This document is part of the online version of the book An Oasis City by Roger S. Bagnall, Nicola Aravecchia, Raffaella Cribiore, Paola Davoli, Olaf E. Kaper and Susanna McFadden, which is available at http://dlib.nyu.edu/awdl/isaw/oasis-city/. It is published as part of the NYU Library's Ancient World Digital Library and in partnership with the Institute for the Study of the Ancient World (ISAW). Further information about ISAW's publication program is available on the ISAW website. Please note that while the base URI of this publication is stable, the specific content available online will be updated to include more links to supplementary material and related digital resources.
The oases of Egypt occupied a distinctive and challenging ecological niche (see §1.3), or perhaps better niches. Of the five main oases, Siwa stood apart for its close connection to Libya and its distance from the world of the Nile valley. Extensive remains of production facilities, however, show that it produced wine and olive oil like the other oases. The Small Oasis, today's Bahariya, seems to have been almost entirely devoted to fruit crops, as indeed it is today. It had close ties to the Fayyum and, especially, to Oxyrhynchos. Much of our evidence for it comes from the papyri from Oxyrhynchos, but active fieldwork in the past decade is starting to enrich our picture of its agricultural, religious, and military dimensions. About Roman Farafra, we know very little, not even what it was called. What today we call the Kharga and Dakhla Oases were administratively combined in the ancient Great Oasis. That linkage, partly paralleled in the contemporary term "New Valley" (which also includes Farafra) has much to recommend it, as Kharga and Dakhla are closer to each other (about 180 km between the chief towns) than to the Nile, are part of the same depression in the desert plateau, were economically very similar, and share many characteristics in their textual and artifactual output. The principal communication routes ran from Dakhla through Kharga to the Nile valley, with an alternative direct route, and from Dakhla through Farafra to Bahariya and on to Siwa. The Dakhla–Kharga–Nile axis was certainly the more important of these.↩#p246
If the inhabitants of the Great Oasis were to live above the level of subsistence farming, they needed many things they did not produce, notably metals, but also finer and harder stones, papyrus, and luxuries like the fine glassware found at Kellis, fish sauce, and other foods not available locally. In addition, the population had to pay taxes to the Roman government, and to the extent that this cash was not locally recycled, it had to be earned. There were only two routes to acquiring the means to pay for those goods and taxes. One was to export something, or things, that were valuable enough to be sold profitably in the Nile valley, even after the cost of transporting them across 365 km of desert had been met; the other was to have a strategic or prestige value that would motivate the ruling power to supply the oases, even at a loss to itself.↩#p247
The second of these certainly explains much of the activity in the Eastern Desert of Egypt during the Roman period: the quarrying of granodiorite at Mons Claudianus, porphyry at Mons Porphyrites, and other expensive stones and gems at other sites, about which much more is known now than was the case a quarter-century ago.1 Acquired at vast expense out of Imperial resources, these stones were needed for Imperial building projects or destined for consumption by the wealthy. The situation as regards the trade routes via the two main ports, Myos Hormos and Berenike, to points south and east, above all to India, is slightly more complex. There was a highly profitable and well taxed luxury trade, in both directions, that furnished a market basis for the large Imperial investment in roads, stations, and ports.2↩#p248
In the Western Desert, there was probably some comparable justification to do with prestige and rarity for the exploitation of the region under the Old Kingdom, when expeditions scoured the landscape for valuable mineral products. How else could the elaborate relays of donkeys necessary to cross such an expanse of desert have been maintained? The best-known Old Kingdom site in Dakhla, ʿAin Asil, was certainly a royal project that at best could have covered its operating costs if its activities had been confined within the oasis. In the Roman period, there is no sign of any Imperial interest in such high-value minerals in the Western Desert. On the other hand, the alum resources in Kharga and Dakhla were exploited—mainly through shallow mining, it appears.3Alum, used as a mordant in cloth dyeing, was an Imperial monopoly, but at the same time it was a highly valuable product enjoying a broad market and may have been a significant component in what made Dakhla profitable at this time. But it is hard to see how it can account for the enormous agricultural growth in the oasis, because the manpower needed for alum extraction would not have been large. Nor, for that matter, is it easy to see why demand should have been so much greater in the Roman period than before, even with a rise in textile production. It hardly seems a good candidate for an important vector of change.↩#p249
The major technological change that made the Great Oasis more viable as a destination and a source of wealth had in fact taken place long before, with the introduction of the camel into the Western Desert. We do not have an exact date for this change. The camel seems to have reached Egypt before the Saite period, but the earliest mention in any text from the Saharan region probably dates to the fifth century BCE, during the Persian domination, in a demotic ostrakon from ʿAin Manawir in the Kharga Oasis.4 We do not, as far as I know, have any secure archaeological confirmation of the camel's presence here at an earlier date, but it is possible that it arrived under the Saites or even earlier. Unlike donkeys, the animal can cross the desert without an elaborate network of stations and wells for frequent watering. A camel could go from the Nile valley to Kharga or Dakhla without any intermediate food or water stops, if none were available. (This is not to say that it will not drink more frequently when possible.)↩#p250
The impact of this fact on the economic realities of crossing the desert was dramatic. Not only does the camel not need to make fueling stops, neither does it need to use part of its payload to carry fuel—that is, food and water—although probably in most cases some fodder would have been brought along. It is not only a matter of having replaced a short-range aircraft with a long-haul jet, but as if one had invented an airplane able to use all of its payload for cargo. The fact that the number and wealth of settlements in both parts of the Great Oasis, as well as in the Small, or Bahariya, Oasis, rise very sharply during the Saite and Persian periods is very likely to be the result in large part of the introduction of the camel as the main freight-carrying vehicle. In Kharga, in fact, growth during the Persian period is even more striking than in Dakhla. But all of this means that the camel was, at all events, no novelty in the Roman period.↩#p251
The camel is thus necessary to explain the Roman takeoff, but not sufficient. This condition had been satisfied more than four centuries before the Romans arrived. We must ask: What did the oases produce, or what could they have produced, that would be profitable when transported all the way to the valley by camel? And, even more important, why did their production of it in Roman times assume so much larger dimensions than in the Saite, Persian, and Ptolemaic periods?↩#p252
A significant part of the answer emerges from the Kellis Agricultural Account Book, the 1786-line account of receipts and outlays over three years, kept by the steward of part of a large estate during the 360s.5 According to my analysis of this account, it showed a two-part estate economy. One part consisted of crops grown for local consumption by humans or animals, or for disbursement in payment for services rendered—mainly wheat, barley, fodder crops, and wine. The other part was provided by high-value tree crops: olives, mainly in the processed form of olive oil; dates; figs; and jujubes. In the account, these last are not expended locally, so this part of the estate's income seems to constitute its effective surplus. What all of these products have in common is that they could have been transported to the valley by camel without their cost to the purchaser there being increased by more than 10–20 percent above the production expense, or at any rate their market value in the oasis, compared to the more than doubling of cost that would have taken place in the case of bulk commodities like wheat or wine. Just like Roman Tripolitania, as analyzed by David Mattingly,6 Roman Dakhla apparently experienced an olive-oil boom. The rich archaeobotanical evidence from Kellis corroborates this picture.↩#p253
Figure 110. An ostrakon used as a tag in a jar stopper (Inv. no 3518).
