(1) Overview
Context
The geospatial data for the Medieval road system in the Levant/Bilād al-Šām were collected in the context of the project ‘VIA-TARIQ: Analysing the long-term change and persistency of the Roman road system in the Levant’. The goal of the project was to 1) analyse the long-term development of the terrestrial transport network from the Roman to the Medieval period, 2) analyse the persistency of the Roman transport infrastructure (roads, stations, etc.) in the Medieval period. Dalgaard et al [1] previously claimed that there is little continuity between the Roman and modern road network in the Middle East and North Africa (MENA) region, based on their analysis of Roman road network digitised from the Barrington Atlas [2, 3] and Medieval caravan routes. Knowing the deficiencies of the datasets that Dalgaard et all used in their large-scale continental-wide analysis (low resolution and accuracy of the geospatial data, biases in data collection and publication of the Barrington Atlas, low representativity of the caravan routes for the complete Medieval road system, etc.) it was decided to take their conclusions as a starting point and to test the hypothesis stating there is low continuity between the Roman and the Medieval road network in the case of the Roman Near East. While the Roman road data were recently compiled and published, resulting in very high-resolution, spatially explicit and richly annotated dataset Itiner-e [4, 5], until now there did not exist a comparable digital and openly accessible dataset for the Medieval (Early-Middle Islamic) road system in the MENA region. An exception is the al-Ṯurayyā Project [6, 7], a digital version of Cornu’s atlas [8] of ‘classical’ Islamic sites and routes (9th–10th c. CE).
Therefore, in order to analyse the long-term development of the road network in the Near East and persistency of the Roman road network and its influence on the Medieval network this digital dataset was conceived. It was decided to apply similar methodology as was used to create Itiner-e by compiling available published historical (Medieval geographers, itineraries), epigraphic (milestones), and archaeological (settlements, road stations, bridges, forts, etc.) evidence to reconstruct the Medieval terrestrial transport network in Bilād al-Šām. This dataset is the first attempt to bring all this evidence together and reconstruct the Medieval road system in high spatial resolution and accuracy and to publish the geospatial data following FAIR and Open Science principles (Figure 1).
Map of the study region and digitised road networks. a) Early Islamic period, b) Middle Islamic period, c) Early Mamluk period, d) Late Mamluk period, e) Study region (marked in black outline) in its geographical context. © Adam Pažout.
Spatial coverage
The spatial coverage of the dataset roughly corresponds to the traditional definition of Bilād al-Šām and ancient Greater Syria from Taurus Mountains in the north to the Red Sea in the south and between the Mediterranean Sea and the Euphrates. It covers contemporary Syria (west of the Euphrates), Lebanon, Jordan, Israel and Palestinian territories, and parts of south-eastern Turkey and Egypt’s Sinai Peninsula.
Spatial extent (EPSG: 3395):
Northern boundary: 4 647 767.7 m
Southern boundary: 3 354 942.7 m
Eastern boundary: 4 564 089.6 m
Western boundary: 3 673 544.7 m
Temporal coverage
The dataset covers the Early and Middle Islamic periods in the Bilād al-Šām, ca. 650-1517 CE.
(2) Methods
The dataset is based on available published historical, archaeological, and epigraphical evidence, and no field survey or field verification was conducted in data collection and compilation. The methodology closely follows the methodology employed in the creation of the Itiner-e dataset [4] as much as it could be followed and adapted to different material base and schemes of Islamic archaeology.
