Space Syntax and Analysis of Complex Spatial Environments

Structures are built to give order and organization to human activities: to prescribe boundaries as to what is possible and permissible, in both the physical and sociocultural sense. For example, cultural decorum may require certain household activities to be strictly divided by gender, age, or rank, and as a result the architecture of the house itself reflects some of those culturally explicit considerations.¹ Once constructed, the spatial arrangement of the house reinforces these social norms, as the walls and doorways create physical limitations on the movement of people throughout the spatial system.

The specific spatial organization of ancient architectural remains can therefore be used as evidence of past social practices. Space syntax analysis is an analytical method designed to compare modes of access within a given spatial system “on the basis of social content of spatial patterning and the spatial content of social patterning.”²It relies on mathematical probability, calculating the statistical likelihood of anyone entering any given room or space in a system, whether in order to perform an activity in that space or to pass through to another space. This statistical likelihood in turn helps to identify socially prescribed attitudes of privacy. For example, a highly connected room, with many entrances and exits, serves as a major thoroughfare, and any activity occurring in such a space is likely to be seen or even interrupted by those passing through. This results in a relatively public characteristic for the space in question. In contrast, a dead-end room with only one way in and out will be more private, as there will be statistically less chance that anyone will need to enter the space.

Space Syntax values can be local or global measures. While local measures are calculated based only on directly adjacent spaces (i.e., a single room away from the given space), global measures account for the entire spatial system, which often includes a multitude of possible paths through the network. As every space has an effect, great or small, on the global system, the identification and calculation of all the relationships in the system is best accomplished by a computer program.³ However, there are a few values that are more easily obtained by analyzing maps and justified access plans (see figure 2.1). The major spatial values discussed here are as follows:

• Depth: a measure of how many other rooms or distinct spaces must be crossed in order to reach a given space. This measure is often synonymous with “threshold depth,” and most commonly is calculated with respect to a single structure and its external environment, such as rooms in a house versus the outdoor world. Justified access plans can visualize depth, showing increasing depth as one proceeds into a house interior. These access plans are perhaps the most simplified and frequently used (though very effective) method for exploring concepts of privacy within ancient houses.⁴
• Connectivity: a simple count of how many spaces are directly connected to a given space (i.e., a local measure). High connectivity suggests a space is an important hub of movement throughout the spatial system, and therefore likely to see more public use than private.
• Symmetry: a global measure of connectivity, calculated as the average number of thresholds or boundaries that must be crossed to reach a given space from any starting point within the system.⁵ The more complex value of real relative asymmetry, or RRA, adjusts for the size of the entire system, permitting quantitative comparisons between separate systems.⁶ As RRA values increase, a system can be said to be less well-connected overall.
• Relative Ringyness: a measure of how many alternative routes exist to reach a specific space, found by counting the number of existing rings that pass through a given space and dividing that number by (p – 1), where p equals the total number of spaces in the system. If there is only one path to a room, as in the “clustered” plan shown above, that space will have a low measure. The relative ringyness of an entire system can be found by counting the total number of rings in the system and dividing by (2p – 5). High levels of ringyness are linked to higher distribution of traffic, lower rates of congestion, and are thus often considered more efficient than linear, non-ringy systems.

Interpreting Archaeological Data with SSA

Space syntax analysis originally developed out of studies of modern architecture and urban planning, and therefore relies on the certainty that the dataset in question is completely known.⁷ In contrast, archaeological datasets always introduce a degree of uncertainty: ancient structures are rarely perfectly preserved, and any spot of low preservation or uneven excavation/recording would introduce great potential for error in the ensuing calculations. For instance, an area of “missing” wall might indicate the presence of an original doorway, but also might be the result of decay and collapse of the structure following its abandonment. In addition, global studies require large continuous areas: not only individual structures, but also the streets and paths between them. For this reason, studies of larger networks of towns and neighborhood analysis have been largely limited to a few well-preserved and excavated examples, including ancient Pompeii, Herculaneum, and Ostia.⁸

Individual ancient structures have proved better suited for syntactical studies, and houses in particular have proved a popular subject.⁹ The syntactical analysis of ancient Roman houses has proved particularly important, as it provides comparative evidence for the rich documentary sources of information on Roman cultural attitudes toward domestic space. For example, the Roman domus is traditionally interpreted as highlighting the house owner’s social status via intentional display of wealth, rank, and cultural taste within the central rooms, the atrium, and peristyle.¹⁰ Syntax analysis of the typical Roman floor plan (see figure 2.2) confirms the importance of these distributive hubs through which most of the household traffic ran.¹¹

Space Syntax Analysis of Ancient Karanis

Roman Karanis is well suited to syntax analysis on both local and global levels: the University of Michigan’s excavation of the site from 1924 to 1935 uncovered more than 180,000 square meters of domestic neighborhoods dating from the first to late third centuries CE. The major publication, an architectural report by Elinor Husselman, included top plans for most excavated areas, drawn into three major phases according to the stratigraphic levels of the town.¹² This publication provided scholars with the opportunity to begin syntax analysis of the ancient housing.¹³