The hundreds of ostraka now yielded by the house of Serenos at Trimithis are generally in accord with this overall picture, although they cannot offer the kind of synoptic picture of estate management that we find in a comprehensive account book. Instead, they show us many dozens of snapshots of individual economic acts: deliveries of hay in one set of accounts; many tiny ostraka used as tags, set in jar stoppers (made of mud) atop containers of wine or oil, giving at most a year, a place, and a personal name; receipts for hay or barley used as donkey feed; letters requesting the provision of olive oil or wine (Fig. 110). Archaeobotanical evidence is not as well preserved at Trimithis as at Kellis and is still under study, but it is in general consistent with what is found at Kellis. In effect, we are seeing much the same agricultural structure and range of crops as in the Kellis accounts, but from different vantage points, and no doubt more of the local-consumption side of the ledger than of the surplus for export.7↩#p254
This agricultural structure included a landlord, in our case Serenos, whose home is referred to as "the house" throughout the texts, along with his wife the oikodespoina, "mistress of the household," never named; a set of landholdings centered around wells in the nearby countryside; and a system for collecting at least part of the produce centrally, evidently at Trimithis itself. At least one ostrakon shows that the wells themselves were an important part of the property portfolio, as had been the case in Dakhla since the Old Kingdom, with farmers paying regular amounts of rental per day of water supply. The exact nature of the settlements around the wells is not usually made explicit, but they are called by the Egyptian word for "water" or "well", pmoun, "the water of," followed by a name or description. The ostraka in the first volume resulting from the Amheida excavations contain more than forty such pmoun names. In many cases, the Greek word hydreuma, "well," appears, rather pleonastically, before pmoun, which for Greek-speaking people no doubt had come to be part of the name rather than a mere descriptor.↩#p255
This description based on the ostraka tallies fairly well with the picture drawn from the Kellis account. There is one difference of standpoint that increasingly strikes me in looking at not only the ostraka referring to the activities of Serenos' household in B1, but also those deriving from other households and used as fill under Serenos' house and the adjoining street. In the Kellis account we see goods of all kinds coming into the possession of the pronoetes, the manager who is the "I" of that account, and only some of them going out as disbursements. In the ostraka from Trimithis, by contrast, the goods that were consumed locally are overwhelmingly dominant—that is, grains and especially wine. The greater visibility of wine is easily explained: grain traveled in sacks, wine in jars with labels set into the stoppers. The dominance of consumables in the Trimithis documentation is probably to be explained by the fact that our ostraka come from Serenos' city house, where the family lived and things were consumed, rather than from a storehouse where goods for export would be concentrated.↩#p256
But since the publication of the Kellis Agricultural Account Book, the picture has also been complicated in a different way. The account book also records cotton, although not in very large quantities. Now, however, we find mentions of rather large amounts of cotton in a few ostraka from Trimithis too, and it is becoming clear that cotton production was no small-scale enterprise. I showed in a recent article that all of our papyrological evidence for cotton production in Roman Egypt comes from or concerns the western oases;8 it was also being grown in Lower Nubia around Qasr Ibrim, as John Peter Wild has demonstrated.9 But the Nile valley in Egypt seems not at that time to show any signs of cotton-growing—and this is actually no surprise but a near-inevitability, for cotton is a summer crop, and Egypt's arable land was not available during most of the summer because it was under water. Only after the Nile's flood receded could the land be sown with winter crops, especially wheat and barley, which grew well in basins moistened by the flood. In the oases, by contrast, as we noted earlier, water flowed, but did not flood, 365 days a year. It would thus have been perfectly possible to grow cotton in the oases during the summer on the land that in winter produced wheat or fodder crops.↩#p257
Moreover, that two-season advantage was not limited to cotton. Tosha Dupras has shown that the bone collagen of Kellis residents buried in the cemeteries of that village contains an isotope of carbon that cannot derive from the staple carbohydrates that are typical of the ancient Old World diet—that is, wheat and barley—but rather must involve maize, which as a New World crop not known in Egypt until modern times is out of the question in this context, or millet.10 And in fact, millet has been found in significant quantities at Kellis, although it is never mentioned in the documents from any of the oases, and only in one passage of Olympiodorus (FHG 4.64.33), pointed out many years ago by Guy Wagner.11 Olympiodorus tells us that millet is sometimes sowed three times a year in the oasis. We do not know if it was eaten by the people of Kellis; it is very rarely mentioned in the papyri from the Nile valley, either, and it does not seem to have been a valued part of the Egyptian diet, despite its high nutritional value and easy cooking. But its consumption could be an oasis peculiarity. Even more likely, perhaps, is that it was absorbed indirectly, through feeding it to animals whose meat or milk was then consumed by humans. But either way, it was obviously another crop that could be grown in the summer and thus allow the land to produce two crops a year, or even, as Olympiodorus says, three.↩#p258
Some interesting conclusions may be drawn. First, the possibility of producing two crops a year could have helped offset the remoteness of the oases from the valley. The additional income would essentially com-pensate for the overhead on oasis production caused by transportation costs, eliminating the built-in disadvantage that the distance of the oases from the valley gave them. Second, neither cotton nor millet is known in Egypt before Roman times, as far as I can determine. Their introduction may be a key, along with the increasing consumption of olive oil, to the attractiveness of the oases as places for investment in land development in the early Roman period.↩#p259
