“396 unknown Roman forts” in the Middle East: a “major discovery” put into perspective.

In this post, I would like to return to a publication cited by the blog The Ancient Near East Today (Nitschke 2023) and relayed by the Agade mailing list as one of the major discoveries of 2023. Here is the summary of the article in question: “Researchers from Dartmouth College led by Dr. Jesse Casana used declassified images from the CORONA and HEXAGON spy satellite programs to identify 396 unknown Roman-era forts. The research shows that hundreds of forts extended over a large area in a general east-west trending line from western Syria to the Tigris River in Iraq. The results challenge previous assumptions that Roman forts in the region followed a north-south line along the eastern frontier of the Roman Empire, and were intended primarily as a defensive measure. Rather, the authors argue that the distribution of the forts suggests that they were intended to support a ‘system of caravan-based interregional trade, communication and military transport’”.

This article, by Jesse Casana, David D. Goodman and Carolin Ferwerda (2023) was published online in the journal Antiquity by Cambridge University Press on the 26th of October 2023. The selection of this article as one of the ten most important archaeological discoveries of 2023 has given it considerable publicity, including in Europe: for example, it is the subject of a presentation on a Youtube channel in Spanish (EXOPLANETAS Noticias Ciencia y Tecnología) entitled ¿EXISTIÓ una GRAN MURALLA ROMANA en el Oriente del Imperio?

The authors are specialists in remote sensing from the Department of Anthropology of Dartmouth College, Hanover, USA. Thanks to their work, the online database corona.cast.uark.edu, which is freely accessible, allows scholars to consult high-resolution and geometrically corrected imagery for the entire Near East. One of their projects involves documenting archaeological sites visible on satellite images from the 1960s and 1970s. To date, they have documented more than 10,000 archaeological sites, or potential sites, and related features in a 300.000 km2 study area extending from the Mediterranean coast of Lebanon and Syria to the Zagros Mountains of northeastern Iraq, registered in a GIS database (Casana 2020). All these features were classified according to morphological criteria. For this article, the authors have selected the morphological class comprising square (or rectangular) constructions, between 50 and 100m per side, i.e. 396 constructions that resemble quadrangular fortifications. They show several plates of Corona image extracts with buildings that are 50-80m per side (Figure 5), 100-200m per side (Figures 6 and 7) and over 200m per side (Figure 8).

The scientific legacy of Antoine Poidebard

The starting point for this study is Antoine Poidebard’s well-known publication, La trace de Rome dans le désert de Syrie (1934), which is abundantly illustrated with high-quality images. 38 (out of 116) of the buildings surveyed by Poidebard have been located. In fact, the authors selected features very close to what Poidebard defined as “forts”, and their illustrations show both buildings identified by Poidebard and “newly documented forts”. The attribution of the buildings identified by Poidebard is neither questioned nor placed in a 1930s research context. Nor is their function, which was considered essentially military – Poidebard himself being a soldier. In the 1930s, the use of what was then a revolutionary method for detecting sites – aerial archaeology – and dating them based solely on written sources meant that a group of morphologically similar sites could be attributed to a single period, without taking into account a larger picture of the evolution of the sites and territories and their long-term history (the question of the reuse of building materials, for example, is a relatively recent one). These issues have since been debated, and a number of researchers have been revisiting Poidebard’s interpretations with the publication of several books based on exhibitions created from the superb documentation bequeathed by Antoine Poidebard to the St Joseph University of Beirut (Nordiguian, Salles 2000; Denise, Nordiguian 2004; Geyer 2000).

Casana et al. make three assumptions: that the structures identified by Poidebard are indeed dated to one and the same period, “Roman”; that a single morphological response corresponds to a single type of settlement and one chronological period of construction/occupation; and that all these constructions are primarily military. As an archaeologist familiar with the use of aerial and satellite photography, I would like to take up these points again.

“Unknown”: the historiographical question

In the paragraph on the working method, the authors state “we first mapped all previously published archaeological sites in this study area, creating a database of approximately 4500 sites” (p. 1522). Were these publications consulted just to put points on a map? Having worked on the Syrian region for many years, how can the authors ignore the historiography of research on this area? It is all too easy to say that “there has been only limited archaeological research on the fortifications of the eastern Roman limes” (p. 1520). Is this because a specific search of bibliographic databases for the term “fortification” has turned up nothing? We may be surprised, for example, to find in this article only the much-quoted reference to Poidebard 1934, but no mention of Mouterde et Poidebard 1945, which is the continuation of this work.

