1The numerous and sometimes exceptionally well preserved fortlets in the Eastern Desert of Egypt make it tempting to revisit a subject already studied in several works by other scholars (Maxfield 1996; Cuvigny et al. 2003; Sidebotham, et al. 2008). The architectural study of these particular Roman military buildings, poorly understood throughout the Empire, merits further attention, and recent excavations carried out by different teams, French, American and British, have shed some light on a number of new elements. In addition, the bibliography is somewhat scattered, and is constantly evolving. It is, thus, the right moment to present a new synthesis on this subject with a unified bibliography.1

2The distribution map of these fortlets (fig. 1), called praesidia in the ostraka (or the same word transliterated from Latin to Greek), has changed very little in recent years, other than around Berenike where the American-Dutch mission conducted numerous surveys that revealed hitherto unknown sites (Sidebotham et al. 2008, fig. 15.1). The roads that cross this desert, however, have functions quite different from one another and this fact alone explains the existence of distinct architectural types. It is important to remember that these tracks were generally not built as roads: they were trails that led from one stopping point to another and whose routes can vary significantly in detail, especially in flat and sandy areas. Other than the road stops (sometimes rock shelters, studded with graffiti and rock inscriptions), where travellers camped, the only infrastructure included the stations built and controlled by the army, where one could rest and refuel, some probably also had wells that were not protected by the praesidia –but for that reason remain little known– or towers that mark the road from Coptos to Myos Hormos, although their function and chronology are unclear (Brun, Cuvigny, Reddé 2003, p. 207-234). The Via Nova Hadriana between Antinoopolis and the coast, well north on Figure 1, seems to be an exception. For much of its ca. 800 km length the surface was cleared and dotted with stone cairns, while the road on its west-east course from the Nile to the coast lacked praesidia, that portion parallel to the coast had them (cf. Sidebotham and Zitterkopf 1997, 1998, Sidebotham et al., 2000; Sidebotham 2008, p. 42-52). There are no known milestones on this or any other major routes in the Eastern Desert. It should be remembered that traffic was controlled there, both from a tax and a policing point of view,2 by a system of “laissez-passer”, at least in some areas.3

Fig. 1

3The stations installed on the tracks that lead to the great quarries in the Eastern Desert (Mons Claudianus, Mons Porphyrites) have a substantially different typology to that of the forts that line the caravan trails leading from Coptos to the two major ports of the Red Sea, Myos Hormos and Berenike. So the praesidium of Umm Balad, south of Porphyrites, has, in its first phase, a plan which is very similar to that of Mons Claudianus (Maxfield, Peacock 1997, fig. 2.3 and 2.44), but with smaller dimensions (fig. 2). None of these fortlets leading to quarries are actually built around a central well, a feature which, on the other hand, characterizes the praesidia on the roads of the two great caravan routes of the Eastern Desert.4 The military character of the stations at the quarries or on tracks that lead to them is not obvious at first, since they essentially housed a civilian population engaged in the extraction and working of the stone, next to some soldiers who policed the area, as evidenced by the extensive documentation provided by ostraka.

Fig. 2

4Fortlets associated with large tracks through the desert to transport goods from eastern trade between the Red Sea and Coptos, on the Nile, are of a different type. They are not known archaeologically before the time of Vespasian (Cuvigny et al. 2003), when there is, however, only evidence on the road from Berenike: a series of inscriptions bearing the name of the prefect of Egypt L. Iulius Ursus and dated AD 76-77 (Bagnall, Bülow-Jacobsen, Cuvigny 2001). Pliny the Elder (HN VI, p. 102-103) describes the initial state of that route, naming the stations whose toponyms are also listed in the Itinerarium Antonini. Pliny, however, says nothing about the route to Myos Hormos, which was still in use during his time, but his testimony does not negate the existence of more ancient archaeological evidence in this desert. Ptolemaic stations appear on the track leading from Edfu (fig. 1)5 and Strabo (XVII, 1, 45) describes of a previous caravan system whose archaeological remains are largely absent today (De Romanis 1996; Cuvigny et al. 2003; Cuvigny 1997). Within this context, it is worth noting the large inscription from Coptos that mentions the construction of lacci at Apollonôs Hydreuma, Kompasi, Berenike and Myos Hormos by a vexillatio composed of legionaries and auxiliaries (Dessau 2483= I. Portes 56). It is usually dated to the end of the reign of Augustus or Tiberius but nothing precludes it from being later. Whatever the meaning of the word lacci (“cisterns” or “wells”),6 the inscription from Coptos shows the involvement of the army in the control and management of routes in the Eastern Desert, from the first century AD. Conversely, a station like that of Dios/Iovis is clearly dated to the time of Trajan by an inscription, probably because the well of the fort originally built in another nearby wâdi, Bi’r Bayza, no longer produced enough water (Cuvigny 2010, No. 1). Perhaps we have the same situation with the station of Phalakron, probably shortly occupied if we consider that it did not, like the others, undergo numerous alterations changing the original plan.

