3D Architectural Simulation and Uncertainty
“Like Magritte’s painting of a pipe that reads ‘this is not a pipe,’ a virtual reality model should carry a warning label that reads ‘this is not the past.” 65
The 3D landscape visualization used for this publication does not claim to re-create the cemetery and ritual site of Saqqara as it existed in the past—a fully impossible goal. The model, like any map, is a simplification and selection of elements that I, as an archaeologist, deem important for my specific focus of study. Indeed, each individual structure envisaged in 3D within the model of Saqqara greatly reduces the complexity of the original monument, abstracting it down to a series of lines or polygons that summarize its general exterior form. This model includes few interior spaces and only the most general color and texture of stones or brick, and it lacks hundreds of details that made up the real spaces of the past. Designed to serve as a landscape model, the level of abstraction for each element was purposefully very high, including only the most basic information key to the scale of inquiry. Beyond that, the model attempts to reimagine the ancient appearance of structures now dramatically altered by time and the elements, aspects that cannot be known with certainty today. Each visualization is based on the work of scholars who excavated and described the monuments, with their conclusions emerging out of the types of partial and incomplete data that typify the archaeological record. (A more detailed description of the construction of the model is the subject of Section 6.)
The visualization of these types of historical architectural models and landscapes in 3D has been met with suspicion in the academy, because any such work on ancient materials requires a great deal of hypothesis and conjecture. The concreteness of the visual image makes some scholars especially uncomfortable, because it does not allow for the same qualifications and silences as textual description affords.66 Some archaeologists consciously reject the term ‘reconstruction’ for historical materials in the digital realm,67 suggesting that the term is problematic and hinders acceptance of this work.68 To re-construct suggests a level of certainty that those working with the partial record of the past can never claim, a re-creation or exact replication of a real past that we can access today.69 Clarke argued that the term 'virtual simulation' better describes the models, which, he suggested, function as a set of abstractions that help us describe and investigate complex historical phenomena.70 Successful simulations are formed by “empirical observations, assumptions, logical interpretation and extrapolation, and creative imagination.”71 Favro similarly defined these models as abstractions based on “reasoned” evidence and suggested that we understand them as “knowledge representations,” a type of surrogate that both aspires to accuracy and accepts error.72 I understand the Saqqara model functioning in this way: not as re-creating, but fully as a construction at each point, one that is shaped by archaeological practice and publishing, the digital environment, and my own priorities and interests as a scholar.
Nevertheless, I maintain that this process is part of all historical scholarship, and that the model should be seen as operating in a system otherwise thoroughly acknowledged by researchers as standard practice. Historians assemble fragments of the past in complex ways, including and excluding the texts and other materials deemed relevant or irrelevant, making sense of the whole based on the patterns they had a hand in devising.73 Archaeologists similarly select and interpret at every stage of the excavation process, deciding what to record, how to represent it and its significance, and how the material traces encountered make sense in toto. All archaeological data is classified, interpreted, and filtered through a system of understanding by the archaeologist, and this is an accepted part of the discipline.74 That visualizations of historical structures and sites are also so produced does not make them problematic as scholarly output; it makes them typical.75
Gaps in the Archaeological DatabaseThe spatial analysis of historical data, in 2D or 3D, digital or analog, presents a specific set of challenges. While the hypotheticals involved with visualizing the appearance and form of aspects of an ancient site discussed above must be considered, more significant for research outcomes might be the overall uncertainty and biases in the archaeological database. The archaeological database, in contrast to the archaeological record (which includes the entire existing corpus of past material culture, including that unexcavated or as yet unidentified), consists of the totality of historical materials identified and available for consideration by archaeologists.76 This database is highly biased based on a number of factors, including the interests and funding priorities of archaeological missions, the types of methods used in data collection, deposition practices of people in the past, modern and ancient land use, site abandonment factors, and more.77 Cowley has pointed out that many scholarly attempts to analyze spatial patterns and distribution of cultural activities across space at sites or on the regional level merely recreate the past biases in survey data instead of any real past activity patterns.78 Indeed, quantitative analysis of past systems, including site development, tomb placement, and visibility studies—all considered in this work—can only be as successful as the data upon which they are based. Yet ancient sites are typically only partially excavated and remains are often ambiguous, so our modern archaeological databases will always inadequately reflect the archaeological record.79 It is crucial, then, that uncertainties in our data be acknowledged and discussed, and not just uncertainties related to the three-dimensional aspects of the project.