Among the results presented, the authors state that “our survey also reveals previously undetected lines of forts extending west-southwest from the Euphrates Valley, towards Apamea and western Syria” (p. 1528). However, their bibliographical knowledge of this region is very poor. For example, the “newly documented” fort CNR5952 (Figure 8A) actually corresponds to a Roman fort, that of Isriya, the ancient Seriane, known since at least the beginning of the 20th century (Todt and Vest 2014: 1735-1736). The site was also visited by Mouterde and Poidebard (1945: 89-94 and pl. LVIII-LIX) and studied by Rüdiger Gogräfe (2016). It is a city founded under Diocletian, surrounded by a rampart and comprising a fortress with a slightly trapezoidal enclosure, measuring 190 x 140m, and a very well-preserved temple. On the same figure, at E, CRN10002, Tell ‘Awan had already been identified by Poidebard (1934, pl. I-III). Altogether, no less than a third of the “newly documented complex forts” in Figure 8 were already known!

“Roman”: an over-simplistic view of dating

Reading the article, one gets the impression that the dating is based on preconceived ideas. In fact, throughout the Mediterranean world, the interpretations of archaeological structures proposed in the last century tended to attribute to the Romans any building or system of urban or rural development that was well preserved or considered to be “well-constructed”.

Discussions with local people show that the hypothesis of an attribution to the Roman period is the most common, “roumani” often being used for “old”. The Roman attribution also confirms the adage “you only lend to the rich”, as it is true that the Roman period left many features in the local landscape that are easily identifiable (blocks of architectural decoration, columns, mosaics, inscriptions, etc.). But it seems to me that we underestimate the persistence of these preconceptions, even in scientific research. The last part of the article deals more specifically with the dating of forts, for which Jesse Casana has developed a method based on the morphology of the sites. He carefully points out that “It is important to note that dating a site based on its morphology requires a robust comparative sample of similar sites that have already been recorded by regional archaeological surveys, ideally in nearby areas” (Casana 2020: 159). In Syria, among the majority of works mentioning the creation of forts from the 3rd century AD onwards, and their occupation until the early Islamic period, only the structures analysed in Jebel Bishri are “securely dated to the Roman and Late Roman periods, with seemingly little later occupation” (Lönnqvist et al. 2011). Except that the ceramic assemblages presented by the Danish mission as “Roman” in support of this dating are more than dubious, since they are not very consistent and include a majority of body sherds that are therefore undatable. For example, the identifiable sherds illustrated in the form of colour images and approximate sketches can be dated to the 6th century at the earliest, with half of the shapes belonging to types characteristic of the 7th-9th centuries, i.e. Early Islamic (Lönnqvist et al. 2011: 283-284, figs. 24 to 28). However, this misattribution to the Roman period has been challenged (Gschwind and Hasan 2006: 368-370 and Gschwind and Hasan 2008: 459, note 27). Marcus Gschwind and Haytham Hasan clearly show the variety of dates obtained by analysing pottery (from excavations or survey) as well as the types of settlement, both civil and military.

A reading of the original publications dealing with the Roman limes could have provided better support for the dating. Furthermore, although the author considers his dating method to be more reliable than that of pedestrian surveys (Casana 2020), the contribution of the latter cannot be so much underestimated. Pedestrian surveys were developed from the 1960s onwards (Wilkinson and Tucker 1995 ; Wilkinson 2003) and made it possible to refine dating by analysing the material collected, usually pottery for the historic periods (Fig. 1). The reliability of survey material has been discussed many times; however, in the case of sites whose visible remains show no major modification, associated with a reduced and repetitive material assemblage for sites of the same type, dating with pottery becomes much more reliable (Rousset 2010).

Fig. 1: Archaeological survey in the western part of Syria. In red, sites documented by the mission Marges arides (Cartographie ANR PaleoSyr)
“Forts”: form and function of the remains identified

According to the authors, the buildings identified in the satellite images are all “forts”, i.e. buildings used primarily for military purposes and/or roadside accommodation. Once again, this does not do justice to the state of research into these structures.

Several examples illustrate this problem. Figure 5 uses Corona images to illustrate some of the buildings identified by Poidebard. However, the building in Figure 5A is in fact the Umayyad residence of Cholle (Konrad 2008: 452; Todt and Vest 2014: 1064), whereas the Roman fort is located in the centre of the town to the north-east (Fig. 2).