5I will now focus on the praesidia of two routes to Myos Hormos and Berenike studied recently by the French mission7 and dated between the end of the first century AD and the first half of the third. Six of them were partially excavated and recorded on the Coptos/Myos Hormos route (fig. 3-8), and they have since been published (Cuvigny et al. 2003); four others have been excavated more or less completely on the route to Berenike (fig. 9-12), but only that of Didymoi was the subject of two monographs (Cuvigny et al. 2011, 2012). This little series allows us to draw some conclusions.

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

Fig. 11

Fig. 12

The plan

6The plans are not standardized and no station is identical, but there are architectural similarities. The fortlets are geometric in shape and more or less regular, usually square, sometimes rectangular, with a curtain wall flanked by projecting round towers at the corners, and near the only gate; intermediary towers are sometimes known. The centre of the space thus created is occupied by a large well, which is usually collapsed in on itself. The barracks are, therefore, placed against the curtain wall. We do not know what buildings existed in the free spaces in the centre, since the nearby buildings collapsed into the wells. The dimensions are about 50/60 m on each side. We can summarize this information in the following table:

7The projecting semi-circular towers are contemporary with the construction of the ramparts and do not constitute later additions. Posterns were observed in four cases. At Didymoi and Dios, where they are original, they were blocked in a second phase. It is interesting to note the irregularity of the height of the walls, always built in dry stone, with a pronounced batter to ensure the stability of the construction. No ditch has ever been observed. The relatively low elevations reflect an absence of significant threats, even if, in the course of the second century, additional defensive constructions and different texts point to a renewed danger (Cuvigny 2005, p. 135-143).

8One of excavated fortlets does not follow this pattern, that of Qusûr al-Banat, on the road from Myos Hormos. This is, in fact, a station with smaller dimensions (c. 39 x 32 m), rectangular in plan, without corner towers, and perhaps without a central well. The single gate is framed by two small rectangular towers. From a chronological point of view, this fortlet is attributed to the beginning of the third century AD according to the ceramic material (Cuvigny et al. 2003, fig. 22).

Internal improvements

Wells and cisterns

9The systematic presence of a large central well in these praesidia reflects the core purpose of these buildings: to control water points and, thus, traffic on the road, to ensure the provision of supplies to travellers and caravans. The water was extracted from the well by a system that excavations have so far been unable to clearly identify, although ceramic pots were found, implying the existence of water lifting devices. The water was quite often stored in large intramural cisterns, built out of fired bricks; at Didymoi, these cisterns were multiplied as a result of several reconstructions, increasing the total storage capacity from 120 m³ to more than 380 m³ (Cuvigny et al. 2011, p. 20-24; fig. 13). However, such cisterns are not known everywhere: on the Myos Hormos road they appear at Krokodilô (Al-Muwayh) but a well-preserved fort like Maximianon (Al-Zarqâ') certainly did not have them. On the route to Berenike, three excavated stations (Didymoi, Dios, Xeron Pelagos) had them, but not Phalakron. This storage system made it possible to water animals that remained outside the enclosures: feeding pipes crossing the rampart have been found at Maximianon, Didymoi and Dios. In the last two cases they were driving water to outside troughs (fig. 14).

Fig. 13

Fig. 14 

The baths

10Several very thoroughly excavated forts revealed small bath complexes, except that of Qusûr al-Banat (Reddé 2009).

Maximianon

11In the northeast corner of the fort a small thermal unit was installed after initial construction (fig. 15).

Fig. 15

12The baths could be entered by a small door, 0.68 m wide along the north curtain wall, and one entered a room 4 x 3.15 m, containing two bathtubs constructed of bricks and carefully coated with lime mortar, each with dimensions of 1.78 x 0.68 m. Only the northern one is still fairly well preserved. The inner face of the rampart has, at this point, at a height of 0.60 m, a coating of ochre lime, which probably marks the upper limit of the bathtub because remnants of a white coating identical to that which seals the installation of the baths appears underneath. Two successive floors were observed.

13This frigidarium opened onto a caldarium room which was circular with a diameter of 1.80 m (room 10) and constructed in shale; it was probably topped by a dome made of fired bricks, of which only the lower course remains (fig. 16). Its floor, originally at the same level as the frigidarium, was then below it after the installation of its second paved floor. The floor of room 10 consists of shale slabs on its southern half, resting on the sterile backfill that goes down to the substrate. No trace of a furnace or a hypocaust has been found.

Fig. 16

Didymoi

14The excavation uncovered a small bath complex next to one of the cisterns of the fort (cistern 2), whose masonry is solid; its construction must go back to that of the praesidium (fig. 17-18).

Fig. 17

Fig. 18

15To the south, a hypocaust room (7a) of 2.70 m (EW) x 2.30 m (NS) was noted. The furnace is to the south-west, preserved under the later masonry of room 9: there is a channel 0.40 m wide, probably broken to the west. Its covering has not been preserved. The eastern wall of the caldarium, built of brick (30 x 15 x 7.5 cm), is well preserved, while the western wall has been partially destroyed by the construction of room 9, and the southern wall destroyed by later facilities. The pilae are made of square bricks 30 x 30 cm. To the west, they are not joined to the wall, but there are half pilae that leave a space for the circulation of warm air between the wall itself and a dividing wall, part of which remains standing. In the north-western corner of the caldarium is a heating duct which looks out over fire-reddened masonry, apparently without pilae. The construction of room 9 destroyed the previous building and no longer allows the plan to be accurately recognized, but one can probably imagine that there was a small hot tub there.