However, the published information only represented a small percentage of the data recorded by the Michigan team; the architectural report only included detailed plans for thirteen mud-brick structures, representing less than 3 percent of the total excavated by the Michigan team. In 2002, a joint project between the University of California, Los Angeles; the Rijksuniversiteit Groningen; and the University of Auckland began to survey, excavate, and conserve the site.¹⁴ As the primary investigator of the Karanis Architecture Project since 2008, I have worked to combine published, archival, and modern survey results to create a fully digital spatial database of the ancient site.¹⁵ This work has greatly expanded what is known of the city, including more than one thousand domestic properties from over 524,000 square meters of area—almost tripling what was recorded and published by Michigan (see figure 2.3). In addition, associated pottery and other archaeological finds have firmly expanded the occupation dates for Karanis beyond the third century and into the sixth century CE.¹⁶

As a result, there are large portions of Karanis that provide a suitable dataset for SSA. One in particular, designated Karanis Central (KAC), includes more than twenty-six thousand square meters of continuously excavated area. The original Michigan excavation identified three distinct phases of occupation: of these, Layer C is the oldest, dating roughly from the late first to early second century CE. Layer B is mid-second to mid-third, after which there was a period of abandonment before the poorly preserved A layer, which probably dates from the late third century until the abandonment of the site sometime in the sixth century.¹⁷

Each property within KAC is clearly defined by perimeter walls, enclosing the open space of the courtyards as well as any associated structures. More than 84 percent of the properties include a house. This has been used as the major criterion for categorizing them as domestic, as opposed to properties that have a more purely commercial function. In all other respects domestic properties share many similarities with economic structures: granaries and dovecotes can be found in properties with houses as well as those that lack associated dwellings. Houseless commercial properties may have been privately owned, but it is also possible that they were under state control and not private at all.¹⁸

Houses in Karanis are easy to identify by their construction characteristics, which remain consistent throughout the C and B layers. They are invariably freestanding buildings, with compact floor plans and walls built to support the weight of upper stories. Property C45 is a typical example of a C layer, second-century domestic property at Karanis (see figure 2.4): the lot is only eighty-three meters square, slightly below the average of ninety. The house itself shares no walls with any of the other structures on the lot. The floor plan only has two rooms, each located off the central stairwell structure that leads to the other stories of the house. Even while Karanis houses are not often preserved today above ground-floor level, the ubiquitous staircase proves the existence of additional stories, or at least a flat roof space that could also serve as an activity area. Ptolemaic and Roman-era documents describe mud-brick houses that stood as many as eight or more stories;¹⁹ the highest preserved house at Karanis was at least three stories, including the ground floor. Basements were also common when the bedrock level allowed for them, and were used for storage of domestic and economic goods.²⁰

The justified access plan immediately makes clear just how essential the stairway is for all movement throughout the house. Most of the rooms have a connectivity of only 1: that is, they are dead-end spaces. These could be used for either private or social activity with little chance of casual interruption, as they have no through-access. Even the upper story rooms are separate cells, dependent on the stairway for access. In contrast, the stairway passage is the highly connected, well-integrated hub of the system. This single central space also keeps overall threshold depth low; every room is easy to reach, with no single space being significantly “further” in threshold depth from the others.

At first the Karanis house appears to reflect a similar “clustered” access plan to the Roman domus, which might initially seem to suggest that both house types have similar functions in social display of the homeowner’s status. Yet compared to the wide-open atrium at the center of the domus, the Karanis house clusters around the restrictive space of the stairway: a narrow passage that is rarely more than 0.6 meters wide. Two people cannot occupy the space at once, or even pass each other at all in their comings and goings, and yet even important guests would be required to squeeze through this space in order to reach any room suitable for hosting company. The physical limitations of the house, both in terms of scale and in syntax, seem designed to discourage substantial social interaction and to create a highly controlled, enclosed, and private interior.

It is possible that the rationale behind such restrictive house structures is due to more than just social considerations: environmental factors are also at play. An enclosed mud-brick house is uniquely suited to the temperature swings of desert climates, staying cool in the day and releasing heat during the nights.²¹ Another serious issue was sand: the Egyptian Fayum towns, lying on the borders of habitable land, faced constant problems from windblown sand and other debris blowing in and filling every possible space.²²

However, there are other indications that privacy and control were intentional aspects of Karanis house construction: the houses are all completely freestanding, independent buildings. While courtyards frequently share perimeter walls with neighbors, houses are never built adjoining other structures.²³ Documentary papyri from Hellenistic and Roman Egypt demonstrate how the structural autonomy of such houses could be considered crucial and was protected in formal agreements between neighbors in times of construction. For example, British Museum Papyrus 10524 preserves a case from 283 BCE Luxor, where a woman named Tahib swore to her neighbor Pleehe that her construction activities would not alter his walls, even though her own house would abut his own house.²⁴ The document is a social contract written with formal legal language, including the monetary penalty and a list of sixteen witnesses to the contract. The papyrus and other surviving documents like it emphasize the importance placed on private property rights during the era, and how the autonomy of the house itself was of the utmost importance.