Seen in this light, the oases can have profited from a two-part strategy: on the one hand, self-sufficiency in essential foodstuffs, of which they grew just enough to feed themselves and their animals; on the other, a highly export-dependent focus that tied them into a globalized economy in high-value tradable commodities: in their case, as we have seen, particularly olive oil, cotton, and the more traditional dates, figs, and jujubes—along with alum, in all likelihood. The ability to implement such a dual strategy would have constituted a remarkable competitive advantage for the oases. The one caveat I would offer is that Bahariya Oasis, the Small Oasis of the papyri, probably was not self-sufficient in cereals. Because it was both much smaller and nearer to the valley than Dakhla, the resulting need to import was not as significant a cost for Bahariya. We have a considerable amount of evidence from customs declarations and letters of the Roman period showing that wheat was sent to the Small Oasis from the Fayyum and from Oxyrhynchos.12 It was more expensive in the oasis than in the valley, but not enough to make the economics of growing wheat in that oasis attractive. Even today, Bahariya is almost totally dominated by fruit crops.↩#p260
An export-based economy located at a great distance from the Nile entailed an extensive land transportation industry. The ostraka are saturated with mentions of donkeys, mainly employed in local transportation aimed at concentrating the surplus in the hands of the landlord and seeing that provisions get to those who need them. They also mention camels and camel drivers often enough to show their presence, but records concerning how the surplus was sold and moved to the valley have not come to light at any of the excavated sites so far, just as they are largely lacking in the papyrus-producing cities of the valley. But just as we glimpse in the ostraka from the Eastern Desert the outlines of a large corps of camel and donkey drivers based in Coptos, so we can infer the existence of a similar body of men who moved the oil, cotton, dates, figs, and alum from the oases to the valley. In the case of Dakhla, this may have meant Lykopolis (modern Assiut), but there were routes also to the Panopolite (now Akhmim), the Hermopolite, and to the Antaiopolite; and it is probable that the cross-desert caravan route sent many fingers out into the valley cities.↩#p261
6.2. A rural settlement in the oasis system
Nicola Aravecchia
The bedrock of this dual agricultural system in the oases was the production center built around a well. Such hamlets have not been the object of serious archaeological investigation until recently, and only in the oases do they survive well enough to be studied in depth. One of these agricultural centers is ʿAin el-Gedida, in the central part of the Dakhla Oasis. It was always substantially smaller than the oasis cities and towns such as Trimithis, Mut, and Kellis. But precisely for this reason, ever since the very first excavations there in the mid-1990s, ʿAin el-Gedida has sparked interest among scholars working in the region. They noticed the unusual layout of the remains and were intrigued by questions concerning the nature of the site and the role it had played within the religious, administrative, and economic systems of the oasis during Late Antiquity.13 These questions led to new excavations, carried out between 2006 and 2008 under my direction. This investigation enabled us to gather a considerable amount of data, which answered some of those questions and added to our knowledge of the social and economic organization of rural Dakhla.↩#p262
Figure 111. ʿAin el-Gedida, Google site view (to the northwest) with archaeological plan superimposed.
ʿAin el-Gedida lies three km north of the village of Masara and is bounded to the north by an escarpment and a narrow strip of desert land, with two rocky mounds as its most striking topographical features. All around are cultivated fields, now dangerously encroaching upon the current edges of the site (Fig. 111). The area is dotted with numerous bushes and trees, including palm trees, which grow there thanks to the easy availability of water. Indeed, the name ʿAin el-Gedida means "the new spring", pointing to the relative wealth of water in the area as the reason for its use as cultivated land. There is a strong likelihood that the modern name coincides with the ancient one, if we may judge from a Greek ostrakon found during the 2008 excavation. This mentions Pmoun Berri, a toponym meaning "the new well", which is the precise equivalent of the modern Egyptian name ʿAin el-Gedida.↩#p263
Figure 112. Site map of ʿAin el-Gedida.
The archaeological remains are spread over five mounds, of which one (Mound I) is substantially more extensive than the other four (Fig. 112). The mounds are close to each other except for Mound V, which lies over 200 m from the main one. It is difficult to establish the overall dimensions of the site. In fact, the adjacent cultivated fields likely cover a portion of the ancient remains. This makes it hard to assess whether the zones between and around the five mounds were densely built up, forming a continuum with the mounds, or whether there were separate clusters of buildings on each mound. Also, the heavily disturbed context of Mound V complicates the situation, making it impossible to establish its outline with any degree of precision.↩#p264
Figure 113. ʿAin el-Gedida, southern half of the main mound (view to the northeast).
We detected and surveyed the remains on all the mounds, but focused our excavating on the main one. The southern half of this mound, investigated in the mid-1990s by an Egyptian mission led by Kamel Bayoumi, gives the impression of having developed from a smaller, centrally located core of buildings into a larger complex extending toward the mound's edges (Fig. 113). We found unambiguous archaeological evidence that architectural features had been added to earlier structures, which had often been substantially altered. These rooms generally demonstrated a very poor building technique and no systematic plan. They did not form separate clusters—in fact, we recognized no easily identifiable domestic units in this area—but were interconnected in a complex network extending over most of the southern part of the hill. The hasty construction of these buildings suggests, as Bayoumi has argued on the basis of the first excavations at ʿAin el-Gedida, that they were designed to satisfy the needs of a rapidly increasing population.14↩#p265
Figure 114. Storerooms in the southern half of the main mound at ʿAin el-Gedida (view to the northwest).
Unfortunately, most of the structures cannot be reliably identified as far as their original function or functions are concerned. More firmly recognizable as storage magazines are a set of three narrow rooms symmetrically placed at the southwest corner of Mound I (Fig. 114). These spaces typically have low walls, whose original height can be inferred from the scanty remains of the vault springings, and clay containers set into the original floor, likely used to store crops.↩#p266
Figure 115. A kitchen in the southern half of the main hill (view to the east).
