Introduction

1The theme of this volume highlights the historical importance of the relationships between people and animals. While many chapters focus on exotic species, this paper concerns the working animals whose capabilities transformed agriculture, transport, trade and war in Maghreb and Sahara. The camel is perhaps the most celebrated of several species of animals that have been used to provide transport and traction from at least the 1st millennium BC; others include horses, mules and donkeys1. The contributions of each of these species to human communities and the emergence of complex societies in the Maghreb and Sahara have been critical. Yet much remains unresolved and uncertain about the processes of animal domestication in Africa, the precise roles and the scale of use of these different beasts of burden and the ways in which societies changed and developed as some chose to adopt and husband these animals (Blench and Macdonald 2000; Gifford-Gonzalez and Hanotte 2011; Mitchell 2015). The horse, camel and donkey have been characterised as ‘the noble, the fabulous and the humble’ and there is certainly a tendency in many accounts of Saharan travel and trade to focus on the first two species to the disadvantage of the more humble donkeys and mules (Bernus 2006, 51).

2In the early 20th century, the prevalent view was that the introduction of the camel was a late Roman innovation designed to shore up the frontier in the 3rd or 4th century AD, but with unforeseen and disastrous consequences once the camel was also taken up by the nomadic peoples of the Sahara (Carcopino 1925; Gautier 1937, 188-201; Leschi 1940). Demougeot offered a partial rebuttal of this model by demonstrating that the camel was first encountered in Africa by Rome as early as the 1st century BC and was not brought there by Romans from elsewhere (Demougeot 1960). By the later 20th century, the colonialist perspective was being increasingly challenged (summarised in Mattingly 1995, 178-180). For instance, Brent Shaw argued that the dromedary was present from a much earlier date in the arid-zone of North Africa, even countenancing direct continuity with wild camelids identified in prehistoric assemblages (Shaw 1979). However, there does seem to be a genuine gap in the record of Saharan camels – although they are present in some phases of the wet Sahara, they are absent from the early rock art and from the faunal assemblages associated with the Neolithic herding and hunting communities. The reappearance of the camel cannot be dated with precision, but it is linked with a number of other significant novelties including horses, wheeled vehicles/chariots and oasis cultivation of a developed package of crops. Whether or not there were relict specimens of wild camelids in some Saharan refugia, the eventual dissemination of camels followed the introduction of domesticated stock from the direction of the Nile along a chain of pioneer oases (Liverani 2000). A recent discussion of the issue has argued for an initial phase of development of Saharan trade primarily focused on donkeys and horses in the early 1st millennium BC, followed by the introduction and progressive increase in use of the camel from the mid-1st millennium BC (Liverani 2006, 450-452). Austen is more equivocal about the date by which camels became important and opened up the possibility of caravan traffic across the Sahara, arguing that this did not occur until medieval times (Austen 2010, 17-19).

3By the time that early European travellers in the 19th century were recording their observations of North Africa, the camel was the sine qua non of Saharan travel and trade2, therefore it is all too easy to assume it enjoyed the same pre-eminence from its initial appearance on Saharan trails3. But was that always the case? We shall argue that it was not and this raises new questions about the reasons behind the eventual predominance of the camel over other beasts of burden and the historical stages of this transformation. This contribution will outline the long-term histories of the species that criss-crossed the Sahara and how these animals are essential to our understanding of Saharan trade.

4After briefly reviewing the characteristics and capabilities of the main transport animals, we shall present the evidence for the use of each species in the pre-Islamic era. Excavations in the Libyan Sahara and at the termini of the Trans-Saharan trade routes provide an opportunity to reassess our understanding of the role of these species in Saharan travel and trade. We shall also examine a variety of other evidence pertaining to transport animals, such as the ostraca of Bu Njem, a range of iconographic evidence including sculptures and rock art. Finally, we shall consider the implications of changes to the weight and volume of Roman amphorae carried deep into the Sahara.

Transport Animals of the Sahara

5It is commonly asserted that the camel was pre-eminently suited to desert travel in comparison to its rivals: horses, donkeys, mules and oxen4. This is certainly true in the most extreme conditions, but in periods of climatic amelioration or on better watered routes the advantages are less obvious. Careful appraisal needs to be made of environmental conditions on each Saharan route to understand the potential use of animals5. In the 1940s, donkeys were primarily used in the central Sahara for drawing well-water, horses were a rare luxury and camels were the essential beast of burden, carrying c. 150 kg (Bernus 2006; Despois 1946, 174-176; Lethielleux 1948, 89-92). Donkeys are generally thought to have roughly a third of the carrying capacity of camels, with a maximum load of about 65 kg (Barker et al. 2007, 344-345).

6The abundant rock art of the Sahara provides a relative, but not closely dated, sequence for the appearance of the key domestic beasts of burden, with successive phases identified as the Bovid, Equid and Camelid periods6. Cattle were domesticated from the 6th millennium BC and the donkey may also have been domesticated at an early date, though images are often described as wild donkeys (Lutz R. and G. 1995, 83, 97). Following major climate change c. 5,000 years ago, travel within the Sahara became much more difficult and the abundant horse imagery is conventionally dated to 1500 BC and later (Camps and Gast 1982; Lutz R. and G. 1995, 63-64), with the camel assumed to appear towards the end of the 1st millennium BC or a little later (Hachid 2000, 225-228; 2006; Lutz R. and G. 1995). However, Liverani has speculated that most occurrences of horses and chariots date to the mid-1st millennium BC and later, a position also adopted for the most south-westerly occurrences in Mauretania (Liverani 2006, 450-452; on Mauretania: Vernet and Naffe 2003). Di Lernia and Zampetti have published the palimpsest paintings from the Ti-n-Anneuin shelter, which include scenes of oasis palms, a chariot and horses, a seated group with a Mediterranean amphora, as well as possible representations of fortified settlements (qsur) and camels (Di Lernia and Zampetti 2008, 92-94). It is certainly true that most images of camels belong to a later rock art style than the equid phase, but there is some overlap and the earliest camels are depicted in the company of horses7.

