Table of Contents
THE BYRSA'S SECOND DEATH:
RECONSTRUCTION AND ERASURE IN THE HEART OF
COLONIAL CARTHAGE
by
Joseph Burkhart
B.A., St. Olaf College, 2017
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
Master of Arts
in
The Faculty of Graduate and Postdoctoral
Studies
(Classical and Near Eastern Archaeology)
The University of British Columbia
(Vancouver)
August 2021
© Joseph Burkhart 2021
The following individuals certify that they have read, and recommend to
the Faculty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:
The Byrsa's Second Death: Reconstruction and Erasure in the Heart of Colonial Carthage
submitted by Joseph
Burkhart in partial
fulfillment of the requirements for
the degree of Master of
Arts
in Classical and Near Eastern
Archaeology
Examining Committee:
Prof. Kevin Fisher; Classical, Near Eastern and Religious Studies; UBC
Supervisor
Prof. Matt McCarty; Classical, Near Eastern and Religious Studies; UBC
Supervisory Committee Member
Abstract
As the spatial, historical, and cultural nucleus of Carthage, the Byrsa
Hill is a stark illustration of the erasure inherent in the city’s colonial reconstructions. In 146 BC, Romans
obliterated the last resisting Punic forces in the acropolis on the hill’s summit, completing the genocide that
was the third Punic War. A century later, Roman colonists radically reshaped the hill into the organizational
nexus of a new colony, removing and burying Punic remains to form an enormous monumental platform . Although
this platform was occupied throughout Carthage’s Roman, Byzantine, and Islamic periods, following its
“discovery” by European explorers in the early 1800s, it became the center of French colonial excavations in the
ancient city. In the last 160 years, the hill has been excavated by many different project directors, and the
published archaeological datasets for the site vary widely in accessibility, quality, and completeness.
In this thesis, I use the Byrsa Hill as a case study to investigate
systemic problems in the ways that archaeologists traditionally organize and publish their data. By integrating
previously published architectural datasets from the Byrsa Hill, I uncover many inconsistencies and gaps,
including vague, contradictory, and missing descriptions of Roman features from older and newer excavations. My
restructured dataset enables new interpretations of the development of the site in the Roman period, in
particular revealing the diversity of construction strategies employed during the complex’s Augustan-era
construction. This
Lay Summary
In this thesis, I investigate the Byrsa Hill, a site in the ancient
Mediterranean city of Carthage (located in modern-day Tunisia). This site was the center of the ancient city for
thousands of years, and, perhaps consequently, it has been a target for erasure by colonizers. Such erasure has
been closely associated with the work of reconstruction, both during the foundation of the Roman colony 2000
years ago, and during the French occupation of Tunisia (1881-1956).
By digitizing previously published maps and profile drawings for the
Byrsa Hill, I draw new conclusions about the site’s early Roman occupation, and I uncover many errors that have
not been pointed out or corrected since the maps and profiles were published. The errors themselves, and the
lack of prior corrections, point to deep and systemic problems in the ways that archaeologists of this site (and
in general) traditionally manage and publish their data.
Preface
This thesis is my own original, unpublished, and independent work. All
maps, diagrams, and other figures were made by me, unless otherwise specified.
All figures which were sourced from previous publications have been
reproduced with permission from their respective publishers. Permission for figures from the volumes Byrsa
I, Byrsa II, and Byrsa III was granted by École Française de Rome. Permission for
figures from (1983) was granted by Cahiers des Études Anciennes. Permission for figures from (1990) was granted by l'Comité des Travaux Historiques et Scientifiques. In the
chapters and appendices below, all of these figures are identified either by a conventional in-text citation
(e.g., 1983, 105 fig. 8), or a Figure ID (e.g.,
In-text citations and the bibliography of this thesis follow a modified
Harvard format. In the interest of clarity, in-text citations referencing Byrsa I, II, and
III use the short title of the volumes, rather than the author and year. Thus, a citation that would
normally read ( 1979, 103) instead reads (Byrsa I, 103). This is to ensure that readers who are familiar with the site and
accustomed to the title-based citation systems common in Francophone archaeological publications can easily
identify these volumes.
A brief note is warranted on the internal-linking functionality
embedded in the digital version of this thesis. In the PDF file, all cross-references are internal
hyperlinks—i.e.,
Table of Contents
Chapter 7 Conclusions: Reforming the
Publication of Archaeological Research
List of Tables
List of Figures
List of Supplementary Files
- AxesOfReconstructionMetadata.csv
- FeatureMetadata.csv
- ProfileMetadata.csv
- ReproducedFigures.zip
- Shapefiles.zip
Acknowledgments
In scale, complexity, and pacing, this thesis has been by far the most
difficult project I have ever done. It is the culmination of five years of working, learning and dreaming, and
yet the vast majority of the project was compressed into the last five months and conducted mostly in solitude.
After losing two vibrant communities and many, many friends to the COVID-19 pandemic, I spent thousands of hours
alone in my bedroom, cataloging and correcting errors in other people's data, and trying to understand why so
many generations of archaeologists have suffered under such similar circumstances and struggled with such
similar problems. Words cannot express how prohibitively solitary the experience has been.
And yet, this thesis would not—could not—exist without the support of
many other people. If not for the love, wisdom, and generosity of my family, my partner, my friends, and my
colleagues, I would not be writing these words today.
Thanks first must go to my family, my partner, and my closest friends:
Chris, David, Rebekah, Gus, Amber, Leif, Walker, Jaye, and Ben. These people loved me, encouraged me, and held
me through some of the hardest days of my life.
Thanks also goes to my faculty advisors and mentors. My advisor, Prof.
Kevin Fisher, and my second reader, Prof. Matt McCarty, provided insightful guidance and patient support
throughout every stage of this project. They believed in me, my ideas, and my
Thanks also are due to my friends from Green College, including Azhar,
Sophie, Alison, Noga, Yotam, Hannah-Ruth, Saori, Sadia, Alicia, Steve, Aisha, Elahe, Alex, and many, many
others. Much of the sentiment and conviction underlying this thesis were born in conversations with these
friends, and their continual curiosity and kindness have been a source of great comfort and inspiration.
No less thanks are due to my friends from the CNERS department,
especially Brianne, Lara, Anna, Gillian, Liz, Safia, and Ben (a different one). No one—not faculty, family, or
(non-grad student) friends—will ever be able to understand what we went through this past year. Their openness,
supportiveness, and utter grit comforted me, grounded me, and gave me the strength to keep pushing my project
forward, even when it seemed that it would never end. A very special shoutout goes to Brianne, Lara, and Liz,
who kept me sane during many long workdays in the Reading Room.
There are so many others I could thank here. My very good friends
Danny, Wed, Anthony, and Jacob also listened to me and supported me and never failed to bring a smile to my
face. UBC's librarians were invaluable—Evan Thornberry, Paul Lesack, Stephanie
I could go on for pages more, but I will end my list here. To all those
whom I've neglected to name, thank you so very much. Your contribution to this project is not forgotten.
for my family
Chapter 1 Introduction: Reforming the Lifecycle of Archaeological Data
“Thus, we find ourselves in the contradictory position of being encouraged to pursue steady fieldwork, reminded in publications on the ethics of the profession that it is our obligation to publish our work promptly [...], yet the system of rewards and constraints under which academics labor pushes us against fulfilling standard publication requirements.” ( 2010, 201)
This project started out as an inquiry into the role that the Roman
reconstruction of the Byrsa Hill played in Augustus’ consolidation of power and legitimacy during the early
years of his reign. The foundation of the Colonia Iulia Karthago was a massive logistical undertaking,
involving
the clearing, sorting, and reworking of vast quantities of stone rubble, and the excavation and
transportation of hundreds of thousands of cubic meters of earth. The Byrsa was at the center of this
tremendous labor: its summit, which had housed the Punic city’s acropolis, was graded down to virgin soil,
and the sides of the hill were buried in a layer of backfill 2 to 3 stories high, creating a rectangular
platform that covered an area of at least 35,000 m2, which is 3 times
the area of the forum of Augustus
1
This monumental complex served as the organizational nexus for the new colony’s uncompromising
orthogonal grid, establishing the axes of the cardo maximus and decumanus maximus and
constraining the dimensions of the insulae. In my original proposal, I laid out a plan for using
the Roman Byrsa complex as a proxy to study the social and political ramifications of the foundation of the
Colonia Iulia in the context of the early Roman empire.
This rough estimate does not include the area west
of the cardo maximus, which was essentially unexcavated prior to Ladjimi-Sebaï's small
investigation in 1994 and whose relationship to the rest of the complex is uncertain (see Byrsa I, 47, 55 and 2005, 94-97). The
construction of a modern road after 1994 has likely rendered the area unsuitable for further
investigation ( 2005, 97).
The COVID-19 pandemic created many barriers to this research. Some
barriers were expected: the Canadian government essentially banned travel to Tunisia, which precluded any
form of ground-truthing, and to Italy and France, which precluded in-person archival research. There does
not seem to be a centralized archive for the archaeological excavations on the Byrsa, not even for the
UNESCO-backed excavations of the 1970s and 80s, and no smaller archive has been digitized or published
online.
2
Thus, I was limited to data which were published in books or
articles—particularly the volumes
Byrsa I, II, and III, which are the primary excavation reports from the 1970s and 80s, and
which lacked the completeness and precision that may have been found in unpublished archives. Other barriers
were unexpected: with the closure of libraries across the world, the Interlibrary Loan service stopped
accepting requests for physical items, and it did not resume accepting requests until January of 2021—thus,
for the first 3 months of this project I was limited to data which was published in books and articles
that had been digitized and re-published According to Mr. Turquin (pers.
comm.), archives manager of l’Ecole française de Rome (“EFR”, publisher of the
Byrsa volumes), the unpublished materials from the excavations (maps, drawings, etc.)
are not housed in the EFR archives. Some of the personal materials of Serge Lancel may be housed
in the Centre Camille Jullian, but they are not mentioned on the organization’s website.
Even as I resolved this pandemic-imposed problem, I began to
encounter other roadblocks to data collection that were inherent in my sources, such as inconsistent data
recording schemes, incomplete and selective site plans, and cursory or opaque analyses. I became
increasingly preoccupied with the causes of those problems, many of which I have come to see as deeply
rooted in the systems that have long structured—and still continue to structure—the lifecycle of
archaeological data. These are systems of inquiry, by which excavations are conceived and planned to achieve
specific priorities (i.e., to prioritize certain kinds of data); systems of excavation, by which data in the
field are collected and organized, and by which the locations and final depths of excavation units are
decided; systems of analysis, by which those data are turned into evidence for specific interpretations or
claims; and systems of dissemination and storage, by which those data, evidence, interpretations, and claims
are entangled together in figures (maps, profiles, photos, etc.), tables, and text, arranged in documents or
other structured media (presentations, archives, databases, etc.) and made available to other scholars or
the general public.
These problems are embedded in the practice of archaeological
research today, particularly as it is conducted by western researchers working in the ancient Near East and
3
Problems of inquiry are evident, for example, in the
neocolonial fixation on narrowly-conceived times
and cultures—often “Roman” or “Greek,” or defined in opposition
thereto—and the desire to selectively
preserve their material remains as “cultural heritage,” to the
intellectual neglect and physical destruction of others that are of more
relevance and interest
to local communities and stakeholders. Given the ubiquity of this
problem, it is impossible to discuss
representative examples in detail within the constraints of this
thesis, but it has been identified in
recent and on-going archaeological projects in Afghanistan, Egypt,
Iraq, Tunisia, and many other countries.
By “western researchers,” I refer primarily to
researchers from the United States, the UK, and France. Following (2010, 197-198), I do not mean to
imply that archaeological practice is homogenous between (or even within) these countries.
Rather, I wish to call attention to recurring elements in the various ways that archaeological
research is practiced.
4
General discussions of this complex problem can be
found in (2010) and
(2018). As (2010,
198-210) notes, this problem continues to be built into the
frameworks of funding, publication, and employment that structure archaeological labor. This
problem is too complex and widespread for a bibliography to be shared for each of the named
countries, but a few illustrative examples may still prove helpful. For commentary on the
problem in response to the destruction of the Bamiyan Buddhas in central Afghanistan, see
(2010). For a discussion of the problem in excavations near the
Egyptian village of Gurna (West Bank), see (2020). (2019) touch on the problem in their
critique of UNESCO’s ‘Revive the Spirit of Mosul’ project, though their focus is on cultural
heritage preservation and policy, rather than archaeological investigation. Some projects which
have notably avoided this problem include Ronayne’s (2006; 2016) study of (and participation in) the campaign against
Turkey’s Ilısu Dam and Forensic Architecture’s (2018)
investigation into ISIL’s destruction of Yazidi temples.
In excavations on the
Byrsa, these problems of inquiry first
manifested in the early excavators’ obsession with the site’s Punic
remains. In their quest to uncover Punic features, especially tombs,
these excavators tore through the
hillside, giving only cursory attention to Roman levels and
essentially ignoring medieval Islamic remains.
5
The later excavators of the 1970s and 80s attempted to rectify these shortcomings: they recorded
Islamic Lancel (in Byrsa I, 13) puts
it plainly: "Il faut le dire tout de suite : à Byrsa comme ailleurs à Carthage, le premier
objectif des explorateurs des ruines de la cité antique fut d'abord, sinon exclusivement, la
découverte des restes de l’époque punique.”
6
and they carefully documented the Roman backfill and occupation layers in the few locations where
their stratigraphy was undisturbed.See e.g., the medieval cistern in reference grid
cells
7
Ironically, these lacunae in Roman and medieval remains at times forced them to focus most of their
attention on Punic features and artifacts, unavoidably reprising the proclivities of their forebears. This
association was strengthened by the tone of their mission, which was situated within UNESCO’s larger
campaign to “Save Carthage” from encroaching urban development.See e.g., discussion on the “Butte of Lapeyre” in
Byrsa I, 97-142
8
On the Byrsa, that meant mostly saving ruins from the Punic city, since prior excavations had greatly
diminished or outright destroyed the Roman, Byzantine, and Islamic remains.
On the inception and development of the UNESCO
campaign at Carthage, see (2005, 45 n. 158) and references
therein. UNESCO’s efforts to uncritically preserve “cultural heritage” according to narrowly
western definitions are well documented ( 2018). The
violence and alienation caused by this work, and the role of archaeologists as perpetrators, has
only recently begun to be examined (e.g., 2010; 2010; 2016; 2016; 2020).
Problems of excavation involve deficiencies inherent in the
techniques and decisions that directly govern the collection and recording of data. In the UNESCO-backed
excavations on the Byrsa, these issues are evident, e.g., in the lack of proper stratigraphic illustrations
in certain sectors (Section 6.1), the lack of comparative data
on the construction of the foundation piles (Section 5.1.2), and
the lack of technical description of the only extant remains of the Augustan apses (Section 5.1.3). As I will explore in Sections 6.3 and 7.1, these
problems are linked to traditional divisions of labor and publication norms in
Problems of analysis involve the opaque creation of “archaeological
fact”, in which the distinctions between “data” and “interpretation” become blurry or disappear entirely.
For example, as I discuss in Chapters 5 and 7, it is sometimes very difficult (if not impossible), in the maps of
Byrsa I, II, and III to see which features were found in situ and which
were speculative or reconstructed. This sort of ambiguity makes it very challenging to independently
evaluate the excavators’ analyses and reuse their topographical data.
Because problems of excavation and analysis can only be identified
through the careful review and analysis of excavation records, they can become obscured by problems of
dissemination and storage, which are more intractable and farther reaching. On the Byrsa Hill, the problem
can be put simply: the results of excavations can be accessed only by a few dozen scattered articles and
several books.
10
These sources, even those from the UNESCO-backed excavations, contain major gaps, and yet there are
no centralized archives for the unpublished excavation records.Traditionally, western archaeologists tend to
distinguish books between “monographs” (one topic, generally one author) and “edited volumes”
(multiple topics, multiple authors). Historically, excavation results have tended to be
published under a single author (typically the site director), but multiple-author volumes did
sometimes occur. Archaeological research on the Byrsa has been published in both—Byrsa
I and II can be considered edited volumes, and Byrsa III a monograph.
11
This problem is situated within a broader “publication crisis” in archaeology, which has remained
entrenched despite numerous attempts to address it over the last 50 years. This crisis is complicated by the
fact that book publication continues to be a crucial (and increasingly difficult) barrier to professional
entry into academic archaeology. For a discussion of the gaps in technical
description of the Roman foundation piles, see Section 5.1.2.
For a discussion of ambiguities and
inconsistencies in the figures of Byrsa
I, Byrsa II, and Byrsa III, see Section 6.1 and Appendix D. Gaps
in sources prior to the UNESCO-backed
investigations are discussed throughout. On the lack of centralized archives for excavations on
the Byrsa, see n. 2.
The published results of the UNESCO-backed excavations are
essentially limited to Byrsa I, Byrsa II, and Byrsa III. Each of these volumes was published no more than 5 years after the last of its data was collected:
Byrsa I (published 1979) covers excavations in 1974–1976, Byrsa II (published 1982) covers
excavations in 1977–1978, Byrsa III (published 1985) covers excavations in 1977–1980. Within the
literature on Carthage more broadly, this timeframe is unusual, as most publications were delayed by 10
years or more. The volumes for the British excavations (1974–1979) were delayed by 5 years (1984) and 15
years (1994).
13
The volumes for the German excavations (1974–1980) were delayed by 11 years (1991) and 17 years
(1997). The volumes for the Swedish excavations (1979–1983) were delayed by 19 years (2002 ) and 35 years
(2018). These large publication delays can be found at many other sites across the Mediterranean,14
and they reflect the fact that it is enormously difficult and expensive to prepare excavation records
for publication as a book.