Compared to the Roman archaeology, there are considerable material, spatial, and temporal biases in addition to biases in intensity, spatial coverage, and focus of the archaeological investigations in the region of interest and on the Medieval road system. Road archaeology and transport networks are notably absent from the Oxford Handbook of Islamic Archaeology [9]. Milestones are confined to the Umayyad and Abbasid periods (7th–9th c. CE). Road stations (khān, caravanserai) appear late in the archaeological record (Ayyubid period, ca. 12th c. CE) and are often studied for their architecture and in the context of social and economic history. Attempts were made to reconstruct the road network based on locations of road stations, traveller’s accounts, and itineraries [10, 11, 12]. Dussaud [13] and the Tabula Imperii Byzantini project [14] provide sweeping overview of the evidence but fail to capture the long-term development and their maps are a palimpsest of roads from the Roman to the Ottoman periods. Petersen [15] focused on the Syrian section of the Hajj route but is mostly preoccupied with road stations and does not deal with the actual course of the road in the landscape. Sauvaget reconstructed the network of the Mamluk barīd (postal and communication system) based on Mamluk sources and research on Syrian road stations [16, 17, 18]. Franz [19] recently reviewed the evidence for Ayyubid and Mamluk communication system and provided a geospatial gazetteer of its identified stations. Bridges received some attention but generally are understudied, and there are difficulties with their dating [20, 21]. The study of Medieval roads in the Bilād al-Šām is therefore skewed either towards particular material evidence (road stations, bridges) or historical sources (Arabic geographers, merchant and pilgrim itineraries) and so biased towards the major communication and commercial routes. Only Major [22] attempts to reconstruct the regional road network in the 12th–13th c. CE northern Syrian coast drawing on wider archaeological evidence for settlement patterns and castles in addition to historical records.
This dataset, therefore, to a certain degree reflects these biases, as it was not always possible to avoid or ‘correct’ them without being too speculative in my reconstruction. There are no comprehensive historical-archaeological gazetteers of the Medieval Bilād al-Šām that would help to ‘fill the gaps’ using settlement contexts to establish the course of a road. The sites compiled in this dataset, then, are in no way representative of the totality of archaeological sites dated to the Medieval period in the study region. This dataset should be treated as preliminary and necessarily incomplete due to those biases and lack of research on many aspects of the road archaeology in the Medieval Bilād al-Šām. It is necessary first step in the future research on the roads and terrestrial transport networks of the Medieval MENA region.
In order to explicitly incorporate the uncertainty introduced by these biases in the dataset, several data descriptors are used. 1) Segment certainty refers to the spatial accuracy of the digitisation of the road segment. 2) Road certainty refers to certainty of the presence of a road in a given period, based on the available sources. Segment certainty has three values: ‘certain’, ‘conjectured’, and ‘hypothetical’. ‘Certain’ refers to road segments that were identified through archaeological investigations, or it was possible to locate them with high spatial accuracy based on the archaeological and historical evidence within +/–50 m in mountain terrain (typically mountain passes) and +/–200 m in flat or moderate topography. ‘Conjectured’ denotes road segments whose spatial accuracy is lower due to insufficient evidence. ‘Hypothetical’ denotes road segments with very low spatial accuracy or roads without a fixed course where multiple layouts could have been used at the same time (such as desert roads), a digitized line then represents one of potential road layouts. Road certainty has three values: ‘high’, ‘medium’, and ‘low’. ‘High’ signifies that the presence of a road segment is signalled by both primary (historical) sources, such as itineraries, and by secondary archaeological and historical-geographical sources. ‘Medium’ denotes road segments for which we lack primary historical sources but there exist secondary sources on them. Finally, ‘low’ denotes road segments for which we lack both primary and secondary sources and their digitisation is therefore based mainly on the settlement context and judgment of the author of the digitisation. The summary statistics of the Segment certainty and Road certainty categories by periods are shown in Table 1.
Table 1
Summary statistics of Segment certainty and Road certainty categories of the road dataset by periods.
| EARLY ISLAMIC | MIDDLE ISLAMIC | EARLY MAMLUK | LATE MAMLUK | |||||
|---|---|---|---|---|---|---|---|---|
| SEGMENT CERTAINTY | ||||||||
| LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | |
| Certain | 1,622.8 | 13.2% | 1,331.9 | 11.3% | 1,339.3 | 11.6% | 1,333.6 | 11.5% |
| Conjectured | 8,412.5 | 68.4% | 9,417.8 | 79.7% | 9,436.4 | 82.1% | 9,614.5 | 82.6% |
| Hypothetical | 2,227.2 | 18.4% | 1,065.2 | 9% | 716.1 | 6.3% | 685.4 | 5.9% |
| ROAD CERTAINTY | ||||||||
| LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | LENGTH (km) | PERCENTAGE | |
| High | 7,656.4 | 62.3% | 7,515.9 | 63.6% | 7,957.5 | 69.2% | 8,104.1 | 69.7% |
| Medium | 1,650.1 | 24.3% | 4,197.7 | 35.5% | 3,500.4 | 30.5% | 3,495.5 | 30% |
| Low | 4,441.5 | 13.4% | 101.3 | 0.9% | 33.8 | 0.3% | 33.8 | 0.3% |
The road dataset is divided into sub-periods that reflect different historical and material sources available for each period and to better understand and analyse the long-term development and change of the road system (Table 2). The periodisation combines a revised archaeological and historical/dynastic scheme applicable to the region [23, 24, 25, 26].