Private Space and Social Interaction: Courtyard and Economic Activities

While Karanis house interiors may have been highly protected and private, they were only a fraction of the total domestic space: every property had some type of unroofed courtyard, its area enclosed and defined by a perimeter wall or series of abutting wall segments. Unlike house walls, these were frequently shared with neighboring properties, but as they were not load-bearing they could be altered or repaired quite easily without threatening the structural integrity of either neighbor’s buildings.

In many ways courtyard space seems to have taken priority over the house itself, as the average courtyard space is at least half of the total lot area. The Karanis house structure makes this arrangement possible, as the compact building footprint maximizes outdoor space. Even the largest properties maintain these multistory plans instead of the horizontal sprawl of, for example, a Roman villa.

In contrast to the standardized floor plans of the Karanis houses, there seems to be no preferred or consistent layout for courtyard spaces: depending on the size and dimensions of the lot, courtyards can vary from wide, undifferentiated spaces to complex networks of passages and outbuildings. Similarly, while interior doors and passageways are narrow and controlled (0.6 meters wide), connections between courtyard spaces are frequently much wider: see, for example, the pathway between courtyard spaces in figure 2.4. In fact, it can be difficult to distinguish between individual courtyard spaces, as there are few formalized thresholds between areas.

In terms of syntax analysis, the courtyards could be considered an extremely distributive clustered plan: a basic open space with many options for movement to and around other loosely defined activity areas. Perhaps the most important SSA feature of Karanis courtyards is that each one has direct access to the public street system, independent of the house. This feature is obviously practical in allowing traffic to bypass the restricted house interiors, permitting freer movement. This arrangement also increases the public nature of the domestic courtyards, making them accessible not only to household inhabitants but also to neighbors and potentially even strangers. In contrast to the restricted-access house interiors, access to courtyards from the public street was rarely controlled via physical barrier: doors or gates are very rarely attested. Therefore, it seems quite likely that these routes were left intentionally “open,” allowing for people outside the immediate household to access the courtyard.

The reason for this accessibility likely lies in the function of the courtyard itself as a space for household economic activity. Excavation of the Karanis courtyards revealed special installations and dedicated activity areas, including threshing floors, flour mills, and cooking stoves.²⁵ Most common of all are bread ovens: by no means did every property have its own, but they were well-distributed throughout the neighborhoods, so each house at least had potential access to one nearby (see figure 2.5). This distribution pattern strongly suggests that many activities associated with food preparation were highly social. Such an arrangement has ethnographic parallels from Pharaonic Egypt, where baking was commonly shared among households, thus optimizing the use of fuel.²⁶ Communal food preparation activities are inherently social encounters of some duration, increasing the likelihood that individuals from distinct households will interact while working side by side. This behavior has the ultimate effect of creating close neighborhood ties that would be maintained through repetition over time.

Moreover, many adjacent properties have direct access between their courtyards—that is, interproperty access points (figure 2.6). In the mid-second century of Layer C, more than a quarter of all properties shared access with an adjacent neighbor, either through a small gate or a simple opening between the property walls. It is impossible to reconstruct the specific social connections between the inhabitants of such properties—perhaps some represent extended family relationships. However, neighbors who were initially strangers could have grown into both economic and social alliances through sharing activity space. Again, bread ovens provide crucial evidence for specific activities that likely united these partnerships: more than 77 percent of these conjoined properties have oven access.

Private Space in the Public Realm: The Use of Shortcuts to Expand the Street System

So far we have considered access to private properties in order to share workspace for social and economic domestic activities. However, there are also more temporary, transient types of access to consider: How often was private property used for wider, more global access through Karanis? The existence of interproperty routes creates the possibility for through-travel in a way that could provide “shortcut” alternatives to the public street system.

The Karanis street system seems designed to prevent the need for such routes—like many other towns of the Greco-Roman Fayum and indeed the eastern Mediterranean, it was built according to a modified Hippodamian grid.²⁷ The intersecting Karanis streets created a network of connected paths, with multiple options for travel along syntactic rings. Each ring encircles a block of private properties, each one directly connected to the public street system and therefore independent of its neighbors for free movement through the town. For this sample neighborhood of Karanis, eight complete rings can be identified, with each individual street connecting to an average of roughly eight to ten private properties. The streets are therefore highly connected distributive hubs of access, controlling movement to private properties much as the stairwells controlled movement to other rooms in the domestic house plans.

However, the great freedom of movement and path choice was even further increased when one includes the interproperty access routes. These create even more potential traffic rings: the eight complete traffic rings created by the town streets alone expand to forty-one when the routes through private properties are included (see figure 2.7).

Obviously, many of these routes are theoretical rather than practical. We can dismiss almost any that run through house interiors: even if the lengths (both metric and syntactically speaking) are “shorter,” the physical limitations of narrow passageways make these routes too restrictive for real traffic. However, some domestic courtyards provide legitimate “shortcuts.” Property C86 (see figure 2.8) is an excellent example of this: it probably dates to the first century, when the area was the northern edge of Karanis occupation, but successive years of town expansion led to it being surrounded to the north and east.²⁸ As other properties encroached, public access became restricted and C86 posed a serious obstacle to the flow of traffic between two large blocks. Perhaps because of social pressures to keep traffic circulating within the expanding neighborhood, the owner of C86 seems to have allowed some of the private property to be used as a virtual roadway, running east to west so it joined two major streets. This completes a syntactic ring with the public system and allows a shortcut: an 18-meter trip across instead of the 104-meter southern route along the official streets.