Two examples from the eastern Mediterranean bear
mentioning. The first example is the royal palace of Ugarit, a massive complex dating to the
Late Bronze Age. Although the palace was excavated in the 1950s under the direction of Claude
Schaeffer, systematic study of its architecture only began in 1989. This new project, which is
under the direction of Jean-Claude Margueron and Olivier Callot, is still ongoing ( 2008, 20). To date, the
project has only produced several preliminary publications ( 1986; 1995).
The reason for the absence of further publications is unclear—Schaeffer’s documentation may have
become misplaced or withheld (as (2008,
30, 48) seems to imply), and after 2011 is it possible that
the Syrian Civil War stymied field and archival research. A second example is the Uluburun
shipwreck off the southeast coast of Turkey. Considered the best-preserved instance of a LBA
shipwreck, the site has been the subject of over 40 years of excavation and research. Despite
this attention, there has been no comprehensive publication of the site, even though it has
produced numerous papers and theses—c.f. the overviews published by site director Cemal Pulak
(e.g., 1998) and Shih-Han Samuel Lin’s (2003) remarkably exhaustive
M.A. thesis, which was written under Pulak’s supervision.
The protracted delays in preparing such books are not helped by the
fact that they are often unsatisfactory, either because they are impersonal ( 1989), or “boring”
15
Prior to the 2000s, these efforts were largely ineffective, because the technologies necessary for
quickly connecting distant researchers with excavation archives were either inaccessible or did not exist
( 2004). As the internet has matured and digital publication tools have
proliferated, some excavations have adopted hybrid research schemes that attempt to balance open or
accessible archiving with more conventional publication media (e.g., 2016), but these are still exceedingly rare. In
general, for a variety of reasons, traditional books remain the standard for site publication.
16
The reasons for this are complex and differ between
countries. As
(2013, 282-283) points out, the permit
requirements of host countries
(among other legal factors) often dictate the pace at which preliminary (i.e., “interim,” but
all too often final) reports are published, and differences in research cultures can
place varying amounts of emphasis on primary data vs. interpretation.
Previous treatments of this issue have largely viewed it through
theoretical and methodological (and sometimes economic) lenses: what is the nature of archaeological data
and interpretation, and how should they be disseminated to be more accessible and useful
18
This is in spite of the fact that a majority of academics do not believe book publication should be a
requirement, nor that a book is necessary to present their scholarship; moreover, failure to complete a
monograph correlates overwhelmingly with departure without tenure (
2003).Estabrook and Warner’s (2003) study
constitutes, as far as I can tell, the only interview and survey-based statistical evaluation of
the role of book publication in RPT requirements. Cronin and La Barre’s (2004) study has a
similar purpose but is based on an analysis of official written criteria, rather than personal
surveys and interviews. While Estabrook and Warner’s (2003) study is
now a bit dated, their results are still readily accepted in broader studies of RPT requirements
in higher education (e.g., 2018; 2021). Estabrook and Warner (2003) interviewed
researchers exclusively in History, English, and Anthropology departments, but their findings
seem applicable to the field of Classics as well.
19
Only among historians did (2003) find a majority of respondents
(65.6%) who favored the book requirement. Interestingly, they found no relationship between
respondents’ age or tenure status and their opinion about the book requirement. Out of the
respondents who left their institution without tenure, 74% (23 out of 31 respondents) had
started but not completed their book (these data are viewable in tabular form in an online
supplementary page, retrievable from
https://web.archive.org/web/20040113221620/http://lrcsurvey.lis.uiuc.edu/surveys/99EZJ2/99EZJ2_0001.html).
I believe this problem can be tied to a lack of training and
institutional support in archaeological data management. Archaeologists deal with an incredibly diverse
array of data types from a myriad of sources and contexts, and they have the singular responsibility, to an
extent perhaps unique among the sciences, to serve simultaneously as excavator, curator, archivist, and
writer—to which may be added programmer or webmaster, if they intend to digitally publish
their excavation archives. These roles are simultaneous, not sequential. The plans of the writer depend on
the plans of the excavator, curator and
In this way, the aim of my project has become two-fold: to use the
site of the Byrsa Hill as a case study for systemic problems in archaeological data management, in order (I
hope) to improve the state of the publication of the site, opening it to wider inquiry. After digitizing
and collating most of the previously published cartographic data, I will integrate the data into a single
georeferenced dataset. From this initial dataset, I will extract the georeferenced extents of the most
important Roman and Punic architectural features, and I will synthesize the existing data on each of them.
This combination of extraction and synthesis will produce a new, restructured dataset of most of the Roman
and some of the Punic features, which will allow me to pose new interpretations and ask new questions of the
site. The entire restructured dataset will be openly published online, enabling readers to replicate my maps
and measurements.
21
My initial and restructured datasets are included
in the supplementary files accompanying this thesis. My restructured dataset is also published
in two repositories: https://github.com/josephburkhart/Byrsa-Archive-Static and
https://github.com/josephburkhart/Byrsa-Archive-Active. The first repository mirrors the
supplementary files—that is, it will not be altered or corrected in any way. The second
repositoty contains an actively updated version of the restructured dataset, to which new
features, corrections, and additional information will be added after publication.
A last note. The colonialism inherent in my project should be
acknowledged plainly. Firstly, I am a white, settler-colonial, cis-het-identifying man, living and working
in Vancouver, British Columbia, on land that was taken by force from the hənqəminəm-speaking xʷməθkʷəyəm
(Musqueam) people, who have lived here for many thousands of years, and whose heritage, knowledge, and ways
of being are still actively being destroyed today. I directly benefit from the oppression of these people,
even though I have tried hard to learn about their history and culture, and to work against the systems
which oppress them. Secondly, much of the data I use in my project was collected within colonialist and/or
racist systems of thought, which saw Greeks and Romans as beacons of civilization enlightening dark and
savage peoples through colonization and assimilation, and which positioned the European colonizers of the
18th, 19th, and 20th centuries as their direct descendants and rightful successors in this quest. These
schools of thought had diminished by the time of the UNESCO-backed investigations of the 1970s and 80s, but
aspects of them still persisted, and continue to persist today. Thirdly, I am using these data primarily to
investigate the social and political implications of imperial Roman colonization. The deaths of Carthage,
first in the genocide of 146 BC, and second in the razing and reconstruction of the imperial colony,
constitute one of the most profound acts of colonial erasure in all of Roman history. As the observant
reader may note in the sections below, the past 160 years of excavation on the Byrsa Hill have in some ways
reprised that second death, largely erasing thousands of years of material history through destructive
excavation and incomplete publication. It is my earnest hope that this project may set at least a small
amount of that
Chapter 2 Notation Systems Used in this Thesis
“Plan, dû à A. Thouverey, des fouilles du P. Lapeyre (Revue Africaine, 1934, pl. 2, entre p. 345 et p. 346)”“Plan, by A. Thouverey, of the excavations by P. Lapeyre (Revue Africaine, 1934, pl. 2, between p. 345 and p. 346)” (Byrsa I, 24)“On trouvera, joint à cette notice, un plan des fouilles (Pl. XI) [sic] Il a été exécuté par M. Bonnet-Labranche, architecte diocésain. Les lettres et les numéros insérés dans le texte renvoient au dit plan.”“A plan of the excavations (Pl. XI) will be found attached to this notice. It was executed by M. Bonnet-Labranche, a diocesan architect. The letters and numbers inserted in the text refer to the said plan.” ( 1893a, 95)“Nous donnons une description sommaire des divers vestiges conservés ou reconnus, en les répartissant par paliers que nous désignerons par le chiffre de leur altitude”“We give a summary description of the various remains preserved or recognized, by dividing them in stages which we will designate by the number of their altitude” ( 1924b, 178).
The three quotes above illustrate some of the different approaches that
previous researchers have taken to identifying figures, features, and feature locations on the Byrsa
2.1 Figure ID System
The past centuries of excavation on the Byrsa Hill have produced a
truly dizzying quantity of maps, sections, and other geospatial illustrations. The fraction of these that saw
the light of publication are scattered across many dozens of articles, bulletins, and books, although the
majority of them can be found in the Byrsa volumes. In order to create an integrated dataset of the
Roman and Punic structures on the Byrsa, I must draw upon well over a hundred of these graphics, considering
their ambiguities and inconsistencies, determining their original geospatial extents, and clearly linking them
to the data and feature traces that I derive from them. Due to publishing restrictions, I cannot reproduce all
of these figures in my thesis; for most of them I can only provide a citation. It is critical that these
citations be meticulous, since my results cannot be reproducible if my dataset is not clear.
Proper citation is made more difficult by
the fact I must identify not
just the publication and the location of the figure within it, but also the person or entity who digitized that
publication. As I discussed in Chapter 1, for most of this work I have been
limited not just to drawings that have been published, but to drawings that have been digitized and
As a result of these requirements, it is difficult to include enough
citation information for every figure while maintaining overall readability. To obtain a balance, I will employ
a figure ID system that clearly identifies a figure's original publication, location within that publication,
and the publication's digitizing entity, all within a unique, compact code. Each code is composed of three
identifiers separated by a full stop:
- source (B1, B2, GD1980, etc.): the figure's original publication. The full citation for each code is given in Table 2.1.
- scanner (J, P, etc.): the person or entity who digitized source. The meaning of each code is given in Table 2.2.
- page (006, 010a, 151-1, etc.): the page number of the figure in
source.22This identifier always begins with three digits (including leading zeros if necessary) to aid in sorting. When the digits are followed by 'a' the figure is a foldout page (that has not been assigned its own number) immediately following the page specified by the three digits; 'b' indicates a second foldout page after 'a'. If there are multiple figures on page, then '-1', '-2', or '-3' will indicate the place of the identified figurePage numbers are used instead of figure numbers because in Byrsa I figure numbers reset to 1 at the start of every chapter—a citation would thus need to indicate both chapter and figure number for Byrsa I, but not for Byrsa II or III. Citing figures by page number allows the same system to be applied to all three volumes, and it is also faster to use.
in the order (by ascending figure number) of figures on that page (in very rare cases, this is used to indicate a specific subsection of a figure—see e.g., , , ; , ; , ). If a figure occupies multiple numbered pages, it is identified by the page on which it begins (this only occurs once: ).
| Source code | Full citation |
| B1 | Byrsa I |
| B2 | Byrsa II |
| B3 | Byrsa III |
| D1893 | (1893a) |
| D1983 | (1983) |
| D1990 | (1990) |
| GD1980 | (1980) |
| G1982 | (1982) |
| G1990 | (1990) |
Table 2.1: Source codes used in
Figure IDs.
| Scanner code | Scanner identity |
| G | Gallica (https://gallica.bnf.fr) |
| J | Joseph Burkhart (author) |
| K | Staff of Walter C. Koerner Library |
| P | Persée (https://www.persee.fr/) |
Table 2.2: Scanner codes used in
Figure IDs.
Several examples may prove helpful.
All figures digitized by Persée (scanner
code P) had their pixel aspect
ratios corrected
2.2 Feature ID System
There are thousands of Punic and Roman features in prior publications
on the Byrsa Hill. Many of them are never identified unambiguously, and many are shown on maps with ambiguous
symbology that makes it difficult to distinguish features that were actually found from features that
were merely reconstructed or hypothesized by the excavators.
I have attempted to resolve these problems in two ways: first, by
limiting myself to digitizing only large structural features (walls, cisterns, etc) and mostly omitting
architectural fragments; second, by creating a new organizational framework in which each feature is given a
unique alphanumeric ID that encodes its ontological category. I am acutely aware that the creation of a new
identification system for a site with decades of data may prove naïve, if not futile, so in designing this
system I have sought to prioritize accessibility to francophone scholars and compatibility with previous IDs.
Each ID is composed of a type code, subtype code (optional), number, and ontological code:
- Type and subtype codes distinguish features by general architectural categories and subcategories, which are indicated in Table 2.3. Each subtype is specific to its parent type. Some types do not have a subtype. Wherever possible, the codes reflect wording that is similar in both English and French, and where this is not possible, the codes reflect the French.
- Numbers are assigned sequentially to features within each subtype from 1 to
9999, and each number is given leading zeros to ensure that there are always 4 digits (this helps with
sorting). If a feature had a previous ID that indicated its order in a type or subtype (e.g., piles
C1, C2, C3, or walls a, b, c of Punic block D), whenever
possible a number was chosen to reflect that previous order (e.g., pile C1 becomes , wall b of Punic block D becomes ). - Ontological category codes indicate whether a feature is extant (e),
reconstructed (r), or hypothesized (h). These terms require some clarification.
- By extant, I mean a feature that is shown in situ in my initial dataset because it was directly described in a publication in a geolocatable fashion.
- By reconstructed, I mean either
- a feature that was shown as a probable reconstruction within my initial dataset,
- a feature that was directly described in a publication, but not in a geolocatable fashion,
- a feature that is neither shown in my initial dataset, nor described in a publication, but whose presence, position, and form can nevertheless be judged to be “probable” from the initial dataset.
- By hypothesized, I mean a feature that is neither shown in my initial dataset, nor described in a publication, whose presence and position is supported by the initial dataset, but whose form can only be judged to be “possible.”
These codes and numbers are combined in the format
[type]-[subtype][####][Ontology], so that, for example,
All of these feature IDs are presented in tabular form with
corresponding maps in Appendix C. In these tables, I identify the map
from which the feature was traced, and any underlying assumptions that were necessary to complete the trace.
In the interest of full transparency, I have also created unique IDs
for the axes I used in my reconstructions. When the reconstruction of a feature involves the use of one of these
| Type | Subtype | English | French |
| CI | - | cistern | cisterne |
| ES | - | stairs | escalier |
| MU | AR | apse rectilinear wall | mur rectiligne d’apside |
| AS | apse semicircular wall | mur semi-circulaire d’apside | |
| FA | façade wall | mur de façade | |
| IA | wall in (Punic) block A | mur de l’Îlot A (Punique) | |
| IB | wall in (Punic) block B | mur de l’Îlot B (Punique) | |
| IC | wall in (Punic) block C | mur de l’Îlot C (Punique) | |
| ID | wall in (Punic) block D | mur de l’Îlot D (Punique) | |
| IE | wall in (Punic) block E | mur de l’Îlot E (Punique) | |
| SA | amphora wall | mur d'amphores | |
| SC | substructural wall C | mur substructurel C | |
| SD | substructural wall D | mur substructurel D | |
| SE | substructural wall Es | mur substructurel Es | |
| SF | substructural wall F | mur substructurel F | |
| SG | substructural wall G | mur substructurel G | |
| SH | substructural wall H | mur substructurel H | |
| SK | substructural wall K | mur substructurel K | |
| SM | substructural wall M | mur substructurel M | |
| SN | substructural wall N | mur substructurel N | |
| SO | substructural wall O | mur substructurel O | |
| SS | substructural wall S | mur substructurel S | |
| SX | substructure wall (other) | autre mur substructurel | |
| PI | B | foundation pile in row B | pile en la rangée B |
| C | foundation pile in row C | pile en la rangée C | |
| E | foundation pile in row E | pile en la rangée E | |
| X | foundation pile (other) | autre pile | |
| RU | - | rudus | rudus |
Table 2.3: Type and subtype codes
Figure 2.1: Diagram of the
feature
location system for major grid cells
axes, I reference the ID for that axis in the
feature trace table in Appendix C. A table and accompanying map of
these axes is also included in
Appendix C.
2.3 Feature Location System
The excavators of the French
Mission created a customized grid system
to allow themselves to specify feature locations in an imprecise but concise fashion. This system divides the
entire site into 2020 m major grid cells (
23
There is very little
discussion of this system in the
Byrsa volumes. Cursory descriptions are provided by Lancel (in Byrsa I, 59) and Gros (in Byrsa III, 5 n. 1).
Illustrations of the major grid cell identification system can be found in
This system is the key to georeferencing most of the maps and profiles
in the Byrsa volumes. Unfortunately, much of the system was not published, and I had to reconstruct it
from the pieces that were. My reconstruction of the grid system will be presented in Chapter 4.
This feature location system works well for concisely (albeit
imprecisely) locating figures on a sitemap, and I have chosen to use it in this thesis in order to ensure
compatibility and comparability with previous scholarship. It must be noted that a drawback of this system is
its opacity to readers who are not already familiar with the site. For example, designating a wall's location as
“
Chapter 3 An Abridged History of Excavations on the Byrsa
“C'est de cette idée de l’interdépendance, non seulement technique, mais monumentale, de l'ensemble des vestiges qu’il convient de partir pour reprendre l'étude de cette colline dont le malheur paradoxal est d’avoir été trop explorée.”“It is from this idea of the not only technical, but also monumental interdependence of all the remains that we should start to resume the study of this hill, whose paradoxical fate is to have been explored too much.” (Gros in Byrsa III, 25)
The history of excavations on the Byrsa Hill spans 160 years and
more than a dozen different project directors. For the first century or so, basically all of the project
directors operated independently, with almost no coordination. They ranged widely over the hillside, largely
avoiding the areas excavated by their predecessors. Their open-air excavations removed huge quantities of
earth and materials, as they searched for features of interest (primarily Punic tombs and inscriptions).
When they documented other features at all, they described them briefly and vaguely. Those features very
rarely survived the destructive combination of natural erosion and spoliation that followed from excavation,
That history is too long and convoluted for a full account to be
included here, but a very abridged account should still prove helpful.
