of Basra, showing the limits of the early Islamic city at the time of its maximal expansion, and the wall line has been identified (fig. 12.1).3 Kufa was not walled but surrounded by a fosse (khandaq) under Caliph al-Manṣūr (fig. 12.2).4 Its maximal area can be seen from the line of a canal, which was located perhaps on the line of the khandaq, and delimited 3 Pellat, al-Baṣra. corona image ds1035–1040da019, dated 23/9/1966. 4 Northedge, Umayyad and Abbasid urban fortifications. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  212 northedge figure 12.2 The site of Kufa showing the limits of the early Islamic city the early Islamic city, larger than 1960s Kufa.5 The areas are: Kufa, 1,375 ha, and Basra, 2,200 ha, approximately. Both were large cities by any standards, even if somewhat smaller than the early days when amṣār were effectively temporary camps. By contrast, we know very little about early Islamic Baghdad, the city that existed before the return from Samarra and before the construction of the Dār al-Khilāfa on the east bank of the Tigris. We do not even know for cer- 5 corona image ds1039–2088df064, dated 28/02/1967. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 213 tain where the Round City was located precisely, though some Iraqi scholars say they do know its location under modern-day Baghdad. It is evident that one cannot know the population of Baghdad, apart from estimates from the texts. There is no other source, as the early Abbasid city has long been built over. The most recent serious estimate is that of Lassner.6 According to Lassner’s estimate, based on the text of al-Khaṭīb al-Baghdādī, the city area was of the order of 70km2 (6,965 ha). Although Baghdad is generally known only from literary sources, there is reason to believe that the statements preserved by the Khaṭīb (d. 1071) on the dimensions of the city are more or less accurate thus justifying the claims of native Baghdadis that their city was more spacious than all oth- ers. One account gives a figure of 53,750 jarib; two others indicate 43,750 jarib—the lower estimate apparently having been obtained from a survey undertaken in the time of al-Muwaffaq (ca. 890). It is significant that the figures cited in this account are not only expressed in terms of jarib but in linear measurement as well. The West Side was found to be 250 habl in length and 105 in width making for a total of 26,250 jarib. The east side was 250 habl in length and 70 in width making for a total of 17,500, and for the entire city 43,750 jarib. Simple arithmetic computation reveals that the figures correspond, and one square habl = one jarib […] the known value of the jarib is 1,592sq. meters. The fact that the dimensions of the city as obtained in three differ- ent accounts, and according to three different metrological systems (habl, jarib, mil), are in agreement, may be taken as an indication of their accuracy.7 Lassner then suggests that this means a population of the order of 280,000 and offers that it could be up to twice as much. A simple length-times-width calculation can hardly be very precise, as Bagh- dad was not a rectangle; in fact, this type of calculation is roughly in accordance with the size of Samarra, that is, Baghdad at its maximum was physically some- what larger than Samarra, which has 58km2 of built-up area.8 Samarra evidently never reached the level of the population of Baghdad, as its economy was entirely based on the support of the state, and the environ- 6 Lassner, Topography 158–60. Lassner, Massignon; Lassner, Notes; Lassner, “Habl.” 7 Lassner, Topography 157–158. 8 Northedge, Remarks. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  214 northedge mental situation was poorer than that of Baghdad. The caliph moved there, because it was not Baghdad, and settled there with his army and the court, which were both financed through taxation. Merchants and other business- men moved there to sell their products to the state. The factor that makes the difference is that the archaeological remains are preserved. It is quite an exceptional case: there is no other large ancient city in the world whose remains have been preserved so well.9 And indeed, much of the plan can be recovered, though, as with all archaeological material, the preservation is incomplete. Even so, it is