Table 2
Periodisation and terminology used in the dataset.
| PERIODISATION USED | DATING | DYNASTY/OTHER TERMINOLOGY |
|---|---|---|
| Early Islamic | 650–950 CE | Umayyad, Abbasid |
| Middle Islamic | 950–1250 CE | Fatimid, Middle Byzantine, Seljuk, Crusader, Ayyubid |
| Early Mamluk | 1250–1350 CE | Mamluk, Middle Islamic |
| Late Mamluk | 1350–1517 CE | Mamluk, Middle Islamic |
Early/Late Mamluk distinction reflects small changes in the barīd network that took place before ca. 1350 CE and the construction of several new road stations during the second half of the 14th c. and 15th c.
Medieval and modern Arabic and Turkish toponyms are used with simplified English transcription (without punctuations and special symbols). Common English names (Damascus, Jerusalem, etc.) are used in certain places.
Steps
The road digitisation was made in three steps: a) road identification, b) road localization, c) road digitisation. Concurrently, sites associated with the road system (start and end points, road stations, bridges, forts, sanctuaries, settlements, and cities) were digitised using the same sequence of a) identification, b) localisation, c) digitisation.
In the first step, roads and sites were identified based on historical sources, previously published atlases, regional summaries, milestones, archaeological surveys, and excavations. The project principally relied on Dussaud [13] and Tabula Imperii Byzantini [14] for Syria and Northern Syria, Cornu’s atlas for the Early Islamic period [8], and Archaeological Survey of Israel and MEGA Jordan databases for the southern Levant (Israel/Palestine, Jordan). Umayyad and Abbasid milestones are known only from several roads south of Damascus, and most were published by Sharon in Corpus Inscriptionum Arabicarum Palaestinae (CIAP) [27, 28], which also includes building inscriptions of khāns. Road stations were collected first by Sauvaget [16, 27], and more recently by Cytryn-Silverman in Israel and Tavernari in all of Bilād al-Šām, who also provides a reconstruction of the road network in the Ayyubid-Mamluk period [10, 11]. Historical sources used cover mainly Early Islamic geographers in Bibliotheca Geographorum Arabicorum (ibn Khurdhadhbih, Qudamah ibn Jafar, al-Maqdisi, ibn Hawqal, al-Istakhri) [29, 30, 31, 32], Naser-e Khosraw for Middle Islamic period [33], and al-Umari and al-Qalqashandi for the Mamluk barīd [19, 34]. The selection of historical sources was limited by availability of the primary editions and works in translation. For the complete bibliography used for road and site identification, see the bibliography.md file in the dataset.
In the second step, the identified roads and sites were spatially located. This proved to be the most sensitive issue in the case of road localisation, as most of the sources used in step one are less concerned with recovering the actual course of the road in the landscape and focus more on routes (itineraries) without specifying topographic details of the road. Where possible, additional archaeological data were collected from archaeological and survey reports and databases such as the Archaeological Survey of Israel and MEGA Jordan to recover regional settlement patterns and propose the course of a road using settlement context. The information obtained from sources in step one was subsequently compared to various planimetric sources, such as modern and historical satellite imagery [35, 36] and topographical maps, e.g., Survey of Western Palestine [37], Carte du Liban [38], Karte des Ostjordanlandes [39], Karte von Kleinasien [40], French and British maps in 1:50,000, 1:100,000, and 1:200,000 scale [41, 42, 43], and 1:100,000 maps of Turkey [44].