24
For the sake of clarity and brevity, I will not discuss every researcher in this chapter, only those
whose findings are pertinent to my datasets. I will also not discuss those findings in detail; such details
will be provided, when necessary, in Chapter 5.For longer, more detailed accounts of this
history, see e.g., Lancel (in Byrsa
I, 13-55),
(2005, 39-101), and most recently
(2020, 72-109).
3.1 Charles-Ernest Beulé (1859)
While the Byrsa Hill was included in a number of loose surveys in
the 1700s and early 1800s (e.g.,
1738; Humbert 1821; 1859), Beulé was the first western researcher to excavate on the site. A French archaeologist and politician
with a thorough education in classical literature and history, Beulé had first made a name for himself by
excavating on the Athenian Acropolis, where he used a large amount of explosives to uncover a Roman
staircase and a Byzantine postern gate ( 2007). During his
single, self-funded excavation season (1859), Beulé made a massive informal ground survey of the hill and
its immediate surroundings, and his team of excavators (mostly local workers) conducted at least 11
different excavations. Beulé published a very inaccurate map of the survey and excavations, along with a
description of their results in Fouilles À Carthage (1861), a rather rather rambling and confused, albeit enthusiastic book,
replete
3.2 Alfred-Louis Delattre (1880-1897)
Almost thirty years after Beulé’s expedition, Delattre began
conducting excavations in many locations on and around the hill. These excavations were conducted under the
authority of the Pères Blancs, an ecclesiastical order that was tied first to the Saint-Louis
Chapel and later also to the Cathedral, and of which Delattre was a member.
28
Delattre had a penchant for bite-sized publication, resulting in a mess of small articles and reports
that are often vague, repetitive, and confusing.29
For my datasets, his most important excavations are those in the complex’s southwest corner, where he
uncovered part of the substructural walls of a temple ( 1893a; note that the excavator mistook these foundations for the walls
of a bastion); the south perimeter, where he uncovered an alignment of apses that originally were continuous
with those found in Beulé’s “Fouilles G” 3.3 Charles Saumagne (1923-1926)
30
In 1923, Saumagne began directing a new archaeological project
focused on the east side of the complex under the auspices of the Direction des Antiquités et Arts de
Tunisie (
1924b, 177; Byrsa
I, 35).
31
The excavations and surveys from this project produced the first map to cover the entire site ( 1924b, 181
plan II), which included very rough elevation measurements. Saumagne continued the project in
1925 and 1926, but he never published the results of these later campaigns (Byrsa I, 35 n. 65); they were finally
taken up and brought to publication over 50
years later in Byrsa I. (1924a) also investigated urban layout of Roman Carthage and its relationship
with the surrounding rural cadastration, but these findings are less relevant for this thesis.This organization seems to be identical to the
Service des Antiquités et des Arts discussed e.g., in (2013, 473). While
(1924b, 177), Lancel (in Byrsa I, 35), and
(2005, 42) refer to the
organization as “la Direction”, (2005, 156, 158) also refers to it
as “le Service.”
3.4 Gabriel Guillaume Lapeyre
(1930-1938)32
Four years after Saumagne’s investigations had concluded, Lapeyre,
another member of the Pères Blancs, began a new project focused on the southern side of the
complex. Over the next eight years, Lapeyre conducted a series of excavations that uncovered numerous
33
The quantity and locations of these excavations are difficult to pin down, because their results were
published very incompletely and imprecisely, when they were published at all (Byrsa I, 25). Lapeyre recorded almost no stratigraphic information, and his
excavations destroyed the vast majority of the Roman strata for the south side of the complex.3.5 Colette Picard (1947)
Intrigued by some features found by Lapeyre under a large building
in the southern part of the complex,
34
Picard, who was then the curator of the site of Carthage, conducted a limited investigation of the
area (
2005, 43; Byrsa I, 31-32). She determined them to be non-funerary walls dating to the late 3rd
or early 2nd century BC, near the end of the Punic period ( 1951).3.6 Jean Ferron and Maurice
Pinard (1950-1958)35
Several years after Picard’s investigation, Ferron (another member
of the Pères Blancs) and Pinard (an architect) began directing a new project which resumed Picard’s
excavations and extended them to the south ( 2005, 44; Byrsa
I, 33). Underneath
36
Afterwards, Ferron and Pinard extended their excavations eastwards, where they found remains of more
Punic buildings grouped into city blocks. The results of these later excavations were published in similar
fashion to the earlier ones ( 1960).This publication notably includes an extensive
finds inventory, which appears to contain mostly ceramics. Unfortunately, the inventory cannot
be easily searched because the openly published digitized version of the publication is missing
critical text-encoding metadata.
3.7 Serge Lancel et al. – The French Mission (1972-present)
By the early 1970s, decades of real estate prospecting and urban
expansion had caused ancient Carthage to be almost completely subsumed under modern constructions. In 1972, at the request of the newly independent Tunisian government, UNESCO launched an
international campaign to excavate remaining parts of the ancient city, to keep it from being completely
destroyed. The Byrsa Hill was re-entrusted to French institutions, and the Académie des Inscriptions et
Belles-lettres created a new project, the Mission Archéologique de Carthage-Byrsa to lead
the new excavations, with archaeologist Serge Lancel as the overall project director ( 2011b). This project, which I will
hereafter
39
began excavating in 1974, and has continued to excavate more or less continuously ever since ( 2011b). The project was divided into
several teams: the first team (responsible for “sector A”) was lead by Serge Lancel with the assistance of
Jean-Paul Thuillier, the second team (responsible for “sector B”) was lead by Jean-Michel Carrié and Nicole
Sanviti (succeeded in 1978 by Jean-Paul Morel), and the third team (responsible for the summit and parts of
the southern slope) was lead by Pierre Gros with the assistance of Jean Deneauve, Marie-France Giacobbi and
Françoise Villedieu ( 2011b). A handful of architects have
contributes to the project, in particular Gérard Robine, as well as the technical artist Philippe de
Carbonnières. Jean-Paul Morel became the director of the project in 1983, a position he still occupies today
( 2011b).It must be noted that there were previous
missions françaises, which were associated with the Pères Blancs.
The primary publications to come out of this project are Byrsa
I, II, and III. Primary excavation results can also be found in several other
articles (e.g., 1980; 1983; 2011a). While there have been several
indications that Byrsa IV and V are in the works, there has been no other major long-form
publications from the project since 1985, when Byrsa III was published.
40
(2011b) notes that Byrsa IV is forthcoming, to be published
by the École Française de Rome, as with the previous three volumes. Morel was awarded a
White Levy grant in 2013 to support the publication of Byrsa V
(https://whitelevy.fas.harvard.edu/people/jean-paul-morel), but I have found no other mention of
it.
Chapter 4 Collating the Cartographic Data
To date, no large-scale map has been
published that indicates
features' dimensions, orientations, dating, and construction for the entire Byrsa Hill complex.
41
Ideally, this map should also show elevation data for the upper and lower extents of the features, or at
least show the extents of other figures in which elevation data are encoded (such as profile drawings—see
Chapter 6). Instead, a multitude of smaller maps have been
published (mostly in the Byrsa volumes), which are drawn in different styles, emphasize or omit
different features, and conform to varying standards of accuracy and precision. In this Chapter, I will show how
I have integrated these smaller maps into a single cartographic dataset, which I will call my “initial
dataset.” In the next chapter, I will show how this integration revealed many ambiguities and
inconsistencies in prior maps, and how I resolved or mitigated them to produce a final dataset that is as
internally consistent as possible.
The first step to producing a unified sitemap is to
“georeference” each map in the initial dataset—that is, to determine the extents that it represents
within a Coordinate Reference System (CRS).
42
Visually, this causes the maps' features to overlap, making it easy In the Byrsa volumes, the site spatial
reference grid system was established on axes of the “Lambert Nord Tunisie” projection
(Byrsa I, 33 and 59, n. 2; Byrsa III, 5, n. 1),
which is used in the projected CRS “Carthage / Nord Tunisie” (EPSG:22391). Grid North
(GN) for the site grid is therefore equivalent to GN for the projection (Byrsa II, 299), which, noteably, is not equivalent to Magnetic
North—hence, bearings
4.1 Basemap
The best candidates for basemap were
43
Both maps displayed large sections of the reference grid, covering most, if not all, of the Byrsa hill,
and both maps also displayed two modern buildings (the Saint-Louis Cathedral and the Carthage Museum), enabling
them to be georeferenced relative to features in modern maps. Another candidate is
44
Second, 45
Third, The caption to
Before the reference grid could be reconstructed, the basemap had to
be georeferenced relative to an external dataset, so that the derived grid would be georeferenced as well.
Conventionally, digital georeferencing is carried out using algorithms that determine the “best fit”
for a map, given some reference coordinates—the algorithm scales and warps the image, and translates it to its
georeferenced location. It is important to give the algorithm as many reference coordinates as possible and to
make sure that they are spread out over the whole map, otherwise the algorithm will determine the best fit only
for part of the map. In
46
Then, it was manually translated and rotated so that the footprints of the modern buildings in the map
overlapped with the corresponding footprints in an external dataset.All georeferencing was performed using the
“Georeferencer” tool in QGIS 3.16.0. The algorithm used for
47
The external dataset used for this purpose was from OpenStreetMap (OSM),Rotation was performed using Guilhem Vellut's
open-source Freehand Raster Georeferencer plugin (version 0.7.1).
48
because its building footprint fit the This data was published by OSM user ulrichm
under the Open Data Commons Open Database License, meaning that it is free to use and re-publish
with proper citation.The data is part of Ideally, the external dataset would have come from a local,
municipal source; such datasets tend to have much higher resolution. Unfortunately, no such sources
appear to be available online.
Figure 4.1: Comparison of the
building footprint of the Byrsa Museum in datasets from OpenStreetMap (blue) and Google Maps (red),
overlaid on
4.2 Reconstructing the Reference Grid
As was mentioned in
Section 4.1, the
reference grid in
The reconstruction of the major grid cells proceeded as follows.
"Horizontals" and "verticals" are the lines of the grid, and are referenced by the cells that they run between
and around (Figure 4.2). Consequently, "the horizontals from
Figure 4.2: Horizontals (dotted
lines) and verticals (solid lines) for major grid cells
The first cells to be
reconstructed were those whose boundaries either
were directly plotted on the basemap, or could be determined by extending existing horizontals and
Figure 4.3: Cells directly plotted
on the basemap
At this point, two blocks of cells
had been reconstructed, leaving a
gap corresponding to cells
Figure 4.4: The reconstruction of
cells
To ensure that no part of the site
fell outside the bounds of the
reconstructed reference grid, two additional rows of major grid cells (
49
Cells Since row XX is south of row I, it is unclear whether
the roman numerals continue to increase past XI to the north. Unfortunately, the Byrsa
volumes do not resolve this ambiguity. In this thesis, I assume that they increase at least up to
XIII to the north.
Because the minor grid cells are not plotted on the basemap, they had
to be entirely reconstructed from their description in the Byrsa volumes: measuring 55 m, they were
created by subdividing each major grid cell into a 44 grid (illustrated in
50
Out of 3808 reconstructed grid cells, just 548 (14%)
had a side length that deviated from 5.00 m by more than 10 cm. None of the cells had a side length
that deviated by more than 50 cm.
Figure 4.5: The reconstruction of
cells
Figure 4.6: Schematic diagram of
the
shape and dimensions (in meters) of reconstructed minor grid cells within a) a perfect 2020 m grid cell
and b) a major grid cell that has been distorted. The distortion in b) has been exaggerated for
illustrative purposes—for the actual reconstructed minor grid cells, side lengths rarely deviate from
5.00 m by more than a few centimeters.
4.3 Other Maps
Every map of the site (that was useful for my purposes) was
georeferenced relative to the reconstructed minor and major grid cells, except for
In total, these maps, which
comprise my initial dataset, cover an
overall area of 25608
51
The extents of these maps are plotted in Figures 4.7, 4.8, and 4.9. Areas are calculated within the Carthage / Nord
Tunisie projected CRS, which uses the Clark 1880 ellipsoid (EPSG:7011). The area for the
hypothetical extents of the complex does not include the region to the west of the cardo maximus,
which is almost entirely unexcavated.
Figure 4.7: Map extents from
Byrsa I overlaid on basemap
Figure 4.8: Map extents from
Byrsa II overlaid on basemap
Figure 4.9: Map extents from
Byrsa III overlaid on basemap
Figure 4.10: Map extents from
other
publications overlaid on basemap
Chapter 5 Digitizing the Features
“on ne dira jamais assez combien l'absence d'archives a été préjudiciable à un développement rationnel de la recherche sur le sommet de Byrsa”“we can never say enough how detrimental the absence of archives has been to the rational development of research on the summit of the Byrsa” (Gros in Byrsa III, 25)“A large part of our failure to publish what we 'know' may result from the fact that we aren't sure what we do know or how we know it.” ( 1996, 42)
Digitizing features from the initial dataset required the resolution
of four significant problems. First was the problem of scale: the dataset covers thousands of Roman and Punic
features, and digitally tracing features on this scale is a major logistical challenge. Second is the problem of
variability: many features are depicted on different maps in slightly different positions and orientations,
requiring the evaluation of the different depictions and the citation of the map chosen. Third is the problem of
ambiguous nomenclature: many features either have no (published) identifier, an ambiguous identifier (e.g.,
there is a “mur a” in both Punic block C and Punic block D), or several identifiers (e.g., “mur D”
is sometimes called “mur Lapeyre”). Fourth is the problem of ambiguous ontology: on many maps, features found
in situ are depicted alongside features that are conjectured
I have attempted to resolve these problems in two ways: first, by
limiting myself to digitizing only large structural features (walls, cisterns, etc) and mostly omitting
architectural fragments; second, by creating a new organizational framework in which each feature is given a
unique alphanumeric ID that encodes its ontological category. The details of this framework are explained in
Section 2.2, but a very brief review may be helpful. The Feature ID codes
are composed of type (wall, cistern, etc.), subtype (substructural wall D, façade wall, etc.), number (0001,
0002, etc.), and ontological codes (E - extant; R - reconstructed; H - hypothesized), which are combined into a
Feature ID code of the form [type]-[subtype][####][Ontology], so that, for example,
A brief discussion about the accuracy and precision of my traces is
warranted. Because I cannot ground-truth my results, I cannot build a reference dataset to compare them to, so I
cannot definitively judge their accuracy. A rough comparison to satellite data from Google Earth (Figure 5.1) indicates that features are within about 2 m of their counterparts
in the satellite images, which is very likely within the error bounds for the satellite imagery.
More useful is to consider the precision of my traces. In the interest
of brevity, my assessment must be qualitative, rather than quantitative. The precision of my results can be
evaluated from the internal consistency of my initial dataset, i.e., to what extent the boundaries of features
shift between maps. Much of my initial dataset is consistent to
Figure 5.1: Outlines of the extant
B and C foundation piles (in red) overlaid on satellite imagery of the Byrsa Hill from
Google Earth. The outlines are about 2-3 m off of their counterparts in the images.
within 10 cm, most of it is consistent to within
20 cm, and all of it is consistent to within 30 cm. This is perhaps reflective of limitations inherent in my
reconstruction of the reference grid (see Section 4.2),
since
the lines of the grid shown in the basemap were approximately 20-30 cm at scale. In summary, my feature traces
are reasonably accurate and precise.
5.1 The Roman Features
The extant and reconstructable Roman architectural features on the
Byrsa Hill mostly fall into two categories: substructures and perimeter structures. The first category includes
substructural walls and foundation piles that supported above-ground structures and helped to contain the
massive embankment created during the complex's initial construction. The second category includes structures
that lined the complex's perimeter, such as the apses
It is often difficult to assign dates of construction to the Roman
features, since most of the time they have already been excavated, then backfilled. Where stratigraphic context
is no longer accessible, dating must proceed based on analyses of materials, construction, and the structural
requirements of the complex, which can only give a general idea of the complex's development.
52
In general, it seems that the excavators of the French
Mission were either unable or unwilling to deconstruct some of the architectural features, which
would have allowed them to search for datable inclusions and describe their construction materials
in detail. Consequently, their analyses are always comparative and point to construction dates that
are very broad.
Features of interest in this thesis can usually be assigned to one of
two phases. Features from Phase 1 were generally built during the initial construction program that began with
the Augustan foundation and lasted for at least a decade or more. From the very beginning of the program,
substructural and structural features were needed to direct, resist, and constrain the pressure exerted by the
Augustan embankment, which generally towered 10-20 m over the level of the streets below. Structures from this
phase were often constructed of rubble stones bound together with clay or earth, rubble stones that were
reshaped and arranged in opus reticulatum, or large sandstone blocks taken from earlier Punic
structures and arranged in opus quadratum or opus africanum. Features from Phase 2 were
5.1.1 The Walls
In the paragraphs below, the construction and dimensions of the
substructural walls are discussed. In the interest of clarity, walls are presented in order from south to north.
In the interest of brevity, only walls which belong to Phase 1, or replaced those walls, or are relevant for the
reconstruction of those walls are presented.
5.1.1.1 Wall D (Figure 5.2)
Wall D (see Figure 5.2)
was first investigated in the early 1930s by Lapeyre, who, under the impression that it was some sort of Punic
fortification, apparently excavated along it for a length of over 80 m (1934, 341-343). When the wall was re-examined
by the excavators of the French Mission
(Carrié and Sanviti in Byrsa I, 106-113; Lancel in Byrsa II, 45-46), they found that
Lapeyre had deliberately misrepresented its construction and
Figure 5.2: Summary map of Wall
D during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
its dimensions in his description and his maps to
better suit his interpretation.