7One of the problems of the delineated sequence of Cattle, Horse and Camel phases in the rock art is that it also encourages us to think about successive stages in pack animal use, with the appearance of new species leading to technological revolutions in possibilities of travel and the transport of goods. In reality, we might think of much greater overlap in the availability of pack animals from the 1st millennium BC onwards and a more gradual shift in the balance of usage (see also Fothergill et al. 2020).

Oxen

8There is limited evidence for the use of oxen as draught or plough animals in the Sahara in antiquity. Although they have a reputation for being slow and dependent on regular water and grazing/fodder, they do not differ much from donkeys in these requirements (Bovill 1968, 16-17). However, in arid environments they do not maintain condition when used to carry loads, which makes them unsuitable for transport over longer distances.

Horses

9The arrival of the horse in North Africa probably occurred at some point in the early 2nd millennium BC and was closely linked to the diffusion of the chariot from Egypt (Camps and Chaker 1993)8. The early evidence for the spread of the horse and chariot relates to the rock art representations that are widely disseminated across the Sahara (Camps 1984; 1982). An equid skeletal find from antenna tomb 00/195bis in the Wadi Tannzuft also bears mention. Here, an equine distal femur fragment was found on a large flat stone covering the human rib cage (Alhaique 2002, 183). Alhaique considers the bone too large to be a domestic donkey and raises the possibility that this is either a wild ass or a horse (based on comparison of size with archaeological specimens from Italy)9. The significance of the horse in Saharan society (Daumas 1968), has been somewhat overshadowed by the prominence of the camel in the early modern period (exaggerated further in popular imagination), but it is evident from a range of material to be examined below that the horse had a particular association with early oasis communities (Duckworth et al. 2020; Sterry and Mattingly 2020). Indeed parts of Mauretania, were known as horse breeding centres in more recent history (Webb 1993; Campbell et al. 2006)10. Horses can carry about a third of their weight, so a typical small North African horse might have had a carrying capacity of around 90 kg (Vigneron 1968). Horses have generally been treated as prestige animals in the Maghreb and Sahara, so may have been favoured for riding and as a commodity of exchange, rather than as pack animals per se. In the 15th to 19th centuries, horses were regularly traded for 10-20 slaves and a thoroughbred might be worth as much as 100 slaves, 100 cattle and 20 camels (Lydon 2009, 132-133; Webb 1993).

Donkeys and Mules

10The wild ass is native to the Maghreb and Sahara and finds of Equus africanus are recorded at a number of Holocene sites across the Maghreb (Marshall 2007, 376-378). For example, in Fazzān, a mandible and teeth were found in corbeille 07/39 C1 of the Middle Pastoral period in the northern Messak Settafet (Di Lernia et al. 2013). This cairn has been dated to 4044-3960 calBC and corresponds with a period in which the Sahara was dominated by cattle-based pastoralism and all other identified elements from the site are of domesticated cattle, sheep or goats.

11The domestic donkey is a hardy equid that is also well-adapted to Saharan conditions (Camps et al. 1988). Bates rated Libyan donkeys second only to the camel for their hardiness and adaptation to desert conditions (Bates 1970, 29). Domesticated donkeys have been identified from contexts dating to c. 4000 BC in Egypt. The first securely dated Equus asinus outside of the Nile comes from Wadi Hariq in western Sudan 2197-1435 calBC and it seems likely that by this point domestic donkeys could have spread across much of the Sahara and Maghreb (Marshall 2007, 385). Indeed, Marshall has drawn attention to the usefulness of donkeys in pastoral societies for bringing water to their herds (ibid., 374-375). It is not clear whether the African donkeys were independently domesticated from the abundant wild population that is attested to by portrayals in Saharan rock art, or brought from Egypt in a domesticated state. However, other possibilities exist as recent DNA research strongly links domestic donkeys with the Nubian and Somali clades (Gifford-Gonzalez and Hannotte 2011, 13-14). Donkeys have a carrying capacity of c. 65-75 kg. A working pack donkey in hot conditions requires c. 10 litres of water per day, but in extreme conditions, they can survive for a day or two with minimal water (Bovill 1968; Sidebotham 2011, 25). Mules are the offspring of a male donkey and a mare; hinnies are the offspring of a female donkey and a stallion. Both possess a useful combination of the characteristics of the two breeds. This makes them more resilient than horses and capable of managing larger loads than normal donkeys, though their infertility presents a substantial drawback. Mules and hinnies have a carrying capacity of c. 70-90 kg.