In step three, each road segment and site were manually digitised following spatial localisation suggested or recovered in step two in a Geographic Information Systems (GIS) platform and stored as a single data record in a vector layer with an associated attribute table. The digitised road sections follow a line connecting known sites associated with a route (settlements, milestones, bridges, road stations, etc.). The digitised paths of such segments are refined by comparing them to existing contemporary and historical roads on georeferenced topographic maps and satellite imagery. The digitised reconstructed segment aims to consider local topography and conform its course to local obstacles (wetlands, rugged terrain, etc.). In many cases, thus ‘conjectured’ road follows a modern or historical road where additional archaeological and historical data are missing. As a rule, if the start and end points of a road remain the same between periods and no other variables are known to influence the course of the road, the road segments are not changed between the periods. This method was applied to the Early Islamic period using the Itiner-e dataset and subsequently to the following periods (i.e., Early Islamic to Middle Islamic etc.). Summary statistics of the total road length and number of sites is in Table 3.
Table 3
Summary statistics of the dataset.
| PERIOD | ROAD LENGTH (km) | NUMBER OF SITES |
|---|---|---|
| Early Islamic | 12,262.5 | 361 |
| Middle Islamic | 11,814.9 | 409 |
| Early Mamluk | 11,491.8 | 429 |
| Late Mamluk | 11,633.5 | 443 |
Geodesic length of each road segment in meters was calculated using the WGS84/World Mercator (EPSG: 3395) coordinate system and stored as an attribute.
Quality Control
Topological consistency of the road segments was checked to ensure that: a) the lines do not self-intersect, b) do not self-overlap, c) all road segments are split at intersections with other road segments or where they intersect with points in _sites datasets.
Constraints
The main constraints of the dataset are related to the accuracy and spatial resolution of the location of roads and road related sites in the archaeological literature. Archaeological Survey of Israel and MEGA Jordan are geospatial databases, and the accuracy of the representative points of archaeological sites is usually good. In other instances, the information obtained from publications was cross-checked with available satellite imagery and topographic maps to locate the reported archaeological remains. Segment certainty field is used to store information about the spatial accuracy of the road data.
Chronological information is often missing in the _sites datasets, as it was not always possible to establish approximate start and end dates of each site based on the available literature, besides general attribution to a certain period. Missing values are recorded as ‘9999’ in the database.
(3) Dataset description
Object name
The dataset consists of separate road and site files for each sub-period following the chronological scheme defined above in (2) Methods section using this naming convention: period_name_roads or period_name_sites. bibliography.md file contains the complete bibliography used for creation of the dataset. readme.md file contains a brief summary and description of the dataset.
Data type
Digitised roads are stored as vector polylines (ESRI Shapefile) or as a FeatureCollection with LineString geometries (GeoJSON). Digitised sites are represented as points and stored as vector points (ESRI Shapefile) or as a FeatureCollection with Point geometries (GeoJSON). All geospatial data is in the projected coordinate system WGS84/World Mercator (EPSG: 3395).
Each road segment and site is treated as a single record (row in the database) in the geospatial database with associated variables stored in the attribute table. The attribute table is structured to hold basic descriptors of each road segment and site, such as unique identifiers, typology, information related to spatial accuracy and certainty, chronological, historical, and bibliographical information, and creator of each record.
Data structure of _roads files is described in Table 4. Data structure of _sites files is described in Table 5.
Table 4
Data structure of the _roads files.