53
For Carrié and Sanviti’s
critiques of Lapeyre, see Byrsa
I, 106, 111-112. His simplistic description (1934, 343) overemphasizes the wall’s toughness: “il est constitué
essentiellement par de fortes chaînes composées de blocs de pierres de taille.” His map 1934, pl.
2; reproduced in
Wall D is of bipartite
construction. The lower section, which
in some places was set in a shallow foundation trench,
54
was constructed using essentially the same materials and methods as the E piles: rubble stones
from the Punic destruction layer bound together with clay.
See e.g., profile
55
, The bottom of this section basically conformed to the topography of the layers
immediately below the
Punic destruction layer,
56
but the way the wall conformed was variable: in some places the bottom course was arranged in a spiked
formation, in others a stepped one (Byrsa
I, 112). The upper section, which was generally thinner
than the lower section (suggesting that it was not molded in a trench), was built heterogeneously, with segments
of large sandstone blocks alternating with segments of rubble joined with clay (Byrsa I, 112-113). The interface between the upper and lower sections varies in
elevation between 52.29 m and 51.11 m for a western segment (See Carrié and Sanviti (in Byrsa I, 112): “le plus souvent, ces particularités s'expliquent par
l'adaptation des premières assises aux mouvements du terrain, qu'ils soient naturels (déclivité du
sol) ou accidentels (fosse sépulcrale).” They further conclude (Byrsa I, 112),
“recherchant pour s'asseoir le niveau de l'argile en place, [le mur] a dû s'adapter au relief
naturel de la colline, et présente donc une élévation variable selon les endroits.”
Based on the stratigraphy, composition, and construction of wall
D
, Carrié and Sanviti
While Lancel’s analysis is convincing for the eastern segments, I do
not think it is necessarily applicable to the western segments, the closest of which is located over 20 m away.
The western segments clearly conform very closely to the level of the pre-Punic destruction layers, and the
interface between lower and upper sections roughly conforms to it, too. If the entire wall was built after the
embankment, why would the builders have contoured the bottom of their trench to the Punic pavement and
sub-pavement layers, and then tailored the level of the interface to match that contouring?
57
The contouring and variable construction of the wall’s bottom is more consistent with a modular approach
to the wall’s construction, in which different teams of workers under different supervision laid segments of the
lower section in different places along a pre-determined centerline. A key advantage of this bipartite approach
is its logistical flexibility: segments of the lower section could be scheduled for construction strategically,
to meet the distribution of available building materials and the local structural demands of the embankment’s
construction.Especially when the rubble linked with clay
construction was clearly capable of standing to a greater height, which is evident in its presence
in the upper section.
58
One of the key logistical challenges of the colony’s
foundation was the procurement of building stone. Because Carthage possessed no nearby
construction-grade quarries, the Punic ruins had to provide much of the stone used in the Augustan
colony’s foundation. Thus, prior to (or, more likely, simultaneous with) construction, the city’s
substantial Punic ruins were completely razed, and rubble from the Punic city’s
59
In summary, the wall was probably constructed in individual segments, more or less in concert with the
embankment, with the specific timing, order, and methods varying between different segments.This hypothesis could be better evaluated if Carrié
and Sanviti had reported the spacing between each of their evaluated segments. They may imply that
the spacing is constant (“l’espacement d’axe à axe entre les éléments doubles est plus proche de 6 m
que de 5.90 m ;” Byrsa I,
110), but if it was not constant then it would suggest that the
upper section was constructed piecewise. Even if Carrié and Sanviti had reported these dimensions,
they would not necessarily be representative of the entire wall, since their survey was limited by
its fragmentary preservation.
The width of wall D varied, but how is not exactly clear. (1934, 342) reports that it is on average 1 m, but
he then reports that the blocks are generally 0.66 x 0.70 x 0.60 m, with some as large as 1.10 x 1.01 x 0.50 m
or even 2 m (whether he means 2 m in length, width, or height is unclear) at the southeastern end—these
dimensions clearly do not reflect an overall wall width of 1 m. Carrié and Sanviti do not report an average
width, but one of their profiles (
60
Unfortunately, Lancel’s cursory description of Although it is obvious from photo
Figure 5.3: Summary map of wall
G during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
even farther east)
does not indicate whether the extant feature preserves the wall’s full width. I will assume that it does.
5.1.1.2 Wall G (Figure 5.3)
Wall G (see Figure 5.3)
is very poorly published: I am not aware of any previous publication which provides even a cursory discussion of
its construction and dimensions, with the exception of (1990, 144), who briefly notes that is is made from the same materials, and using the
same methods, as wall D and the E piles.
61
On at least one large map Wall G is also shown in photos of excavations
of Punic block C in
Figure 5.4: Summary map of wall
H and an associated wall during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
of the site,
5.1.1.3 Wall H and an Associated Wall (Figure 5.4)
Wall H (see Figure 5.4) is mentioned only cursorily in the Byrsa volumes, and
subsequent publications have added little to the picture. (1990, 149) describes its construction as irregular stones (presumably rubble) joined
with earth, which is clearly similar to the construction of wall D, and consequently he dates it to the
Augustan foundation. It had a foundation pit nearly 1 m deep (Byrsa I, 94). There are only two
extant segments of H in my initial dataset (
The inner westward spurs of wall H are interrupted by a set of
very large sandstone blocks (up to 2.5 m long, almost 1 m wide, almost 1 m high), which are linked together in
courses (three are extant) by dovetail seals resting on bare earth (Gros in Byrsa I, 273-280; 1990, 157). Two isolated pairs of similar
blocks to the west of these courses indicate a westward continuation. Judging from map
Figure 5.5: Summary map of wall
K and its mirror during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates
True North.
reconstructions. The
extant and reconstructed sections of this wall are shown in Figure 5.4.
5.1.1.4 Wall K (Figure 5.5)
On the other side of the stylobate wall, wall
K (see Figure 5.5) perfectly resumed the track of the northern
westward spur of
62
Deneauve interprets this dimensional There is another “wall K” mentioned by
Deneauve (in Byrsa I, 172) to the north of the Phase 2 temple in the southwest corner of
the complex (see Section 5.1.1.7). Deneauve indicates that this
wall is labeled on figure 1 of that report, but it is not. Judging from map
63
Following the same assumptions of symmetry as he used for the stylobate wall above, Deneauve posits an
identical wall that mirrors wall K about the temple's central axis. I retain this reconstruction.This Phase 1 substructural wall is
64
The extant and reconstructed sections of this wall are shown in and Figure 5.5.
The Phase 2 substructural wall is
5.1.1.5 Wall M and Three Associated Walls (Figure 5.6)
Wall M (see Figure 5.6) was a substructural wall in opus caementicium running
basically parallel to the cardo maximus about 28 m from the western side of the complex.
65
Extant segments of this wall can be seen e.g., in maps Wall M is sometimes labelled as two adjacent
walls M1 and M2 (see e.g.,
66
Unfortunately, the extant segments of the third and fourth walls do not appear to have been published in
a reasonably georeferenceable fashion. They are shown on several low-quality This building has been preliminarily judged to be a
library.
(1990, 154) proposed this
assignment based on the building's location and the thickness of the the east and west walls (i.e.,
wall M and the second associated wall). (1997, 349) took up this assignment and provided some textual evidence for
it.
(2005,
73) and (2020, 93) retain this assignment.
Figure 5.6: Summary map of wall
M and associated walls during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
georeferenceable maps (of which the
best is
67
but these maps are invariably printed at a very small scale,
and they diverge from the rest of my initial
dataset by at least 1-2 m.The extant segments of
the western and northern walls
of the enclosure are shown in
68
I do take the western extent of the third wall from The agreement between my initial dataset and
I reconstructed wall
M as follows. I assume that the northern
side was flush with the line of the southern façade of the eastern basilica in
69
and that it was 4.0 m wide (following 1983, 105 fig. 8). I reconstruct the intervening segment between This line is referred to as
70
.
This line is referred to as axis of reconstruction
I reconstructed the second (southern) wall as follows. I assume that
the west side was colinear with the west side of the third wall, and that the wall maintained a constant width
of about 2.1 m, based on the width of the west side of
I reconstructed the third (western) wall as follows. I assumed that
its south side abutted the north face of the second wall, and that its north side abutted the south face of the
fourth wall. I assume the west side followed the line shown in
71
I assume the wall had a constant width of 4.4 m, following (1983, 95).I reconstructed the fourth (northern) wall as follows. I assumed that
the north side was flush with the line of the south façade of the basilica in
72
I assumed that the east side had a width of 2.1 m as shown in (1983, 105 fig. 8). I assumed that the west side was colinear with the west side of
the third wall, and that its width was equal to that of the east side.5.1.1.6 Walls D', F', and G' (Figure 5.8)
Deneauve conducted excavations in a
handful of locations on the north
side of the central square (very near the side walls of the modern museum), which uncovered the remains of
substructural walls that paralleled walls D, F, and G, which he termed walls D’,
F’, and G’ (see Figures 5.7 and 5.8). The results of these excavations were only published in one
cursory article ( 1983),
73
which provides very little information about the walls’ locations, dimensions, and construction. The
schematic site plan included in this article ( 1983, 105 fig. 8) is drawn to scale (specifically 1/200), but it is discontinuous and
it has no reference grid or other feature which could be used to georeference it.The excavations in this area are very occasionally
mentioned elsewhere, e.g., Byrsa
III, 46 n. 6, 140.
74
In Deneauve’s map, the excavation trench at the
southwest corner of the basilica’s walls does not seem to be included on any other published map of
this area, even though the excavation there was conducted in
Figure 5.8: Summary map of walls
D', F', and G' during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
The excavators uncovered the
remains of five segments of wall
D’. Of these, the westernmost abutted the north wall of the large enclosure adjacent to the cardo
maximus (
75
The width of wall D’, judging from the plan, varies between 1.0 and
1.2 m; in my reconstruction, I
assumed a width of 1.1 m. No information is given about the construction of this wall, but Deneauve dates it to
Phase 1 (1983, 96).The connection between these two walls is ambiguous,
in part because it is not clear precisely what feature Deneauve found in the southwest of the
basilica. Deneauve only says “un sondage a permis de retrouver les tranchées de fondations
correspondant à l'angle sud-ouest de la basilique exactement à l'endroit prévu, mais surtout de
constater que celle de son mur latéral sud se prolongeait vers ‘l'ouest au-delà dé l'édifice.
L'extension des recherches n’étant pas possible à cet emplacement, un autre sondage a été ouvert à
quelques mètres, à l'ouest. Il a confirmé l'alignement de la tranchée de fondation de ce mur D’,
celui de l'égout qui longeait le portique et celui des murs F’ et G’” (1983, 97; italics are mine). It is not clear if Deneauve is talking about
a foundation pit in negative (i.e., whose original contents were removed prior to excavation), or a
foundation constructed in the form of a trench, essentially the same as a substructural wall; I
assumed that it was the latter. It is also unclear whether these foundation trenches in the
southwest corner of the basilica belong to the Antonine basilica itself, or to an earlier
construction. On Deneauve's plan (1983, 105 fig. 8), wall
D’ and the trenches appear to be perfectly aligned and of the same width, a point which
Deneauve also notes (1983,
98).
The excavators uncovered the remains of two segments of walls
F’ and G’, which ran parallel to each other (F’ abutting the north side of G’) to
the south of and parallel to wall D’.
76
For the western segment (which according to the plan was located 9.0 m east of wall M), judging
from the plan, the interior distance between D’ and F’ was about 9.5 m, F’ was about 1.5 m
wide, and G’ was about 1.7 m wide ( 1983, 105 fig. 8). For the eastern segment (which according to the plan was located
10.0 m east of the line The eastern segments appear fragmentary in Deneauve's
plan, but he does not comment on this in his discussion.
5.1.1.7 Other Walls
On the north side of wall D’, the
excavators found opus
signinum paving (
1983, 97). On the eastern side of wall
D’, the excavators found that this paving was delimited by a foundation pit (Figure 5.9), which (1983, 97) notes is aligned with similar remains found 60 m to the west during repair
work to the museum.
77
Judging from the plan, the interior distance between this pit and wall D’ was about 10.1 m and the
width of the pit was about 2.3 m ( 1983, 105 fig. 8). Deneauve hypothesizes that the enclosed area constituted a portico
opening onto the forum (1983, 97). No information is given about the construction of this wall. (1983, 98) does not directly suggest a function,
but implies that the wall may have served as a stylobate. To the west of Deneauve’s excavation, more remains
(probably) belonging to this substructure were found in 
Figure 5.9: Summary map of portico
substructures during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
Consequently, I consider this wall to be reconstructed from this point
east (
The remains of a parallel
substructure, which probably served as the
stylobate for another portico, were found adjacent to the northwest side of the basilica ( 1983, 98; Byrsa III, 140). No map of these remains has been published. The width of this
substructure is given as 1.75 m ( 1983, 98), which yields an interior distance of 58.94 m between it and the stylobate
of the southern portico.
78
To the west of Deneauve’s excavation, more remains (probably) belonging to this substructure were found
in (1983, 98) provides a slightly different interior distance between the
stylobates of 58.78 m. While the difference between these two distances is within my margin of
error, an explanation is included here for the sake of transparency. This interior distance assumes
that the north and south stylobates are perfectly parallel, that their interior dimensions are
identical, that the distance between the north face of D’ and the north face of the southern
stylobate is 12.25 m, and that the interior distance between the north and south walls of the
basilica is 83.28 m. I retain the first two assumptions. I do not retain the third
assumption—instead I retain my measurements of the interior distance and stylobate width from the
plan ( 1983, 105 fig. 8), yielding a
figure of 10.1 + 2.3 = 12.4 m; while Deneauve’s figure of 12.25 sounds more precise, he does not
include the individual measurements from which he calculated it. I do not retain the fourth
assumption, instead opting for the figure yielded by my georeferenced map
A little over 10 m past the eastern ends of this northern portico,
79
in This measurement is based on the reconstruction of the
basilica shown in
Figure 5.10: Summary map of
basilica
substructures during Phase 2. The dating of
adjacent later constructions. I assume that the
north and south ends of this
wall originally abutted the north and south façade walls of the basilica (see Section 5.1.3.4). As far as I am aware, there is no published description of
the wall’s mode or date of construction;
80
I will assume that the wall dates to Phase 2 and that it was constructed similarly to those another
substructural wall which was found about 21 m farther east. This eastern wall is structurally related to the
apses of the eastern perimeter, so it will be discussed in Section 5.1.3.4.On the one hand, the lack of a date is not entirely
surprising, since the excavators, working in the garden of the modern museum, would likely not have
been allowed to dig as low as the foundation trench. On the other hand, the lack of a description of
construction is very disappointing, since it was obviously known to the excavators and it could have
provided a rough idea of the dating.
There are several additional sets of substructural walls that deserve
mention. According to Deneauve’s reconstructions (
81
Two walls are adjacent to the southern staircase (It is not totally clear if these staircases should be
considered “hypothesized” or “reconstructed”—see n. 142.
82
Based on The southern pair is also visible in basically the
same position and orientation on
Figure 5.11: Summary map of walls
next to staircases during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
Figure 5.12: Summary map of
southwest temple substructures during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
publication does not mention them.
83
In Deneauve’s Phase 1 reconstruction (Map
More certain than those pairs of walls are the substructures which
appear to have
84
These walls were constructed in opus caementicium, with an opus quadratum crosswall
(which the excavators call Es) giving them the shape of an H (Byrsa I, 168-172). The
UNESCO-backed excavations in this area divided the structure into 5 walls: O (west), N (north),
S (south), S1 (smaller wall to the south abutting O), and N1 (smaller wall to the
north abutting O). Based on maps Finally, there was a retaining wall that was braced by the apses along
the south side of the complex. Since this wall is in close structural relation with these apses and other south
perimeter structures, it will be discussed in Section 5.1.3.2.
5.1.2 The Foundation Piles
Excavations on the southern slope
of the Byrsa have uncovered three
rows of Roman foundation piles (B, C, and E) running parallel to the axis of the decumanus maximus and
spaced 11-12 m apart.
85
Each row is composed of up to a few dozen piles, spaced 4-5 m apart, thrust into (or through) the
Augustan embankment. These piles served two primary purposes: they assisted in retaining the massive weight of
the Augustan embankment (which would otherwise have overwhelmed the support structures lining the decumanus
I south), and they served as substructures for walls and other constructions on the surface of 11.40 m between the rows
B and C, 12.10 m between rows
C and E (measurements taken from Byrsa
I, 132).
These piles stretch almost the entire length of the southern slope of
the hill, and given that almost no above-ground Roman structures remained when archaeological excavations began,
the piles were likely some of the first Roman features to be discovered.
86
It is unfortunate, then, that no comparative study of the individual piles has ever been published.
Instead, brief descriptions are scattered across the excavation reports of Lapeyre, Ferron and Pinard, and the
authors of the Byrsa volumes. In the following paragraphs, I draw from the general observations given
in these reports to summarize the existing data for the piles’ location, composition, and chronology.The piles have been attested in publications since at
least the 1930s, but they were probably known much earlier. Seventeen of the C piles (and
apparently two of the B piles) were known to Lapeyre—see Thouverey’s map in Lapeyre (1934, pl.
2; this map is reproduced in Byrsa
I, 24). It is very likely that Beulé encountered piles
during his
excavations in the southeast corner of the complex, but he does not mention them and as far as I can
tell, no evidence of them has survived.
5.1.2.1 The E Piles (Figure 5.13)
The northernmost row contains six clearly
extant piles (E1
through E6) and physical traces of three others (Eh, El, Ee), and a very likely
location for one more (Ea).