Camels

12Camels are the pack animal most suited to desert extremes and have a higher tolerance of heat and water shortage than donkeys, mules and horses11. The key characteristics of the camel that have made it such a desert specialist relate to its ability to conserve water and to control its body temperature across a wider range of extremes than most mammals. It can sustain fluid loss equivalent to 30% of its body weight and is also able to replenish fluids (when available) more rapidly and more safely than other animals. Endurance tales are often exaggerated, but the camel has the capacity to go 6-7 days without water in cool weather and 3-5 days in high temperatures. The soft-padded, splayed foot is especially good on sandy terrain and the heavily-lashed eyelids have adapted to resist sand and dust. Camels are comparatively slow travellers: 4-5 km per hour and average day marches of 25-30 km are a common speed estimate in accounts of the great age of Saharan camel caravans in the 19th century (Lydon 2009, 209-210). Pliny mentions a caravan from Southern Arabia to Gaza that crossed over 2000 km in 65 days (averaging 34 km per day) (Pliny, N.H. 12.32.64-65). Camels are relatively amenable to being herded and can be roped together in long files. The largest caravans of 1000 camels or more would extend across several km of terrain. Further advantages of the camel are that it could also provide valuable resources for sustenance (especially milk as a result of an extended lactation period and meat, though generally only in extremis) and shelter (hair and skins). In traditional Saharan caravans, camels were more generally used for transporting large loads of up to 150-200 kg, rather than for riding (Lydon 2009, 210).

13Disadvantages of camels include their speed (slow) and motion (sea-like rolling), their proneness to lameness on rocky terrain and vulnerability to fly-borne diseases in more humid zones; this last factor makes camels rare in the northern Maghreb (Lydon 2009; Johnson 1969, 8; Mitchell 2005, 145-146; Shaw 1979). Camels are mainly kept in zones with less than 250 mm of yearly rainfall. Large differences in vegetation in the Maghreb/north Sahara and south Sahara and sub-Sahara have led to the evolution of regionally-specialised camels12. There is a high attrition rate if camels are taken directly across the Sahara and historically the norm has been to change camel teams in the central Sahara using beasts adapted to the environmentally-distinct zones to the north and to the south.

Faunal Evidence

14As part of this research, we have collected data from all published faunal assemblages in North Africa and the Sahara. In total, 1191 equid elements and 247 camel elements have been recorded across 43 sites and 125 assemblages (fig. 1-2; table 1). Additionally, a further 14 associated bone groups relating to equids originate from the Circus Cemetery at Carthage.

Fig. 1. Map of sites with evidence of equid bones along with rock art areas (excluding the Nile Valley).

Fig. 2. Map of sites with evidence of camel bones along with rock art areas (excluding the Nile Valley).

Table 1. Sites with evidence of equid and camel bones (excluding the Nile Valley).

Saharan sites

15Faunal evidence for draught animals in the Sahara is somewhat rare for several reasons. Equids and camels were not primarily raised for meat (although there is some evidence for camel butchery at all Saharan sites at which they appear), hence their bones are far less common than other domesticates such as cattle, pigs and sheep/goat. Secondly, the assemblages we do have are generally fairly small (on average around 400 bones per phase) and those that we have are generally poorly preserved and heavily fragmented. Moreover, equid bones often have a secondary value as a material for craft work and tool production. Nonetheless, the assemblages we have do give some indications as to how widespread these different animals were (see Fothergill et al. 2020).

16In the 1st millennium BC, there is evidence of equids from the Early Garamantian hillfort of Zinkekra in the form of an equid mandible, though it is of fairly small size and can therefore be classified as likely or probable donkey (perhaps not surprising given that they are located 50 m or more above the main oasis floor on the rocky slopes of the Messak) (Van Der Veen and Westley 2010, 521; Fothergill 2012). There are numerous rock art images of men on horseback in the Garamantian heartlands of the western Wadi al-Ajal, including from Zinkekra itself (Mattingly, Fazzān 1, 2003, 287-317). No equid or camel bones are known from the later site of Fewet, although the assemblage contains only just over two hundred identified elements (Alhaique 2013).

17In the 1st millennium AD, we have the first evidence of camels within a Saharan settlement at both Aghram Nadharif and Jarma, the Garamantian capital (Alhaique 2006; Holmes and Grant 2013; Holmes 2013); these can be no later than the 2nd century AD, although the possibility remains that they would have been present earlier. Further evidence from this period comes from Saniat Jibril (1st-4th century AD) which has evidence of both camels and equids (Van Der Veen and Westley 2010, 521). A potentially important, but poorly published site is the necropolis of Oued el Hamara on the northern fringe of the Algerian Sahara. Roffo claims that in this cemetery there were many animal bones (of which camel predominates, including a secondary inhumation of a phalange, metatarsal and mandible), camel skin fragments and straps and cords made from camel leather (Roffo 1937, 207). There is little dating evidence although the burial rite is clearly pre-Islamic.

18Somewhat less clear, is the interpretation of a piece of textile and two cords made from camel hair, found at the surface of T01 in the Eastern Tanzzuft (Maspero et al. 2002, 160). The tomb itself dates to the Late Pastoral/Early Garamantian period, but there is no direct dating of the objects, nor is it clear how long a disjuncture there was between the construction of the tomb and the placing of textile.

19It is not possible to say with certainty if horses were kept in these settlements, but donkeys were certainly present in all settlements apart from Fewet and camels in all sites dating to the 1st millennium AD. With such small numbers, it is risky to try to reconstruct too much about population, but camel elements appear in either similar (Jarma from 2nd century, Saniat Jibril) or greater numbers (Jarma from 3rd century onwards, Aghram Nadharif) to equids which would appear to suggest a rapid adoption of camels as draught animal. There are few unfused equid bones and no unfused camel bones reported, which suggests that these animals were generally reared away from the settlements. The lack of definitively-identified horse in all assemblages may indicate that those prestige animals were highly valued and were not normally entering the food chain or else were primarily owned and used by other communities (i.e. pastoral groups), though this seems unlikely. The widespread rock art evidence of images of horses, while not closely dated, is suggestive of a strong horse presence by the 1st millennium BC, if not even earlier. As we shall see below, the Garamantes also represented horses and camels as common elements in their repertoire of figurine art at their oasis settlements. Because of their greater resilience, camels were the more likely species to have been raised and herded at a distance from the oases.