16Just to the north, between the caldarium and cistern 2 is a cold pool (7b) of 1.20 m (NS dimensions) x 1.45 m (EW dimensions), carefully coated with fine white lime mortar, in the interior, and with a step width of 0.30 m to the south. A duct, visible in the masonry at the southeast corner, joins the pipe b.

17With no access possible to the north, due to the presence of cistern 2, or west, because of the praefurnium (6), the entry must be either from the east or the south. This last hypothesis is the most probable: we can see, inside the later masonry, a floor of bricks coated with lime mortar and a bulge, which attests the presence of the wall, to the east. Unfortunately, an old survey carried out in this space (6) prevents us from observing the architectural connections. This room could be interpreted as the ruined apodyterium. A supply pipe passes underneath. If this hypothesis is correct, one could enter, from an apodyterium to the south, into the caldarium with a small labrum in the northwest corner, then directly to a small frigidarium. There was no tepidarium.

Dios / Iovis

18The excavation revealed, along the southern rampart, the presence of a small well-preserved bathhouse, installed near a series of large cisterns located in the south-western corner of the praesidium. Its construction seems to be contemporary with the construction of the fortlet. There are three rooms (fig. 19-20). Room 43 (5 x 3 m) has walls constructed of blocks of sandstone bonded with mud. The ground is still paved with large stones preserved towards the east. Access, no doubt on the side of the cistern, is no longer preserved. This is definitely an apodyterium.

Fig. 19

Fig. 20

19One goes into room 42 through a small door, 0.70 m wide, located against the rampart. Room 42 (3 x 2.80 m) is delineated by walls of rough stones, sealed with clay. In the north-eastern corner is a bathtub (1.95 x 1 m) constructed of fired bricks sealed with lime mortar and carefully covered with a thin layer of white plaster. The floor of the room is paved with bricks arranged in a herringbone pattern. It has undergone a major renovation to the south, on the side of the rampart to allow the deep installation of a drainage pipe whose outlet is located in the alley between the cistern and the bathhouse.

20One passes into the room 41 by another small door, about 0.60 m wide, situated against the rampart. Room 41, which corresponds to the caldarium, includes two elements: 41a, which constitutes a small rectangular space of 2.95 m x 1.90 m and 41b, which forms a space of 4 m x 2.70 m, oriented perpendicularly to the other one. In elevation, the walls are constructed of rough stones sealed with clay; below floor level, the construction consists of fired bricks moulded using a single module (0.26 m x 0.13 m x 0.065 m).

21Room 41a has kept its floor and its suspensura intact; it has been built along the peripheral walls with a foundation of fired bricks forming a step towards the interior of the room and by a row of pilae in the centre. The latter, of square form, are built with two bricks placed side by side, over a total height of 0.63 m (9 rows of bricks). Above are the bipedales, 0.60 cm square (thickness 0.10 m). These are in turn coated with a fine white lime concrete. In the south-eastern corner of the room the suspensura forms a funnel extended by a vent pipe consisting of a base of a broken amphora.

22Lime whitewash covered the faces of the walls of room 41a. The excavation revealed the presence of green glass window panes, one of which preserved the seal around it for attachment to a window frame.

23Room 41b, unlike the previous one, is ruined because of the collapse of the suspensura. Only the traces of a bathtub made of fired bricks, at the south-western corner of the room, are preserved. The suspensura comprises pilae, each formed of two rectangular bricks. Many repairs can be observed both on the floor, incomplete and uneven, and in the suspensura.

24A particular device deserves to be noted in the eastern wall of the hypocaust, towards the north-eastern corner. The elevation of a thin wall the thickness of a brick (0.16 m) is pierced at two points by a square ventilation hole which emerges towards the outside of the wall of the hypocaust. This partition is built between two brick pilae, which form a right angle towards the east. It is probably a chimney. The fireplace, however, was not preserved.

25These small baths, with a total area of about 48 m², are the largest observed to date in praesidia of the Eastern Desert of Egypt. Their plan seems easy to reconstruct and I suggest that room 43 was an apodyterium; room 42 a frigidarium with a bath; room 41a a dry room, certainly warm, thus a tepidarium; and room 41b a wet warm room, a caldarium with a bath. The whole forms an L-shaped plan, with only one entry/exit point at one end.

Xeron Pelagos

26The bath complex is up against the eastern curtain wall of the fort.8 It occupies a 45 m² surface (10 m N/S x 4.50 m x E/W) and is organized from north to south into four rooms (spaces 32-35: fig. 21, 22). Its entrance was positioned to the north, but it disappeared into the collapsed well.

Fig. 21

Fig. 22

27The walls of the bath are poorly preserved: in the frigidarium, they are made of rough granite blocks sealed with mouna (mortar/plaster made out of clay); in the caldarium, the interior facing of the walls is made of fired bricks, while the exterior facing of the south wall is made of fired bricks and sometimes granite blocks sealed with mouna. All the bricks used in the bath are of the same module: 26 x 13 x 6 cm.