Particularly worth mentioning is a large room set in a central position immediately north of the three rooms just described. This space, which seems to have been roofless even in antiquity, was once connected via a long corridor to the central and north part of the mound, although at some point a staircase obstructed the passage into this corridor.15 The room was identified as a kitchen after the discovery of substantial remains of at least three bread ovens (Fig. 115). These were circular ceramic features, built on a raised earth platform and surrounded by mud-brick partition walls. The dimensions of the kitchen/bakery and the presence of the ovens suggest that this facility satisfied not just the needs of one family, but likely those of a fairly large group of people.↩#p267
Figure 116. Plan of the main hill at ʿAin el-Gedida.
Our 2006–2008 work focused on the central and north parts of the main hill. A surface clearance of this area revealed a network of several buildings, of various sizes and some interconnected. Although the general layout gives the impression of a rather confused spatial arrangement, we identified traces of more regular planning. Sets of interconnected rooms, some opening onto spaces that seem to have been inner courtyards, formed larger, roughly rectangular blocks divided by streets (Fig. 116). A remarkably large rectangular building is located at the north edge of the hill. This structure, which was not excavated, comprises two rectangular rooms lying at the center of a wide, also rectangular, feature. Comparing this with similar buildings in Dakhla, for example at Kellis and ʿAin es-Sabil, suggests the complex was a columbarium, or pigeon tower, a typical feature of the oasis landscape in Roman times and during Late Antiquity, just as their descendants are today.16↩#p268
Figure 117. An excavated area to the south of the church complex (aerial view to the southwest).
A few meters to the south of the columbarium, and centrally positioned within the mound, lies the church complex discussed in Chapter 5.17 Excavated between 2006 and 2007, it represents one of the site's most significant archaeological discoveries. The investigation also extended to the area immediately to the south and east of the church, with the goal of ascertaining the topographical relationship of the complex with the surrounding buildings and passageways. Among these is a long street, running from north to south along the east side of the church complex. The street narrows where the apse, built against the east wall at a later time, protrudes into the street itself. Further east is a small open-air industrial area, with evidence of several bread ovens—not dissimilar from the large kitchen with multiple ovens found in the southern part of the mound. A barrel-vaulted passageway runs east–west along the south wall of the church, intersecting, at its east end, with the same street. An open courtyard is situated there, showing signs of clay and mud-brick features possibly associated with feeding animals (Fig. 117).18↩#p269
We retrieved a substantial amount of pottery, including thousands of fragments and some complete vessels.19 Small objects such as cups, bowls, and plates prevail over larger containers, although some of the larger ones were found in most contexts. All these vessels share a character consistent with a domestic assemblage in a fairly poor rural environment, which is also suggested by the other archaeological evidence. The ceramic material, with few exceptions, points to a fairly homogeneous chronological framework spanning the early fourth to (at the latest) the beginning of the fifth century CE. The documentary evidence, in the form of Greek and Coptic ostraka and more than one hundred and fifty coins found at ʿAin el-Gedida—again with very few exceptions such as earlier coins found in contexts of dubious reliability—supports a fourth-century dating.↩#p270
Figure 118. ʿAin el-Gedida, remains of the temple/ceramic workshop (view to the north).
Although ceramic and numismatic remains pointing to phases of occupation earlier than the fourth century were scanty, to say the least, it became clear that the previous evidence did not give the full picture when in 2008 a large building complex was investigated near the west edge of the main hill. This complex lies to the west of a room that was used, at least in its latest phase, as a dump for domestic waste. The building measures 18.5 m north–south and 7.1 m east–west; its east walls are preserved to a maximum height of over 2 m, while the north and west parts have been subject to particularly severe destruction and erosion (Fig. 118). The inside face of the east wall is characterized by ten regularly spaced niches; it is likely that the west wall was niched as well, but it is not preserved to the height at which the niches would have been placed. The discovery of bins and large basins inside the largest room, as well as of partially worked clay, several sherds of unbaked pots, and two large fragments of potter's wheels, identified the complex, in its late phase of occupation, as a pottery workshop.↩#p271
However, a study of the original layout, which did not include the internal partition walls in the large courtyard, led to a different interpretation for the earlier phase. In fact, the building was originally a small mud-brick temple, with a large courtyard characterized by long niched walls leading northward into a sequence of pronaos and naos, both flanked by two longer rectangular rooms symmetrically arranged. Abundant comparative evidence exists to substantiate this identification. One of the closest parallels is the unexcavated temple of El-Qusur, at Tineida in the east of Dakhla,20 which shows a very similar layout to the original plan of the west complex of ʿAin el-Gedida, including long niched walls in the large rectangular courtyard.↩#p272
The discovery of a pagan temple at ʿAin el-Gedida, later converted into a small industrial establishment, was a significant event. Indeed, it undoubtedly testifies to a longer history of occupation of the site, which must have begun at a time when traditional Egyptian religion was still in full flower. It is obvious that the temple must have been built some time before its conversion into a pottery workshop, this latter probably datable to the same time as the church complex, that is to say, the first half of the fourth century. This means that the site must have been occupied since at least the second century CE, and certainly by the mid-third, when the building of pagan temples seems mostly to have ended in Egypt.↩#p273
Why was ʿAin el-Gedida abandoned sometime during the last thirty years or so of the fourth century or the beginning of the fifth? The evidence suggests no episode of violent destruction that might have led the inhabitants to leave the site abruptly. No objects of significant value were found in the rooms that were excavated, suggesting that the buildings in the settlement had been emptied by their owners of anything worth taking. Perhaps the site was abandoned not because of some sudden, to us unknowable, incident, but was a planned departure—arranged quickly but not so quickly that the villagers could not sort their possessions and take with them anything they wanted. This is something that we see at several other sites in the oasis. Scholars have debated possible causes, such as climate change, economic depression, or political unrest. At any rate, notwithstanding possible economic and social changes taking place at ʿAin el-Gedida, the reasons that led all its inhabitants, as well as those of other sites, to abandon their homes and move elsewhere—where, we do not know—are not yet clear.↩#p274