not possible to know what the population of Samarra really was. There was no census of the population, as occurred in China under the Sui and T’ang dynasties.10 De la Vaissière, in a study of Central Asian popula- tions, was able to reach a degree of precision with regard to eastern Turkestan (Xingjian) and their Turkish populations, but was limited to approximations for the western side.11 Certainly there were surveys, as discussed by al-Qāḍī, but they were mainly concerned with registering populations and land for tax purposes.12 However, it is possible to advance the question, even if one cannot reach a satisfactory conclusion. Estimates of medieval populations commonly depend upon a calculation of population by density and area. For example, the work of Dijandjgava on Fusṭāṭ and Cairo relies on textual estimates and very gen- eralized statements from surviving cities whose old centers furnish evidence of early modern times, presumed to be similar to the density of medieval cit- ies.13 Samarra is a very different case, as it was much more spread out, with a popu- lation spread visible in the archaeology, as well as being mentioned in the texts. “The people spread out in their construction at Surra Man Raʾā more than they had done at Baghdad, and they built broad residences.”14 In this sense, Samarra is atypical for a medieval city, though it may have been quite similar in style to Abbasid Baghdad. The issue of Abbasid city versus medieval city was tackled 9 That is why we embarked on recording the remains, see Northedge, Historical topography; and Northedge and Kennet, Archaeological atlas. Much today has been destroyed. The idea was to recover as much as possible from old air and satellite imagery (mainly 1953) so that we could know as much as possible about what the city had been like. 10 Bielenstein, Census; Pulleyblank, Registration. 11 De la Vaissière, Early medieval. 12 Al-Qāḍī, Population census. 13 Dijandjgava, Ways of estimating. 14 Yaʿqūbī, 263. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 215 by Lassner in his article ‘Massignon and Baghdad: The complexities of growth in an imperial city’, where Massignon is presented as explaining Baghdad as a typical walled medieval Islamic city, whereas in the early days it evidently was not. However, Baghdad did finally become such a typical walled city in the later period, from the seventh/thirteenth century onward. 1 The Urbanism of Samarra Evidently, when al-Muʿtaṣim’s city Surra Man Raʾā was founded in 221/836, the main city was laid out with the Dār al-Khilāfa in the north and a main avenue running south from the gate of the palace for over 7 km, with a single bend adjacent to the first congregational mosque, now lost, where the markets underneath the center of the modern city were also placed (figs. 12.3–4).15 This avenue was later called Shāriʿ Abī Aḥmad and was subsequently supplanted as the main avenue by the Shāriʿ al-Aʿẓam. To one side or the other of the avenue were placed cantonments (qaṭāʾiʿ) of the lesser military, such as Waṣīf, the Farāghina, the Maghāriba, the Jund, and others, while major figures and their followings were placed outside the main city, the Usrūshaniyya under al- Afshīn Khaydar b. Kawūs al-Usrūshanī outside the village of al-Maṭīra to the south, and the Turks under Ashnās outside the small town of Karkh Fayrūz to the north. This initial urban plan was similar, but not identical, to that of Baghdad. There was no fortified Round City; rather, the Dār al-Khilāfa was placed at the head of the avenue, and the mosque in the center of the city. It was indeed the model of the rabaḍ of al-Ḥarbiyya with its blocks of houses for the military. The idea of the fortified royal city was abandoned until later in the fourth/tenth century.16 Nearly all the military cantonments (see fig. 12.5), and the subsequent ca- liphal city of al-Mutawakkiliyya, were built on the same model of a palace for the leader,17 from which an avenue led away, with the larger houses of the major figures on the avenue, and blocks of houses laid out behind (see fig. 12.6). 15 Northedge, Topography, ʿAskar. 16 Northedge, Umayyad and Abbasid urban fortifications; Northedge, Early Islamic urban- ism. 17 The leader was not necessarily the war chief. Ashnās, for example, never led the Turks in war but was the civil governor of the cantonment, as was Ḥarb, the governor of al-Ḥarbiyya in Baghdad. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  216 northedge figure 12.3 The site of Samarra, showing the extent of the ancient remains For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 217 figure 12.4 The division of the archaeological site into zones For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  218 northedge figure 12.5 The disposition of the military cantonments For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 219 figure 12.6 Al-Mutawakkiliyya, showing the classic layout of a cantonment (qaṭīʿa) at Samarra For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  220 northedge The list of cantonments (see fig. 12.5): 1. The cantonments of al-Dūr, defined as Area U in the archaeological evid- ence (Northedge, Historical topography fig. 78). 2. The cantonments of al-Karkh, situated outside the pre-Islamic town of Karkh Fayrūz and identified as Area F (Northedge, Historical topography fig. 76). 3. Area G, identified as al-Wazīriyya (Northedge, Historical topography fig. 61, 146). 4. The cantonments of al-Jawsaq, Area X (Northedge, Historical topography fig. 60, 144–146). 5. The “old” cantonment of Waṣīf, situated in Area H. 6. The cantonment of the Maghāriba, sited on the Shāriʿ al-Khalīj in Area J (Northedge, Historical topography, fig. 80). 7. The cantonments of the Jund and Shākiriyya, situated in the avenues (Area J). 8. The new cantonments of the east side (Shāriʿ al-Askar and Shāriʿ al-Ḥayr al-Jadīd) (Northedge, Historical topography fig. 81). 9. The cantonment of al-Afshīn at al-Maṭīra (Area K) (Northedge, Historical topography fig. 82). 10. The cantonment of Balkuwārā (Area R) (Northedge, Historical topography fig. 84). Evidently these cantonments cover the whole period, not just the initial found- ation. Then, subsequently, as the city began to expand, a greater and greater civil- ian population began to settle in the central area and took over parts of the central cantonments. In particular, the political Turks wanted to be near the caliph. Large mansions were built near the Dār al-Khilāfa, and much of the military was moved out, including the Turks of Waṣīf, who moved to the Sogh- dian cantonment of al-Maṭīra. In addition, the new congregational mosque was built by al-Mutawakkil “in the direction of al-Ḥayr,” and some important fig- ures, in particular the kuttāb, heads of the dīwāns, wished to live near the new mosque, as indeed the archaeological evidence shows continued to be the case long after the abandonment of Samarra by the caliphs. In addition, there was the new caliphal city al-Mutawakkiliyya, which was built and abandoned in two years 245/859–247/861 (see fig. 12.6). The useful- ness of this short-lived construction is that it shows in fully preserved detail what the caliph of the time would want, although conceivably it was never fully finished.18 18 For example, the zone around the Abu Dulaf mosque, where one would expect the mar- For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 221 2 Approaches to the Population If the calculation of the population from the surface area is not going to work in this case, one can ask what other methods can be used. 2.1 The Mosques of Abbasid Samarra In the Muslim context, one traditional approach is to calculate the size of the mosque or mosques.19 In the earlier Abbasid period, we are still in the period when there was a single congregational mosque in a city, where the entire adult male population was supposed to hear the khuṭba on a Friday. In fact there were four jawāmiʿ in Samarra. First, the unfinished mosque of al-Rashīd’s octa- gonal city al-Mubārak, construction of which was abandoned in 180/79620 and was never used. Second, al-Muʿtaṣim’s mosque, which has now disappeared, was abandoned as too small according to al-Yaʿqūbī. Third, al-Mutawakkil’s mosque, with the Malwiya, which continued to be the congregational mosque from 237/851–852 until some point probably in the sixth/twelfth century. And fourth, the mosque of Abu Dulaf, which was used for prayer in 246/860 and the following year and then said to have been abandoned, although it probably continued to be used by the inhabitants of the longer-term town of al-Māḥūza (Area C).21 Thus, the mosque of al-Mutawakkil