87
These piles are roughly constructed from irregular chunks of Ariana limestone and other stones from the
Punic destruction layer that are joined together with clay.There is a gap in between
E6 and E1 (at
grid cells
88
Piles E1, E2, and E3 For a cursory description of the stones used in
construction, see Carrié and Sanviti (in Byrsa
I, 129, 132-133)
and Deneauve and Villedieu (in Byrsa
I, 191).The identity of the bonding agent is
more ambiguous:
Figure 5.13: Summary map of the
E piles during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
| Feature ID | Previous name |
|
|
pile E1 |
|
|
pile E2 |
|
|
pile E3 |
|
|
pile E4 |
|
|
pile E5 |
|
|
pile E6 |
|
|
pile El |
|
|
pile Eh |
|
|
pile Ee |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pile El |
|
|
|
|
|
|
|
|
pile Eh |
|
|
|
|
|
pile Ee |
|
|
|
|
|
|
|
|
pile Ea |
|
|
pile E3 |
|
|
pile E5 |
Table 5.1: Feature IDs and
previous
names for the E piles.
display a bipartite morphology in which their lower sections
appear roughly constructed but relatively straight, while their upper sections appear more exactly constructed
but lean towards the east (Byrsa I,
132-133). This suggests that the E piles were constructed by digging
foundation pits into (and in the case of Eh, through) the Punic destruction layers, filling the pits with
irregular chunks of Ariana limestone and other stones from the Punic remains, intermixed with dirt or clay, and
then placing stones in more regular patterns on top of the filled pits. The rough mode of construction, which
closely resembles that of wall D, prompted Carrié and Sanviti to conclude that the E piles are
contemporary with wall D and shortly predate the Augustan embankment (Byrsa I, 132-133).
In modeling the Augustan Byrsa, it is necessary to reconstruct some
additional piles and sections of piles in this row and to hypothesize several more. The extant, reconstructed,
and hypothetical features in row E are summarized in Table 5.1 and Figure 5.13. Several of the extant E piles are clearly fragmentary, so I
have correspondingly reconstructed some sections: these are
89
we can hypothesize 4 additional piles to the west of El (It is unclear from Deneauve's description if these
extents are purely hypothetical, or grounded in some material remains.
Figure 5.14: Summary map of the
C
piles during Phase 2. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
5.1.2.2 The C Piles (Figure 5.14)
The central row contains 21 extant piles
(C1 through
C21), which are constructed of opus caementicium with small re-used rubble stones whose dressing
resembles that of opus reticulatum facing.
90
This composition is consistent with a date after the initial Augustan For the identification of
opus caementicium,
see Lancel (in Byrsa I, 51-52) and (1934, 343). For the stones see Byrsa I, 74. This
assessment is somewhat tenuous. From Lancel’s description (in Byrsa I, 74), it is
unclear if these re-used stones are inclusions or facing: “les piles […] comportment dans leur
maçonnerie des éléments de remploi, notamment des petits moellons de parement d’opus
reticulatum.” Many photographs of the C piles appear to show masonry facing, or at
least the imprints of facing that has been
| Feature ID | Previous name |
|
|
pile C2 |
|
|
pile C2 |
|
|
pile C3 |
|
|
pile C4 |
|
|
pile C5 |
|
|
pile C6 |
|
|
pile C7 |
|
|
pile C8 |
|
|
pile C9 |
|
|
pile C10 |
|
|
pile C11 |
|
|
pile C12 |
|
|
pile C13 |
|
|
pile C14 |
|
|
pile C15 |
|
|
pile C16 |
|
|
pile C17 |
|
|
pile C18 |
|
|
pile C19 |
|
|
pile C20 |
|
|
pile C21 |
|
|
pile C8 |
|
|
pile C16 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Table 5.2: Feature IDs and
previous
names for the C piles.
construction of the Byrsa complex
(Byrsa I, 74; 1990, 144). The stratigraphy and morphology of the piles support this hypothesis by
helping to post-date the piles relative to the Augustan embankment. Studies of the stratigraphy around the piles
show no evidence of foundation pits distinct from the piles themselves, and the piles tend to widen slightly
toward the top (Byrsa I, 74).
91
Both of these observations suggest that the piles were constructed by digging deep shafts through the
Augustan embankment and pouring a mix of concrete and stone inclusions into the shafts.
Some of the photographs in the Byrsa volumes
contradict this assessment of the pile morphology. For example, photo
The C piles were (originally?) joined together with additional
stonework to form substructural wall C.
93
The piles are contiguous with the H-shaped foundations on the SW corner of the hill, which also post-date
the Augustan embankment and likely belong to Phase 2.This wall was clearly visible to Lapeyre, and is shown
in his plan to run unbroken for a length of almost 100 m (1934, pl. 2; reproduced in
94
It is possible that the southwest foundations and the C piles are contemporary; These foundations are often called the “quadrilateral”
in older publications. This is the same period of
95
and refer the reader to the following chapter (a report on the excavations on the SW corner) for further
discussion (Byrsa I, 132). Unfortunately, while that report acknowledges that the C piles are
contiguous with the SW foundations (Byrsa
I, 170), it does not discuss their relationship.Carrié and Sanviti (in Byrsa I, 132): “en ce qui concerne
le mur C, sa fonction et sa date ont été
éclairées par le dégagement d’un temple (à l’ouest de notre secteur) dont l’alignement des piles C
prolonge la fondation nord.”
It seems likely that the C piles originally extended farther
eastward, since the last extant one, C21, is over 40 m away from the east side of the complex. I
hypothesize that there were an additional five C piles to the east of C21, sized about 1.8 1.8 m, spaced
about 4.4 m apart, and following the same alignment.
96
This is the average spacing of the five easternmost
C piles. The alignment used was based on the line between the centers of C17 and
C21. The sizing is the same as that assumed for the hypothesized B and E piles.
5.1.2.3 The B Piles (Figure 5.15)
Table 5.3: Feature
IDs and previous names for the B piles.
| Feature ID | Previous name |
|
|
pile B1 |
|
|
pile B2 |
|
|
pile
B3 |
|
|
pile B4 |
|
|
pile B5 |
|
|
pile B6 |
|
|
pile B7 |
|
|
pile B8 |
|
|
pile B9 |
|
|
pile B10 |
|
|
pile B11 |
|
|
pile B12 |
|
|
pile B13 |
|
|
pile B14 |
|
|
pile B15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pile B1 |
|
|
|
|
|
pile B15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The southernmost row contains at least 15
extant piles (B1
through B15), with two other possible piles to the east of B15.
97
These were constructed differently from the piles in the other two rows, with small pieces of rubble
(including some stones that seem to come from destroyed Punic buildings) surrounded by white-grey mortar (Byrsa I, 133).There are two oblong features in grid cells
98
Their morphology is also quite different, flaring widely at the base (see photos
Carrié and Sanviti actually say that the stones are
joined with cement (Byrsa I,
133): “petits moellons noyés dans un ciment gris
blanchâtre.” In contrast, Lancel (in Byrsa
I, 74)
characterizes the bonding agent used in these piles as “mortier” (mortar). As with the
C piles, (2005,
47) claims that the B piles are constructed in opus
incertum. I follow Lancel and assume that the bonding agent is mortar.
99
Although it is possible that this
flaring is actually
earthen supports left by the excavators (see note 90).
These piles are much shallower than the E and C piles; some do not even make contact with
the Punic destruction layers, indicating that the piles post-date the Augustan embankment. Carrié and Sanviti
infer that the Punic inclusions were used in other Roman structures before the B piles, prompting them to
conclude that the B piles were constructed from the remains of decommissioned Roman structures after the
complex's initial construction (Byrsa I,
133). More precise dates for the piles have been proposed, but these are
controversial.
100
Carrié and Sanviti decline to offer a more precise
date, but suggest that the construction of the piles resembles that of the later Roman walls found
on the area of preserved stratigraphy ironically termed “Butte Lapeyre” (Byrsa I, 133). In their discussion of these walls—which they describe as
stones joined with lime mortar (c.f. their description of “ciment” in the B
piles; see n. 98)—they suggest a dating “postérieurement à
l'abandon de la plateforme supérieure de la colline comme centre politico-religieux de la cité, et à
la désaffectation des édifices de type public,” but also see no relation “avec la
période tardo-romaine et byzantine”
Figure 5.15: Summary map of the
B
piles during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
While the C piles likely
replaced earlier Augustan foundations,
the case is not so clear for the B piles. As Deneauve and Villedieu point out (in Byrsa
I, 171), supports would have been necessary from the
beginning of the complex to hold back the tremendous mass of the embankment—not just in the southwest corner,
but also all the way along the southern slope. Out of all of the foundation piles, the B piles had to
make up the greatest difference in elevation, which ranged from 13 to 20 m over the eastern third of the south
slope.
101
The idea of constructing piles, from unworked rubble stones, on a footprint of about 1.8 1.8 m, to a
height of 13-20 m before surrounding them with an embankment seems rather impractical. I think it is reasonable
to suggest that the B piles were built during or immediately after the construction of the Augustan
embankment, in order to minimize the risk of structural failure. The fact that some of them do not touch the
Punic destruction layers has no bearing on their dating, nor does their morphology. In the absence of more
information about the pile's inclusions and the strata located below them, I think an Augustan dating is
possible, and perhaps probable.
See topographical points e.g., in
If we date the B piles to Phase 1, then it is reasonable to
assume that they would have originally extended to some point close to the extrados of the southwestern apses,
since
Figure 5.16: Map of the
hypothetical
extents of the Phase 1 B piles, compared to the extents of the Phase 2 southwest temple
substructures. Piles outlined in red would have been removed during the construction of the temple. CRS
is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
some sort of substructure was necessary to help retain the
embankment (Byrsa I, 171) and the substructural walls in opus caementicium had not yet been
built. This westward extent of the B piles can be indirectly hypothesized based on the layout and
construction of these later walls, despite significant gaps in their published data (see Figure 5.16).
102
It is difficult to determine the development of this
set of substructures because of problems of publication. First excavated in the early 1890s by
Delattre, who considered them some sort of bastion (see Byrsa I, 51 n. 58),
they were cleared in places below the level of the foundations, causing some of the walls to
collapse. During the UNESCO-backed excavations of the 1970s, they were re-excavated and documented
more systematically, albeit very sparsely.
It seems clear that the opus quadratum wall Es was
constructed before the other opus caementicium walls N, S, and O, and that its
north and south ends, which were ultimately encased in walls N and S, originally spanned both of
their widths.
103
While it is likely that wall Es served an active substructural role in the final building above
it, its first purpose might very well have been as a temporary retaining wall. The pouring of walls N and
S could not proceed without the destruction of vaults, façade walls, rectilinear walls, and semicircular
walls in the first and fourth apses of the southwestern alignment (Byrsa I, 169),This is evident in maps
104
which would have destabilized the second and third apses,Specifically, of Series 1 (see Section 5.1.3.1), counting southwards from the northernmost
apse.
105
compromising the integrity of the entire Phase 1 support system in an area where the embankment rose 5 –
6 m above the level of apse floors.Again, of Series 1 (see Section 5.1.3.1), counting southwards as before.
106
The installation of Es would have alleviated that problem. Walls N1 (For elevations see the reconstructed profile
107
It is unclear if this dismantling happened before or
after the construction of walls N1 and S1, but after seems more likely to me, since
that would have ensured the continued structural integrity of the semicircular walls. Deneauve and
Villedieu (in Byrsa I, 169) suggest that the semicircular walls were left in place to help
withstand the embankment, and while the bracing function of N1 and S1 seem to support
this, I think it is more likely that their primary role was to support the structure above. The
pressure from the embankment between them and Es would not have been very great, and it would
have decreased dramatically once their interiors were filled up with earth.
Based on the published data, It seems that walls N and S
may each have been constructed in two sections: on map
Why was wall Es placed where it was, about 2.5 m back from the
extrados of the southwestern apses? The simplest answer would be that it was placed according to the initial
plans of the 2nd-century architect who designed the temple, but this answer seems unlikely. The architect must
have been aware that there were substructures embedded under the Augustan platform, but they likely did not know
their precise locations and dimensions. These would have needed to be determined before they could begin laying
the foundations. I hypothesize that the B piles originally extended to a point abutting the east side of
Es, whose location was chosen to mark the end of that extension. The architect could then have been
confident that Es would function as a temporary replacement for the
108
It is possible that the same would be true for the Phase 1 substructure that was replaced by the C
piles.This alignment is termed “Series 1” in
Section 5.1.3.1.
109
In maps
Notably, this hypothesis would align the westernmost B pile
(and perhaps Phase 1 C pile) with the westernmost E pile. While this certainly does not prove the
hypothesis, it does make it more attractive, from an engineering standpoint. Thus, I follow this hypothesis and
hypothesize an additional seven piles in row B, each about 1.8 1.8 m and spaced about 3.9 m apart
(consistent with the extant piles near them), following the same alignment.
It seems likely that the B piles originally extended farther
eastward, since the last extant one,
110
This is the average spacing of the five easternmost
B piles. The alignment used was based on the line between the centers of B1 and
B7. The sizing is the same as that assumed for the hypothesized C and E piles.
5.1.2.4 Other Piles
During the excavation of the northern portico of the central square
(see Section 5.1.1.6), (1983, 96) found the remains of a column sub-base about midway between walls G'
and D', leading him to suggest that the northern portico had a central row of
Figure 5.17: Summary map of the
other piles. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
columns similar to the southern portico (see Figure 5.17). This sub-base is not reliably
geolocatable. It is somewhat unclear if Deneauve thought there was a row of foundation piles in the northern
portico—he makes no such hypothesis in his text, but his plan (1983, 105 fig. 8) specifically labels
this sub-base as part of a hypothetical row of E' piles. This hypothesis seems to be retained in
Deneauve's Phase 1 reconstruction of the site (
111
Consequently, I will assume that Deneauve hypothesized a row of foundation piles in the northern portico,
whose dimensions and spacing were very similar to the E piles. I retain Deneauve's hypothesis, and I
further hypothesize that these piles were basically square, about 1.8 1.8 m, stretched between the eastern and
western limits given in It is possible that these squares are meant to
indicate column bases, but Deneauve does not say one way or another. Notably, the squares in the
northern portico are not all aligned with each other: the third and sixth squares from the eastern
end are shifted to the south. The spacing between these two squares is a little over 16 m, which
seems roughly equivalent to the spacing between the two squares shown in Deneauve's plan (1983, 105
fig. 8). This suggests that Deneauve may have actually found
two sub-bases, rather than the one that he reports (1983, 96). One problem with
this theory is that the distance between the eastern sub-base and the line of the western façade of
the basilica in the plan (about 9.8 m) does not match the distance in the reconstruction (about
14.7 m).
5.1.3 The Perimeter Structures
No Roman features have been more enduringly popular in the minds of
western archaeologists than the structures which lined the perimeter of the Byrsa complex. This has created
several notable conditions. Temporally, the perimeter structures have the greatest depth of scholarship, ranging
from Beulé's foray in 1859 to the first—and, so far as I can tell,
This task is very difficult to do at all, let alone concisely. In
the
interest of clarity, this section will be organized spatially. After a brief overview of the sorts of structures
that will be considered, I will discuss the apses on the south side of the western slope, then the apses of the
west side of the south slope, continuing in a counterclockwise fashion until the apses of the northeastern
corner. The perimeter structures of the northern slope and the western slope are mostly unexcavated and have
never been published, so they are omitted. For each section of the perimeter, I will summarize the existing data
for the dimensions, construction, and dating of the structures therein.
The perimeter of the Byrsa
complex was essentially comprised of façade
walls, staircases, rectangular rooms, and apses that lined the complex's edges along the cardo IV east,
decumanus I south, and cardo maximus. A brief overview of the apses is warranted.
112
While these apses may have performed various functions during their lifetime,
113
this thesis is concerned only with their structural purpose. They served as the support system of the
outermost retaining wall for the Augustan embankment, a role that was essential from the very beginning of the
complex’s construction.
(2005,
90) follows (1980, 324) in suggesting that the
southwestern apses functioned as small shops, but this assignment seems purely speculative. The
function of the eastern apses is also uncertain, but it seems that the apse on the axis of the
decumanus maximus was at one point designed to house a statue (
2005, 288-289; Byrsa III, 29). Gros (in Byrsa III, 36-37)
suggests that it may also have stored books.
5.1.3.1 The Southwest Perimeter
The Roman features of the
southwestern section of the perimeter are
primarily described in the 1974-1975 site report of Deneauve and Villedieu (Byrsa I, 143-182), and some
subsequent findings are published in (1980). These
researchers identified a series of five apses (hereafter “Series 1”) along the southern part of the
western side, of which remains of three are extant. The other apses were either destroyed by prior excavations
or incorporated into a later set of substructures (walls N, S, and O; see section 5.1.1).
114
The extant, reconstructed, and hypothesized figures in Series 1 are summarized in Figure 5.18. These apses were first encountered in the late 1880s
by Delattre, who thought they were Roman cisterns (not discussed in publication, but marked on his
map in 1893a, pl.
XI). Subsequent excavations in this area by Lapeyre (around
1934) cleared the apses below the foundation level, causing the first one north of the SW corner to
collapse (Byrsa I, 167). The first architectural survey of these apses was conducted
in 1974-5 as part of the UNESCO-backed excavations. By this time, most of the structures were
destroyed, and it was no longer possible to date the remains through stratigraphy.
Figure 5.18: Summary map of Series
1
(in the southwest perimeter) during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
The rectilinear walls rested on a
foundation composed of two courses
of rectangular stones (Byrsa I, 168).
115
The types (sandstone or limestone, or a combination thereof) and dimensions of these stones have not been
published, but rough measurements taken from map The only extant example is the northern wall of the
first apse north of the corner. Only one block of the second course remains (Byrsa I, 168).