Since the Egyptian team's excavations of the mid-1990s, another possibility concerning the nature of the site has been mooted. As already mentioned, the south half of the main hill is characterized by an unusual configuration comprising a network of closely interconnected buildings around a central open-air kitchen. This arrangement suggests a social structure based on communal living rather than separate households. The discovery of the church complex provided additional information in this regard. The existence of another large kitchen near the complex, and especially of the large assembly hall capable of seating a considerable number of people, seems also to point to the presence there of a community—people who were not organized in the manner of a family, either nuclear or extended. However, these data provide no clue as to who they were or where they came from.↩#p275
Neither the Egyptian nor the more recent excavations found any sizable structures at ʿAin el-Gedida that could be identified as large dormitories—which one might have expected in light of the evidence for communal living just described—but neither did they discover incontrovertible evidence of houses for single families. The paucity of domestic-architecture remains at the site could be explained by the relatively limited area under investigation compared to the overall extent of the ancient settlement. Indeed, it cannot be ruled out that most people may have lived on the other mounds, while on the main hill stood mostly buildings of a more communal kind such as the church complex, small industrial installations including the large kitchens/bakeries, the storage rooms, and the ceramics workshop within the remains of the mud-brick temple.↩#p276
Obstacles to fully understanding the physical and social structure of the settlement apart, the current state of research does strongly suggest that ʿAin el-Gedida had at this time an agriculture-based economy. As noted above, archaeological investigation of late antique Egyptian villages and hamlets has been very limited. On the other hand, documentary sources abound, shedding light on the economy, society, and daily life as well as on the links that country dwellers had with the rest of Egypt, especially with larger towns and cities.↩#p277
Roger Bagnall has analyzed the many aspects of life in fourth-century villages, basing his study on the information supplied by written documents, on ostraka or papyrus, especially the archives of people involved in managing land.21 The picture that emerges is of a dynamic world heavily engaged in agriculture, involved in the economic, social, and political affairs of the time, and also in religious matters. Villagers would have worked in small industrial or craft areas related to the agricultural activities, which played a primary role in the economy of the rural settlements. One could usually find, for instance, granaries, pigeon towers, open-air bakeries, and workshops for manufacturing objects of daily use, such as pottery. All these features have been identified on the main hill at ʿAin el-Gedida, although some of them only tentatively. The kind of layout described above appears to have been fairly standard in Egyptian villages of the time.↩#p278
We know from the papyri that the rural settlements in late antique Egypt were not all full-scale villages. An alternative but associated type, attested to by numerous documentary sources although not yet by substantial archaeological evidence, consists of epoikia: that is to say, small rural centers connected to the management of large agricultural estates.22 Here, a workforce could be employed, full-time or seasonally, to work the land under the direction of overseers, in many cases alongside the tenants. It may be that workers moved to the estate and lived there for the duration of their contract. The spatial arrangement of these epoikia is unknown, because none has ever been identified and excavated. According to a reconstruction made by Dominic Rathbone, chiefly based on documentary evidence, epoikia consisted of buildings functionally associated with the agricultural activities carried out in the farmstead.23 Some Egyptian epoikia were created as isolated entities, later developing into regular villages, while others seem to have been integrated, from the start, into the territorial economies of existing villages.↩#p279
We should not assume that the people involved in the epoikia system necessarily led a fully communal lifestyle. In fact, it cannot be ruled out that wage workers may have moved to these rural settlements with their families, occupying houses similar to those found in other types of settlement. At ʿAin el-Gedida, the south half of Mound I might well reflect the spatial arrangement of part of an epoikion, comprising not its residential quarters but a sector where the buildings more closely associated with agriculture were concentrated, including installations such as bakeries, built to satisfy the needs of a relatively large community. Within the context of an epoikion, we should not be surprised to find a church at the center of Mound I, largely consisting of utilitarian public spaces. Indeed, written sources attest to the possibility that churches were associated with this type of rural settlement.24↩#p280
More evidence prompting us to identify ʿAin el-Gedida as an epoikion comes from an ostrakon found at the site in 2008 (already mentioned), which acknowledges the payment of money by someone described as "from the georgion of Pmoun Berri," the latter being, we may suppose, the likely Egyptian name of ʿAin el-Gedida in the fourth century. Here georgion appears to refer to a farmstead or agricultural settlement and, if indeed it refers to ʿAin el-Gedida, thus establish the basic nature of that settlement. The evidence testifies to a vibrant rural community, well adapted to the local environment, taking advantage of what the surrounding land had to offer and processing its products on site. Furthermore, the small industrial workshops uncovered at ʿAin el-Gedida shed light on a society whose involvement in the local economy extended beyond those activities strictly related to agriculture. Interestingly, it is also probable that, due to its smallness and its proximity to Kellis—and also its similar chronology—ʿAin el-Gedida was administratively dependent on its more sizable neighbor. Indeed, some documentary sources suggest there were strong links between the two settlements, just as there were with the somewhat larger ʿAin es-Sabil.↩#p281
Our excavations at ʿAin el-Gedida offer a direct, although partial, glimpse into several aspects of land management and exploitation and, more generally, into rural life in fourth-century Dakhla. Moreover, the discovery of what may be the first archaeologically attested epoikion has a potential significance stretching well beyond the oasis boundaries and giving us new and valuable insights for the study of rural society and economy in late antique Egypt.↩#p282
6.3. Government and military
Roger S. Bagnall
The countryside that produced Dakhla's crops was from an administrative standpoint organized much as was the rest of Egypt, into subdivisions called toparchies and then, from 307/8, into pagi. In each of these districts there were a number of villages, kômai. Kellis itself and its important counterpart Mesobe, the location of which is not known, both had the juridical status of headquarters of a toparchy, and some other villages are mentioned more casually in the Kellis papyri without any clear indication of their status other than being villages. It is likely that Trimithis was also, during some part of its history, a village at the head of a toparchy, although it seems that it was a polis by the fourth century—and proud of it, to judge from the personification of polis (city) standing in the corner of Room 1 in the house of Serenos (see §7.2) and watching the divine shenanigans going on in the paintings there.25↩#p283