Maghrebian sites

20In the Maghreb, equids appear in some of the earliest settlement assemblages at Althiburos, Carthage and Mogador13. In most cases, it is not possible to differentiate between donkey and horse, but both are certainly present by early/mid-1st millennium BC. As with Saharan sites, it is likely that the size and context of assemblages reduces the likelihood of these taxa appearing and there is good reason to think that both species could have been present at an earlier date. As we have already noted, the donkey was likely to have been widespread amongst pastoralists at the point at which the first sedentary settlements were constructed. For the horse, the data are too few to date its introduction, but its early presence on the Tunisian Steppe perhaps favours dispersion from the Nile rather than the Levant (via the Phoenicians).

21From the 2nd century BC, all three beasts of burden, including the earliest definitively-identified camel, are represented in the Maghrebian faunal material and most assemblages of reasonable size contain both donkey and horse, the ratio of which generally varies from 3:1 to 2:1. Most assemblages were excavated from urban contexts on or close to the Mediterranean, yet around 25% contain camel, generally no more than a handful of bones. In northern Tunisia, the climate is ill-suited for dromedaries, so their presence is most likely indicative of their use in transport. The post-military settlement at Gheriat al-Garbia has one of the largest assemblages of dromedary bones with the 104 bones accounting for 10.3% of the overall NISP (Pöllath and Rieger 2011, 178).

Sub-Saharan evidence

22Camel and equid bones are rare in south of the Sahara for the entirety of the 1st millennium AD. Cubalel and Siouré, Senegal, have the earliest camel and equid bones with dates of AD 1-250 and AD 250-500 respectively, though the earliest horses come from Akumbu (AD 600-1000) (Macdonald K.C. and R.H. 2000, 139-142). Special mention should be made of the site of Aissa Dugje in which at least 8 horses (including a juvenile) have been identified in deposits dating as early as calAD 630-877 (MacEachern et al. 2001)14. The small size of these horses perhaps suggests that they had been bred to have some resistance to Trypansomiasis; hence, the date of the earliest horses in the Sahel should be considered at least several generations earlier. These few sites provide strong evidence for the presence and use of these animals in Trans-Saharan trade from the early 1st millennium AD.

Population profile

23When the collated data are analysed, there are peaks in the number of both equids and camels from AD 100 to AD 400 and a trend of increasing camel numbers from AD 800 onwards (fig. 3). It is probably not coincidental that the first peak in numbers coincides with the evidence for increasing Saharan trade in Garamantian Fazzān and it is also tempting to link the second peak to the resurgence and extension of Saharan trade in the early Islamic period. Camels and donkeys can easily perform different roles and require different husbandry strategies with the former lending itself more easily to large-scale herding whilst the latter were often left to roam in a semi-feral state.

Fig. 3. a) Graph showing the absolute numbers of equid (circle) and camel (square) elements over time from archaeological assemblages in North Africa; b) Graph showing equid and camel bones as a percentage of main domesticates from archaeological assemblages in North Africa.

24Prior to the early modern period (Phase 1, 1-2 and 2), camel elements make up only a very small proportion of the overall assemblage from Jarma (although this increases over time; by Phase 3 they make up 11% of major domesticates). There are no bones of juvenile age in these earlier phases (Holmes 2013, 856, 863). The predominance of adult bones and the presence of later-fusing element portions such as the distal and proximal femur and the proximal tibia may indicate that camels were not bred in the immediate vicinity and that individuals were sourced as adults from elsewhere. However, immature bones are particularly vulnerable to taphonomic degradation and may therefore be under-represented in the sample. The adult camel population could suggest that these were pack animals rather than producers of milk or meat (Horwitz and Rosen 2005, 129). Unfortunately, distinguishing between the sexes from skeletal material is difficult in camels and the small assemblage size combined with high levels of fragmentation prevent definitive conclusions with regard to husbandry strategy.

25Transport was not the only role that camels fulfilled at Jarma since butchery marks were present on a range of elements in multiple phases (Holmes and Grant 2013). It is likely that some were eaten soon after they became available in the Sahara. This is perhaps not surprising, as the consumption of camel flesh is an ancient and widespread, if not a universally agreeable practice. Whilst Aristotle found both camel flesh and milk delicious and palatable, Galen was of the opinion that only “mentally and physically camel-like” people would eat camel meat (Dalby 2003, 71; Aristotle Historia Animalium 578a, 10-16; Galen, De Alimentorum Facultatibus 6.664). Butchery marks have also been noted on camel bones from other contemporary desert sites such as Dûmat al-Jandal, a Nabataean urban centre in the Al-Jawf oasis of Saudi Arabia (Monchot 2014).

26Equid bones from Jarma date to AD 100-200 (Phase 6/7) and some of these bore signs of butchery (Holmes and Grant 2013). Although butchered equid bones have been noted at the Avenue du President Habib Bourguiba, the Byzantine cemetery at Carthage and at Sidi Khrebish, these have generally been considered the product of bone-working or tool-making activities rather than slaughter of horses for meat (Schwartz 1984, 237-240; Rielly 1988, 303-304; Barker 1979, 22). The marks are thought to be indicative of carcass disarticulation and subsequent sawing off of the ends of long bones for use in craft work. The butchery marks present on the equid bones from Jarma do not fit with this model as they are present on elements which are not routinely used for bone-working, such as mandibles and attempted removal of long bone ends is not evident (Holmes 2013).

