Saturday, 7 October 2017

Saharan and trans-Saharan contacts and trade in the Roman era

The following post offers a brief discussion of Saharan and trans-Saharan contacts in the Roman era along with a distribution map of Roman finds made beyond the southern boundary of the empire. Although such Saharan/trans-Saharan contacts have often been assumed to be a primarily medieval and later phenomenon, recent archaeological work in the Sahara and West Africa suggests that there was, in fact, a significant degree of interaction taking place from at least the first century AD through until the seventh century. This interaction is thought to have been primarily driven by a trans-Saharan trade in slaves that was largely organized and controlled by the ancient Garamantes of the Libyan Sahara.

Roman and early Byzantine finds from Saharan and sub-Saharan West Africa, after Wilson (2012), MacDonald (2011), Magnavita (2009 & 2013), and Fenn et al (2009); also shown  are a selection of Saharan trade routes that may well have functioned in antiquity after Wilson (2012), with a possible western addition from Boone et al (1990), and the location of the Garamantian capital of Garama. Click here for a larger version of this image; note, the black dotted line represents the approximate normal southern edge of the Roman Empire in the second century AD. Image drawn by C. R. Green using a public domain basemap from NASA/Wikimedia Commons.

Undoubtedly the most significant archaeological evidence for Roman interaction with the regions to their south comes from the Libyan Sahara, in particular the Wadi al-Ajal (Fazzan) area once occupied by the Garamantes, around 1,000 kilometres south of Tripoli. Although the Garamantes are referred to by a number of classical authors from Herodotus onwards, it is only in recent years that the scale and significance of both the Garamantian civilisation and Roman trade and contacts with them has been recognized. In particular, research by the Fazzan Project and the Desert Migrations Project has demonstrated that the Garamantes made use of elaborate underground irrigation systems known as foggaras in the Fazzan area of Libya in order to create a prosperous oasis civilisation in the Sahara desert, with several small planned towns and a capital, Garama (modern Germa/Jarma).

At its height, Garama was home to around 4,000 people, with a further 6,000 living within 5 km in surrounding satellite villages and many more—perhaps up to 100,000 in total—living across the Garamantian territory as a whole, and the archaeological evidence accumulated over the last generation or so from this area indicates that there were, in fact, significant Roman influences on both Garamantian architecture and culture, despite its situation so far to the south of the Roman border. So, for example, monumental public buildings and the grander houses of the Garamantes from the first century AD were built using ashlar stonework, rather than mudbrick, with colonnaded courtyards, Mediterranean-type wine presses, and even hypocaust fragments, marble veneers and hydraulic cement indicative of a Roman-style bath-house all being in evidence. Likewise, significant quantities of Roman imports have been recovered from over 200 sites in the Wadi al-Ajal and southern Fazzan, including Roman finewares such as African Red Slip Ware; amphorae that once contained wine, olive oil and fish products; and lamps, jewellery and glassware. As Andrew Wilson has emphasised in an important survey of the evidence,
The apparent ubiquity of imported pottery (finewares and transport amphorae) suggests that imports from the Roman world were not simply restricted to an elite few, but were fairly widely available in Garamantian society, both in the Wadi al-Ajal and the Murzuq depression.
The peak of this exceptional Roman contact and trade with the Saharan Garamantes, suggested to have required a caravan trade 'numbering in the hundreds of camel loads per year', appears to have come in the late first to early fourth centuries AD, but Late Roman and early Byzantine imports continued to arrive in this region through into the fifth, sixth and seventh centuries, albeit in lesser quantities. This relative decline in trading across Late Antiquity is thought to have been mirrored by the failure of the underground irrigation systems that supported the Garamantes' civilisation and its transit trade (due to the water table that the foggaras tapped falling below an economically exploitable level), a process that may well have been completed by the time of the first Arab incursions into the region in the mid-seventh century and which arguably led to the recorded political instability in the northern Sahara and along the Roman frontiers during Late Antiquity.