The rectilinear walls themselves have survived just barely, if at all,
but judging from Beulé and Delattre's descriptions of the southern apses (apparently confirmed by Deneauve's
excavations—see 1980, 322), they
would have been constructed in large-block opus quadratum.
116
The semicircular walls were composed of small pieces of stone rubble bound together with lime and lined
with opus reticulatum (Byrsa I,
167). As with the foundations, the types and dimensions of these stones have
not been published, but map Deneauve and Villedieu (in Byrsa I, 168) describe a series of blocks that serve as a foundation for
some sort of jamb in one of the rectilinear walls. but it is unclear if the researchers are consider
the blocks a foundation proper, or else part of the rectilinear wall above them. I assume it is the
latter, since the rectilinear walls of the south slope are described as predominantly opus
quadratum resting on rudus foundations (see e.g.,
1980, 322).
The ceilings of these apses have not survived—they were very likely
destroyed during the construction of substructural walls O, N, and S. Deneauve and
Villedieu report a large fragment of a vault in opus caementicium inside the southernmost apse, but it
is unclear if this came from the apse itself or from some other structure (Byrsa I, 168).
These apses likely had a façade wall, as indicated by the shape of the
surviving rectilinear wall foundations, but none of this wall appears to have survived (Byrsa I, 168). Its
As Gros and Deneauve reconstruct them (
To the south of Series 1, Deneauve and Villedieu found the remains of
a rectangular corner room constructed of façade walls is opus quadratum and opus reticulatum
(Byrsa I, 164-167). The room measures approximately 8.9 m in length, and its width is
about 5.8 m at its west side and 5.5 m at its east side). Based on its construction, this room likely dates to
Phase 1, although later alterations (including the installation of some cisterns) have made its original
function unclear. This room is also shown in Figure 5.18.
To the east of that corner room, researchers identified a series of at
least six apses (of which remains of three are extant) along the western part of the southern side (hereafter
“Series 2”).
119
The extant, reconstructed, and hypothesized features from Series 2 are summarized in Figure 5.19. In construction, these apses mostly mimic the previous It is possible that parts of the rudus of one
or two other apses are extant, but these pieces cannot be identified on Gros and Deneauve's map
(
Figure 5.19: Summary map of Series
2
(in the southwest perimeter) during Phase 1. Five adjacent apses to the east are also shown. CRS is
Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
alignment: the semicircular walls were lined with opus
reticulatum (which presumably served as facing for a matrix
of rubble joined with lime), the rectilinear walls were build in opus quadratum with stones from the
Punic destruction layer, and there was a façade wall (opus quadratum or perhaps opus
reticulatum) ( 1980, 322).
120
Gros and Deneauve reconstruct façade walls between 0.70 and 0.75 m thick, rectilinear walls about 0.60 m
thick, and semicircular walls about 1.0 m thick ( (1980, 322) do not describe the
construction of the semicircular walls beyond noting that they include opus reticulatum.
Judging from map
Gros and Deneauve's map (
121
By the time of his excavation, (1893a, 102) says that his trenches
covered 8 extant apses (though only 7 are shown in his map) in a non-continuous series, leaving two
gaps. The gap to the east corresponded with a section of the embankment that Delattre left
unexcavated which probably contained most of the Series 3 apses discussed below; it was
approximately 4 apses wide. The gap to the west, which was approximately 3 apses wide, was excavated
by Delattre. The absence of the 3 apses here was likely caused by later modifications that
122
At least one of the apses was lined along its interior with opus reticulatum ( 1893a, 102).
The specific term Delattre uses here is “moellons,”
which can refer to many different kinds of stonework.
123
Delattre’s (1893a, 102) measurements of these apses
suggests an apse width (interior diamter + widths of the side walls) of 3.20 m, which is rather less than that
of Series 2. This figure is very imprecise because it is an average of Delattre's hypothesized original number
of apses over the approximate length of Delattre's excavations (Byrsa I, 48 n. 48). I retain the
reconstruction of the two apses from Delattre's map can be roughly georeferenced relative
to walls S, N, and Es. In the georeferenced map, feature VI (which seems to be
one of the pieces of Delattre's alignment of apses) overlaps very well with
5.1.3.2 The South Perimeter
Such variability grew more
pronounced about 8 m farther east, where
the UNESCO-backed excavations uncovered a series of about five apses (hereafter “Series 3”) that
were originally continuous with those of Series 2.
124
The extant, reconstructed, and There are 4 fully
preserved apses (i.e., including the
entire semicircular recess and remains the full width of the semicircular walls) shown on the maps
of the Byrsa volumes (
Figure 5.20: Summary map of Series
3
(in the south perimeter) during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates
True North.
hypothesized features from Series 3 are summarized in Figure
5.20. They are likely the apses that were left buried by Delattre's
excavations, but which are still shown in dotted lines on his map (1893a, pl.
XI). As far as I am aware, the only published descriptions of these apses are from Lancel (in Byrsa I, 48), who notes very briefly that they are constructed from small pieces of
rubble (I assume joined with lime) inset with large stone pillars in opus quadratum, and (2011a, 352),
who mentions (also very briefly) that the semicircular walls are in opus caementicium with numerous
ceramic and stone inclusions that appear to be Roman in origin, suggesting a later construction date.
125
Because I have not found any confirmation of Morel's
cursory assessment, I retain the late 1st-century BC
construction date accepted by Lancel, Gros, Deneauve, and Ladjimi-Sebaï. The semicircular walls can be seen on
map Lancel’s exact description (Byrsa I, 48): “quelques vestiges des autres absides subsistent encore (fig.
3), ils présentent un appareil de petits moellons dans lequel étaient intercalées des jambes de
pierre de grand appareil. Les blocs ont été arrachés mais leur empreinte reste bien visible.”
Morel's exact description (2011a, 352): “en revanche, dans le
secteur B, ce mur, désormais
dépouillé de son revêtement et dont beaucoup des blocs de grand appareil qui en constituaient
l’armature ont été récupérés par la suite (sans doute à l’époque arabe), incorpore dans son opus
caementicium de nombreux vestiges manifestement romains – pyramidettes d’opus reticulatum
précisément, fragments de tuiles, morceaux de marbre (dont certains moulurés) – qui donnent à penser
que cette section du mur n’a pas été édifiée entièrement, il s’en faut, dès les tout premiers temps
de la Carthage romaine.” On Morel's map (2011a, 327
fig. 1), dotted lines in between the apses seem to indicate
that the rectilinear walls narrowed when they were no longer set into the semicircular walls. I do
not retain this hypothesis.
Among the four apses whose recesses were completely preserved,
interior diameter was between 2.2 and 3.5 m. If we assume that the outer and inner edges of the façade were
colinear with those of Series 2 (meaning that the façade walls for Series 3 were 0.75 m thick), then interior
length varied between 5.3 and 5.9 m. The rectilinear wall
The mode of
construction for Series 3 was markedly different from that
of Series 2. The rectilinear walls were much thicker and set deeply into the rubble-lime matrix of the
semicircular walls. These semicircular walls were also much thicker and were reinforced by thick stone insets,
and they tended to be “flatter,” that is, the central angle of the arc of their intrados is smaller.
The transition between these two modes of construction can be observed in the westernmost apse of Series 3,
whose semicircular wall does not have a stone inset, and whose eastern rectilinear wall does not appear to be
set into the rubble.
126
The
second observation could be caused by a trick of
perspective in the published map, so further confirmation is wanted. Such confirmation may be
partially supplied by Delattre's map (1893a, pl.
XI), which appears to show the very transition I note here;
apses on the west side of the gap of 4 appear not to have blocks inset into the semicircular walls,
while apses to the east of the gap do.
The extrados of the Series 3 apses appear to be contiguous with a
thick Roman retaining wall (
127
Lancel suggests that the wall was destroyed only within the limits of
Beulé’s “Fouilles G,”
meaning that the extant It is very difficult to determine the state of this
wall's preservation from the few mentions of it in prior publications. At one point, Lancel seems to
say that Beulé’s excavations (and the spoliation that followed) completely destroyed this wall along
the southern slope (Byrsa
II, 85), but contradicts this assessment later
on in the same volume, where he notes that his own excavations encountered the wall just southwest
of Punic block B (Byrsa
II, 143).
128
Unfortunately, Lancel did not plot the extant limits of the former on any of his maps.
129
I will assume that the limits of the extant section of the wall
are even with the limits of Series 3. To
the east and west of these apses, the limits of the rest of the retaining wall thus be hypothesized. I
hypothesize that the eastward limit was dictated by the wall of amphorae found by (1861, 39) and (1894, 89-91), and that the western limit was
imposed by the extrados of Series 1. This is consistent with the hypothetical(?) extents shown on maps Although the northern side of this wall is plotted on
many maps (e.g.,
Based on my above assumption, Beulé's “Fouilles G” began
at the western limits of Series 3.
130
These excavations stretched for about 40 m to the east and unearthed another series of (likely) six apses
(hereafter “Series 4”).It is possible that this assumption conflicts with (1893a), whose map indicates 2
excavated apses between the gap of four and Beulé's “Fouilles G.” If That gap of four
really does correspond with the four fully preserved apses of Series 3, and if Delattre meant that
Beulé's excavations began at the eastern edge of those 2 apses, then they would begin one apse to
the east Series 3, rather than zero as I have assumed. Ultimately, there is not enough information
to resolve this delimma, so I will follow my initial assumption.
131
The extant, reconstructed, and hypothesized features in
Series 4 are summarized in Figure 5.21. Beulé describes their
construction thus: the apses had an
interior diameter of 3.30 m, an internal length (from the apex of the recess to the inside of the fronting wall)
of 4.20 m, a rectilinear wall For the length of the excavations, see (1861, 46, 59, 79). The number of
apses warrants some qualification. On Beulé’s map (1861, pl. I; reproduced in
Figure 5.21: Summary map of Series
4
(in the south perimeter) during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates
True North.
thickness of 1.10 m, and a fronting wall thickness of 1.00 m.
Five apses were
constructed (both rectilinear and semicircular walls) from large stone blocks measuring as big as 1.50 x 1.25 x
1.00 m ( 1861, 60). For the rectilinear walls,
these appear to have been laid in the regular, interlocking courses of opus quadratum ( 1861, 60).
132
For some courses, perhaps specifically those of the semicircular walls and/or the foundations, the blocks
were not regularly shaped, but still fit very closely together ( 1861, 61 and pl. II fig. 2).Though Beulé does not specifically say opus
quadratum, it seems proper based on his illustrations (1861, pl. II fig. 2).
133
Mortar was found in some external joints and on some uppermost courses, but
nowhere else ( 1861, 63). The sixth apse was constructed
mostly in opus reticulatum (Beulé notes this is one of several, the rest being presumably unexcavated),
and at the time of excavation it still stood 8 m high ( 1861, 63).Beulé’s description here is ambiguous: “quoique les
assises paraissent, au premier aspect, réglées, elles ont cependant des saillies et des retraites,
des tenons et des mortaises , qu'on dirait empruntés à l'agencement des charpentes” (1861, 61). “Assises” could refer generally to some of the stone courses,
or it could refer specifically to the foundations of the walls. In Beulé’s illustration (1861, pl. II
fig. 2) of one of these apses, it appears that the
rectilinear wall is opus quadratum, while the semicircular walls are in irregular
stonework. (1980, 323)
identify this irregular stonework as “une sorte d'opus
africanum intercalant des jambes de pierre dans un opus incertum.”
134
Six meters above the floor, the walls of the recessBeulé’s description is ambiguous. In his initial
assessment, it seems that he is referring to a “cintre” (arch) in opus reticulatum, rather
than an apse, but in a later elaboration he refers to it as a cul-de-four, which properly
is a term for semi-dome but is also commonly used to refer to the vault formed beneath a semi-dome
( 2006). Deneauve (in Byrsa I, 48) seems to
think that Beulé was describing an apse—i.e., both the rectangular room and the recess (see n. 112)—but I think it is more likely that he was describing only the
recess, that is, the vault beneath the semi-dome, and I retain this assumption going forward.
135
were pierced with square holes which (1861, 64) surmised once held cross-beams. Beulé does not say whether the other apses
also had these holes, but he does note that none of them survived to a height greater than 5 m (1861, 64).
Here, Beulé uses the term cul-de-four to mean
the vault below the semi-dome.
Some aspects of Beulé's description warrant further scrutiny. Judging
from the variability observed in the dimensions of the Series 3 apses, Beulé very likely homogenized
136
Nevertheless, his measurements are the only ones we have from this point onward, until the apses of the
eastern basilica: all of the apses in Series 4, all of the apses between Series 4 and the southeast corner of
the complex, and all of the apses between that corner and the basilica have been destroyed or incorporated into
modern constructions, and no measurements or maps of them were ever published. Consequently, I retain Beulé's
measurements for Series 4, and I will assume that they are applicable to all of the apses after Series 4 as
well. All of these apses will be considered hypothesized.Or he may have only measured sections of each apse,
and then combined the measurements together. His map (1861, pl. I) shows some
sections of the apses in dark shading, and other sections in light shading. These differences in
shading may have been intended to distinguish excavated sections from unexcavated sections.
Ultimately, it is impossible to tell one way or another. His description of the apses includes no
mention of the different shading, and his schematic drawing of the apses (1861, pl. II
fig. 1) shows patterns of shading that are different from
the patterns in his map (some dark shaded sections appear to have grown larger, while others appear
to have shrunk). Subsequent reproductions of this map (e.g., Byrsa I, 15 fig. 1; 2020, 77 fig. 32) seem to
have basically homogenized the shading.
The dimensions and construction that Beulé reports for Series 4 are
noticeably different from those of Series 3 (Figure 5.21). Aside from a
slight widening in the rectilinear and façade walls, Beulé's apses are significantly shorter, with an interior
length of just 4.2 compared to the 5.7-6.0 m of the eastern apses of Series 3. This means that their extrados
would (just barely) not touch the intrados of the nearest apses of Series 3. Moreover, Beulé makes no mention of
the south slope retaining wall that was observed to be contiguous to the extrados of Series 3, even though he
had to have encountered it, since its presence in this area is attested in the negative (Byrsa II, 85 n.
2).
137
I propose that these discrepancies may be caused by two simple problems: first, that the two easternmost
apses of Series 3 were originally lined with semicircular opus quadratum walls that have Lancel maintains that the retaining wall passed
through this area, even though his excavations found no positive remains of it—see n. 127.
138
Contrary to what appears on his map, it is certain that he did not clear all six apses down to the level
of their bases; by his own admission (1861,
59), temporal and financial constraints forced him to simply follow the walls
with discrete trenches. This strongly suggests that he would have disregarded adjacent walls of much rougher
rubble-and-mortar construction, since the features that captivated him were almost entirely constructed in
large-block opus quadratum.
At one point in Beulé's excavations of this area,
progress was halted by the presence of large sections of walls built in mortar and small stones,
which Beulé thought dated to the time of Theodosius, and which were “à peu de profondeur,
renversées par pans énormes, couchées en terre dans toute leur longuer” ( 1861, 50). I have to wonder
if these sections of wall may not have been part of the upper section of the retaining wall.
Ultimately, the question is moot. When neither Beulé's workers nor Beulé himself could break through
the walls, he destroyed them with explosives ( 1861, 50).
Consequently, I hypothesize that the apses excavated by Beulé were
actually constructed with a 2.0-1.5 m thick rubble-and-mortar wall lined with a 1 m semicircular fronting wall
of opus quadratum, and with rectilinear walls extending past the intrados of the fronting wall to meet
the southern edge of the rubble and mortar wall. I retain this hypothesis for all of Series 4 and the rest of
the apses between it and the start of the amphorae wall.
Figure 5.22: Summary map of the
southeast perimeter during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
5.1.3.3 The Southeast Perimeter
Several decades after Beulé,
Delattre (1893b; 1894) excavated an unspecified number of apses to the east of Series 4,
whose contiguity therewith is unknown. These features are summarized in Figure 5.22. He notes that
one of these apses had interior facing in opus reticulatum (1894, 89-90); the rest presumably were
lined with large-block stonemasonry, as was the case for Series 4.
140
Leaning against the extrados of these apses, Delattre found a very large wall of stacked amphorae
(Beulé's vague descriptions of the stonemasonry were
clarified by Gros and Deneauve—see n. 133.
At the southeastern corner of the complex,
141
the perimeter turned approximately 90 degrees to the north. There were probably apses along the perimeter
here—this is strongly implied by Delattre's (1894) description and it is retained in 142
His Phase 1 reconstruction has the eastern end of this staircase flanked
by two pairs of apses (which
terminate their segments of the perimeter), and the western end of the staircase founded on a short
substructural wall.
143
Deneauve's reconstruction calls for the dimensions of
the staircase to change between Phase 1, and Phase
2, but due to a lack of evidence and some key inconsistencies, These are
144
In
Deneauve's Phase 2 reconstruction, this staircase's
width decreases by the width of wall F, and the northern pair of flanking apses is shifted
south accordingly. This arrangement seems purely hypothetical to me, and I do not retain it. I have
found no evidence at all that wall F continued past
At each corner formed by the 90 degree turns in this section of the
perimeter, Deneauve (
Finally, I assume that the flanking apses terminated the substructural
walls (
145
Or, if they didn't cut off the openings, that they
formed the façade; the rough construction of walls D and G would not have been
suitable for this purpose. It is completely unclear which of these two options Deneauve meant to
convey, because his very rough illustrations do not follow any reliable symbology, and because he
does not discuss them in enough depth in the text.