is the only one whose area is useful. There was a considerable question of access, as many of the outlying districts were too far for people to travel for the Friday prayer, and certainly the Turks of al-Karkh were probably not expected to travel to the mosque of al-Mutawakkil, as there was a mosque in the courtyard of the palace of Ashnas, where the Turks assembled for a letter of complaint made to al-Muhtadī in 256/870.22 However, the pattern of prayer typical of later times was not yet established: there are no mosques for local quarters. If one did not pray in the jāmiʿ, one prayed in the masjid of one’s chief, as in the Umayyad period. Architects’ guides today for the design of mosques allot 0.85 m2 for the indi- vidual at prayer, 60–80cm width for sitting shoulder to shoulder, and 1.2m for fore and back to enable the person to make the prostration. Probably it is better kets to have been located, is empty and largely unbuilt, apart from a row of shops on the avenue. The market would, at the time of the occupation of the city, have been open-air and not yet built up. 19 E.g., Dijandjgava, Ways of estimating. 20 Northedge, Historical topography 93–94. 21 Northedge, Historical topography 216. 22 Al-Ṭabarī, Taʾrīkh iii, 1,799. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  222 northedge table 12.1 Areas of the Samarra mosquesa Congregational Mosque of al-Mutawakkil (H3) Inner Mosque 37,440m2 Inner Mosque plus roofed ziyādas 66,468m2 Outer Enclosure 165,886m2 Abu Dulaf Mosque (T1) Inner Mosque 29,110m2 Outer Enclosure 124,536m2 Mosque of Ashnas (F1.1) 11,383m2 a Details on the individual buildings can be found in Northedge and Kennet, Archaeological atlas, passim. to say 1 m2. One could presume then that the internal area of the mosque corres- ponds to the maximum envisaged by the architects, evidently not necessarily the number who regularly attended. There were, of course, also some unusable areas in the mosque. It is unlikely that the outer uncovered enclosures were ever intended to be used for prayer, but the roofed ziyādas outside the mosque certainly were. A modern-day equivalent can be seen in the Mosque of the Prophet in Madina, which in the version of the mosque in the 1980s had ziyādas very similar to those of Samarra. There is also the question of the reception capacity of the muṣallās intended for the festival prayers, where the male population would have been expec- ted to attend the festival prayers. There were five in Samarra. One is located behind the mosque of al-Mutawakkil, evidently intended for the same pub- lic as the mosque, and one to the east of the Dār al-Khilāfa. Then there is one apparently intended for the Turks of al-Karkh and al-Dūr, and one or two for al-Mutawakkiliyya. All are substantially larger than their corresponding con- gregational mosques. These figures suggest that the architects were expected to accommodate up to approximately 60,000 adult males in the mosque of al-Mutawakkil, but it would be surprising if as many as 30,000 ever prayed in the Abu Dulaf mosque, as the mosque was not in reach of so many, though there was at least one day when people crowded in to present the caliph with petitions.23 More likely, the 23 Al-Ṭabarī, Taʾrīkh iii, 1,453–1,454. Northedge, Historical topography 223. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 223 table 12.2 Areas of the muṣallas at Samarra Y19 Muṣallā adjacent to the Congregational Mosque 197,47m2 Y6 Muṣallā east of the Dār al-Khilāfa 122,57m2 U337 Muṣallā of the Turks in al-Karkh 200,25m2 T1042 Muṣallā of al-Mutawakkiliyya 182,82m2 S11 Muṣallā east of the Raṣāṣī canala 210,75m2 a The public for whom S11 was intended is uncertain. It could have been intended for the inhabitants of al-Muḥammadiyya, built earlier than al-Mutawakkiliyya, or as a second muṣallā for the new caliphal city. architects provided more space than necessary, for reasons of prestige. And the same is true of the muṣallās. 