These challenges are certainly relevant for this publication. Although the site of Saqqara has been the subject of intensive excavation for almost 200 years, large areas of the site remain poorly documented. Early publication often did not meet the standards of today, and basic knowledge about some areas of the site have been lost, with archaeologists making attempts today to relocate tombs from early site plans.80 Other areas have never been scientifically excavated. For example, survey performed in the 2000s by the Saqqara Geophysical Survey Project over an extensive portion of the central Saqqara site shows dense areas of rectangular structures in zones in the Abusir wadi and the area immediately west of the Dynasty 1 elite tombs.81 Jacques de Morgan identified more than fifty tombs in the latter area on his 1879 plan as dating to the Old Kingdom and New Kingdom.82 While both of these areas are blank spaces on contemporary maps of the necropolis,83 it is clear that there was concentrated occupation in these locations at some point in the past. Our spatial understanding of the site therefore remains incomplete.
The gaps, holes, and different resolutions of data collected from the nineteenth century through today are addressed in a variety of ways in this work. In many cases, the existence of structures is mentioned in the text and considered in descriptions but not included in the model. This is especially true for data from the nineteenth and early twentieth centuries, much of which cannot yet be reconciled with GIS maps of the cemetery. Other information has been incorporated into the model for direct consideration in the 3D environment. This especially includes the Old Kingdom mastaba tombs discovered by de Morgan, Mariette, Quibell, Firth, and others. These tomb locations are identified in modern maps, many of them including the dimensions of their original superstructures. Procedural models, which generate a simple 3D shape based on the length and width of the monument footprint (discussed in Section 6), represent these structures, even if little is known about their original appearance. These models are generalized approximations of the exterior of a typical mastaba tomb from the period, used in order to represent the presence of these tombs without necessitating the custom design of each monument. I have also attempted in a few cases to include multiple visualizations that offer alternatives or varying levels of detail. While the web-based interactive model in Section 2 offers a single narrative reconstruction of the cemetery as a general overview, the research model allows for more nuance and layering, and I have attempted to consider this legacy data carefully in the analysis in Sections 3 and 6.
Chronology and Temporality and the Ancient Egyptian SystemTemporal studies of archaeological landscapes obviously hinge on a basic understanding of cultural chronology. Yet a firm correspondence between relative and absolute chronology for ancient Egypt has not been established. The absolute dates used for the model are based on the preferred chronology of the UCLA Encyclopedia of Egyptology (UEE).84 Their concordance with historical events remains imprecise, with the earlier periods containing hazier dates than the later ones. For example, in PeriodO,85 a digital gazetteer providing period definitions for historical eras, the start dates for the Old Kingdom, compared across six sources, vary seventy-plus years, with the proposed end of that same period spanning sixty years. The beginning of the New Kingdom, however, varies only twenty-three years, with most sources agreeing on a three-year span (1547–1549 BCE). Scholars have previously tied relative royal chronologies and Egyptian civil calendars to astronomical events (whose position has been calibrated within the modern Western dating system) in an attempt to produce absolute dates for royal reigns.86 This works only to varying degrees of success, and the synchronization is still controversial.87 Although radiocarbon dating and other scientific methods have been applied to Egyptian materials, the results have not yet been definitively reconciled to relative chronologies.88 Due to a large number of textual sources with overlapping information from both inside and outside Egypt (including Greek texts), Dynasty 26 has the only solidly confirmed dates before the advent of the Greek Ptolemies.89 The absolute dates used in this project for the 3D model should therefore be understood as approximate.