Beyond the larger villages, as already noted, were numerous places named after their well, found in both the Kellis account book and the ostraka from Trimithis, as well as in the Kellis papyri and ostraka. We know that some of these were referred to as epoikia, "settlements", using a term well attested in the valley and discussed in the previous section with regard to ʿAin el-Gedida.26 But the term epoikion is not always used even for places that we know were so designated, and it is entirely possible that some, even many, of the pmoun names in the documentation were in fact classified as epoikia in more official parlance.↩#p284
The oases today still show many remains of smaller rural settlements of the Roman period. As these have in general not been excavated, we are usually lacking the wherewithal to link them to the place names in the documents or to connect a particular physical manifestation with a verbal description. In §6.3 we describe the one such settlement to have been uncovered in Dakhla, ʿAin el-Gedida. These settlements seem from a governmental viewpoint to have been dependent on the villages, but we have no direct evidence of how they were governed.↩#p285
Amheida certainly had such settlements in its vicinity, whether large or small. Many of them lurk in the pmoun names. Vestiges of these settlements can be found on the ground in the environs of Amheida. Not many km away is a field full of farm buildings, pigeon houses on the upper floors and storage underneath (see figure 22). ʿAin es-Sabil, where the Ministry of Antiquities has found another fourth-century church (see §5.3), may have been a large example of the genre, if it was not a village; we do not yet know its ancient name—we have only scratched the surface in this respect.↩#p286
These oasis settlements certainly differed from their counterparts in the Nile valley, if only because their existence depended, unlike subsidiary agricultural enterprises in the valley, on having their own water source; and their size was essentially a product of the volume of that water source's flow. But the same questions arise, especially whether they were fundamentally operated by wage labor, or by tenants, or by a combination. The communal facilities at ʿAin el-Gedida would certainly tend at first sight to push us in the direction of seeing this place as populated by wage laborers rather than permanently resident tenant families. As mentioned earlier, it is not yet clear to us what parts of the site were sleeping facilities and how these were configured, but there is little to remind us of the family-style houses we find at Kellis or Trimithis.↩#p287
One other point must be borne in mind. In Dakhla, the ezba, as such settlements are called in Arabic, was associated until relatively recent times with seasonality; the work there was done by people who lived in a larger village during most of the year but moved to the ezba without their families for the periods of intense seasonal agricultural work—planting, vineyard and orchard cultivation, harvest, the processing of fruit, and the storage of the products. Perhaps the people of ʿAin el-Gedida did not form a discrete population at all but were in the main residents of Kellis who spent only part of the year in the georgion. In the case of the many pigeon-house/storehouse complexes near Amheida, there is nothing still visible that we might recognize as habitation, so they were quite likely dependent on a larger village or even the city itself to house most of the people, most of the time.↩#p288
The Great Oasis had an Imperial presence apart from the civilian officials who governed it. We know from the late antique list of provinces, officials, and military units known as the Notitia Dignitatum that Trimithis was the base of a cavalry unit called the Ala I Quadorum ("first wing of Quadi" [a people of the Danubian region recruited into the Roman army]); and the military camp, ta kastra (the Greek word is a borrowing of the Latin castra), is mentioned in a fair number of documents from Kellis and in a graffito from a side-room of the pillared hall at Amheida (B6), excavated in 2011. But there is nothing at Amheida that looks anything like a military camp, and there was long among archaeologists working in Dakhla an instinct to think that the camp should have been at nearby El-Qasr—even though we know that the word qasr, although also derived from the Latin castrum, was applied to many places in the eastern Roman world that were not in fact castra, so it cannot be taken for granted that a Qasr had once been a castra.↩#p289
In 2006, Fred Leemhuis, who has for several years been doing conservation and reconstruction work on two of the larger Ottoman houses in El-Qasr, made an important discovery. He was walking one day toward an area where he was doing some trial excavation, in what he thinks is the oldest part of the town, when he noticed that a large and rather eroded wall along his path, probably originally part of the town wall, was built in bricks of a different size from that typical of the Islamic town. He had seen the wall many times, but it had taken a slightly different light to bring the lines dividing the bricks to his attention.↩#p290
The brick size was Roman, and he soon found a couple of other massive bastions built of the same bricks. More have now come to light elsewhere in this part of El-Qasr, and it seems to us beyond doubt that the reason we have not so far spotted anything at Amheida that looks even remotely like a fort for the Ala Quadorum is simply that the unit's headquarters were located at El-Qasr, just a few km from the main urban district of Trimithis and obviously considered part of it, as well as closer to the main road to and from the Farafra Oasis. Moreover, the oldest part of El-Qasr looks like a Roman fort in plan. Further excavations by the Egyptian authorities have found more of the wall, and it now appears that El-Qasr was a fort of dimensions and shape more or less identical to those of El-Deir in Kharga, and indeed to other tetrarchic forts in Egypt. It was called pkastron nperro ("the imperial camp") in Coptic, as we know from some ostraka excavated by our Egyptian colleagues.27↩#p291
At the same time, we can now say with confidence that the garrison of Dakhla was more complex than the entry in the Notitia Dignitatum would suggest. As is increasingly acknowledged to be normal Roman practice, it included detachments from other units. One of these, the Tentyrites, turned up in an ostrakon found at Amheida in 2008; the second is the Legio II Traiana, represented by the mention of an "optio [a military rank] of Asphynis" in an ostrakon discovered in 2009. Asphynis, in the Latopolite nome in the Nile valley, is where this legion was stationed in the fourth century, as we know from a centurion's will in the Columbia University papyrus collection.28↩#p292
To these may be added an ostrakon found at ʿAin el-Gedida, not far from Kellis, which is a receipt for annona (rations) for the hippotoxotai, horse-mounted archers, stationed at Mothis. So at least one of these units contained cavalry archers, no doubt useful in assuring the security of the desert roads. Perhaps these were the Tentyrites, since the Notitia indicates that equites sagittarii indigenae ("native cavalry archers") were stationed at Tentyra. It is noteworthy that Dakhla has so far given little sign, on the surface, of the numerous small forts characteristic of Kharga. But this may be deceptive. Like the fort at El-Qasr, they may yet be found.↩#p293