28The bath underwent several phases of operation and was abandoned before the final abandonment of the fortlet. Some of its facilities, including the furnace, are later than the first phase of the fortlet.

29The northern rooms of the bath (32-33) experienced at least two construction phases. During the second phase a north/south corridor of 2.30 x 1.80 m and a basin of 1.05 x 0.60 m were constructed. The two rooms are linked by a 45 cm wide door. Room 34 (3.35 x 2.25 m) is accessible from the north by a 0.55 m width door. A gap in the north-western part of the room shows that the floor was made of an apron of fired bricks arranged in a herringbone pattern, covered with a lime-mortar screed, which was remade at least once.

30The final bath room (35) is the largest of the three; it measures 2.90 m x 3.60 m and links with room 34 by a very narrow door, barely 40 cm wide. It has a large bathtub (1.85 x 0.75 m) completely occupying the eastern side of the room. The room was heated by a hypocaust whose pilae, arranged regularly, are made of square bricks with sides measuring 27/28 cm. Two slabs of suspensura are preserved to the northwest: they are 33 cm square and 7 cm thick. The bathtub is not mounted on pilae, but rests on a stack of granite blocks, covered with thick lime concrete, topped with an apron of fired bricks.

31These four very different examples show the use of architectural designs that do not seem to belong to a unified construction programme. At Maximianon the creation of the bath was done in a second phase, whereas it seems to have been planned from the beginning at Didymoi. Heating is also achieved using different technologies: a calorific supply must have come from outside the baths at Maximianon, which suggests the use of basins of hot water for ablutions after raising the temperature of the room with braziers, while we see the more classic device of a hypocaust at Didymoi, Dios and Xeron Pelagos. In the latter two cases, the bathing complex is sufficient for two or three dozen men and travellers who may have gone there. At Maximianon and Didymoi, on the other hand, we should highlight the relatively small size of these baths. The lack of water and scarcity of fuel must have seriously curbed the use of these thermal complexes. Unfortunately, there is not a single ostrakon informing us about the system of labour that managed the operation of these desert baths, contrary to what is observed at Bu Ngem, Libya, and there is even less information on how often they were open (Marichal 1992).

The praetorium

32An ostrakon from Maximianon tells us that the four corners of the fort had a particular function: the ostrakon mentions “the corner of the horreum” the “corner of the praetorium”, the “corner of the bath”, the “corner of the pipe” for the supply of external troughs (O. Max 783; Cuvigny et al, 2003, p. 218-219). As we know from excavations the position of the bath and that of the horreum and the pipe, it is easy to identify the praetorium of Maximianon, which comprises only a single room, a little larger than the others at the north-eastern corner of the fort. However, this plan is not standard: at Dios, we have also identified the praetorium in a corner of the fort, because of its mosaic floors (a black and white decoration with local stone, certainly, but “luxurious” and exceptional in the desert). This was, however, in a state of reconstruction, and we do not know where the original building was (fig. 23). In other forts, this type of room has never been identified to date.

Fig. 23

The barracks

33The barracks from the initial phase of the fortlets are quite difficult to identify archaeologically, since they were rebuilt in most cases. The two best examples, little affected by alterations, are those of Maximianon and Qusûr al-Banat: the rooms are positioned up against the curtain wall. The Maximianon plan reveals a series of regular rooms, well built and about 6 m x 3 m, without an antechamber, along the western rampart. Few internal developments have come to light. At Qusûr al-Banat, probably Severan, these dormitories are smaller (about 5 m x 3.5 m) and placed against each side of the station. There too, antechambers are missing. The room on the southwestern corner revealed an internal partition and a bench (bed?) (fig. 24).

Fig. 24

34The three most thoroughly excavated praesidia (Didymoi, Dios, Xeron Pelagos) do not have the same architectural regularity and their plans, at first glance, do not seem “military” (fig. 25). This is, in fact, the result of a long evolution; many successive alterations affected the original plan to the point of hardly leaving any remains, or even traces. It is, therefore, impossible at present to identify the first phase, but it seems certain that a military architect was involved. At Dios, the presence of such a specialist is confirmed by a dedication (in Greek) of a certain M. Antonius Celer, architect of the Cohort I of the Lusitanians, at Zeus Helios Megas Sarapis. The inscription was discovered in a context of reuse, and we do not know, unfortunately, if this is the unit that built the fort in 114/115, but the hypothesis is tenable (Cuvigny 2010, No. 2).

Fig. 25

35The surface levels uncovered at Dios, Didymoi and Xeron Pelagos are the latest and we can date them at the earliest to the Severan period, given that the material found dates to the middle of the third century. It is certain, however, that the praesidia were no longer occupied roughly after the time of Gallienus.9 This phase is interesting because it allows us to observe a very significant evolution in the rhythm of daily life inside these forts. This change results in a range of phenomena that can be summarized as follows:

– proliferation of construction, with many phases of successive reconstruction, without regular plans. The buildings are of mediocre quality.

– multiplication of bread ovens. The Dios station had an impressive number of these (fig. 26), and many silos (fig. 27).

Fig. 26

Fig. 27

– pens for small livestock and/or poultry at Dios in the free space west of the well (fig. 10 and 28).