2.2 Houses and Their Occupants In 2001, Derek Kennet published a paper entitled ‘The form of the military can- tonments at Samarra and the organisation and size of the Abbasid army’, the latter part of which was largely based on calculations from archaeological evid- ence.24 The chapter aimed to calculate the size of the army, not the general population of Samarra, and was naturally restricted to discussing the role of the male members of the army. The point of departure for calculating the popula- tion density was the size of the cantonment at Balkuwārā, which was settled with 12,000 Arabs, Ṣaʿālīk,25 and others by al-Mutawakkil under the nominal command of his son al-Muʿtazz, a young child at the time.26 The number of houses in the Balkuwārā cantonment seemed to correspond with the number of 12,000 Arabs, if the figure of an ʿarāfa of 10–15 men accommodated in a unit with four rooms and a stable at Malatya in 140/757–758 under al-Manṣūr were used, as cited in al-Balādhurī’s Futūḥ:27 “There was built for the jund that they settled there for each ʿarāfa two lower bayts and two upper bayts above them and a stable (and the ʿarāfa is 10 persons to 15 men).” The unit would correspond to a small house at Balkuwārā. However this would imply that the soldiers were single, unmarried men, as 10–15 men with their wives and families could not be accommodated in four rooms. While it 24 Kennet, Form. 25 Ṣaʿālīk (vagabonds) were tribal Arabs, perhaps those dropped from the dīwān 25 years earlier, or their descendants. Northedge, Historical topography 171. 26 Kennet, Form 171. 27 Al-Baladhurī, Futūḥ 187. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  224 northedge table 12.3a Typology of buildings relative to the question of residence Total Block Group of small Abbasid houses, commonly in a rectan- 1,874 gular block. Blocks Area of unclear block construction, which would 82 include several “block” units. House Site excavated as a “house” by the German Expedition 43 or the Directorate-General of Antiquities. Mansion 1 Very large residential building other than that defined 16 as a palace. Mansion 2 Large residential building greater than 8000m2. 96 Mansion 3 Residential building 5,000–8,000m2. 54 Mansion 4 Residential building 2,000–5,000m2. 46 Mansion 5 Individual residential building less than 2,000m2. 12 Palace Monumental residential building identified as belong- 23 ing to a member of the Abbasid family.a a The terminology defining a palace corresponds to the definition given in Northedge, Palaces. is likely that some soldiers were unmarried and could be accommodated in a barracks-like environment, it is probable that many were married and would require more space, there being no process of retirement from the dīwān,28 and each family may have included more than one generation of soldiers. As there are only 1,596 cantonment houses, there probably never were 12,000 at Balkuwārā—either some were quartered elsewhere, or the hoped-for recruit- ment of 12,000 was not achieved, or the figure included entire families. In any case, al-Balādhurī’s expression for the allocation of 10–15 men sounds more like a block of houses than an individual building: the word bayt at this time nor- mally indicates a multiroom apartment rather than a single room. For example, al-Muntaṣir, when walī ʿahd, had a bayt in al-Jaʿfarī, a secondary residence unlikely to have been small. In addition, many of the house blocks include a yard at the end for a stable. In fact, when the cataloguing of the inventory of buildings took place,29 an approximate typology was made, covering the entire site. 28 Kennedy, Armies. 29 As published in Northedge and Kennet, Archaeological atlas. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 225 table 12.3b Buildings that might or might not be residential Building Any archaeological definable building remains, not oth- 1,860 erwise specified. Modern structures and ancient tells are excluded. Complex Irregular complex of archaeological building remains, 182 apparently belonging to a single unit, but not with an identifiable residential plan. Complex Regularly planned building with complex plan. 121 building In fact there has been a tendency, when thinking of Samarra in army terms, to count only the masses of small houses in each cantonment (qaṭīʿa), but the society was an integrated one, from lord to vassal. The populations of the larger houses must have been considerable. Kennet concentrates on the small houses, which were most often organized in blocks of 6–25 houses (see fig. 12.7), sometimes with an additional communal yard that could have been used for a stable. Although the houses of each block were commonly