27Equiphagy (eating equids) has a long history. Though Galen did not find the domestic donkey at all worth eating, he did think that wild asses were delectable (Galen, De Alimentorum Facultatibus 6.664) and Liutprand of Cremona claimed that Byzantine emperors maintained entire herds of wild asses for culinary purposes (Liutprand of Cremona, Embassy 38; Norwich 1993). Horse meat was purportedly not eaten by the Greeks or Romans except at times of extreme hardship (Dalby 2003), but source material on Saharan practices is lacking. Horse meat is currently considered to be “makruh”, or avoided if at all possible by the tenets of Islam (Von Bargen et al. 2013), though the extent to which this may have affected past populations in Fazzān is not known. The equids at Jarma are not definitively assigned to horse, mule or donkey due to the high amount of fragmentation in the assemblage (Holmes and Grant 2013); however, metrics of the six measurable elements present are more consistent with donkey than mule or horse in terms of size.

Pathologies

28Palaeopathology presents another way of accessing past relationships between humans and other animals. Draught, riding and packing work all take their toll on the skeletal system and these activities are often linked to the manifestation of certain lesion types. These pathologies are mentioned in reports from Jarma and Carthage as affecting both camel and equid bones (Holmes 2013, 864; Schwartz 1984, 240). Kevin Rielly has identified five instances of pathology in equids at the Byzantine cemetery in Carthage which he connects to “load and/or traction stress” (Rielly 1988, 305). Although formal criteria for identification of osteoarthritis are published differential diagnosis cannot be undertaken without comprehensive pathological description (Baker and Brothwell 1980, 115); unfortunately, images of these lesions are not often provided and descriptions are limited. Holmes describes a “polished proximal surface” in the humerus of an equid from Jarma (Holmes 2013, 864). This seems likely to be a description of eburnation, a form of bone degeneration at the site of cartilage erosion. Although this lesion is potentially diagnostic of osteoarthritis, other components of the disease (e.g. pitting at the articular surface, additional bone formation near the joint) are not described and thus any diagnosis is speculative. Problematically, normal ageing processes can affect the skeleton in similar ways15, but these lesions do nonetheless indicate that these animals led productive and probably long working lives.

Literary and Iconographic Evidence for Transport Animals in 1st Millennium BC/AD

Literary evidence

29Guédon provides the most recent detailed discussion of the literary evidence for animals used for transport in Roman Africa (Guédon 2010, 80-88). Carts and wagons were probably primarily a feature of the northern Maghreb, but representations of two-wheeled carts (drawn by a camel and a pair of oxen) are known at Ghirza in the Libyan pre-desert (Brogan and Smith 1984, 154). In desert regions, pack animals were the most common transport option. For convoys of pack animals, 30 km per day is indicated as the norm in Roman legal sources (Gaius, Digest 2.11.1; Ulpian, Digest 38.15.2.3 and 50.16.3). This relates to the walking pace of pedestrians accompanying the animals and this must often have been the limiting factor on caravan and pack transport in all periods: the speed and endurance of humans determining the journey length and time more than the capacity of the animals.

30Comparisons with eastern and western desert routes of Roman Egypt show a range of animals were used together for carrying goods, especially camels and donkeys16. One of the peculiarities of the ancient sources is the relative lack of firm attestations of the different beasts of burden; we read many accounts of journeys in North Africa, but all too frequently without the mode of transport being specified, with a 400-year gap in literary references to camels for instance between 46 BC and the late 4th century AD (Guédon 2010, 294, n. 297). However, there are indications that a mixture of animals was employed, rather than one species more exclusively. Comparisons with eastern desert routes of Roman Egypt show a range of animals were used together for carrying goods, especially camels and donkeys (Adams 2001; Cuvigny 2006, 399-426; Sidebotham 2011, 170-173). The different carrying capacity of various animals is indicated by a transit tariff from Maxula in northern Tunisia, with rates being highest for camels and drivers, then horses, then mules (Guédon 2010, 82; CIL 8.24512).

31There is ample attestation of the importance of horse-rearing in the Maghreb and Sahara. Most celebrated is Strabo’s comment that 100,000 colts were bred each year (Strabo 17.3.19). The North African peoples were famed for the equestrian skills in warfare and this is reflected in imagery of mounted warriors from both Maghreb and Sahara (Laporte 1992; Barnett 2019; Mattingly, Fazzān 1, 2003). The earliest literary reference to camels relates to the capture of 22 dromedaries by Caesar from the army of Juba I (Caesar, Bellum Af 68.4). Bovill argued that the first introduction of camels can be traced in the account of a Roman military convoy that attacked the Garamantes in AD 69, in that this expedition pioneered a new and faster route to the Garamantian heartlands (Bovill 1956). Though this is not generally accepted today as relating to the first introduction of the camel to North Africa, it is certainly possible that larger numbers of camels being available may have facilitated the use of this route, thought to cross over the water-scarce terrain of the Hamada al-Hamra. Certainly, by the 3rd century AD camels were integral to the supply of the garrison forts in this region and by implication to Saharan trade. The Bu Njem ostraca mention a statio camellariorum, as well as a regular work duty ad camellos (Marichal 1992, 100-106, 122, no. 5; for the ad camellos p. 120-121, 127, 157, no 3, 4, 9, 42). Four Libyan men are described in ostraca as being a camellarius/camel driver (ibid., 183-188, 196, nos 76, 77, 78, 79, 80, 88).

32However, it is apparent that the camel had not become the ubiquitous transport animal at this date. In another two ostraca we appear to have references to Garamantes approaching the fort with consignments of barley on three mules and four donkeys (ibid., 177, no. 71.2, 179, no. 72.1). There are also lots of references to horses, though the garrison was part mounted.