One central question about the military is this: What was it doing in the oases? There is, as far as I know, no evidence at all for its presence in the Great Oasis before the Tetrarchy (Diocletian's four-emperor regime, in office 293-305), and probably not in the Small Oasis either. It appears that the building of a series of fairly uniform forts at Qaret el-Tub in Bahariya, El-Deir in Kharga, and El-Qasr in Dakhla is part of the same wave of military construction in the late 280s, and continuing for a couple of decades, that we find elsewhere in Egypt.29 Nowhere is it clear whether there was any direct threat that we would describe as military, or whether instead we are dealing essentially with the assertion of control over lines of communication, plus perhaps a policing function. The building of forts in the Eastern Desert under Vespasian in the mid-70s, where before there had been unfortified stations, has been connected with the increasing nomadic raiding that is recorded in the ostraka from the station at Krokodilo, on the road from Coptos to the port of Myos Hormos.30 It is possible that the same thing was true of the Western Desert in the later third century—a much more difficult environment for nomads, but also for defenders, than the Eastern Desert, because water sources do not lie as near the surface in the Western Desert as in the Eastern, meaning that a dense network of stations based on wells is not feasible. The exposed lines across the desert were much longer in the west, and the forts are in any case in, or sometimes on the edges of, the oases rather than out in the open desert of the high plateau.↩#p294
It is, then, natural to wonder if the steep decline of the oasis settlements in the second half of the fourth century and beginning of the fifth is connected with greater insecurity. It will be clear that only secure transportation lines across the desert to the Nile valley can have enabled the existence of the economy that I have suggested earlier. There are, of course, other possible explanations for the decline, which have been canvassed from time to time. These include dwindling water supplies, the invasion of sand dunes, or the possibility that the decline is illusory and produced by our failure to recognize the later settlements archaeologically and to excavate them.↩#p295
There are good arguments against all of these hypotheses. First, the water supply is not likely to have failed in widely dispersed locations at the same time, and Mut and Amheida even today have fairly high humidity near the surface, even though throughout the oases water must in general be sought from much deeper wells than in antiquity. Second, sand dunes are local and never cover more than a fraction of any settlement at a given moment; one would not react to a dune invasion by abandoning a city or village, but by moving house to another neighborhood. Third, Dakhla is well surveyed, and the post-fourth-century pottery is recognized. There is no good reason to think that our perception of occupation is seriously flawed. There are, after all, sites such as Deir Abu Matta, where occupation into the fifth century and perhaps later is known from the ceramic finds. El-Qasr seems to have been inhabited continuously from the third century until the present day.↩#p296
It is possible that the question of security, in this case, affected not so much the security of the oasis itself as that of the roads between the oasis and the valley. If we remember that the flourishing of the oases in Roman times seems likely to have depended very heavily on the ability to export goods in the production of which the oases had an advantage over the valley, we can see that even a modest reduction in the safety of the desert roads could have had a devastating impact on the oases' economic viability.↩#p297
In his book on the oases, Guy Wagner cites a passage of the Christian writer John Moschus (112; PG 87C, 2976–8) in which a number of monks in the oasis were taken prisoner by marauding tribesmen called Mazikes.31 The dramatic date of the episode can be no later than the reign of Tiberius II (578–582). One of these monks was taken to Hibis, the capital of the other part of the Great Oasis (modern Kharga), in order to try to scrape together 24 solidi (one-third of a Roman pound of gold, or about 108 grams) to ransom a group of elderly and ailing monks. But the bishop there could find only 8 solidi, which the Mazikes refused, grabbing another, senior, monk instead. We have no way of assessing how typical this sixth-century vignette is, but it may point to a greater level of insecurity in late antiquity and, eventually, to an inability to sustain a society that depended on the security of its communications with the world of the Nile valley, hundreds of km away.↩#p298
6.4. Oasis society
Roger S. Bagnall
The ostraka of Trimithis do not, at least so far, form a rich body of material for understanding the society of the oasis. Highly formulaic and almost without exception limited to one sort or another of economic affairs, they tell us nothing directly about most aspects of human life. Even the papyri from Kellis are uninformative about such basic questions as demography, marriage, and divorce. Nonetheless, the structural facts of economic life visible in both texts and archaeology force one to confront some essential aspects of oasis life during the Roman period, and to think about the consequences.32↩#p299
At the most fundamental level, oasis society must have been highly unequal, almost certainly more unequal than the society of the Nile valley. The main reason is that the Roman development of the oasis required large amounts of capital deployed in expensive assets owned by private parties, and it was they who controlled the flow of water from the wells and thus the land made green by that water. Other people worked for them. We have little direct evidence of the relationship between the investors—some of whom, at least, were absentee owners—and those who worked their land. From the Kellis Agricultural Account Book it is clear that this relationship was constructed around the payment of rent, thus on a model of landlord and tenant. But the juridical nature of this landlord–tenant situation and how it was represented in legal documents are unknown; there are no land leases preserved among the Kellis papyri, nor do we have any labor contracts. Also, the ʿAin el-Gedida ostraka tend to suggest rent rather than wages as the main payment stream, but we cannot hope to recover any precision about the ways and means.↩#p300
Dakhla thus seems to have had little equivalent to the small and middling landholders who formed the core of Egyptian village society in the valley. This is not to say that the valley did not have its share of landless tenants, but that Dakhla was home to very few, if any, of the historic "royal" farmers, tenants of pharaoh's land who had long-term possession and who probably, in the main, at some point under Roman rule, came to own that land.↩#p301
Before we conjure up too stark a picture of oasis society as a grossly unbalanced barbell, however, we need to turn the spotlight on some other groups. One such group was engaged in animal husbandry, management, and transportation. As is clear from the oasis landscape and the physical relationship of the oasis to the valley, a very large number of donkeys and camels were required for the entire system to function. People had to raise these animals, take care of them, and drive them on their appointed rounds. There were a lot of onelatai, donkey drivers, and kamelitai, camel drivers, in the oases, far more than in most valley communities. Second, agricultural commodities needed to be processed. With food crops, most of this work was surely carried out by the farmers, the same people who raised the crops. But there must also have been some professional millers. Even more demanding was the processing of cotton and probably of some quantity of flax as well. These products in their raw state are bulky and awkward to transport, and flax needs processing soon after harvest. The work needed to be done locally. Spinners and weavers were required.↩#p302