The model is therefore organized fundamentally on the dynasty system, used by the field of Egyptology to describe the political chronology of the Pharaonic state. This relative system is credited to the writings of the Egyptian priest Manetho from the third century BCE, during the Ptolemaic Period, but the ancient state itself does not seem to have used this dating system.90 Like our modern periodization systems, it manifests the same problems in trying to create a narrative history of a culture at that time already distant past, with gaps and inconsistencies.91 The earliest preserved written native chronologies (including the Palermo stone, purportedly recording events from Dynasties 1–5) are annals that document yearly events, each associated with the reign of a king. These so-called king lists (including the Turin Canon) offer sequential listings of the Egyptian kings, seemingly with unbroken strings of orderly royal succession, which conceal the much more complex political realities reconstructed by modern historians.92 A number of these documents leave out kings later deemed problematic, and others include divine figures considered part of the mythological foundation of the state. Their function as religious and political propaganda therefore makes them challenging to use for basic chronological reconstructions. However, decades of study with overlapping king lists, annals, and other records have allowed historians to create a generally accepted chronology of royal reigns during those periods of major state unity. When the specific order is unclear, rulers can usually be assigned to a specific dynasty, providing their general relative position within the overall dating system. Because monumental architecture was most frequently constructed during periods of state power, the relative chronology of most of the kings represented in the materials at Saqqara is firm.
Egyptologists working in the Saqqara region assign monuments under study to the reign of a specific king (usually based on inscribed material that is either part of or associated with the structure) or to a specific dynasty, if possible, positioning a structure within the established relative chronology described above. As the field’s understanding of ceramic sequences and architectural development has improved, especially for the earliest periods of state unity, some burials lacking inscriptional material and dated by early excavators have been assigned new dates.93 For a model spanning more than 2000 years, the dynasty was chosen as the smallest practical unit of displayed time. This choice was made to incorporate not only those monuments that can be dated to a specific royal reign (which are many) but also those dated only by architectural style or ceramics, which can rarely be linked to the reign of an individual king.
Time-Sliders and Simulation Slices in Temporal 3D ModelsTemporal models cannot hope to reflect the full process of change over time in the past. Most model visualizations choose to represent one moment, or a series of moments, termed ‘simulation slices’ by Clarke.94 These slices of time represent “a reasoned interpretation” of what was present in a given place at distinct points in time.95 The model of Saqqara includes eighteen such time slices, each corresponding to a dynasty with significant monumental construction at the site.
During a time-slice, architecture appears, remains, alters, or disappears, suggesting construction, use, deflation, or removal. That simplifies what was obviously a much more complex pattern, with varying intensity and type of use at different times at Saqqara, and, in most cases, some modified version of robbed, reused, modified, or deflated rather than total destruction, disappearance, or removal. The model highlights larger-scale processes, leaving out many of the aspects of short-term use and reuse that Egyptologists examine for individual buildings.96 That the model represents removal or deflation makes the 3D objects better reflect the temporal complexity of the site, where structures are displayed as having not a single moment but multiple moments of change across time.97 It makes clear that the individual lifecycle of a monument can differ from that of another, and that rates of change at the site are not simultaneous but fluctuate and overlap. The rates of an individual monument’s deflation or the date of its removal is rarely known, and the model represents deflation and removal only in a handful of cases, with deflation expressed in highly simplified ways. (See Section 6 for discussion of the life span of individual monuments and their visualization in the model.)