Even though it provided such access, the space remained part of the private courtyard property: instead of optimizing control over the rest of the property by constructing a privacy wall along the south edge of the ersatz street, the area remained open, suggesting trust that users would stick to the shortest path rather than wander through the courtyard. This openness is particularly interesting as C86 is one of the largest and most economically complex properties of KAC: courtyard installations include bread ovens, an olive press, and several large stone granary bins. Notably, these are all located away from the “path” portion of the property, kept private and nonaccessible through both physical distance (they are not along the shortest, most expedient route) and by obstacles: actual gates and doors that could be bolted and locked. This arrangement demonstrates the careful negotiation the property owners must have made between social cooperation and protecting their own private interests.

It is crucial to realize that interproperty routes were created and maintained through the agreement of the landowner(s) and only lasted for as long as they were considered beneficial. The archaeology at Karanis indicates that shared property-access arrangements were far from permanent, as the overwhelming trend in third century adaptations of older architecture is to block up shared access points. Thirty percent of the shared interproperty routes from the C layer were in fact blocked by the mid-third century B layer, and by the end of the third century, 62.5 percent of all original routes were blocked.

The decline in shared access seems to coincide with socioeconomic changes, including depopulation and especially economic downturn.²⁹ The situation may have caused inhabitants of Karanis to change many of their own domestic activities, and to protect private economic interests at the expense of broader social cooperation. Evidence of this can be observed in the change in distribution of bread ovens (see again figure 2.6). In the B layer, fewer households have direct access to an oven, or are even in proximity to an oven, and the ovens that remain are clustered onto properties that seem more like large-production bakeries than domestic courtyards. These properties often lack traditional house structures entirely, and they have highly controlled access points, emphasizing their economic rather than domestic function. The evidence suggests that food production practices may have undergone a widespread cultural shift away from household cooperation to more focused commercial endeavors.

The closing of a privately owned access route was likely an unpopular move, but the action could not be legally stopped: in ancient property law, occupants had almost complete legal authority over who could access and use their private land, as long as their activities did not obstruct the fair use of the streets or of other people’s property.³⁰ In fact, the right to privacy in domestic space was considered so essential under Roman law that even landlords could be denied access by their tenants.³¹ So when occupants exerted these rights, blocking private access, the civic government had no obligation (or even authority) to get involved, unless it was a threat to public access as well.

Instead, the inhabitants of Karanis and other communities of the time would have relied on social agreements to give and receive access to private land. These could have taken many forms and varied widely depending on the relative statuses of those involved. For example, a property owner might have allowed casual foot traffic through his courtyard, but no carts or beasts of burden. Even more likely, access could be limited to known individuals—given explicitly rather than assumed for a broader public.

While many access agreements may have been “handshake deals” between individuals that left no lasting trace, papyri of the time preserve some examples of more formalized arrangements. The Greek-language documents employ the phrase “eisodos kai exodos” in referring to access into and out of private property.³² Such clauses were often included in bills of sale, seen as servitudes on the property in perpetuity—to be honored by later generations of owners.³³ Papyri could also specify individual users, even which paths were to be made available and what areas were off-limits.³⁴

These documents were codified forms of social agreements, created with the formalized language of legal documents of the time, some with accompanying witnesses and monetary penalties for infractions.³⁵ They demonstrate the serious social authority given to interpersonal contracts, and the lengths individuals would go to in order to secure their property rights and their access over time.

The Privatization of Public Space: Encroachment and Annexation

While locals were maneuvering to secure through-access to private land, property owners and occupants were also frequently shoring up their bargaining positions and increasing their own authority, not only over their own land but over adjacent public spaces as well. By the mid-third century CE, the public street system itself was facing serious encroachment from private construction. Most of these B-layer features can be characterized as simple structures attached to the houses like steps and windscreens. These privately built features were often considered necessary for facilitating access to houses, especially as the street level of Karanis rose over the years.³⁶

Such encroachments into public street space are not unique to Karanis, of course: studies of Pompeii and Herculaneum show benches, steps, and ramps that restrict street space in the name of providing access to private houses.³⁷ The frequency of the archaeological finds suggests such encroachments were commonly tolerated. However, they were also obstacles to public movement, narrowing the width of streets to points that could restrict cart- or even foot-traffic and cause serious bottlenecks to the neighborhood systems.³⁸ As such, they were nuisances to public order, and technically illegal.

In its most extreme form, street encroachment ended through-traffic entirely and effectively annexed public space to private properties. Barriers across whole streets converted the public space of a street into a restricted linear route: a dead end or a cul-de-sac, with higher privacy for the property or properties at the end of the street segment. As the third century progressed, many of the Karanis streets were transformed, at least in part, to private courtyard space, either enhancing existing properties or creating new ones altogether.