Fig. 28

– in the final phases rubbish is no longer thrown outside the fort and instead it is found in the unoccupied rooms, usually on the side opposite the entrance. At Didymoi the excavation unearthed many layers of dumps and animal litter inside the fort, which indicate a smaller “garrison” that no longer occupies all the space. The cisterns were partially filled, an indication of a smaller population. The thermal baths seem to have been abandoned.

The aedes

36In an article based primarily on excavation data from praesidia on the Myos Hormos route, I suggested that the military shrines were located in front of the gate against the opposite rampart (Reddé 2004). The argument then developed was based as much on theory as on paltry archaeological remains. This reasoning is still valid, but recent excavations of the fortlets on the Berenike route make it necessary to qualify it and to consider new data.

Dios/Iovis

37Excavated at the end of the 2008 campaign by E. Botte and J.-P. Brun, the aedes of Dios was published as a preliminary report in Chiron in 2010. H. Cuvigny wanted to edit rapidly a series of oracular ostraka discovered during this excavation, placing them in context (Cuvigny 2010). I rely here on her description to summarise briefly the main elements necessary for understanding what follows, leaving the excavators to outline further details in future publications. The following points will, therefore, be considered:

1- The praesidium was founded in AD 114/115, evidenced by a Latin inscription found in the passage of the gate (Cuvigny 2010, No. 1).

2- The aedes backs onto the curtain wall, on the right, after the entry passage. But there is a secondary development that took the place of the earlier barracks. The original sanctuary was not found, the first interior arrangements of the fort having been completely restructured several times. On the other hand, two inscriptions were recorded in front of the entrance: the first, in Greek, which mentions the architect who built the station under Trajan, is dedicated to Zeus Helios Megas Sarapis (Cuvigny 2010, No. 2); the second, in Latin, includes an erased name indicating a damnatio memoriae, most likely that of Caracalla or Alexander Severus rather than Commodus (Cuvigny 2010, No. 3). It is probably at the beginning of the third century that the redevelopment of the area took place, with a transfer of the aedes away from its original location.

3- The new sanctuary takes the place of room 49 of the fort, and measures 4.25 x 3.5 m. It includes a series of specific facilities (fig. 29).

Fig. 29

38At the bottom, facing the main entrance, and up against the wall, is a wide podium, fully preserved (L= 2.25 m x l= 0.745 x H= 1 m), built with reused bricks from baths or cisterns and from limestone blocks, sealed with mortar. Decoration resembles an opus sectile of soapstone and limestone slabs. Different graffiti and proskynemata adorn the facade. This podium is connected, at the rear, by a staircase of three steps leading to a side door in the north-eastern corner of the room (fig. 30).

Fig. 30

39Excavators recorded several statues, some on the ground, others attached to the podium itself: there is an enthroned god (Zeus Helios Megas Sarapis?), a Cerberus head of unfired clay, the statue of a Graeco-Roman god, upright and still sealed to the podium and the statue of a standing Egyptian god (fig. 31). Other divine representations, which almost all refer to the cult of Sarapis are present, as well as inscriptions and graffiti.

Fig. 31

40The top of the podium went through several phases of remodelling: at first, it seems to have been flat, but perhaps hosting some of the above mentioned statues; in a second phase, the surface appears to have been reworked with a mortar aggregate, which still seals the foot of one of the statues and four wooden uprights (6 to 7 cm x 3 cm) which form a square framing a hole in the podium (fig. 32). After a violent episode of destruction, which occurred in a third phase, the podium was restored with reused materials, and at the time of discovery, still had, from north to south, an oil lamp, the knees of the seated statue of Sarapis (?) and the statue sealed on the top. Behind it, a small statuette of Jupiter holding a bolt of lightning was discovered. At this point the passage to the side door was completely blocked, and a fired brick construction added on the northern side of the podium. The staircase, now useless, contained a filling of sand and rubble where the oracular ostraka were found (Cuvigny 2010). At the southern end of the podium, there was a sort of tabernacle opening towards the exterior on top of the repairs on the podium (fig. 32).

Fig. 32

41In front of the podium, the room was, in its initial phase, decorated with a black and white pavement of mosaics (shale and quartz) forming different geometric patterns: a checkerboard on one half of the room, beyond the door, a floor uniformly white in the part located in front of the podium, with a black frame surrounding it, in the axis, and a white centre.

42From both sides of this framework appear two square bases of fired, whitewashed bricks, resting on the mosaic. At the foot of the podium in the centre axis is a third, smaller square cube.

43This space had several successive refurbishments: in a second phase two “pools” were inserted, placed on the mosaic floor, along the northern wall. These basins probably received containers whose traces can be seen as depressions in the ground. The closest podium blocked the passage to the side door and the stairs behind the altar. In a third phase, a floor of fired bricks covered the mosaic and basins, now hiding the central cube No. 10 (fig. 33).

Fig. 33

44Details of these changes cannot be accurately dated, but must fall within the third century, as we noted that these fortlets do not seem to have any material from the fourth century, except sporadic finds indicating the limited use of the track but not a permanent occupation.