Given these assumptions and hypotheses, and based on the dimensions
reported by Delattre, the extents of the amphora wall can be posited with a reasonable degree of confidence.
This is shown in Figure 5.22, where I have extended the amphora wall
to an interior length of 50 m. It is notable that the amphora wall seems to replace both the retaining wall and
the rubble-and-mortar wall reconstructed and hypothesized for Series 4. Delattre clearly says that the amphorae
leaned against the extrados of the apses, and
One last topic must be discussed before I leave behind the apses of
the southern perimeter: dating. It is almost universally agreed that the apses along the southwest, south, and
probably southeast slopes date to the Augustan foundation. This dating is based primarily on their construction
and partially on their context. The use of opus reticulatum and opus quadratum composed of
stones taken from previous Punic structures is commonly found in other Roman structures which date to the
colony’s first few decades. Along the southeastern slope, the wall of amphorae could be narrowly dated to
between 43 and 15 BC on the basis of stamps ( 1894, 91). While this logic may not seem watertight—comparative architectural dating is rather imprecise, and the
amphora wall did not stretch across the entire south slope—the destruction of the southern apses and the lack of
5.1.3.4 The East Perimeter
The extant, reconstructed, and
hypothesized features of the east
perimeter are summarized in Figure 5.23. Based on Deneauve's
reconstructions (
148
Since the topography of this stretch of the perimeter would probably have been comparable to that around
Series 4, it is reasonable to hypothesize that these apses had roughly the same dimensions as Series 4, and were
similarly backed by a rubble-and-mortar wall followed by a secondary retaining wall. Following Deneauve's
reconstruction, I assume that there was a corner room at this northern turn, and for both rooms' dimensions I
follow the same assumptions that I made for the southeast perimeter. This section As with the previous staircase, I have been unable to
find any published evidence for this one. I also have been unable to find evidence for the southern
pair of flanking apses. Consequently, I consider these features to be hypothesized. Deneauve's
reconstructions (
Figure 5.23: Summary map of the
east
perimeter during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True North.
of the east perimeter was in Beulé's time
occupied by an old French cemetery and now is completely buried under
modern constructions, so it
has never been formally excavated.
On the north side of this
staircase lie the remains of the last
documented section of the complex's perimeter, where Beulé (1861, 67-74) found an alignment of five
apses (of which he excavated two) on the eastern slope of the hill, beginning a little over 100 m north of the
complex’s southeast corner to avoid the old French cemetery. His sparse description of their construction is
superseded by the modern survey conducted by Pierre Gros as part of the UNESCO-backed excavations.
149
The extant, reconstructed, and hypothesized features of the northeast perimeter
are summarized in Figure
5.23Notably, Gros (in Byrsa III) presents his
results thematically, rather than organizing them by field survey, as was done the previous
Byrsa volumes. As Gros himself acknowledges (Byrsa III, 3), while
this makes it much easier to understand and visualize the development of the structures that he
investigates, it tends to blur the distinction between observation and analysis, making it harder to
critique or reproduce his conclusions. This organization is reflective of Gros’s broader
investigative approach, which proceeded according to “le principe de la cohérence et de la
repartition symétrique des structures” (Byrsa
III, 25).
Gros identified nine apses in Beulé’s alignment (four on either side
of a central apse that was on the axis of the decumanus maximus), and two additional pairs of apses
oriented perpendicularly, bordering the alignment to the north and south.
150
Originally, according to Gros, the apses appear to have been basically homogenous in construction,
although they were decorated differently and some were modified in later periods.Lapeyre seems to have seen, or at least assumed, the
perpendicular apses, but never published on them (Byrsa III, 7 n. 5). (1982, 637-640 n. 5) says that while
they appear “sur certains plans anciens,” it is impossible to tell whether they were
actually seen or just hypothesized.
151
Focusing most of his attention on the central apse, due to its superior preservation, he described their
construction as follows.Regarding decoration, cf. the cul-de-four of
the central apse, with its caissons, and its neighbors, which have none (Byrsa III, 9). Regarding modification, see e.g. the apse immediately south of
the central apse, which had large horizontal “indentations” cut into its semicircular wall (Byrsa III, 9).
152
I must point out that one of the great shortcomings of
Byrsa III is its complete lack of a systematic architectural inventory for the retaining
wall and associated support structures. The partial inventory
Figure 5.24: Summary map of the
northeast perimeter during Phase 2. The dating of
The rectilinear walls
rested on earthen berm (“banquette”)
foundations—presumably cut into the elevation of the hill as it climbed to the north—while the paved interior of
the rectangular rooms had more complex foundations composed of a layer of pebble-cement on top of a layer of
rubble.
153
The foundation of the semicircular walls does not seem to have been directly investigated, probably to
minimize the risk of collapse, but it is reasonable to assume that it was identical to that of the rectilinear
walls.
The only
rectilinear wall foundation examined by Gros
is the wall of the apse immediately north of the central apse—presumably the southern wall, although
Gros does not specify (Byrsa
III, 33). The foundation of the interior
paving is only examined (and perhaps only extant?) in the central apse (Byrsa III, 29). Gros seems to say at one point (Byrsa III, 34) that there were also
some foundations in blocks of Hamilcar
sandstone, but he does not specify where. The central apse seems to have blocks only from El
Haouaria (Byrsa III, 7), while the apse immediately to the north has blocks from both El
Haouaria and Hamilcar (Byrsa
III, 33).
The walls themselves were of bipartite construction. The bottom
section was El Haouaria and Hamilcar sandstone in opus quadratum, with block width usually 0.40-0.50 m,
height 0.50-0.90 m (shorter for higher courses), and length around 0.50-1.2 m.
154
The blocks had 2.5-5 cm thick joints of beige mortar, and they all bore marks from lifting clamps (Byrsa III, 7). There were five courses for the rectilinear walls and six courses for the
semicircular walls. On top of the opus quadratum there were three courses of stones shaped to look like
bricks (specifically bipedales), some of limestone and others of sandstone, which The central apse seems to have blocks only from El
Haouaria (Byrsa III, 7), while the apse immediately to the north has blocks from both El
Haouaria and Hamilcar (Byrsa
III, 33). Block height is reported by Gros
only for the blocks of the central apse. Block lengths are not directly reported but can be roughly
determined from map B3.P.64a. Block width can be estimated by halving the width Gros gives for the
rectilinear walls, and this estimate is confirmed by measuring blocks in
The layer of brick-like stones served as an impost for the vaulted
ceiling of each apse, which was constructed in opus caementicium (Byrsa III, 7-8). Over the semicircular
recess, this was a cul-de-four, and over
the rectangular room it was a barrel vault. The first northern apse may have had a different ceiling system, but
it is not clear if Gros accepts this (Byrsa
III, 10-11).
The apses were partially enclosed by a fronting wall composed of
stonework, presumably in opus quadratum like the rectilinear walls. It was joined to them with the same
mortar, had a width of around 1 m, and may have extended for about 1.6 m on either side of each rectilinear
wall, creating openings of about 2.75 m for each apse (Byrsa III, 33).
155
The reconstruction of the openings is very tenuous,
since by the time of Gros’ excavations the fronting wall was almost entirely destroyed down to the
foundations. The reconstruction hinges on connecting the only remaining above-ground piece of an
anta (located just south of the mouth of the central apse) with an adjacent block (embedded in the
ground along the same alignment), and then hypothesizing a symmetrical structure on the other side
of the apse mouth (Byrsa
III, 33). In reconstructing the missing
segments of façade wall, I assume that each segment is 4.20 m long, stretching about 1.6 m on each
side of the rectilinear wall.
Behind the apses, and operating in conjunction with them, was a
massive substructural wall. This wall served to support the basilica’s eastern alignment of columns, and its
eastern face appears to have braced the extrados of the apses (Byrsa III, 19) and the eastern
Gros pointed out that the stonework used in the apse walls closely
resembled that of other structures dating to the Antonine period, prompting him to assign a date of construction
in the second half of the second century AD.
156
He suggested that these Antonine-era supports replaced analogous Augustan-era structures, whose
dimensions and construction would have been identical to the apses on the southern (Byrsa III, 27,
34) and southwestern (Byrsa III,
34) slopes.Gros first attempted to date the apses
stratigraphically, by excavating below the foundation level at the mouth of the central apse.
Unfortunately, later modifications had completely disturbed this area down to the sterile soil
layer. It is not immediately clear why Gros did not excavate through the floor of the central apse
or one of its neighbors, since the preservation of the original flooring strongly suggested the
presence of datable material below. Perhaps he was stopped by restrictions from the local government
or personal concern for the structural or aesthetic integrity of the last intact apses in the
eastern alignment. Gros’ date of construction is retained by (1990) and (2000), but
apparently critiqued by (2005,
75 n. 292)—see discussion below. Gros himself acknowledges that
his date wants for more solid confirmation, since “il s’agit d’une hypothèse, que rien de décisif ne
saurait fonder; elle s’appuie plutôt sur une convergence de données qui rendent cette identification
possible, sinon probable” (Byrsa
III, 37).
Gros' dating warrants further scrutiny. It is almost universally
agreed that the extant apses along the northeastern slope date to the second half of the 2nd century AD. This
dating is based entirely on their construction (particularly the distinctive type of opus quadratum and the use
of stones to imitate brick-work) since the UNESCO-backed team of excavators did not find any intact stratigraphy
that predated them. Consequently, the
157
This model was featured on the cover of CEDAC 15 and is
still housed in the Carthage Museum ( 1997, 347 n. 3), but aside from a brief
article in a popular news magazine ( 1996), it remains
unpublished. (1996, 54) note that this model was
constructed by C. Redler and J.-L. Terrer of “Espace Graphique” (presumably an architecture firm)
based on plans and elevations that were created by architects Gérard Robine and Philippe de
Carbonnières, and by archaeologists Deneauve and Gros. Redler, Terrer, Robine, and de Carbonnières
are not credited in any other mention of this model that I have found (with the excaption of 2020, 92 fig. 44). Such
reductivity around authorship and responsibility remains commonplace in archaeological publication,
since those who physically collect data in the field (field school students, local laborers, etc.)
are very rarely credited by name in the publication of a field report (noted also in 2010).
158
In (1996), all of
the photos are credited “Centre Camille-Jullian, CNRS.”
Perhaps the Centre Camille-Jullian houses other materials on the Byrsa as well? See n. 2.
Problems with these assumptions can be seen in comparison with a
similar building project that took place on a much smaller scale in the southwest corner of the complex. The
construction of the southwest temple took place with an elevation level difference (platform level – street
level) ranging from about 5 to 7 m, along a relatively limited façade (~16 m), in a corner of the platform that
probably was not previously occupied by a large structure. The construction proceeded by encasing the existing
supports with new foundations and filling the voids with more soil, allowing the embankment front to advance
forward. In contrast, the construction of the basilica took place over an elevation difference of at least 7 m,
over a façade more than 80 m wide (almost half of the eastern side of the platform), in a part of the complex
that was very likely occupied by a major, albeit smaller, structure with another structure immediately west of
it.
159
It seems unlikely that the construction of the Antonine basilica would have pushed the embankment
backward, since, if the ground shifted or settled, it would potentially compromise the structural integrity of
other foundations nearby.In Deneauve (1990), the first structure is labeled
“basilica?”, while the second structure is labeled “arc.” The absence of a site-wide architectural
inventory makes it difficult to find the material evidence for these structures in published
records.
Moreover, it seems unlikely that the Antonine apses would have
followed a different plan from the Augustan ones, since during the construction of the southwest temple they
were incorporated into the new foundations, where they continued to play an important structural role. By
extension, if the Antonine apses of the eastern slope were replaced by new apses following a different plan, we
would expect to see some trace of them in the new foundation structures. We should not assume that all of the
Augustan apses were
Accordingly, I hypothesize that the Augustan apses followed the same
footprint of their Antonine successors.
160
I further assume that the substructural wall abutting the extrados of the eastern apses dates to Phase 1.
These hypotheses are summarized in Figure 5.25.The question of the layout of the eastern alignment in
the Augustan era could be further clarified if the construction date of the retaining wall running
tangent to the Antonine apses was known, but as I noted above, this does not seem possible, given
the current state of the published data.
5.2 The Punic Features
During the initial construction of the Byrsa complex, existing Punic
features were either cleared during the leveling of the summit, destroyed to make way for new Roman
substructures, or else simply buried under the growing embankment. The architectural features that are most
relevant to this thesis are the walls of the five Punic city blocks (A, B, C, D, and
E) on the southern slope, since they occupied the most volume and therefore held the most influence over
the energetic cost of the reconstruction.
The original goal of this section was to document the digitization of
all of the Punic walls, cisterns, tombs, and other features on the Byrsa Hill. Unfortunately, time and size
Figure 5.25: Summary map of the
hypothesized northeast perimeter during Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
Figure 5.26: Summary map of Punic
block A during Roman Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
Figure 5.27: Summary map of Punic
block B during Roman Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
Figure 5.28: Summary map of
Punic
block C during Roman Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow indicates True
North.
Figure 5.29: Summary map of Punic
blocks D and E during Roman Phase 1. CRS is Carthage / Nord Tunisie (EPSG:22391). Arrow
indicates True North.
constraints have made made complete digitization impossible
within the scope of this project. As a preliminary
effort, I have digitized the walls of the five Punic city blocks on the southern slope, as well as a few
cisterns and tombs. The digitized features are summarized in figures 5.26, 5.27, 5.28, and 5.29.
My digitization of these features must be regarded as preliminary. For
the walls, I have not recorded the construction methods, as I did with most of the Roman features, and the
boundaries of my feature traces only reflect contiguity and source map, not actual context—in other words, the
traces of walls are only divided when the map from which they were traced indicates a gap, or when I switch to a
new map for tracing.
161
This means that the boundaries between walls—and their Feature IDs—could change in the future, as they
are refined to better reflect their discrete archaeological contexts. For cisterns and tombs, I have only traced
those features that are included in profiles published within the Byrsa volumes (see Chapter 6).For a few features, it is impossible to reliably trace
a feature from a single map. These are indicated in the relevant tables in Appendix C.
162
As time and resources permit, I intend to digitize
additional cisterns and tombs. These will be made available in the updated repository at
https://github.com/josephburkhart/Byrsa-Archive-Active (see n. 21).
The locations and dimensions of these features are given in maps and
tables in Appendix C, but some additional observations are worth
noting. For block A, all features have been traced from map
Chapter 6 Digitizing the Topographic Data
In the last chapter, I
showed how I digitized all of the Roman
features and some of the Punic ones, and I gave a full synthesis of the existing data for features relevant
to the site’s first phase. This information can offer some insight into the site’s layout, but it does not
reveal anything about the site’s topography, which is essential for understanding the site’s planning,
development, and function, particularly in the Augustan era. As I mentioned in Chapter 1, the Augustan construction of the Byrsa complex required the complete
reshaping of the entire hill: the summit was ground down to sterile soil, and the slopes were buried under
thick layers of soil and debris. Consequently, the design and construction of many of the complex’s
substructures and perimeter structures were largely governed by the initial topography of the hillside.
An archaeological site’s topography can now be reliably recorded
in great detail using remote sensing techniques (e.g., photogrammetry, LIDAR), but these have never been
applied on the Byrsa Hill. Consequently, the hill’s topography can currently only be reconstructed from
published elevation measurements. Such measurements can be found in older publications, such as (1893a, pl.
XI) and
(1934, pl. 1, 2), but their low accuracy
and poor geolocatability renders them basically useless for my purposes.
163
One exception to this is Beulé’s figure of 19 m
for the height of the platform in the southeast corner. This figure seems to have been accepted
by Lancel (in Byrsa I,
14), likely because there has not been an
164
The raw data for these elevation measurements have
never been published, and I am not aware of any published description of the methods used to
calculate them. The latter is a major cause for concern: are the elevations reported relative to
the mean sea level of the nearby coast, or to some other standard? Without a comprehensive
description of the reference points used in these calculations, it could be very difficult to
revise or supplement them with new measurements in the future. As sea levels continue to rise in
the next several decades, this problem will only grow worse.
Elevation measurements in the Byrsa volumes are given
primarily in maps and profiles, although some can be found in the text. While measurements in maps are very
easy to digitize, provided that the map can be georeferenced, it is often difficult to determine what layers
or features they are describing, and how the elevation of that layer changes over space. Hence, profiles are
more useful for my purposes, because they record elevations that change continuously over space and link
them to distinct stratigraphic contexts. In this chapter, I will use the term “profile” to indicate any
illustration which has a vertical axis, even if it is a reconstructive or composite illustration, rather
than a drawing of a trench wall.
The original goal of this chapter was to digitize all of the
elevation measurements that were encoded in profiles in the Byrsa volumes, and to supplement them
where necessary with measurements from maps. Due to time constraints, I was only able to digitize the
extents of most of the profiles, but these should still prove useful to other scholars who wish to better
reconstruct the Byrsa hill’s topography in the future. Moreover, the process of digitizing the extents
revealed numerous errors and omissions in the profiles and maps, which seem to have gone essentially
unremarked and uncorrected in the 40 years since their publication. While some of these mistakes are minor
and can be corrected without much trouble, others can render profiles very difficult or impossible to use.
A full explanation of how I corrected each profile would be
prohibitively dense and is outside the scope of this chapter. On the other hand, readers who want to use my
digitized profile extents may need to see those explanations, in order to evaluate my results for
themselves. Consequently, while I take a mostly synthetic approach in this chapter, I have put full
explanations of my corrections for the most important profiles in Appendix D. In the sections below, I will present an overview of the most
common types of mistakes, and then discuss two example profiles in detail. In the next chapter, I will argue
that these mistakes are reflective of deeper problems in the systems by which archaeological data is
published.