Roman law was clear on the subject: public space was protected and must be left available to all. Local laws and settlement charters repeatedly emphasized the importance of maintaining street spaces.³⁹ Yet despite the law, the archaeological record preserves many examples of street encroachments and obstacles that were allowed to persist, suggesting it was rare to succeed in getting one removed once constructed. Scholars have suggested this is direct evidence of decreasing civic power over time, particularly in the eastern Mediterranean during the later Roman Empire. However, as we have already seen, encroachment was hardly limited to weakly controlled provinces, but could be found in Italian towns during the early imperial period.⁴⁰ This suggests that civic authorities had little interest in enforcement, even when their power was strong.⁴¹ It is also possible that certain acts of encroachment or annexation were allowed because of monetary incentives: either to the magistrates themselves, as bribes, or as semilegal “sales” of public land.⁴²

For Karanis and the other towns of the Fayum, the Late Roman phases are often said to have “lacked internal political and social structures as much as their villages lacked spatial articulation.”⁴³ However, these arguments often focus on the theory that Karanis and other villages of the Fayum were depopulating and even being abandoned by the late third century. In fact, the archaeological evidence now proves that such villages often survived well into Late Antiquity—Karanis itself until at least the early sixth century.⁴⁴ The very persistence of the Fayum towns, despite the lack of ideal syntax plans, proves they were still functional, if not optimal, systems, or else existing legal or social pressures would have rectified them.

It is also possible the lack of heavy administrative oversight was viewed as positive rather than neglectful. It gave individuals and local communities more direct agency in managing their public access systems and their own property. And as the public street system grew more restricted by encroachment and annexation, interproperty access routes became even more important factors for travel within neighborhoods. Occupants of important properties like C86 would have faced huge social pressures to keep them open, at the expense of their own privacy and even potentially their economic security. However, they also stood to benefit socially and economically from strategic negotiation of access rights. In fact, the syntactic design of domestic properties makes this feasible, separating and insulating the private house interiors and economic areas away from the more open, accessible space used by interproperty access.

In fact, as we have seen before with eisodos contracts, social negotiation was potentially even more effective than legal action as a means of managing conflict. Builders of street obstructions were probably seen as defying not only the law, but also social conventions of propriety and fairness, and therefore faced negative opinions and even censure from the public.⁴⁵ The power of social pressure and public censure was also recognized within legal codes across the Roman Empire: the Tabula Heracleensis, for example, compelled residents not only to fix roads they damaged, but allowed the charges against them to be posted publicly both in the forum and at the property itself, publicly identifying them as lawbreakers.⁴⁶ This implies that public censure and shame may have motivated change more efficiently than pure legal proceedings, effectively picking up the slack caused by negligent aediles and other magistrates. The legal system could also be used as leverage in private negotiation, as the threat of lengthy lawsuits and official interference might have driven more expedient solutions derived by social or interpersonal agreements.

Changes to the Access System of Karanis: Practical and Social Consequences

As a result of all this action by private individuals, the entire public street system of Karanis changed drastically over a period of a hundred years. Routes between large town blocks became impassable, essentially combining and creating huge areas that had to be navigated around (see again figure 2.8, and figure 2.9); the average block perimeter increased eighteen meters by the mid-third century, and by the early fourth century A layer, so many former public routes had been converted to private land that none of the original streets remained wholly intact.⁴⁷ A path that had formerly been one long street was replaced by a series of short segments angling around the private architecture; the combined effect was to increase mean depth for the system. In the mid-second century C layer, KAC was composed of thirty-five street spaces; by the mid-third century B layer it had risen to fifty-one.

In comparing SSA values between the two time periods (see figure 2.10), we must therefore consider the different scales of the systems. Of all the calculated values, only RRA and relative ringyness directly account for these differences. The apparent increase in mean depth over time is particularly misleading, because the large increase in total spaces has driven up the average. In comparison, the syntactic length (and therefore mean depth) of the traffic rings has markedly increased by the B layer: C-layer rings were simple, often consisting of only the four or five streets that enclosed a block. In contrast, B-layer rings are oblong paths of many streets (see again figure 2.8), twisting and turning around irregular conglomerations of houses. Relative ringyness of the overall system is in fact one of the major changes in Karanis, decreasing from nearly 30 percent to only 10 percent in a century.

These changes to access created and, in turn, reinforced changes in the socioeconomic cohesion of Karanis. With fewer ringy options, traffic congestion would have increased on the remaining routes. Quiet neighborhood streets could be transformed into thoroughfares, especially if they were wide enough for carts and commercial traffic. At the same time, the system became more complex in terms of wayfinding, as the complicated, twisting routes might be unfamiliar to infrequent users. Locals would know the way, but outsiders to the neighborhood—even those from other parts of Karanis itself—might have trouble navigating the winding paths. Commercial traffic may therefore have settled on wide and easily traversed routes rather than the shortest paths. The existence of interproperty routes would provide additional options to locals who had permission to use them. As a result, some neighborhoods became even more “private,” a modified spatial “domain of the inhabitant,” resulting in very localized communities within the larger town.⁴⁸

In space syntax terms, B-layer Karanis was an inefficient system, and one developing further and further away from the ideal. The reliance on interpersonal social negotiation to manage access only operated on a case-by-case basis, with little apparent consideration for the long-term system that stronger civic enforcement could have provided. Yet it does not necessarily follow that the ancient inhabitants saw their streets as inherently disorderly or completely dysfunctional. As we have seen throughout the Roman world, maintaining an ideal syntactic system was simply not a priority. While private construction that caused minor restriction to public access was widely tolerated, both legal and social mechanisms existed to effect necessary changes if the system was ever judged to be truly threatened or in need of intervention. The results were a flexible strategy for maintaining public access and a street system that functioned for six centuries: in terms of sheer longevity, it must be termed a success.