Fig. 2: Cholle. a: CORONA image (Corona Atlas of the Middle East), b: plan of the site (Konrad 2008 p. 452), c: umayyad residence CORONA image (Casana et al. 2023, fig. 5A)

Site CRN6070, shown in Figure 7D, corresponds to the birka – water reservoir in Arabic – at Qudaym (to confirm that it is indeed a basin, it is sufficient to locate the qanat that feeds it from the south). This hydraulic installation is associated with a square building, 700m to the north-east, measuring around 100m on each side. This complex, which was published by Poidebard in 1945 (Mouterde and Poidebard 1945: 109-113 and pl. LXXI-LXXIII), has been studied again by Denis Genequand, who has shown (with plans and ceramics) that it was an Umayyad aristocratic settlement with a mosque at its centre (Genequand 2012: 174-180). This complex is located 10km north of the village of Qudaym, which corresponds to ancient Acadama (Fig. 3).

Fig. 3: The birka (water reservoir) of Qdaym with its feeding channel. a: CORONA image (Casana et al. 2023, fig. 7D), b: GoogleEarth image 2022-03, c: Mouterde et Poidebard 1945, pl. LXXII.

Another basin is illustrated in Figure 1B: this is the birka (basin on the qanat) of Qasr al-Hayr al-Gharbi (the famous Umayyad “desert castle”), as indicated in the legend to the original plate (Poidebard 1934, pl. XXXVI; Schlumberger 1986: 3-4; Genequand 2012: 280-300).

Since the 1990s, the work of Denis Genequand (2012), among others, on Umayyad residences in present-day Syria and Jordan has shown that the prestige buildings of the 8th century were strongly influenced by the Roman fort architecture, with the purpose of highlighting social status and legitimising power (Genequand 2010). In the more intensively surveyed region, the authors discovered a high density of “forts”. However, it turns out that the vast majority of these ruined buildings, originally built in mud-bricks and either isolated in cultivated areas or within ruined or inhabited settlements in the 1960s, are the same size as the farms that were in use at the time the Corona photographs were taken. So why wouldn’t they be farms, which could be dated to any period from Antiquity to the early 20th century? The location of buildings (p.1524 “These buildings are often isolated, far from other obvious archaeological features, and frequently located in marginal environments with little other evidence of ancient or modern settlement”) corresponds to different types of locality and farming methods. Hence the importance of studying the influence of environmental conditions on the location of sites, especially in marginal environments. This is why it is so important to study both geography and archaeology. The settlements and buildings of a given period, however numerous, cannot remain just a series of points or lines on a map. The military explanation cannot be the only interpretation. The availability of water and food for people and animals, the local resources that could be used to generate income from farming or livestock rearing, and the way in which they were exploited, among others are all inseparable from the geo-strategic position of the settlements.

One of the main conclusions of the article is that the eastern line of defence of the Roman Empire was not north-south, as accepted by specialists of the period, but “Instead, we show that the forts form a roughly east-west line following the margins of the inland desert”. Here again, a minimum of geographical analysis and perspective shows that the region under study corresponds to the northern part of the Fertile Crescent and that the most easily visible sites are concentrated on the margin between the agricultural regions and the steppe zone, where they are also better preserved because they have been less affected by cultivation and the reorganisation of landholdings since the Ottoman period. This may be due primarily to a bias resulting from the nature of the studied area.

Fig. 4: Late hellenistic to byzantine fortifications in the ‘Marges arides’ area (©mission Marges arides O. Barge, M.-O. Rousset)

This raises the question of how much credence should be given to this publication and these “discoveries”. While some of the forts can indeed be considered as such, others are basins (fig. 1B and fig. 7D) or residences from the early Islamic period (fig. 5A). Others could also have had many other functions, such as fortified churches like the one at Zabad, south of Lake Jabbul, monasteries, or simple farms. The results of the North Syrian Arid Margins programme, based primarily on field surveys and the building of a dated corpus (of more than 1.000 sites dated with lithic material, inscriptions, pottery, etc.), have given rise to numerous publications by its members (Geyer 2011; Geyer, Rousset 2001; Geyer et al. 2006; Rivoal 2011; Rousset 2010; Rousset 2011; Genequand, Rousset 2016; Rivoal, Rousset 2019…). These researchers have identified a widespread settlement system for the region and described its evolution from the end of the Hellenistic period to the end of the Byzantine period (Rousset 2011). Among a variety of facilities, military buildings account for no more than 4% (Fig. 4). Some of the buildings that they identified in the field, as captured by the Corona images, strongly resemble the “forts” of the Dartmouth team (Fig. 5).