Kite photograph of the archaeological remains at Germa, Libya, capital of the Garamantes; click for a larger version of this image. The building in the bottom half of the image had stone footings and was excavated in the 1960s; it was fronted by a broad set of steps and incorporated columns in its facade. (Image © Toby Savage, used by kind permission). 

Beyond the probable territory of the Garamantes there have been further finds of Roman material, although the quantities involved are much smaller than those encountered in the Fazzan region of Libya. Within the Saharan desert, there is a scattering of Roman material to the west and south-west of the Garamantes, which have been recently mapped and briefly discussed by Andrew WilsonKevin McDonald and Katia Schörle. For example, a Roman oil lamp, a glass goblet and the imprint of a coin of Constantine on gold leaf were found in the fourth-/fifth-century 'Tomb of Tin Hinan' (Abalessa, Algeria) in the Central Sahara, and a painted Latin inscription and coins have been found in the same area of southern Algeria at Ti-m-Missaou, whilst sporadic Roman finds from sites such as Hassi el-Hadjar and Fort Miribel further north in the Algerian Sahara have been interpreted by Wilson as reflecting the development of a small-scale western route through the desert by around the third century AD.

In addition to these central and northern Sahara finds, a small number of items of Roman manufacture or origin are also known from the southern shore of the Sahara and the semi-arid grasslands of the Sahel. From the far west, in southern Mauritania, there are a handful of coins dating from the first century BC to the third century AD, including two of Severus Alexander from Nouakchott and Tamkarkart. Perhaps more significant, however, are a number of fascinating finds from sites in Burkina Faso and Mali. As was noted in a previous post, a fourth- to seventh-century cemetery site at Kissi, Burkina Faso has produced cowrie shells from the Red Sea or further afield, carnelian and glass beads imported from both Egypt/the Levant and the Sasanian Middle East, and copper-alloy items made from metal that was imported from the Eastern Mediterranean and Britain, probably via Carthage. Likewise, there have been finds of amphorae rims which seem to be imitative of North African amphorae of Late Roman/early Byzantine date from three sites in Mali, to the west of Kissi, one from a context dated c. 450–600 AD. Other arguably relevant items include a number of early beads from Djenné-Jeno, Mali; metallurgical debris and ingots from Marandet, Niger, which match with some of the imported metal found at Kissi; and a second-century AD Janus statue from Roman North Africa found at Zangon Dan Makéri, southern Niger.

Brass anklets found in a fifth- to seventh-century AD grave at Kissi, Burkina Faso (West Africa), made with metal probably imported from the Eastern Mediterranean and Britain via Carthage; see further Fenn et al, 2010. (Image: B. Voss/Afriques, CC).

Turning to the question of what all this means, both David Mattingley and Andrew Wilson have argued that the sheer quantity of available evidence indicates that there must have been a very substantial degree of Saharan and trans-Saharan trade taking place in antiquity. This was probably primarily mediated via the Garamantes of Fazzan, given that the vast majority of Roman exports were concentrated in the hands of—and consumed by—the Saharan inhabitants of that region, with only a very small proportion of this rich array of Mediterranean goods being subsequently traded on into sub-Saharan West Africa, and Wilson thus suggests that we are probably dealing with a network of interlocking sub-systems of short-, medium- and long-distance exchange in and across the Sahara rather than a single trans-Saharan trade route. As to just what was being traded northwards and southwards via these networks, natron, cotton, and gemstones were probably minor components in the Saharan trade with the Roman Empire, but these and other local Garamantian products are almost certainly insufficient to explain the substantial quantity and ubiquity of Roman imports now known from the Libyan Sahara. As such, it is generally thought likely that the primary commodity exchanged for Mediterranean products by the Garamantes originated further south and was obtained by them either by trade in return for Saharan salt, alum and perhaps grain, or by force of arms (something hinted at by our textual references to the Garamantes raiding their southern neighbours)—in either case, the primary commodity in question is believed to be enslaved people.