the same, house sizes in different blocks were varied, ranging from around 95 m2 to 1,300m2.30 The house would normally be of a single story of four–seven rooms on one or two sides of a courtyard, built of earth. Although only one case of such a house being excavated is known to the author (M8, pos- sibly excavated by the German Expedition in 1911–1913 at al-Iṣṭablāt), no details of how it was lived in are known—the walls were simply cleared. The only comparative evidence comes from the Umayyad houses of the Amman Citadel, destroyed with their contents by earthquake a century earlier, probably in 131/749.31 The two houses excavated, one in 1949 by Lankester Harding and the other in 1976–1977 by the author under the direction of C.- M. Bennett, were each composed of seven rooms around a courtyard, with a lat- rine and cistern. Though architecturally different from Samarra, being built of stone, it was obvious that they were inhabited by the entourage of the uniden- tified Umayyad prince who built the palace and citadel, quite possibly the jund al-Balqāʾ mentioned in the arrest of the Abbasid Ibrāhīm al-Imām in 130/747. The house excavated in 1949 on the museum site may have been inhabited by a single familial unit, but the Area C house excavated in 1976–1977 seemed to 30 Kennet, Form. 31 Northedge, Contents. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  226 northedge figure 12.7 Examples of blocks of cantonment houses have two hearths, which suggests two family units (see fig. 12.8: rooms B and F). This could represent two separate individuals in the dīwān put in the same house by the administration, with their families or not, or it could be that the eldest son of the family was married and required separate accommoda- tions. Evidently it is impossible to say how many lived in a cantonment house. They could have been used for one or more families, including male descendants recruited to the dīwān, or for a group of unmarried men. However, the design is oriented toward the settlement of families, as is also indicated by the caliphal policy of buying Turkish slave girls to marry to the Turks. Hilāl al-Ṣābiʾ, in Rusūm dār al-khilāfa, has an entertaining text on how to calculate the average population of a house, which he unfortunately combines with the number of baths to reach 96 million as the total population of Bagh- dad:32 “We find that one house may contain 20 people, while another may contain two, three, or more or less than that. However if we desire a middle and reasonable course, which leaves no room for doubt, we take half the twenty, and double the three, and by adding these we have sixteen. When we take half the sixteen, we come up with eight people, including men and women, old and 32 Dijandjgava, Ways of estimating, gives an alternate source in Ibn al-Faqīh for parts of this story. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 227 figure 12.8 Plan of Building B at Amman, destroyed in 131/749 northedge, contents young.”33 Probably the upper figure of 20 refers to a multigeneration Baghdad house, as has survived at least until recently in al-Karkh and Ruṣāfa. While this is absolutely no basis for calculating household size, it is not a bad hypothesis; eight could have been the average household size. The figure from the Chinese census in Xinjiang at the time was five.34 The total number of identifiable small or cantonment houses is 19,571, in- cluding identifiable apartments in the palaces. The division of small houses into the different areas is given in Table 12.4. The overall number represents the maximum of all periods, including the brief occupation of al-Mutawakkiliyya and al-Jaʿfarī, when Surra Man Raʾā was said to have been emptied. If we subtract them, the number of houses at max- imum occupation is 16,428. 33 Al-Ṣābiʾ, Rusūm dār al-khilāfa 22. 34 De la Vaissière, Early medieval. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  228 northedge table 12.4 Total number of small houses by zone A Al-Jaʿfarī 199 B Sub-palaces of al-Jaʿfarī 10 C Al-Māḥūza 328 D Shaykh Walī/Karkh Fayrūz 126 E 393 F Qaṭāʾiʿ al-Karkh 3,287 G Al-Wazīriyya 621 H Central City, northern half 659 J Central City, southern half 4,623 K Qaṭāʾiʿ al-Afshīn 1,645 L Al-Maṭīra unidentifiable M Al-Iṣṭablāt (al-ʿArūs) 315 Q Al-Musharrahāt (al-Shāh) 253 R Balkuwārā 1,599 T Al-Mutawakkiliyya 2,934 U Qaṭāʾiʿ al-Dūr 1,496 X Qaṭāʾiʿ al-Jawsaq 1,083 Total 19,571 If we were to hypothesize a household size of eight, this would give a pop- ulation size of 156,560. Excluding al-Mutawakkiliyya as double counting, that would mean a maximum of 131,400. If a household size of five, the figure would be 97,855, and without al-Mutawakkiliyya, 82,140. 