The goal of linking the models to a time-slider based on the dynastic system lies in making clear the rate of archaeological change over time at the highest temporal resolution feasible and practical. The dynasties varied in duration and experienced varying amount of activity at the site. This allows the reader to trace the pace of change from one dynasty to another, and not merely across larger political periods (e.g., Old Kingdom, First Intermediate Period, New Kingdom), designations that were devised in the nineteenth century CE and reflect conceptions of strength of state control.98 Those monuments that cannot be dated to a specific dynasty are placed in an unclassified time-slice appearing after the last dynasty of the larger period (Early Dynastic Period, Old Kingdom, etc.). This choice was made so that the model could incorporate as much information as possible about the overall form of the site without excluding monuments with less secure dating.
The use of a simple time-slider linked to absolute dates presupposes a linear understanding of time. The Egyptians had multiple concepts of temporality, including a cyclical concept (called neheh) and an atemporal concept (called djet). The cyclical scale was tied to the solar cycle and the concepts of renewal and rebirth. Djet, often translated ‘eternity’, can be described as a state of sacral permanence.99 Scholar of Egyptian religion Assmann argued that: “[t]here is no place for history in the system of neheh and djet. Each in its way denies the very idea of history: neheh by emphasizing reversibility, djet by emphasizing immutability.”100 Despite these fundamental differences from contemporary timekeeping, the Egyptians also created chronological king lists, annals, and a civil calendar that assigned linear dates to each year of a king’s reign (with the calendar resetting to year 1 at the start of a new reign). The basic unit of each dynasty, the royal reign, was therefore based on what we would term a ‘linear structure’. While the linear reigns and annals thus make a general match to our absolute dating system possible, it is clear that the time-slider in many ways stands at odds with the emic Egyptian concept of time.
The reconciliation of the time-slices to a time-slider necessitated the creation of clear boundaries for the start and end of a dynasty or period. These start and end dates were based on the UEE chronology, but they were adjusted and adapted to fit that need for standardization. This creates a number of problems. One, it engenders a false sense of clear and concrete or defined eras or epochs in Egyptian political history. In many cases, dynasties overlapped, especially when state unification was weak and rulers in different regions of the state ruled at contemporary times. It fashions bounded or distinct eras that were not acknowledged in contemporary times.101 This needs to be stressed, as the resulting model form is heavily influenced by this choice. Two, while such an invented linear progression actually aligns well with the ancient royal mythology of succession presented in the Egyptian king lists (discussed above), both the modern and ancient means of organizing time overly simplify many complex historical situations.102 Three, the time-slider cannot allow for overlapping dates from one dynasty to the next (otherwise model geometry would conflict), so the start and end dates for each period must have at least a one-year gap. The form of the time-slider itself therefore adds a false periodization of which the reader should be aware.
More precise and less wedded to such chronological definitions would be to link each building’s changes to the reigns of individual kings (which would, however, also need to be linked to absolute dates) so that the time-slider would display pace of construction within and not merely between dynasties. Such a time-slider would be exponentially more complex, including over forty kings with individual time-slices, as well as additional time slices for each dynasty with architecture not linked to a specific reign. The complexity and new levels of uncertainty brought into the project through such a dramatic increase in the resolution of the time-slider would be significant.103 Even this more complex system would still be a simplification of historical reality (when monuments were built at various times during and across the reigns of various royal reigns), and it must be stressed again that the model immensely reduces the actual complexity of the historical data. For this project, I cast broad strokes to consider large-scale patterns at the site across long periods of time; different research questions would necessitate the different structuring of the model layers or time-slices.
The environmental changes in the area, including the movement of the Nile River and the building up of sand and debris across the site (raising the ancient ground level) are also linked to the model time slices. That such changes were in fact gradual and progressive in many cases, and not linked to contemporary political change, should be underscored. Terrain and Nile alterations linked to the time-slices (based on dynasties) in the model are therefore represented as more abrupt and defined than the historical reality. The basis for the environmental simulations are discussed more thoroughly in Sections 2 and 6. The boundaries of the Nile River would, of course, also change dramatically across a single season, as the river flooded and then receded each year. The model does not attempt to visualize temporal change at the scale of single years, so such annual temporalities of the ancient landscape are also missing.