Fig. 5: Square buildings in the area East of Hama documented by the mission Marges arides. a: roman fort at ’Umm Wādī 3; b: roman fort of Minṭār al-Muġīra; c: forts at Wādī al-ʿAzīb 2 (south: roman, north: byzantine); d: late roman fort at Tabārat al-Saḫāna 1; e: byzantine fort at Rasm ‘Abd al-Ǧabbār; f: fortified church of Zabad; g: cathedral and castron (built 558-559) at al-Andarīn; h: farm of Ḫirbat Hrībǧa (pottery: byzantine, umayyad); i: birka and buildings at Birkat al-Ġadāmsa (pottery: umayyad, abbassid); j: the abbassid residence of Rasm Abū Miyāl 1; k: byzantine monastery of Wādī ’Iṯriyā 4 (CORONA imagery courtesy US Geological Survey).

Archaeology is a multi-disciplinary science par excellence, and remote sensing cannot be sufficient on its own. There is no shortage of texts on the Roman period (Procopius, Notitia Dignitatum, etc.) and historical topography has been a field widely explored by historians for several centuries (from Dussaud 1927 to the Tabula Imperii Byzantini, for example, Todt and Vest 2014). Anyone wishing to study the archaeology of the Roman period has to take their work into account. As the late Tony Wilkinson was fond of saying, today’s research, enhanced by the use of “new” technologies, must not lead us to forget the more traditional but essential approaches, such as relying first and foremost on the fieldwork and taking into account the work previously carried out by colleagues.


Casana J. 2020. Beyond survey boundaries: satellite remote sensing-based classification and dating of archaeological sites in the Northern Fertile Crescent, in D. Lawrence, M. Altaweel & G. Philip (ed.) New agendas in remote sensing and landscape archaeology in the Near East, Oxford: Archaeopress, 154-174.

Casana J., Goodman D. D. et Ferwerda C. 2023. «A Wall or a Road? A Remote Sensing-Based Investigation of Fortifications on Rome’s Eastern Frontier», Antiquity 97 (396), 1516-1533. [En ligne] https://doi.org/10.15184/aqy.2023.153

Denise F. et Nordiguian L. 2004. Une aventure archéologique. Antoine Poidebard, photographe et aviateur, Arles: Ed. Parenthèses.

Dussaud R. 1927. Topographie historique de la Syrie antique et médiévale, BAH 4, Paris: Geuthner.

Genequand D. 2010. Formation et devenir du paysage architectural omeyyade : l’apport de l’archéologie, in A. Borrut & P. M. Cobb (ed.), Umayyad Legacies: Medieval Memories from Syria to Spain, Leiden, 417-473.

Genequand D. 2012. Les établissements des élites omeyyades en Palmyrène et au Proche-Orient, BAH 200, Beyrouth: Institut français du Proche-Orient.

Genequand D. et Rousset M.-O. 2016. Résidences aristocratiques protobyzantines et omeyyades des Marges arides du nord de la Syrie, in M.-O. Rousset, B. Geyer, P.-L. Gatier et N. Awad (ed.), Habitat et environnement : prospections dans les marges arides de la Syrie du Nord, Lyon: Maison de l’Orient et de la Méditerranée, 207-265. [En ligne] https://books.openedition.org/momeditions/471

Geyer B. 2000. Des fermes byzantines aux palais omayyades, ou l’ingénieuse mise en valeur des plaines steppiques de Chalcidique (Syrie), Aux origines de l’Archéologie aérienne. A. Poidebard (1878-1955), Beyrouth: Presses de l’USJ, 109-122.

Geyer B. 2011. The steppe: human occupation and potentiality the example of Northern Syria’s arid margins, Syria 88, 7-22. [En ligne] https://doi.org/10.4000/syria.888

Geyer B., Besançon J. et Rousset M.-O. 2006. Les peuplements anciens. In R. Jaubert et B. Geyer (ed.), Les marges arides du croissant fertile : peuplements, exploitation et contrôle des ressources en Syrie du Nord, TMO 43, Lyon: Maison de l’Orient et de la Méditerranée, 55-69. [En ligne] https://www.persee.fr/doc/mom_1955-4982_2006_thm_43_1_2264

Geyer B. et Rousset M.-O. 2001. Les steppes arides de la Syrie du Nord à l’époque byzantine ou « la ruée vers l’est », In B. Geyer, Conquête de la steppe et appropriation des terres sur les marges arides du Croissant fertile, TMO 36, 111-121. [En ligne] https://www.persee.fr/doc/mom_1274-6525_2001_thm_36_1_1168

Gogräfe R. 2016. Isriye-Seriana(Itriyā). Heiligtum, Siedlung und Militärstation in Zentralsyrien von der frühen römischen Kaiserzeit bis in die mamlukische Epoche, Damaszener Forschungen 17, Darmstadt.