As Wilson notes, such a trans-Saharan trade in enslaved people certainly operated in the medieval and early modern eras, and our available classical textual sources do indeed imply that the Garamantians were engaged in slave-raiding to their south. Needless to say, his consequent argument that the Garamantes were controlling an earlier form of the trans-Saharan slave-trade operating at a similar or greater scale to that of the medieval and early modern periods (c. 5,000–10,000 slaves per year, across all routes) not only makes sense of the otherwise inexplicable and exceptional concentration of Roman imports in the Libyan Sahara, but moreover explains where the necessarily massive workforce required to dig and maintain the hundreds of kilometres of underground foggara irrigation channels that supported the Garamantes' oasis culture originated too. It has also been argued that the evidence for the presence of slaves of 'sub-Saharan' African ancestry in the Mediterranean world, as most recently surveyed by Elizabeth Fentress and Kyle Harper, potentially supports the existence of such a Garamantian-controlled trade—for example, Fentress notes that there are an increasing number of images showing African subjects in servile positions over classical antiquity, most especially involving children, which she associates with the Garamantes, and a third-century AD inscription from Hadrumentum (modern-day Sousse, Tunisia) does, in fact, directly refer to a black slave as faex Garamantarum, the 'dregs of the Garamantes'. In this light, it is perhaps additionally worth pointing out that there is physical evidence for the likely presence of people of 'sub-Saharan' African ancestry (albeit of varying statuses) in both the Garamantian kingdom and the Roman Empire beyond too. For example, Mattingley notes that a skeletal analysis of Garamantian graves in the Fazzan area of Libya indicates that the people buried there included a significant proportion who seem to be of 'sub-Saharan' ancestry, with one woman recently excavated at Taqallit being furthermore found buried with a sub-Saharan-style lip plug. Likewise, recent work at Roman York, London and Leicester has suggested that a notable proportion of the people interred in the second- to early fifth-century AD urban cemeteries there—respectively c. 11%, 24% and 6% of the total examined—are likely of 'sub-Saharan' African descent, as is the famous third-century AD 'Beachy Head Lady' from East Sussex, although it does also need to be observed that there is no evidence that any of these people were themselves enslaved or of Garamantian origin and a number are, in fact, thought to have been of probably high social status and/or born in Britain.(1)

In sum, the available evidence seems to point to a substantial degree of Saharan/trans-Saharan contact and trade in the Roman era, potentially equal to or even greater than the level seen in the medieval and early modern periods, given the quantity and ubiquity of Roman imports now known from the Libyan Sahara. This trade appears to have been primarily mediated via the Garamantes, who had established a prosperous desert oasis civilisation in the Fazzan area around 1,000 kilometres to the south of Tripoli. The peak in this trade came between the first and early fourth centuries AD, although it continued into the sixth and seventh centuries, and it seems probable that the Garamantes were, in the main, exchanging Roman goods and luxuries for enslaved people obtained to their south, either through trading or by raiding, although other goods may well have played a minor role too. Relatively few Roman exports made it across the Sahara into the Sahel, suggesting that unlike in later eras trading was not directly trans-Saharan but conducted through a network of interlocking trading sub-systems.

Unloading camels in Egypt, from the Late Antique (sixth-century?) 'Ashburnham Pentateuch', BnF NAL 2334, f. 21r. (Image: BnF, Public Domain).


1.     Note, the higher proportion at London may reflect the fact that only 17 skeletons were subjected to ancestry analysis, compared to 83 at Leicester and 85 at York. For details of the cemeteries at York, London and Leicester, see S. Leach et al, 'Migration and diversity in Roman Britain: a multidisciplinary approach to the identification of immigrants in Roman York, England', American Journal of Physical Anthropology, 140 (2009), 546–61, online at; R. C. Redfern et al, 'Going south of the river: a multidisciplinary analysis of ancestry, mobility and diet in a population from Roman Southwark, London', Journal of Archaeological Science, 74 (October 2016), 11–22, online at; and M. Morris, 'Between road and river: investigating a Roman cemetery in Leicester', Current Archaeology, 319 (2016), online at, & 'Leicester's Roman skeletons have "African links"', BBC News, 2 December 2016, online at

The content of this post and page, including any original illustrations, is Copyright © Caitlin R. Green, 2017, All Rights Reserved, and should not be used without permission.