45The existence of a podium in a military aedes has already been discussed several times in the scholarly literature, due to various finds which T. Sarnowski listed and that should now be updated (Sarnowski 1992, note 9): the author cited Castell Collen (Nash-Williams 1969, p. 159), Caerleon (Arch Cambrensis 1970 p. 14, fig. 2), Chesterholm (Birley et al. 1936, p. 229), Risingham (Richmond 1940, p. 110) and Aalen (Planck 1984, p. 22). The best preserved example architecturally is that of the fire station at Ostia, which has, as at Dios, stairs with lateral access (Sablayrolles 1988, fig. 4). This is also the case in Risingham and we know that this architectural feature was still in place in the Lejjun legionary camp (Jordan) during late antiquity. Here, as in Caerleon, the podium also runs along the sides of the room, and is serviced by both a side staircase and a central staircase in front of the apse (Parker 2006, Fig. 4.5 A). In the case of Novae, T. Sarnowski suggested the hypothesis of a wooden platform, traces of which exist on the ground (Sarnowski 1992, p. 226), and it is possible that such devices were common, but often missed by previous researches. This may be the type of element that Tacitus refers to in a frequently quoted passage that tells us about the advent of Otho, placed in suggestu in the middle of insignia of the castra praetoria, the very place where the gilded statue of Galba was previously located (Hist. I, 36). The exact interpretation of this passage remains controversial because it is linked to the idea that this is due to the presence of the imperial statues in the principia and especially in the aedes since the beginning of the principate (Stoll 1992).

46How do we explain the presence of brick cubes in front of the podium at the fortlet of Dios/Iovis? The central one, smaller than the other two, cannot have had the same architectural features as the others. It is difficult to estimate the height of the pillars standing on these three cubes. The most likely solution is that the central one was an altar. We can now safely suggest that the other two were column supports. But two isolated columns have little meaning; so I think that on top of the podium there was a canopy resting on one side of the columns and on the other side on the corbels of the rampart, even on a simple system of beams fixed to the wall and resting on the columns, at the front. A recently discovered inscription found in Alba Iulia gives weight to this hypothesis, as it mentions the construction of a tetrastylum with a silver eagle by a centurion of legio XIII Gemina (Moga, Piso, Drîmbarean 2008). Later on, a similar device was found in the Tetrarchic camp at Luxor, in the former shrine of the Royal Ka which became the military aedes of the new legio III Diocletiana (Reddé 1986). I have suggested elsewhere, contrary to the initial opinion of the editors and relying only on the case of Dios, that the tetrastylum of Alba Iulia was built to house the new insignia installed in place of the previous one (Reddé 2011).

47Was there, by analogy based on what we know of all these buildings, an insignia in the praesidium de Dios/Iovis? Common sense suggests so, because we cannot imagine a station of the Roman army, even in the middle of the Eastern Desert of Egypt and even partially populated by civilians, without a signum, but clearly it could not be a legionary eagle. For a small detachment like that of the fortlets, a vexillum was enough. The study of the recently discovered insignia collected by K. Töpfer with a comprehensive iconography provides extensive research, though much of the documentation, never previously summarised, was already well known to specialists (Töpfer 2011). When they did not accompany the troops, vexilla, such as eagles, were indeed fixed to pedestals, as shown on the sarcophagus of Matteoti of Modena (Toepfer, 2011, pl. 103, SD 49). Two iconographic representations from Hadrian's Wall, one from Corbridge, and the other from Vindolanda (Töpfer 2011, pl. 121, RE 13 and pl. 122, RE 21) also show vexilla framed by two columns (fig. 34): it could not be closer to the reconstruction that I have suggested for the aedes of Dios/Iovis.

Fig. 34

48Was there a place on the podium of this praesidium, for such an insignia? The Vindolanda plaque shows that the vexillum is implanted into a sort of box, probably of wood, which props it up. We have exactly the same device at Dios since four wood studs surrounding a hole in the podium were found. H. Cuvigny thinks this was a device for fixing a statue of Sarapis. I, however, think this was the place where the vexillum was fixed.

49The presence of statues on a podium, including divine statues, or altars in no way contradicts the hypothesis of a military aedes such as those known in other camps from the West. Once again the investigation conducted by T. Sarnowski is enlightening and I refer to his table 5 which shows, even in the aedes, the presence of different dii militares or even local gods (Sarnowski 1989). It is associated, of course, with that of Jupiter, supreme God of the army, and a representation of the reigning Emperor, usually in a small format (Sarnowski 1989; Stoll 1992; Reddé 2004).

Didymoi

50At Didymoi the aedes was also moved, probably at a late date (during the third century) and installed in a location of the fort where we would not expect to find it, on the site of former barracks. Only a segment of an apse, useless and destroyed by refurbishments, exists from the earlier phase, along the curtain wall which faces the entry (fig. 35).