While large-scale social organization drove the physical development of the Karanis streets, individuals could work to improve their own access opportunities, by obtaining permission to use private interproperty routes that provided advantageous shortcuts and detours. Strategic inhabitants would carefully manage relationships within their neighborhoods to maintain such access. This in turn strengthened the social cohesion of localized communities within Karanis and assured that the resulting systems of access functioned smoothly. Rather than characterize Karanis as a town on the edge of chaos and abandonment as Late Antiquity approached, it is crucial to recognize the archaeological remains on both private and public property as evidence of a competent system of social organization.

Notes

1See Bourdieu 1977; Blanton 1994.

2Hillier and Hanson 1984, xi.

3See Turner 2004.

4Grahame 2000; Alston 2002; Spence 2010.

5Grahame 2000, 34.

6Hillier and Hanson 1984, 112.

7Hillier and Hanson 1984.

8See Perring 1991; Weilguni 2011; Poehler 2017.

9See Spence 2005; Nevett 1999; Grahame 2000.

10See Clarke 1991, 9–10; Wallace-Hadrill 1988.

11See Grahame 2000.

12Husselman 1979.

13See Alston 2002, 55–56; Bagnall 1993.

14See Cappers et al. 2013; Barnard et al. 2015; Wendrich et al. 2013.

15See Simpson 2014; Wendrich et al. 2013.

16Simpson 2014; Barnard et al. 2015.

17These dates were published by the original Michigan team (Boak and Peterson 1931, and especially Husselman 1979), but relied heavily on papyrological and numismatic evidence, without considering any pottery seriation. Efforts to adjust the chronology are ongoing, and include Pollard 1998; Gupta-Agarwal 2011; Cappers et al. 2013; Barnard et al. 2015.

18See Husselman 1952 and 1953; Tassinari 2009.

19Nowicka 1969, 108; Husson 1983; Lehman 2013.

20Husselman 1979.

21For technical studies of the properties of mud-brick architecture, see Jerome et al. 1999; Kemp 2000. For ethnoarchaeological accounts, including other regions of the Near East, see Oates 1990; Van Beek et al. 2008; Homsher 2012; Correas-Amador 2013; Lehmann 2013; Love 2013.

22See further Davoli, chapter 1. For the specific effect of windblown debris at Karanis, see Husselman 1979, 8, 29, 67.

23Cf. Alston 2002, 53, who seems to be describing perimeter walls, which were shared, in contrast to the house itself. Freestanding and independent houses are rare in many crowded Roman cities, and surviving legislation suggests the management of shared constructions was complex (Taubenschlag 1955; Owens 1991, 168). In contrast, houses of the Roman Fayum are usually “self-contained” (Davoli 2011, 81); for Karanis itself, there is only one known example of a house that shared an exterior wall with another property, and this was an alteration made a century after initial construction (Simpson 2014, 187).

24“And I will build my house from my southern wall to my northern wall up to thy wall, provided that I do not insert any timber in it …. If I fail to act according to everything aforesaid I will pay thee 5 silver pieces.” P BM 10524; trans S. Glanville 1939.

25Husselman 1979.

26Samuel 1999.

27Owens 1991; Gates 2003; Davoli 2011.

28Husselman 1979, 12.

29Boak 1955 and 1959; Geremek 1969; Keenan 2003.

30Owens 1991, 156–67, 168; Saliou 1994.

31Taubenschlag 1955, 361.

32Taubenschlag 1955; Husson 1983, 65–72.

33Taubenschlag 1955, 259; Kelly 2011.

34Taubenschlag 1955, 257.

35For discussion on their status as legal documents, see Taubenschlag 1955; Youtie 1971; Kelly 2011. Kelly 2011, 71, studying the different levels of the courts that petitions reached before being withdrawn or dismissed, suggests that the contracts were equally useful outside the courts as leverage in interpersonal negotiations. See also PMich V.226.

36Husselman 1979, 8, 29.

37Saliou 1994; Jacobs 2009. Hartnett (2017, 211) notes that many of the encroachments in Pompeii and Herculaneum were in fact not strictly necessary for access, thus throwing into question any legal grounds for their existence.

38Simpson 2014, 157–61.

39Lex Coloniae Genetivae Iuliae (FIRA 21) and see Hartnett 2017, 211; Crawford 1996, 393–94. The Tabula Heracleensis ll. 68–67, Lex Iulia Agraria 4 (FIRA 12) in Campania, and Spain’s Lex Coloniae Genetivae Iuliae 104 in Spain (FIRA 21) give magistrates the power to intervene in cases of obstruction; see also Hartnett 2017, 211. For details on the limitations on porticoes and benches, see CJ VIII, 11, 20 [AD439] and discussion in Saradi 2006; Jacobs 2009, 205.