33Another possible reference linking the Garamantes with camels is the comment by Vibius Sequester concerning the “Cinybs” river region, where gems were mined and carried on camel back to an oppidum Thybrestum (Shaw 1979, 699; Vibius Sequester, p. 147). In late antiquity, there are indications that the numbers of camels present in the lands bordering the northern Sahara numbered in the 1000s. Ammianus Marcellinus recounted the demand of Count Romanus for 4000 camels to support his military campaign in the AD 360s, while Synesius wrote of 5000 camels rustled to carry off booty seized by the Ausuriani during raids on Cyrenaica at the start of the 5th century (Ammianus Marcellinus 28.6.4-6; Mattingly 1995, 177; Synesius, Catastasis I, col. 1669). It was probably for their utility as pack animals that the desert peoples made increasing use of camels. The Tablettes Albertini mention a via de camellos by the Tuletianos estate in the pre-desert lands of south-western Tunisia (XXI, 6; Courtois et al., 1951). Vegetius mentioned the Urceliani and Mazices as desert peoples who used camels, noting their suitability for harsh and waterless terrain (Vegetius, de Re Mil. 3.23). Corippus and Procopius both mention camels primarily as pack animals, accompanying Libyan forces comprising equine cavalry and infantry in the 6th century AD (Mattingly 1995, 178-180).

Iconographic evidence

34The earliest literary reference of 46 BC to a camel in Africa is matched by a representation on a Cyrenaican coin of the same date (Robinson 1927, pl. XII, 1-5). From Tripolitania, there is a well-known series of representations of camels dating to the 2nd-4th centuries AD, including repeated images of ploughing from the pre-desert17. There are also several reliefs of camel caravans and pack camels (Brogan and Smith 1984, 220-221, pl. 67b, 110b; Brogan 1954, 129, pl. XVIIa). There are also a number of images of particular relevance to this paper in which a camel is depicted with two or four amphorae on its back (Brogan and Smith 1984, 154) (fig. 4)18. It could be argued that these images, while ostensibly showing just one flank of the camel, in fact represent the normative total load of two amphorae, which would in reality have been carried one amphora on each side of the animal. However, the alternative interpretation, that the visible load was balanced by two more amphorae on the hidden side, can by no means be discounted for camels. The amphorae depicted in some of the images are quite small in size and a charge of four small amphorae would be within the capability of a camel19.

35Two of the images of camels with amphorae (fig. 4a/b) come from a series of terracotta figurines from Hadrumetum (Sousse)20. Both of these seem to indicate camels carrying pairs of amphorae, as does a 4th-5th centuries picture lamp of camel, also from Tunisia (Camps et al. 1996) (fig. 4c). The Tunisian camel figurines come from funerary contexts, but their significance is disputed. The two relief carvings of camels with amphorae (fig. 4d/e) come from mausolea in Tripolitania, one near Mizda and the other at the well-known site of Ghirza (where alongside a camel carrying two elongated cylindrical amphorae on its left flank, a third largish amphorae is being carried by a man)21.

Fig. 4. Camels transporting amphorae: a) figurine from Sousse Museum; b) figurine from Sousse museum; c) Late Roman lamp from Tunisia; d) relief carving of a camel from a mausoleum near Mizda; e) relief carving of a camel bearing two amphorae from Ghirza Tomb North B.

36A further relief carving of a camel carrying amphorae from near L’Aquila in Italy has been associated by Tchernia with a trading family, the Peticii, who had interests on desert transport in Egypt and perhaps elsewhere, as the name also occurs on an amphora at Carthage (Tchernia 1992)22. Camels also feature on mosaics from North Africa, for instance in one of Silenus from El Jem (Lassère 2015, pl. 2).

37A central Saharan element can now be added to this iconographic debate from excavations at the Garamantian capital at Jarma and Italian excavations at Fewet, near Ghat. Low-fired or baked-mud figurines of anthropomorphic and zoomorphic form have been found at both sites dating to the early centuries AD. The larger assemblage at Jarma includes at least two horses and five camels (Townsend 2013; Townsend 2013; Mattingly, Fazzān 4, 2013a, 517-518) (fig. 5). Zampetti interpreted a camel head from Fewet as a “toy”, but the context of the Jarma material is suggestive of a different interpretation (Zampetti 2013, 153-154). The Jarma material was mostly found buried in a series of pits directly behind a building identified as a Garamantian temple and a religious use is thus likely. Other zoomorphic figurines tentatively identified as horses are known from Aissa Dugje, near Lake Chad (MacEachern et al. 2001, 65). It is interesting that two of the key animals used in Saharan transport and trade should be represented in figurines with some ritual connotations.

Fig. 5. Garamantian figurines: (above) camel; (below) horse (note the pricked indication of bridle).

The significance of amphora size

38Amphorae seem on occasions to have been carried far overland in North Africa (Bonifay 2013). A further approach to the issue of determining the species of pack animals used in Roman-era Saharan transport relates to the size of amphora found in Garamantian burials in Fazzān. For reasons of weight distribution, it is likely that amphorae were normally carried in pairs, one on each side or two on each side as we saw in the imagery of camels carrying amphorae. Indeed it is improbable that on such long journeys animals would have carried either unbalanced loads or seriously excessive loads, so our default assumption should be that amphorae were normally carried as closely matched pairs in terms of size/weight, with a total weight within or close to the notional load capacity of each species. Consideration of the combined weight of a pair of full amphorae should thus in theory give an indication of the sort of animal employed in their carriage. The more than 100 intact Punic and Roman amphorae from Fazzān are thus an important additional proxy marker for the species of transport animals employed. Although exact volumes (and thus weights) are known for only a small sample of the excavated amphorae, approximate values can be deduced and these compared to the notional load capacities for donkey of c. 65 kg, for a horse or mule of c. 90 kg and for a camel of 150 kg (table 2).