Fig. 35

51The new sanctuary is not precisely dated, but J.-P. Brun placed construction of the building after the reign of Marcus Aurelius. It has revealed evidence of a device that demonstrates a cult to Sarapis amid more strictly military images (Brun in Cuvigny et al. 2011, p. 33-47). The aedes located in the northwest corner of the station had three quadrangular niches inserted into the bottom wall, with benches along the side walls and an altar outside it (fig. 36-37). Such a device is reminiscent of those found in Maximianon and Qusûr al-Banat (Reddé 2004) (fig. 38). Painted decoration was still visible under the south niche, probably featuring a line of soldiers marching to the right, as known from the shrine of the imperial cult in the Tetrarchic temple of Luxor (Deckers 1979; Reddé 1986 pl. XXII). A soapstone slab found on the ground shows an armoured cavalryman overlapping to the right, perhaps one of the Dioscuri (fig. 39). Remnants of lorica squamata were unearthed in this context. Other objects seem clearly linked to the cult of Sarapis. This is the case of a stone head (fig. 40), fragments of a terracotta statuette probably representing the god standing, a sphinx, a painting perhaps representing Harpocrates, and two offering tables of a traditional Egyptian type. In a lateral space located immediately east of the niches in the background, a secondary place of worship with a cupule was fitted in a later phase (fig. 41).

Fig. 36

Fig. 37

Fig. 38

Fig. 39

Fig. 40

Fig. 41

Xeron Pelagos

52I identify a room located along the south-western curtain wall of fort as an aedes (fig. 42-43). It measures about 5.20 x 2.60 m and occupies a previous room that was refurbished and raised. The room abuts the rampart, slightly formed in the shape of a niche. The entrance is to the northwest. It is reached by a small staircase which gives access to an L-shaped corridor paved with slabs of shale (fig. 43) that runs along the wall and leads at the bottom, near the rampart, to a square hollow loculus. The northern side of the passage is bordered by a basin (fig. 42c) and a small open area on the corridor (fig. 42d). North-west of the loculus (a), against the wall, a raised niche (b) tops a basin (c).

Fig. 42

Fig. 43

53The loculus (a) consists of four low walls of unfired bricks, 20/24 cm wide, forming a quasi-square (1.10 x 1.20 m), partially embedded in the rampart (fig. 44). These smooth low walls are carefully coated in white, inside and out. The western, southern and eastern walls are broken in elevation, while the low northern wall is intact, as evidenced by the coating that covers its upper edge and the wooden threshold is still visible. Given that this threshold is significantly lower than the perimeter walls, we can envisage that a wooden door closed the space. The base is also coated with lime mortar, but one sees, in the centre, the circular traces of a sealant. The fill included:

– a terracotta headless Minerva (h. preserved to 12.5 cm l. preserved to 9.5 cm). The statuette is hollow, with black and red highlights on a white background (fig. 45). The left forearm, holding the shield, is broken. The right hand, disproportionate, holds a spear. The goddess wears the Gorgon on her breast. An ostrakon found right next to it mentions the existence of an Athenadion and we know from other ostraka that the genius loci of Xeron was Athena.

– a fragment of the handle of an amphora reworked as an incense burner. Incense remains are still preserved on top.

Fig. 44

Fig. 45

54Access to this space has been altered on several successive occasions:

– in a first phase, the corridor (e) terminates in a square space coated in lime mortar.

– in a second phase, this space is divided into two and a step that provides more convenient access to the loculus (a) is added.

– in a third phase, the entire space is raised and a platform is built, whose northern part is decorated with a “mosaic” of white (quartz) and black (shale) tesserae, and shale slabs. Between the mosaic and the loculus (a) is a platform about 40 cm deep (fig. 44). The last layer of the corridor shows a sandy elevation, hardened and trampled.

55East of the access stairway, fitted one inside the other, are two wine amphorae which form a wide discharge pipe (fig. 46). It passes through the north wall of room 38, and leads to a small channel dug in the soil that drains water to the north, where we lose sight of it. Sixty cm before the beginning of this staircase a block of fired bricks, recovered from the bath or the adjacent cistern, appears, in the form of a 50 cm sided cube, with a height of 0.14 m.

Fig. 46

56For interpreting such an assemblage, for which I know no parallel either in the West or in the East, we must consider several factors:

– Although the praesidium of Xeron has been extensively excavated, we are not aware of any other aedes there, while all the other stations have a religious space from the beginning.

– The room just described could be neither a barracks nor a storage space or workshop, because of its internal arrangements. No function other than a religious one seems to fit.

– This is a construction made at an advanced stage of the life of the fort, not an initial construction of the station, as is the case with other examples on this road.

57I, therefore, suggest identifying it as the aedes of the praesidium. The cube, in front of the stairs, could be an altar (or the support of an altar). The existence of external altars for the aedes has been observed at Didymoi (supra). It is then entered by a little staircase, before going into the tiled hallway leading at the bottom to the loculus (a), access to which was marked by steps (then with a mosaic). This loculus was closed, on the side of the corridor, by a shutter (or a wooden door) opening at mid-height. The interior includes traces sealant that could have been for a pillar or something else. Given Roman religious practices, it seems unlikely to have been for a statue. However the device could match that of an arca, supporting a podium for the signa. This is a preliminary hypothesis that needs to be supported by further studies. The basin (c) may find parallels in the aedes of Dios where basins were also found. The need to evacuate liquid is certainly proven by the apparatus seen along the stairs. The presence of a statuette of Athena, genius loci of Xeron reinforces this assumption.

58These various findings raise a number of issues:

– It is the first time to our knowledge that topographic deplacement of the aedes has been documented in military camps. Although we do not know archaeologically the location of the original shrine, it was not present amongst the features we excavated.