2.3 The Elite and Commanders It is likely that that the elite housing was not merely a nugatory addition to the population of the city, bearing in mind the numbers of the household and servants who lived on site (see fig. 12.9). There were 289 palaces and other houses of the elite, among whom could be counted the palaces of the Abbasid family, the commanders of the army, the kuttāb (secretaries) responsible for the dīwāns, and other leaders, such as the Twelver Imams transplanted to Samarra and the Tahirid Isḥāq b. Ibrāhīm, who maintained a house at Samarra.35 A portion of the structures simply denomin- ated “building,” 1,860 in total, may also have been residences. 35 Northedge, Historical topography 189. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 229 figure 12.9 Examples of elite housing from al-Mutawakkiliyya The typology was set out as: table 12.5 Typology of elite housing Type Description Total Hypothesized inhabitants Palace Monumental residential building identified as 23 500 belonging to a member of the Abbasid family.a a The terminology defining a palace corresponds to the definition given in Northedge, Palaces. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  230 northedge table 12.5 Typology of elite housing (cont.) Type Description Total Hypothesized inhabitants Mansion 1 Very large residential building other than that 16 300 defined as a palace. Mansion 2 Large residential building greater than 8,000m2. 96 250 Mansion 3 Residential building 5,000–8,000m2. 54 100 Mansion 4 Residential building 2,000–5,000m2. 46 50 Mansion 5 Individual residential building less than 12 30 2,000m2. House Site excavated as a “house” by the German 43 30 Expedition or the Directorate-General of Antiquities. Equivalent to Mansion 4 or 5. table 12.6 Zonal division of elite housing P M1 M2 M3 M4 M5 A Al-Jaʿfarī 2 1 B Subpalaces of al-Jaʿfarī 3 C Al-Māḥūza 4 D Shaykh Walī/Karkh Fayrūz 1 1 E 2 1 F Qaṭāʾiʿ al-Karkh 2 7 G Al-Wazīriyya 1 1 1 H Central City, northern half 11 4 18 9 9 2 J Central City, southern half 5 17 7 1 K Qaṭāʾiʿ al-Afshīn 1 4 1 1 L Al-Maṭīra 1 M Al-Iṣṭablāt (al-ʿArūs) 1 3 4 R Balkuwārā 1 1 1 T Al-Mutawakkiliyya 1 34 16 19 10 U Qaṭāʾiʿ al-Dūr 2 2 7 15 X Qaṭāʾiʿ al-Jawsaq 1 3 2 Totals 20 16 96 55 47 12 For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 231 One can hypothesize a rather greater number of inhabitants than for the smaller housing, between 30 and 300 a house. If this were so, the elite housing would add around 48,300 to the population. If, as with cantonment housing, we exclude al-Jaʿfarī and al-Mutawakkiliyya, as they were intended replacements of Surra Man Raʾā but only served for a short time, there are 160 elite residences, perhaps serving a population of 32,000. 3 Conclusion Adding the cantonment housing and the elite housing together, we arrive at a total of 204,868 with al-Mutawakkiliyya, or 164,034 without it, that is, at the period of maximum expansion under al-Mutawakkil between 232/847 and 245/859. If a household size of five, it would be 146,155, or 114,750 without al- Mutawakkil’s folly of his last years. The size of the congregational mosque, suggested to be capable of accommodating a maximum of 60,000 adult males, is not out of line with this sort of calculation. There is no way of knowing how close these figures are to the truth, but indeed it could be elaborated by further closer analysis of the data, adding more allowance for destroyed areas. But it does seem perhaps not far from the truth. Bibliography Sources al-Balādhurī, Kitāb Futūḥ al-buldān, ed. de Goeje, Leiden 1866. al-Ṣābiʾ, Customs of the caliph’s palace (Rusūm dār al-khilāfa), trans. E. Salem, Beirut 1977. al-Ṭabarī, Taʾrīkh al-rusul wa-l-mulūk, eds. M.J. de Goeje et al., 16 vols., Leiden 1879–1901. al-Yaʿqūbī, Kitāb al-Buldān (bga 7), ed. de Goeje, Leiden 1892. Studies Adams, R.M., Land behind Baghdad, Chicago 1965. Bielenstein, H., The census of China during the period 2–742ad, in Bulletin of the Museum of Far Eastern Antiquities 19 (1947), 125–163. Creswell, K.A.C., Early Muslim architecture, vol. 2, 1st. ed., Oxford 1940. De la Vaissière, E., Early medieval Central Asian population estimates, in jesho 60 (2017), 788–817. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  232 northedge Dijandjgava, G., Ways of estimating population numbers in medieval Islamic cities as exemplified in the case of Fustāt Cairo, in Al-Masāq 5 (1992), 65–69. Durand, J.D., Historical estimates of world population: An evaluation, in Population and development review 3 (1977), 253–296. Pellat, Ch., al-Baṣra, ei2, i, 1,085–1,087. Kennedy, H., The financing of the military in the early Islamic state, in The Byzantine and early Islamic Near East, iii, ed. A. Cameron, Princeton 1995, 361–378. Kennedy, H., The armies of the caliphs: Military and Society in the early Islamic state, London 2001. Kennedy, H., The feeding of the five hundred thousand: Cities and agriculture in early Islamic Mesopotamia, in Iraq 73 (2011), 177–199. Kennet, D., The Form of the military cantonments at Samarra and the organisation and size of the Abbasid army, in A medieval Islamic city reconsidered (Oxford studies in Islamic art 14), C. Robinson (ed.), Oxford 2001, 157–182. Lassner, J., Notes on the topography of Baghdad: The systematic descriptions of the city and the Khaṭīb al-Baghdādī, in jaos 83 (1963), 458–469. Lassner, J., The “habl” of Baghdad and the dimensions of the city: A metrological note, in jesho 6 (1963), 228–229. Lassner, J., Massignon and Baghdad: The complexities of growth in an imperial city, in jesho 9 (1966), 1–27. Lassner, J., The topography of Baghdad in the early Middle Ages, Detroit 1970. Lassner, J., The shaping of Abbasid rule, Princeton 1980. Lawrence, D., et al., Long term population, city size and climate trends in the fertile crescent: A first approximation, in PLoS one 11 (2016): e0152563. doi:10.1371/journal .pone.0152563. Le Strange, G., Baghdad during the Abbasid caliphate, London 1900. Northedge, A., Archaeology and new urban settlement in early Islamic Syria and Iraq, in Studies in late antiquity and early Islam, vol. 2: Settlement patterns in the Byzantine and early Islamic Near East, G.R.D. King and A. Cameron (eds.), Princeton 1994, 231– 265. Northedge, A., The palaces of the Abbasids at Samarra, in A medieval Islamic city recon- sidered (Oxford studies in Islamic art 14), C. Robinson, Oxford 2001, 29–67. Northedge, A., The historical topography of Samarra (Samarra studies 1), London 2005. Northedge, A., Remarks on Samarra and the archaeology of large cities, in Antiquity 79 (2005), 119–129. Northedge, A., Umayyad and Abbasid urban fortifications in the Near East, in Die Gren- zen der Welt, Arabica et Iranica ad honorem Heinz Gaube, L. Korn, E. Orthmann, and F. Schwarz (eds.), Wiesbaden 2008, 39–64. Northedge, A., The contents of the first Muslim houses: Thoughts about assemblages from the Amman citadel, in Proceedings of the 7th international congress on the For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti  calculating the population of samarra 233 archaeology of the ancient Near East, ii, R. Matthews and J. Curtis (eds.), Wiesbaden 2012, 633–661. Northedge, A., ʿAskar al-Muʿtaṣim: The central city of Samarra, in Beiträge zur Islamis- chen Kunst und Archäologie, iv, J. Gonnella (ed.), Wiesbaden 2014, 38–62. Northedge, A., Early Islamic urbanism, in A companion to Islamic art and architecture, vol. 1: From the Prophet to the Mongols, F.B. Flood and G. Necipoglu (eds.), New York 2017, 155–176. Northedge, A., and D. Kennet, Archaeological atlas of Samarra (Samarra studies 2), 3 vols, Oxford 2015. Pulleyblank, E.G., Registration of population in China in the Sui and T’ang Periods, in jesho 4 (1961), 289–301. al-Qāḍī, W., Population census and land surveys under the Umayyads (41–132/661–750), in Der Islam 83 (2008), 341–416. Susa, A., Rayy Sāmarrāʾ fī ʿahd al-khilāfa al-ʿabbāsiyya, 2 vols., Baghdad 1948–1949. Töllner, H., Die Türkische Garden am Kalifenhof von Samarra, ihre Entstehung und Machtergreifung bis zum Kalifat al-Muʿtamids, Bonn 1971. Torres Balbás L., Extensión y demograffa de las ciudades hispanomusulmanas, in si 3 (1955), 37–59. For use by the Author only | © 2022 Maaike van Berkel and Letizia Osti