Table 2. Amphora sizes and carrying capacity of animals used in Saharan transport.

39What is evident is that there was, in broad terms, an increase in amphora size over time. Some of the earliest amphorae found are extremely small, such as the Van der Werff 1, Mau 35 and so-called amphoroid flagons, generally under 20 litres capacity and weighing c. 30 kg when laden23. Early Tripolitanian amphorae were also of small dimensions, under 1 m tall, with one defined group having a capacity of c. 25 litres and a weight of c. 40 kg when full; these were probably beyond the load capacity of donkeys for a pair of amphorae. There is a second group of these Tripolitanian amphorae that were of c. 40 litres capacity and of 55 kg weight when full. Carried in pairs, the combined weight of c. 110 kg indicates the probable use of camels.

40Many of the medium size group in the late 1st and early 2nd centuries AD are examples of the Tripolitana I or III oil amphora, or Tripolitana II wine amphora though they do not exactly conform with the classic examples of these amphorae known from Mediterranean contexts, leading to some confusion in their categorisation in the Fazzān type series (Mattingly, Fazzān 2, 2007, 348-350; Fazzān 4, 2013a, 337)24. In the Mediterranean region, Tripolitana I was a slightly smaller and earlier (broadly 1st-2nd centuries AD) oil amphora than its close relation Tripolitana III, which seems to have largely taken over by the 3rd century AD. The Tripolitana II was 1st-4th centuries, but more common in the 1st-3rd. Only intact versions of Tripolitana I survive in Fazzān burials (though there are certainly Tripolitana II or III rims). It appears to have been made in two variants (average 28 litres and 40 litres), both of which are smaller than the normal vessels of this type exported to Mediterranean markets (average c. 56 litres at Pompeii). That suggests deliberate manufacture of amphorae intended for the Saharan trade routes that were scaled down in size to suit the intended beast of burden. The smaller size of Tripolitana I was just in the range of horses and mules and could comfortably have been managed by camels; the larger size was more suited for camels alone (Jerray 2015, 190-191; Panella 1977; Peacock and Williams 1986, 52). Standard-sized Tripolitana I and later Tripolitana II or III amphorae would have been towards the upper limit of a camel’s range. If the examples from Fazzān that are similar to the Tripolitana I-III series were in fact specially made to different size specifications for Saharan trade (this can be no more than a plausible hypothesis at present), it also raises the question of what the different sizes indicated. It does not follow, for instance, that what look like Tripolitana I rims indicate imports of olive oil in every case. One possibility might be that amphora contents on the Saharan routes were more denoted by size (25 litres amphorae perhaps for wine, 40 litres amphorae perhaps for oil?).

41However, there are also a few amphorae in Fazzān of even larger dimensions whose combined weight (for a pair) must have exceeded 150 kg. Examples include late antique Tunisian cylindricals (Keay 61 and 62), which individually weigh c. 80 kg when full and must have been pushing the load limit of what even a camel could manage (Mattingly et al. 2010, 121-122).

42The increasing size over time of the thousands of Mediterranean amphorae found in Fazzān may well indicate that the camel played a progressively more important role in Trans-Saharan trade over time, perhaps gradually taking over from donkeys, mules and horses. However, if we examine individual tomb assemblages another aspect is revealed. Many individual tombs, especially in the early Roman period, did not contain uniform sets of similarly sized amphorae, but rather a range of different sized containers. While there is a general increase in the average size of amphorae over time, some late 1st century tombs contain a mix of amphorae of notably different sizes, implying that caravans made use of a range of different pack animals, each carrying vessels appropriate to the animal’s stature. The mix of amphorae sizes in burials such as the late 1st century AD tomb 17 at Saniat Bin Huwaydi (fig. 6) is surely an indication that various pack animals were used together on the Saharan trails at that time, each carrying loads appropriate to their strength (Mattingly, Fazzān 3, 2010, 332-336).

Fig. 6. Amphorae from Tomb 17 at the Garamantian cemetery of Saniat Bin Huwaydi (late 1st century AD) showing the wide range of forms – these can be fitted into two separate weight/size categories of around 30 kg and 40 kg.

43Thus, although the evidence clearly suggests the involvement in this carriage of animals of smaller capacity in the early period, it seems probable from the range of sizes that camels were present alongside the other beasts of burden from the 1st century AD onwards. While table 2 demonstrates that donkeys could have been used only for the smallest of the amphorae imported to Fazzān, it does not follow that this size of amphora was exclusively carried by donkeys. As the various relief representations of camels carrying amphorae show (fig. 4), camels may well have frequently been loaded with up to four amphorae, two on each size. However, the small size amphorae alongside others of about double the weight is surely an indication that caravans in the Garamantian era were expected to comprise a mixture of different pack animals.

44There is also an interesting correlation between the period of peak amphora imports to the central Sahara and the faunal evidence for the highest frequency of both equid and camel remains from a number of sites25. This would seem to confirm that the heyday of the Garamantes in the early centuries AD marked a significant moment in pre-Islamic Saharan trade and that they were utilising a wide range of pack animals, not solely the camels which were certainly available to them by this date.