– In Egypt, assimilation between Zeus, Helios and Sarapis has obviously led to a shift away from “classic” military religion, as we know it from western literature (Domaszewski 1895; Birley 1978). One of the main questions we must ask concerns the date of this shift: is it the result of a late evolution, after the Severan period, or the consequence of a local particularism, due to the Egyptian situation and the fact that these stations were not only occupied by soldiers?

– The Jupiter honoured at Dios is not a purely Roman god, since it is a Zeus Helios Megas Sarapis (Cuvigny 2010, No. 2), and in fact, all the artefacts found in the aedes and the oracular practices point to a cult of Sarapis, which could, in principle, cast doubt on the purely military interpretation of the plan uncovered. Is it, therefore, a Roman sanctuary? A temple of Sarapis? Or both?

59G. Tallet recently proposed that the figure of Zeus Helios Megas Sarapis was an Egyptian god created in an Alexandrian milieu for the Romans, notably the soldiers. The author shows that the distribution of this new divine entity is actually linked to the network of military garrisons and probably began under Trajan (Tallet 2011). The diffusion of Egyptian cults in the military camps is yet more ancient, even in western garrisons, and I have reported elsewhere that it can be found in the Rhineland from the time of Nero (Reddé 2014). It finds parallel in the East, as shown by O. Stoll’s comprehensive investigation (Stoll 2001). Thus, in the oasis of Jawf, where the presence of soldiers is well documented, there is a dedication pro salute domm(inorum) nn(ostrorum) (duorum) I(ovi) O(ptimo) M(aximo) Hammoni and sancto Sulmo (Speidel 1984). Dedicated by a centurion of the III Cyrenaica, the inscription dates to the Severans. Around the same time, even at Bostra, a cornicularius engraved his wish to Io(ui) Op(timo) Max(imo) Genio Sancto Hammoni (AE 1952, 248). At Sûr, in the Lejaa region, a tabula ansata bears a dedication Iovi Hammoni, written by a questionario (CIL III 13604). This type of divine association has become so ordinary in the third century army that a papyrus from Dura indicates that the password of the day (May 27-28, AD 239) is Iupiter Dolichenus. We should remember that this religious contamination in the Eastern army is ancient, since, on the morning of the second day of the Battle of Bedriacum, in AD 69, soldiers of the legio III Gallica greeted the rising sun Syrian style: “Undique clamor, et orientem solem ita in Syria mos est tertiani salutauere” wrote Tacitus, Hist. III, 24. This custom is again confirmed by Dion Cassius (64.14, 3) and Herodian (IV, 15). As noted rightly by R. Haensch, the legionaries were still, at that time, from a typically Roman setting, probably colonial.10 One could multiply the examples. Above all, and this is one of the strong points of O. Stoll's argument, which here meets the opinion of ND Pollard, there is no reason to consider the military and civil religious spheres as radically separate , their relations oscillating permanently between “Integration” and “Abgrenzung”, particularly in the East (Pollard 1996; Stoll 2001).

60The very fact that the honoured deity in the praesidium of Dios is Zeus Helios Megas Sarapis does not make it a foreign god in a Roman military environment and should not lead us to believe that the cultic arrangements observed in this aedes did not meet the ordinary religious practices of the camps. All facilities uncovered find parallels in western contexts, justifying that the shrine at the station of Didymoi on the same road from Coptos to Berenike, is qualified as πρινκίπια τῶν κυρίων that can be decorated with garlands for a celebration that is probably within the Roman religious calendar (Cuvigny 2012, No. 31, s. 5). It should not be forgotten that the deity of the place was also Sarapis to whom one could come to look into the future during oracular practices well attested by ostraka11 (Cuvigny 2010). It is only in modern minds that these different religious practices radically exclude one another. At Didymoi the facilities are much closer to a classic Serapeum, even if decorated with a painting that shows a line of armed soldiers and a shale slab with one of the Dioscuri in military costume.

61The discovery of Xeron Pelagos is a problem of a different order. Does the presence of Athena in this aedes suggest that this goddess has tutelary power in the camp, instead of Jupiter, which would be an absolute novelty in a military milieu? Certainly the Roman Minerva is actually part of militares dii and is often attested epigraphically in the camps, but she never takes the lead role (Domaszewski 1895; Birley 1978). Her presence in the aedes of Xeron Pelagos is not incongruous, but she could only be seen here as an associated deity, probably the genius loci.

62The good state of preservation of these praesidia is not only instructive for our knowledge of the local religious community; we also learn a lot about the general evolution of cultural practices within the Roman army in Egypt in the course of third century.

Conclusion

63These brief reflections show the usefulness of the documentation of the Eastern Desert of Egypt for our understanding of these types of small forts, which are very poorly documented in the West. One should not, however, generalize too quickly conclusions that can be drawn from studying them: we noted that their architecture was linked to their local function and that the stations of the quarries could have other characteristics. A broader study, extended to the Levant, thus in a different geographical and cultural environment, would certainly show other types (Poidebard 1934; Gregory 1995-1997). This diversity of architectural, cultural and religious practices in a military environment is particularly evident in the East. It also tends to relativize our often “normative” view of the Roman army.