More visible to us in the texts, however, are the managers. These men are the keys to the functioning of the entire system. They use a language of collegiality in addressing one another, but one should not mistake this for an absence of hierarchy. Even the top echelon of oasis society, like Serenos, who was a member of the local civic elite, were estate managers for someone. The absentee Faustianus, son of Aquila, the landlord of the property at the center of the Kellis Account Book, lived in Hibis; whether he was in turn the manager for someone richer still and perhaps resident in the valley, we do not know. A host of lower-level managers turn up in the ostraka from Trimithis, Kellis, and ʿAin es-Sabil. It was they who assured the communication of orders from the main population centers to the hydreumata, the transportation of goods in both directions, and the record-keeping necessary for these large enterprises to thrive. This means, among other things, that there was a substantial middle social stratum that had to know how to read and write fairly well. To produce such men required education.↩#p303
A key characteristic of oasis society must have been absence, created in large measure by the caravan traffic to and from the valley. Each round trip must have meant close to four weeks away from home. It is likely that some men were away half of the time, or more. And the letters from Kellis show, moreover, that apart from those involved in the transportation business, many oasites inevitably spent a considerable amount of time in the valley on all sorts of other business. The traces of these absences—for example, to go and register documents in Alexandria—can be found as well in the archive of the nekrotaphoi, the funerary workers of Kysis, which I have assembled and hope to publish before long. Long absences were part of the price to be paid for the economic advantages I have enumerated. And many families had relatives living in the valley, whom they no doubt visited from time to time.↩#p304
We know little about the consequences for oasis society resulting from this structural situation. Although we move very much into the realm of speculation here, it seems worth adducing the presence in the nekrotaphoi archive from Kysis, in the south of the Kharga Oasis, of several people without patronymics, whose only identification is their mother's name—individuals who are elsewhere often referred to by scholars as apatores ("fatherless"), although that term never occurs in the Kysis dossier.33 Apart from one man who as a slave could not have had a legal father, there are at least five instances of non-paternal identification among the identifiable members of the nekrotaphic families. These stand against twenty-six individuals with patronymics and twelve for whom the parentage is not preserved. The fatherless thus come to 16.1 percent of those whose parentage is attested. By contrast, Herbert Youtie estimated from the Karanis money-tax registers compiled by the collector Sokrates in the second century that the percentage of apatores there was in the range of 2.1–2.6 percent. Some self-consciousness of fatherless status is shown in the nekrotaphic archive, despite the absence of the term apator, by the fact that in every case except that of a woman appearing as a grandmother, the word huios (son) or thugater (daughter) is added before the mother's name, where otherwise a father's name would be given without any relational word; by contrast, metros ("of mother") is used only where a patronymic is also given.↩#p305
This last point is also true of the ostraka from Douch, where the five volumes published to date do yield a number of people with mothers' names but no patronymics; any attempt at quantification there is handicapped by the fact that the ostraka come from a military milieu, where ranks and titles are often given instead of patronymics. A count of the names in the Dakhla documentation, as collected by Klaas Worp in his repertory of names from the oases, and including the first volume of the Trimithis ostraka, yields about 4.64 percent of mothers' names; Trimithis alone gives 5.3 percent, but this difference is probably not significant. A figure around 5 percent seems an upper limit, and the percentages in the second volume of Trimithis ostraka are lower. The numbers thus seem higher than those of the earlier Karanis registers, but much lower than in the fairly small sample of the nekrotaphoi.↩#p306
Interpreting this information, even apart from legitimate questions about the statistical value of such small numbers, is difficult.34 We may reasonably be reluctant to revert to the kind of view that Youtie rightly characterized as too judgmental, seeing illegitimacy as "a natural consequence of the loose morals of a low-class population." But given the date and the social milieu, we can exclude the reasons responsible for the kind of illegitimacy Youtie was most interested in, namely the technical lack of a legal father that resulted from Roman restrictions on legitimate marriage between some groups: the fact that soldiers on active service could not marry, and that the Gnomon of the Idios Logos—an elaborate code regulating legal status—prohibited marriage between some classes of the population. The nekrotaphic archive belongs entirely to the period of universal Roman citizenship after the Antonine Constitution of 212, and after the end of the marriage prohibition for soldiers, under Septimius Severus (ruled 193–211)—hence the termination of the two regulations that Youtie cited as artificial generators of legal illegitimacy.↩#p307
Absence may not be the only aspect of oasis society to come into play, however. Illegitimacy in late antique Egypt, when none of the legal requirements limiting intermarriage any longer applied, is probably in large part the result of concubinage, enduring sexual relationships not formalized as marriage. There are reasons to think that concubinage is in essence a product of social inequality, of power in the hands of some men to enter into such relationships even with women of free-citizen birth, on the strength of economic disparities between the parties. Given the economic stratification that seems likely to have existed in oasis society, it is perhaps not surprising that the better-off, mostly the managerial class, would have been able to use that power to enter into unequal unions, perhaps with women from the tenantry. The men who are identified by their mother's name in the Trimithis ostraka are, as far as we can see, members of that rent-paying class. This may be no coincidence. It seems likely that the inequality of oasis society permeated every aspect of personal relationships.↩#p308
Chapter 6 Notes
1 See Bagnall and Rathbone 2004: 285–8; Sidebotham, Hense, and Nouwens 2008.↩#footnote-6.1
10 Dupras, Schwarcz, and Fairgrieve 2008. The difference between American usage of "corn" to refer to maize and British usage to mean "grain" can give rise to confusion, but maize has never been found in an archaeological context in Egypt.↩#footnote-6.10
11 Wagner 1987: 288, pointing out that millet was still grown in the Great Oasis, especially in south Kharga, in the twentieth century.↩#footnote-6.11
25 See Bagnall and Ruffini 2004. The date at which Trimithis became a polis is not known, and it may have come significantly before the fourth century.↩#footnote-6.25
26 Note Thio (P.Kell. 45), Pmoun Tametra (P.Kell. 41), and a couple of others whose names are not preserved (P.Kell. 8, P.Sijp. 11a).↩#footnote-6.26
27 On the fort at El-Qasr, see Kucera 2012; Gardner 2012.↩#footnote-6.27
28 For a more detailed discussion of the garrison of Dakhla, see Ast and Bagnall forthcoming.↩#footnote-6.28
32 Some issues and observations in this section are drawn from part of a chapter I wrote for a forthcoming book on Kellis edited by Colin A. Hope. For Egyptian society in late antiquity more generally, see Bagnall 1993.↩#footnote-6.32
33 This archive is only partly published to date; I am preparing a complete edition and reedition. About apatores, it should be noted that Herbert Youtie pointed out in his celebrated article on the subject (Youtie 1976) that this word becomes scarcer as the third century goes on, then disappears after 314, by coincidence the end-date of this archive.↩#footnote-6.33
34 For what follows, see in more detail Bagnall forthcoming.↩#footnote-6.34