Conclusions

45In this paper, we have reviewed several types of evidence pertaining to the introduction and use of a range of pack animals in the Sahara and Maghreb. There remain significant lacunae in the data, which future faunal studies will hopefully fill. Our principal conclusions are that there was some change in the nature of Saharan transportation over time. The earliest Saharan transport and trade in the 1st millennium BC was probably more dependent on the donkey and, to a lesser extent, other equids. By the late 1st millennium BC, the donkey, mule, horse and camel were all available to and employed by communities in Maghreb and Sahara26. In the Islamic era, camels gradually became more predominant as the transport animal par excellence of the Sahara. However, in place of prevailing models which tend to emphasise a straightforward evolutionary sequence from donkeys, to horses and mules, to camels, we suspect that there was a considerable lag before the camel became the dominant long-range pack animal and that for a long period of time a mixture of all these animals were used in combination together27. The selection of pack animals in different historical periods will have related to a range of factors.

The technological shelf idea

46Modern western science places an emphasis on invention, revolution and replacement in technologies. However, traditional societies have more commonly embraced appropriate technology in making technological choices across a range of available solutions – what can be termed the technological shelf (Greene 1994; Mattingly 1996). The slower-than-expected uptake of a supposedly superior technological invention can often be explained by the fact that it was initially in competition with a range of socially embedded and effective alternatives that functioned adequately in local conditions. This may well have been the case concerning the preferred pack animals of Maghreb and Sahara in the 1000 years preceding the Arab conquests. The combinations of pack and draught animals selected in different sub-regions and on different routes probably varied greatly. Certainly, even on the Saharan trails, the equids were not instantly swept aside by the arrival of the camel. Its eventual dominance on these routes followed a long period when it seems to have been utilised alongside large numbers of equids.

“Horse (and Camel) Nations”

47Mitchell has recently drawn attention to the transformative role of the introduction of horses (and we might reasonably extend this to camels too) not only for transport and combat, but also for social relations (Mitchell 2015). When horses were introduced to the Americas, southern Africa and Australia they regularly attained important symbolic value for the groups keeping them, appearing in rock art, as figurines and in funerary practices. Similarities can be seen in the Sahara for all of these practices and whilst we cannot draw direct analogy with such disparate locations it directs our attention to what changes these animals may have made to everyday lives, power relations within societies and the spiritual world. We may consider the possibility of new societies emerging whose lifeways were based on these animals. A second point relates to the differential uptake between societies and the inevitable imbalances that this would have caused within a few generations. Trans-Saharan trade may have changed substantially as a result of groups establishing new routes and centres or making use of horses and camels to expand, assume control of or raid existing trading structures.

Speed is not the key factor in trade

48While horses and mules have greater speed than camels, caravans always move at the speed of their slowest members; in many periods, slaves have been a key item of trade. In general, human cargo has been forced to march on foot, making their speed and endurance the determining factor in the overall rate of progress of the caravan. In the early modern period, slavers were obliged to force the pace on the longest and most waterless stretches, with horrendous consequences for the slaves, whose bones eventually littered the trail (for example, Denham and Clapperton 1826, 17-18). Regardless of the pack animals used, many Saharan caravans have travelled at human walking pace. On the other hand, raiding parties often combined the use of camels and horses, with horses permitting sudden attack and withdrawal and the camels carrying the larger part of accumulated booty, water and supplies.

The relative abundance of water sources and distance between wells along a given route

49If water is plentiful and the gaps between wells short, then horses and donkeys will compete ably with camels because of their docility and the fact that they can be ridden more comfortably. There is strong evidence to show that since the last major climatic change c. 5000 years ago non-replenished groundwater levels in the Sahara have been gradually but progressively dropping. It is thus probable that wells were more regular and more productive 2000 years ago than 1000 years ago28. The camel’s advantages for Saharan traffic have become more apparent over time and were particularly crucial on stages with limited water availability, where their greater endurance and load capacity enabled them to carry not only marketable cargo, but also vital water, food and fodder for the caravan. The feed requirements of different animals are not so widely different. For example, donkeys require c. 2-3 kg of dates per day, camels 3 kg, plus some straw if no grazing available (Despois 1946, 175). The crucial difference is the requirement for equids to be watered regularly, especially in hot temperatures. In harsh conditions, a rule of thumb is that each horse will require a camel just to carry the necessities to keep it alive on a long waterless stage of travel. In the long term, with the depletion of groundwater resources along the Saharan trails, the balance has tipped increasingly towards the camel.

The availability of potential pack animals

50This is in part linked to their other uses in society and who controlled the supply of animals. Donkeys have been the archetypal beast of burden in oasis communities, with large-scale camel rearing more often the preserve of pastoral communities. The iconographic evidence from the Tripolitanian pre-desert indicates that the communities there used camels for ploughing and as pack animals and that their range may have extended beyond the Roman frontier zone (where they were certainly involved in supply of the Roman forts like Bu Njem) to Saharan trade with the Garamantes. The Garamantes can likewise be connected with a range of equids and camels and are also named as active traders on Saharan routes. How the Garamantes accessed camels is uncertain; it is possible that pastoral groups assimilated to the kingdom were responsible for providing both camels and their drivers and desert guides to supplement horses, mules and donkeys provided by the oasis dwellers. The lack of evidence of young camels at Jarma would suggest that camels were bred and raised at a distance normally. Horses have also commonly been prestige animals, ridden in warfare and raiding in the Sahara and Sub-Saharan zones. However, merchants may have preferred the comfort of a ride on horseback when accompanying caravans. They have for these reasons also been a valuable object of trade and may often have been used for carriage of goods and supplies of a caravan en route to the intended market. A consequence of the increasing importance of the camel in Saharan trade in Islamic times may have been a tilt in the balance of power away from sedentary oasis dwellers and more towards the camel-herding and pastoral desert communities – a state of affairs well represented in the later history of the Tuareg and their power over some oasis communities29.