40Saliou 1994.

41In Rome, public access management was the job of the aedile, but holders of that office were notoriously unmotivated to fulfill their duties. See Cass. Dio 49.43.1 on the astonishing productivity of M. Agrippa’s time as aedile as the exception to the general rule.

42Jacobs (2009, 223) notes that in the case of annexation, the income generated by taxing the new private property could be seen as a benefit for municipal coffers.

43Bagnall 1993, 114.

44For third-century depopulation and “abandonment” of Fayum towns, see Boak 1955 and 1959; Van Minnen 1995; cf. Keenan 2003, Cappers et al. 2013, Simpson 2014; and Barnard et al. 2015.

45Hartnett 2017, 136–37.

46Tabula Keracleenisis 11.32–45, and see Hartnett 2017, 125–26.

47Simpson 2014. See also Husselman 1979, 26.

48Hillier and Hanson 1984, 17.

Works Cited

Alston, R. 2002. The City in Roman and Byzantine Egypt. London: Routledge.

Bagnall, R. 1993. Egypt in Late Antiquity. Princeton: Princeton University Press.

Barnard, H., W. Wendrich, B. Nigra, B. Simpson, and R. Cappers. 2015. “The Fourth Century CE Expansion of the Graeco-Roman Settlement of Karanis (Kom Aushim) in the Northern Fayum.” Journal of Field Archaeology 84: 84–100.

Blanton, R. 1994. Houses and Households: A Comparative Study. New York: Plenum Press.

Boak, A., and E. Peterson. 1931. Karanis: Topographical and Architectural Report of Excavations During the Seasons 1924–28. Ann Arbor: University of Michigan Press.

Boak, A. 1955. “The Population of Roman and Byzantine Karanis.” Historia 4 (2): 157–62.

Boak, A. 1959. “Egypt and the Plague of Marcus Aurelius.” Historia 8 (2): 248–50.

Bourdieu, P. 1977. Outline of a Theory of Practice. Cambridge: Cambridge University Press.

Cappers, R., E. Cole, D. Jones, S. Holdaway, and W. Wendrich. 2013. “The Fayyûm Desert as an Agricultural Landscape: Recent Research Results.” In Das Fayyûm in Hellenismus und Kaiserzeit: Fallstudien zu multikulturellem Leben in der Antike, edited by C. Arl and M. Stadler, 35–50. Wiesbaden: Harrassowitz Verlag.

Clarke, J.R. 1991. Houses of Roman Italy 100 B.C.–A.D. 250: Ritual, Space, and Decoration. Berkeley: University of California Press.

Correas-Amador, M. 2013. “Ethnoarchaeology of Egyptian Mudbrick Houses: Towards a Holistic Understanding of Ancient Egyptian Domestic Architecture.” PhD diss., Durham University.

Davoli, P. 2011. “Reflections on Urbanism in Graeco-Roman Egypt: A Historical and Regional Perspective.” In The Space of the City in Graeco-Roman Egypt: Image and Reality, edited by E. Subias, P. Azara, J. Carrusesco, I. Fiz, and R. Cuesta, 69–92. Tarragona: Institut Calalà d’Arquelogia Clàssica.

Gates, C. 2003. Ancient Cities: The Archaeology of Urban Life in the Ancient Near East and Egypt, Greece, and Rome. London: Routledge.

Geremek, H. 1969. Karanis: Communauté rurale de l’Égypte Romaine au IIe-IIIe siècle de notre ère. Archiwum Filologiczne, 17. Wrocław: Zakład Narodowy Imienia Ossolińskich Wydawnictwo Polskiej Akademii Nauk.

Glanville, S.K. 1939. Catalogue of Demotic Papyri in the British Museum. Vol 1. A Theban Archive of the Reign of Ptolemy I, Soter. London: Trustees of the British Museum.

Grahame, M. 2000. Reading Space: Social Interaction and Identity in the Houses of Roman Pompeii. Oxford: Archaeopress, British Archaeological Reports.

Gupta-Agarwal, S. 2011. “The Final Curtain Call: The Abandonment of Karanis in Light of Late Roman Amphorae.” Coptica 10: 6–76.

Hartnett, J. 2017. The Roman Street: Urban Life and Society. Cambridge: Cambridge University Press.

Hillier, B., and J. Hanson. 1984. The Social Logic of Space. Cambridge: Cambridge University Press.

Homsher, R.S. 2012. “Mud Bricks and the Process of Construction in the Middle Bronze Age Southern Levant.” Bulletin of the American Schools of Oriental Research 368: 1–27.

Husselman, E. 1952. “The Granaries of Karanis.” Transactions and Proceedings of the American Philological Association 83: 56–73.

Husselman, E. 1953. “The Dovecotes of Karanis.” Transactions and Proceedings of the American Philological Association 84: 81–91.

Husselman, E. 1979. Karanis Excavations of the University of Michigan in Egypt 1928–1935: Topography and Architecture; A Summary of the Reports of the Director, Enoch E. Peterson. Ann Arbor: University of Michigan Press.