Gschwind M. et Hasan H. 2006. Die spätrömisch-frühislamische Zivilsiedlung Tall ar-Rum und die spätantike Besiedlung des Euphrattales zwischen Zenobia und Circesium, Damaszener Mitteilungen 15, 321-382.

Gschwind M. et Hasan H. 2008. Tall al-Rum. A Late Roman to Early Islamic settlement on the river Euphrates, in K. Bartl, A. Moaz (ed.), Residences, Castles, Settlements. Transformation Processes from Late Antiquity to Early Islam in Bilad al-Sham, Orient-Archäologie 24, 455-472.

Konrad M. 2008. Roman Military Fortifications along the Eastern Desert Frontier: Settlement Continuities and Change in North Syria, 4th–8th centuries A.D., in K. Bartl, A. Moaz (ed.), Residences, Castles, Settlements. Transformation Processes from Late Antiquity to Early Islam in Bilad al-Sham, Orient-Archäologie 24, 433-454.

Mouterde R., Poidebard A. 1945. Le Limes de Chalcis: organisation de la steppe en Haute Syrie romaine, documents aériens et épigraphiques, BAH 38, Paris: Geuthner.

Nitschke J. 2023, ANE Today – Ten Exciting Discoveries in Near Eastern Archaeology in 2023, American Society of Overseas Research (ASOR), 27 décembre 2023. [En ligne] https://www.asor.org/anetoday/2023/12/ten-exciting-discoveries

Nordiguian L. et Salles, J.-F. 2000. Aux Origines de l’archéologie aérienne: A. Poidebard (1878-1955), Beyrouth: Presses de l’Université Saint-Joseph.

Poidebard, A.  1934. La trace de Rome dans le désert de Syrie. Le limes de Trajan à la conquête arabe, Recherches aériennes (1925-1932), Paris: Geuthner.

Rivoal M. 2011. Entre steppe et plateaux basaltiques : l’occupation de la Syrie centrale à la période byzantine, Syria 88, 141-163. [En ligne] http://journals.openedition.org/syria/907

Rivoal M., Rousset M.-O. 2019. Les aménagements agricoles dans les Marges arides de Syrie du Nord (5e-10e siècles). Des témoins des modalités d’appropriation et d’exploitation des campagnes, in A. Delattre, M. Legendre et P. Sijpesteijn (eds), Authority and Control in the Countryside. From Antiquity to Islam in the Mediterranean (6th-10th Century), Leiden Studies in Islam and Society 9, Leiden – Boston: Brill, 206-260. [En ligne] https://brill.com/view/book/edcoll/9789004386549/BP000009.xml

Rousset M.-O. 2010. Deux assemblages de céramique hellénistique et romaine de la steppe syrienne. In Entre nomades et sédentaires. Prospections en Syrie du Nord et en Jordanie du Sud, TMO 55, 127-145. [En ligne] https://www.persee.fr/doc/mom_1955-4982_2010_mel_55_1_3182

Rousset M.-O. 2011. Le peuplement de la steppe de Syrie du Nord entre le IIe s. av. J.-C. et le IVe s. apr. J.-C., Syria 88: 123-139. [En ligne] https://doi.org/10.4000/syria.905

Schlumberger D. 1986. Qasr el-Heir el-Gharbi : texte et planches, BAH 120, Paris: Geuthner.

Todt K.-P. et Vest B.A. 2014. Syria (Syria Prote, Syria Deutera, Syria Euphratesia), Tabula Imperii Byzantini 15, Wien: Verlag der Österreichischen Akademie der Wissenschaften.

Wilkinson T. J. 2003. Archaeological Landscapes of the Near East, Tucson (Ariz.): University of Arizona press.

Wilkinson T. J. et Tucker D. J. 1995. Settlement Development in the North Jazira, Iraq: A Study of the Archaeological Landscape, Iraq Archaeological Reports 3, Warminster: Aris and Phillips.


Marie-Odile Rousset est archéologue, Directrice de recherche au CNRS, spécialiste du Proche-Orient médiéval. Elle travaille sur le peuplement diachronique des Marges arides et du Massif calcaire de Syrie du Nord. UMR 5133 – Archéorient, Maison de l’Orient et de la Méditerranée, Lyon.

Citer ce billet
Marie-Odile Rousset (2024, 8 mars). “396 unknown Roman forts” in the Middle East: a “major discovery” put into perspective. ArchéOrient - Le Blog. Consulté le 19 avril 2024, à l’adresse https://doi.org/10.58079/vzkf

1 réponse

  1. 16 mars 2024

    […] “396 unknown Roman forts” in the Middle East: a “major discovery” put into p… […]

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search