| FIELD NAME | MEANING | TYPE | DESCRIPTION |
|---|---|---|---|
| Id | Id | integer (Long Integer) | Unique numerical identifier |
| name | name | string (text) | Name of the road segment, typically in the form ‘Start point-End point’ using Medieval or modern names depending on whether the Medieval name is known. Arabic forms are preferred, English forms that are commonly used (Jerusalem, Damascus, etc.) are employed |
| type | type | string (text) | Type of the road segment, values: ‘Main Road’ and ‘Secondary Road’ |
| typeCode | type code | integer (Long Integer) | Road type numerical code, values: 2 = ‘Main Road’ and 1 = ‘Secondary Road’ |
| itinerary | itinerary | string (text) | Name of the known itinerary associated with the road segment, e.g., ‘Hajj Route’, ‘Barid’ etc. |
| source | source | string (text) | Medieval source that (probably) references the road segment, short reference in the form ‘Name Page Number’ of the standard primary edition or translation (see bibliography.md file for the full reference) |
| biblio | bibliography | string (text) | Reference to a secondary literature that was used to identify, localise, and digitise the road segment. Short reference in the form ‘Name Year’ (see bibliography.md file for full reference) |
| segmentCer | segment certainty | string (text) | Spatial accuracy of the digitisation of the road segment. Values: ‘Certain’ (+/– 50 m in mountainous terrain, +/–200 in flat terrain), ‘Conjectured’, ‘Hypothetical’ (mostly unfixed desert routes) |
| roadCer | road certainty | string (text) | Certainty of the existence of the road segment based on the available primary and secondary sources. Values: ‘High’ (primary and secondary sources, or good archaeological/historical-geographical secondary literature), ‘Medium’ (based on archaeological and historical-geographical secondary literature), ‘Low’ (based only on settlement context or archaeological secondary literature) |
| geoLength | geodesic length | integer (32-bit Float) | Geodesic length of the road segment in meters |
| citation | citation | string (text) | Author of the digitisation |
Table 5
Data structure of the _sites files.
| FIELD NAME | MEANING | TYPE | DESCRIPTION |
|---|---|---|---|
| Id | Id | integer (Long Integer) | Unique numerical identifier |
| name | name | string (text) | Medieval name of the site. Arabic forms are preferred, in some cases Frankish/Latin names are used |
| modernName | modern name | string (text) | Modern name of the site |
| type | type | string (text) | Type of the site. Values: ‘city’, ‘settlement’, ‘station’, ‘fort’, ‘place’, ‘sanctuary’, ‘bridge’ |
| minDate | minimum date | string (text) | Earliest date the site can be dated to. For unknown 9999 is used |
| maxDate | maximum date | string (text) | Date of the destruction/abandonment of the site. For unknown or continuous activity until the present 9999 is used |
| umayyad | Umayyad period | string (text) | Whether site exists in given period. Value: ‘yes’, ‘no’, ‘unknown’, ‘possible’ |
| abbasid | Abbasid period | string (text) | Whether site exists in given period. Value: ‘yes’, ‘no’, ‘unknown’, ‘possible’ |
| fatimid | Fatimid/Middle Byzantine period | string (text) | Whether site exists in given period. Value: ‘yes’, ‘no’, ‘unknown’, ‘possible’ |
| ayyubidCru | Ayyubid-Crusader period | string (text) | Whether site exists in given period. Value: ‘yes’, ‘no’, ‘unknown’, ‘possible’ |
| mamluk | Mamluk period | string (text) | Whether site exists in given period. Value: ‘yes’, ‘no’, ‘unknown’, ‘possible’ |
| citation | citation | string (text) | Author of the digitisation |
| biblio | bibliography | string (text) | Reference to a secondary literature that was used to identify, localise, and digitise the site. Short reference in the form ‘Name Year’ (see bibliography.md file for the full reference) |
bibliography.md (Markdown file) is divided into sections “Sources”, “Roads”, “Sites”, and “Other” (i.e., publications consulted during the creation of the dataset but not directly used in data collection and compilation).
readme.md (Markdown file) contains information about the creator, license, and data structure with a brief summary of the dataset.
Format names and versions
.shp; .geojson; .md
Creation dates
Between September 2025 and June 2026.
Dataset Creators
Adam Pažout designed the database and collected the data.
Language
English
License
CC-BY 4.0 International
Repository location
Publication date
12/08/2026
(4) Reuse potential
This dataset for the first time integrates historical and archaeological data related to the Medieval road and transport network in the Levant/Bilād al-Šām into a geospatial database. By making this data FAIR it fosters interdisciplinary engagement between archaeology, history, geography and economy. It can be reused as a reference point for further studies on the Medieval road system and terrestrial transport network in the region and for large-scale quantitative geospatial, network, and econometric analyses and comparison with road systems in other parts of the world and periods. Additional historical-archaeological validation of parts or of the whole dataset might be performed. It can be reproduced for educational and dissemination activities that focus on the Medieval history and Islamic archaeology. The dataset is relevant for studies on economic, social and political history, Islamic archaeology, development of transportation network and technology, and comparative history.