Husson, G. 1983. Oikia: Le vocabulaire de le maison privée en Egypte d’après les papyrus grecs. Paris: Publications de la Sorbonne.

Jacobs, I. 2009. “Encroachment in the Eastern Mediterranean Between the Fourth and the Seventh Century AD.” Ancient Society 39: 203–43.

Jerome, P., G. Chiari, and C. Borelli. 1999. “The Architecture of Mud: Construction and Repair Technology in the Hadhramaut Region of Yemen.” Association for Preservation Technology International Bulletin 30: 39–48.

Keenan, J. 2003. “Deserted Villages: From the Ancient to the Medieval Fayyum.” Bulletin of the American Society of Papyrologists 40: 119–39.

Kelly, B. 2011. Petitions, Litigation, and Social Control in Roman Egypt. Oxford: Oxford University Press.

Kemp, B.J. 2000. “Soil (including Mud-Brick Architecture).” In Ancient Egyptian Materials and Technology, edited by P.T. Nicholson and I. Shaw, 73–103. Cambridge: Cambridge University Press.

Lehmann, M. 2013. “Skylines, Bridges and Mud in the Delta and Elsewhere. A Comparison of Egyptian and Yemeni Tower Houses.” Paper presented at the Delta Survey Workshop, Cairo, March 22–23.

Love, S. 2013. “The Performance of Building and Technological Choice Made Visible in Mudbrick Architecture.” Cambridge Archaeological Journal 23: 263–82.

Nevett, L. 1999. House and Society in the Ancient Greek World. Cambridge: Cambridge University Press.

Nowicka, M. 1969. La maison privée dans l'Égypte ptolémaïque. Wrocław: Zakład Narodowy im. Ossolińskich.

Oates, D. 1990. “Innovations in Mud-Brick: Decorative and Structural Techniques in Ancient Mesopotamia.” World Archaeology 21: 388–406.

Owens, E.J. 1991. The City in the Greek and Roman World. London: Routledge.

Perring, D. 1991. “Spatial Organization and Social Change in Roman Towns.” In City and Country in the Ancient World, edited by J. Rich and A. Wallace-Hadrill, 273–93. New York: Routledge.

Poehler, E. 2017. The Traffic Systems of Pompeii. New York: Oxford University Press.

Pollard, N. 1998. “The Chronology and Economic Condition of Late Roman Karanis: An Archaeological Reassessment.” Journal of the American Research Center in Egypt 35: 147–62.

Saliou, C. 1994. Les lois des bâtiments: Voisinage et habitat urbain dans l’Empire Romain; Recherches sur les rapports entre le droit et la construction privée du siecle d’Auguste au siecle de Justinien. Beyrouth: Institut Français d’Archeologie du Proche-Orient.

Samuel, D. 1999. “Breadmaking and Social Interaction at the Amarna Workman’s Village, Egypt.” World Archaeology 31 (1): 121–44.

Saradi, H. 2006. The Byzantine City in the Sixth Century: Literary Images and Historical Reality. Athens: Society of Messenian Archaeological Studies.

Simpson, B. 2014. “Neighborhood Networks: Social and Spatial Organization of Domestic Architecture in Greco-Roman Karanis, Egypt.” PhD diss., University of California, Los Angeles.

Spence, K. 2010. “Settlement Structure and Social Interaction at El-Amarna.” In Cities and Urbanism in Ancient Egypt, edited by M. Bietak, E. Czerny, and I. Forstner-Müller, 289–98. Vienna: Verlag der Österreichischen Archäologischen Instituts in Kairo.

Tassinari, C. 2009. Il Thesauros di Bachkias: Rapporto Definitive. Bologna: La Mandrago Editrice.

Taubenschlag, R. 1955. The Law of Greco-Roman Egypt in the Light of the Papyri. 332. B.C.–640 A.D. Warsaw: Panstwowe Wydawnictwo Naukowe.

Turner, A. 2004. Depthmap 4—A Researcher’s Handbook. London: Barlett School of Graduate Studies.

Van Beek, G.W., and O. Van Beek. 2008. Glorious Mud! Ancient and Contemporary Earthen Design and Construction in North Africa, Western Europe, the Near East, and Southwest Asia. Washington, DC: Smithsonian Institute Scholarly.

Van Minnen, P. 1995. “Deserted Villages: Two Late Antique Town Sites in Egypt.” Bulletin of the American Society of Papyrologists 32 (1): 41–56.

Wallace-Hadrill, A. 1988. “The Social Structure of the Roman House.” Papers of the British School at Rome 56: 43–97.

Weilguni, M. 2011. Streets, Spaces, and Places: Three Pompeiian Movement Axes Analysed. Boreas 33, Uppsala Studies in Ancient Mediterranean and Near Eastern Civilizations. Uppsala: Uppsala Universitet.

Wendrich, W., B. Simpson, and E. el-Gewely. 2014. “Karanis in 3D: Recording, Monitoring, Re-contextualizing, and the Representation of Knowledge and Conjecture.” Near Eastern Archaeology 77 (3): 233–37.

Youtie, H.C. 1971. “Agrammatoi: An Aspect of Greek Society in Egypt.” Harvard Studies in Classical Philology 75: 161–76.