6.1 The Main Problems with the Profiles
There are 99 distinct profiles
in the Byrsa volumes, of
which I was able to georeference around 60 (see Figures 6.1,
6.2, and 6.3).
165
The other profiles were left out for a variety of reasons, but mostly because their reference
elevation could not be determined, because they could not be georeferenced, or because they did not seem
relevant for my period of focus, the Augustan era. The profiles that I did georeference are summarized in
Tables 6.1 (Byrsa I), 6.2
(Byrsa II), and 6.3 (Byrsa III).
Note that this quantity corresponds to the number
of profiles that are non-continuous. Some profiles (e.g.,
The problems with the published profiles may be divided into three
main categories: problems with reference elevation, problems with mapped extents, and problems with
symbology.
Figure 6.1: Summary map of the
profiles from the Byrsa I.
Figure 6.2: Summary map of the
profiles from the Byrsa II.
Figure 6.3: Summary map of the
profiles from the Byrsa III.
Table 6.1:
Summary Information for Byrsa I profiles. Reference elevations in
parentheses are adjacent to the profile. Map IDs in parentheses indicate that the
profile is located on the map, but its extents are not plotted or labeled.
| Profile | Drawn On map | Labeled on map | Map | Ref. Elev. | Notes |
|---|---|---|---|---|---|
|
|
51.92 | ||||
|
|
(51.92) |
Reference elevation given in |
|||
|
|
51.83 | ||||
|
|
x | x |
|
49.13 | |
|
|
x | x |
|
50.00 | |
|
|
x | x |
|
50.00 | Map extent markers are imprecise |
|
|
( |
(50.00) |
Reference elevation can be determined from |
||
|
|
( |
(49.10) |
Reference elevation can be determined from
|
||
|
|
x | x |
|
51.72 | |
|
|
x | x |
|
51.61 | Profile is split into two pieces that have some overlap |
|
|
x | x |
|
50.33 | |
|
|
x | x |
|
49.85 | |
|
|
x | x |
|
50.70 |
Extents also shown incorrectly in |
|
|
x | x |
|
53.40 | |
|
|
50.00 | ||||
|
|
x | x |
|
||
|
|
x | x |
|
||
|
|
x | x |
|
||
|
|
x | x |
|
||
|
|
x | x |
|
||
|
|
x | x |
|
||
|
|
x |
|
|
||
|
|
x | x |
|
56.00 | |
|
|
x | x |
|
56.00 | |
|
|
x | x |
|
48.53 | Multiple other reference elevations are given; width in georeferenced map has only one discontinuity (following the map symbology, which is incorrect - there should be two discontinuities) |
|
|
x | x |
|
44.95 | |
|
|
x | x |
|
(48.75) |
Reference elevation given in |
|
|
( |
47.47 |
Table 6.2:
Summary Information for Byrsa II profiles. Reference elevations in
parentheses are adjacent to the profile. Map IDs in parentheses indicate that the
profile is located on the map, but its extents are not plotted or labeled.
| Profile | Drawn On map | Labeled on map | Map | Ref. Elev. | Notes |
|---|---|---|---|---|---|
|
|
( |
51.90 | |||
|
|
46.35 | Multiple other reference elevations are given, but they are illegible | |||
|
|
( |
48.53 | Another reference elevation is given | ||
|
|
( |
(49.70) |
Reference elevation is given in |
||
|
|
x | x |
|
46.80 | Multiple other reference elevations are given |
|
|
( |
45.05 | Another reference elevation is given | ||
|
|
x | x |
|
52.73 | |
|
|
x | x |
|
52.66 | |
|
|
x |
|
(48.30) |
Reference elevation can be determined from
|
|
|
|
x |
|
(49.98) |
Reference elevation can be determined from
|
|
|
|
x | x |
|
(50.50) |
Extents in |
|
|
x | x |
|
50.40 |
Extents in |
|
|
x | x |
|
49.30 |
Extents in |
|
|
x | x |
|
49.11 |
Extents in |
|
|
x | x |
|
49.26 | |
|
|
x | x |
|
49.26 | |
|
|
x | x |
|
49.13 | |
|
|
x |
|
49.18 | ||
|
|
x |
|
49.18 | ||
|
|
x |
|
49.43 | Multiple other reference elevations are given | |
|
|
( |
48.93 | |||
|
|
( |
49.02 | |||
|
|
x |
|
55.98 | ||
|
|
x |
|
55.98 |
6.1.1 Problems with Reference Elevation
If a published profile does not explicitly tie some locus to a
reference elevation, it can be very difficult, if not impossible, to obtain the elevation data encoded in
the original profile. Sometimes, it is possible to get a replacement reference elevation from another
profile that is adjacent (e.g.,
| Profile | Drawn On map | Labeled on map | Map | Ref. Elev. | Notes |
|
|
x | x |
|
49.14 | |
|
|
x |
|
(48.67) |
Reference elevation can be determined from |
|
|
|
x |
|
48.67 | ||
|
|
x |
|
(50.55) | Multiple other reference elevations are given | |
|
|
x |
|
(54.56) |
Reference elevation given in |
|
|
|
x |
|
(55.61) |
Reference elevation given in |
|
|
|
( |
(54.90) |
Reference elevations are given in |
||
|
|
( |
(54.90) |
Reference elevations are given in |
||
|
|
49.14 | Multiple other reference elevations are given | |||
|
|
x |
|
(56.06) |
Reference elevation can be determined from
|
Table 6.3: Summary Information
for Byrsa III profiles. Reference elevations in parentheses are adjacent to the profile.
Map IDs in parentheses indicate that the profile is located on the map, but its extents are not
plotted or labeled.
gets 49.70 m from
166
or not at all (as is the case e.g., for Some of the illegible elevation points in this map
can be recovered from
6.1.2 Problems with Mapped Extents
Many profiles do not
have their extents clearly plotted on a map.
Generally, their extents can still be determined from a map, e.g., in the boundary of a trench (as with
profiles
167
or the extents on In these
example profiles, the errors are in the
grid cell labels, rather than the point labels. That is, for example,
168
Finally, for significant number of profiles, the extents are not
drawn at all (e.g., As noted in Table 6.1,
the extents of
169
Notably in this group
are many profiles from
Byrsa III that I was unable to georeference:
6.1.3 Problems with Symbology
Sometimes it is difficult
to determine a profile’s extents from a
map because the profile is actually more of a perspective drawing, i.e., it is a projection of features from
multiple locations and angles onto a single plane. Such drawings can be completely continuous (e.g.,
170
or plotted on a map but not labeled (as with See n. 168.
171
Sometimes, discontinuities are This last little profile,
6.2 Two Big Examples
To illustrate how the problems
discussed above can significantly
limit the available elevation data, it is helpful to consider the example of two profiles in particular.
B1.P.200 is the third-widest profile in Byrsa
I after B1.P.134 and
172
See Table 6.1
for the exact widths.
6.2.1
The extents of profile
173
while the This measurement is almost identical to the width
of the mapped extents on the printed (i.e., non-digitized and non-georeferenced) map, which is
23.95 m.
Figure 6.4: Map
Figure 6.5: Profile
printed width of the profile is 22.91 m. Second, the symbology of the mapped extents
indicates a discontinuity about 8.25 m northwest of the south endpoint, but the profile only shows a
discontinuity 18.75 m northwest of the south endpoint. Third, the grid cell labels given in the profile do
not match the grid cells that the mapped extents pass through on the georeferenced map. I found that the
most direct way to resolve these three problems was to focus on the third problem, so that is the approach
that I will take in this section.
The mismatch in grid cell labels can only be caused by one of
three things: a) the grid cells in map
-
The grid cells in map
are not actually labeled, 174but the labels can be determined and applied by comparison with mapThe figure caption of maprather vaguely assigns the map to “secteur H III 2, H III 5,” but these grid cells are not identified on the map. ; 175the resulting labeled map is shown in Figure 6.4. Therefore, the grid cell labels inThere is a single grid label “H” at the bottom of the map, which is basically illegible. It is possible—but much more difficult—to make determine the grid cell labels using mapinstead of . The grid cell labels of this very large site map ( ) are exceedingly difficult to interpret, but they can be clarified by comparison with maps or . There are no other Byrsa I maps that overlap with and contain grid cell labels. I am amazed at this predicament: in order to simply identify the grid cells in , readers of Byrsa I had to consult at least two different maps, printed in different sizes, styles, and contexts in totally different sections of the book, or else wait for Byrsa II to be published 3 years later. Given the problems I’m discussing in this section, the effect of this barrier seems obvious to me: almost none of the readers, if any, bothered to check the identity of the grid cells. are correct. - The positions of various features along the profile extents plotted in map
can be compared to the positions of corresponding features in profile . In , the distance between the southern side of wall a and the next wall to the south (the ID for this wall is not immediately clear) is about 8.17 m—this is consistent with the distance evident in , omitting the discontinuity indicated by the map symbology. This observation indicates that the discontinuity shown in the map was mistakenly omitted from the profile.
In, the distance between the southernmost wall (ID again unclear) and the left endpoint is about 4.70 m, which is greater than the distance of about 3.10 m in . This difference would be eliminated if I assumed that the south end of the extents continues another 1.5 cm under and past the map’s smudgy grid label, which is not unreasonable.
Finally, in, the distance between the northern side of wall a and the right endpoint of the profile is about 7.80m. This is significantly smaller than the distance of 10.31 m in . 176In the map, if I omit the segment stretching betweenThis distance in mapis measured from the point where the plotted extents intersect with the north side of a reconstructed segment of wall a—the extant segment is actually 0.84-1.31 m southwest of the plotted extents. and 10, which is parallel to the discontinuity farther south, the distance in the map becomes 6.24 m, which is much closer to the distance of 7.80 m in the profile. The remaining difference of 1.56 m is eliminated if I assume that the north end of the extents continues to the middle of the triangle next to the profile extent label A'. These observations indicate that the discontinuity shown in the profile was mistakenly omitted from the map.
Based on these observations, the profile extents plotted inare basically consistent with the features shown in profile , although some adjustments must be made to the north and south endpoints. - According to the grid cell labels in
, the profile bisects cistern E1 (labeled only in the profile) longitudinally in . 177Judging from the maps on which this cistern is plotted (e.g.,It is clear that the bisection is longitudinal because the distance between the cisterns walls in profileis about 1.47 m, which corresponds to the cistern's longitudinal length in map . and ; ), this situation is impossible: the cistern must be bisected in cell . From this, it is clear that at least some of the grid cell labels in are incorrect, rendering the rest of the labels suspect.
Based on the above considerations, it is clear that the profile
extents plotted in
6.2.2
Profile
178
On close inspection, it becomes clear that features within the first 22.40 m of the profile
(beginning from the left endpoint) overlap perfectly with features in The other two are
Two main problems make the rest of profile
179
First, the extents are not plotted on any map. This problem is unexpected, because the profile
clearly has point labels, and essentially every other profile with point labels in the Byrsa
volumes has its extents plotted on some map. Second, the profile’s symbology does not indicate any
discontinuities over the segment that is colinear with This profile took at least 20 hours to
georeference, spread out over several months.
To make the georeferencing process easier to follow, I broke the
portion of the profile that was not colinear with
Figure 6.6:
6.2.2.1 Segment 5
Segment 5 (Figure 6.7) shows a
transverse cut through a cistern (width 2.24 m), almost immediately adjacent to the base of a wall whose
width (0.95 m) and top elevation (53.60 m) almost perfectly match those of
Given those assignments, it is to determine the identity of the
remaining feature. About 0.71 m to the left of the cistern, segment 5 shows a longitudinal cut of a tomb
(width 1.92 m). Multiple tombs can be seen around this area in maps
The assignment of the tomb reveals several important points.
Firstly, the cut through the tomb is not parallel to the cuts through
180
Secondly, the north side of This contrasts with what was observed with
Figure 6.7: Segment 5 of
Figure 6.8: Segment 4 of
discontinuity.
6.2.2.2 Segment 4
Segment 4 (see Figure 6.8) shows
a wall (width 0.69 m) immediately adjacent to (and left of) a longitudinal cut through a cistern (width 6.62
m), which is almost immediately adjacent to (and right of) a second wall (width 0.66 m). Simply put, this
configuration is almost identical to the configuration of walls (
Figure 6.9: Segment 3 of
Notably, the profile omits the piece of
6.2.2.3 Segment 3
Segment 3 (see Figure 6.9) shows
a transverse cut through two parallel and adjacent tombs (combined width 1.77 m), located about 2.55 m from
wall
The profile indicates a discontinuity at the left edge of
6.3 Deeper Problems
The last two sections have made it clear that the profile drawings
of the Byrsa volumes suffer from numerous problems that make them harder to geolocate and
interpret. These problems can obscure or erase the elevation measurements for critical areas of the site,
e.g., as happened with
How did these problems happen? Why were they not caught upon
publication, or corrected in the 40 years since? I do not believe that I have the final answers to these
questions, but I do have some ideas.
Some fault must lie with the
strategies and decisions that
determined how the Byrsa volumes' text, maps, and profiles were first composed. When an entire
excavation report is published without any elevation measurements in its maps or profiles (as with the
southwest corner; see Section 6.1.1), or when
an
entire collection of profiles is drawn without
Some fault also must lie with the publisher's editorial decisions,
which determine not just the style and shape, but also (to some extent) the content of the final
(yet “preliminary”) site publications. The example of
181
From a financial perspective, the omission of the mapped extents and the shrinkage of It is also possible that the profile was
deprioritized because its extents were too difficult to understand. As I showed in the sections
above, there are numerous hidden discontinuities and some of the segments are not parallel to
one another. Moreover, much of the profile seems to have drawn from profiles published elsewhere
in the Byrsa volumes, This combination of inaccessibility and redundancy might also
explain why there is a lack of engagement with
Fault aside, why have these problems not provoked prior criticism?
I have been unable to find any publication to date that offers such critique of the Byrsa volumes.
Many of the problems I have pointed out seem obvious, given the data that I have shown. And therein, I
think, lies the answer: it took five to six months of work for me to be able to begin digitizing the profile
extents and collecting those data. Granted, three of those months were spent diagnosing and correcting
distortions in Persée’s digitized copies of Byrsa I and III, but even so, the remaining
two to three months represent around 300-500 hours of work, several weeks of which was spent reconstructing
the reference grid. I think that the main reason no one has previously published such a critique of the
Byrsa volumes is that no one else has had the necessary tools, time, and interest. The data in
these books had to be fundamentally re-structured before problems could really be identified. This is a
major barrier to research on this site.
The existence of this barrier points, I think, to an even deeper
issue with how archaeologists manage and publish their data. This issue is tied to the division of labor
between archaeologists and technical illustrators, and to the publication paradigm that structures the
relationships between archaeologists and their publishing editors. This is the issue that I will discuss in
my next and final chapter.
Chapter 7 Conclusions: Reforming the Publication of Archaeological Research
“Final publication begins the day the first pick goes into the ground, indeed, before that, in our conception of the project” ( 1996, 43)“L'Afrique carthaginoise est devenue française : c'est nous, à présent, qui sommes les Romains.”“Carthaginian Africa has become French: it's us, now, who are the Romans.” ( 1896, 116)
My collation, digitization, and re-organization of the archaeological
data on the Byrsa complex have highlighted how recurrent problems in archaeological data management have
significantly limited our understanding of the site, particularly in the Augustan period. These problems are
long-standing, and they are persistent: while the systematic methods and analyses of the French Mission
represent a great improvement over the piecemeal efforts of Beulé, Delattre, Lapeyre, and Ferron and Pinard,
their publications suffer from some of the same fundamental problems as their predecessors. At times, these
problems overshadow the improvements, because they make it difficult—and sometimes impossible—to critique or
confirm old interpretations, or to re-use the data to answer new
Before I begin my critiques, I must acknowledge that much of this work
would not have been possible without some major advancements in the open publication of old archaeological
publications on the Byrsa. Without the digitization and online re-publication of (1861), Delattre (1893a; 1893b; 1894), (1924b), (1934), and Ferron and
Pinard (1955; 1960), most of which are out of
print and not commonly held by anglophone universities, my dataset, analyses, and interpretations would have had
very little historical depth and the scope of this project would have been greatly reduced. Links to these
sources are included in my Bibliography.
7.1 Old and Common Problems
Most of the problems with the management and publication of
archaeological data on the Byrsa Hill fall into three categories: problems related to the distance between
archaeologists and technical illustrators, problems related to the distance between different archaeologists,
and problems related to the distance between archaeologists and publishing editors. I will discuss each of these
problems in turn.
7.1.1 Distance Between the Archaeologist and the Technical Illustrator
The most visible problems that my study has highlighted concern the
collection of published maps and the profiles that make up my initial dataset. Put simply, the geolocatability,
accuracy, and symbology of the published maps and profiles are often compromised, making them harder for other
scholars to understand and re-use. I believe this results from a general disconnect between the (archaeological)
illustrator and the archaeologist (writer). This disconnect prompts the question of what the role of a map
should be in an archaeological publication.
None of the early
maps of the Byrsa Hill follow a coordinate system,
nor do they provide reference datums—perhaps unsurprising, since most of them were produced by architects (e.g.,
Bonnet-Labranche, Thouverey), rather than geographers or cartographers. Consequently, they are not directly
geolocatable.
182
In contrast, the maps of the French Mission were created within a standard geographical coordinate
reference system and linked together with a single, site-specific coordinate system, but, as I showed in Chapter
4, the second system was not fully published, and the two systems were
never linked in the publications. The biggest improvement in the maps of the French Mission is their Still,
some maps can be georeferenced indirectly, if
their features also appear in another directly georeferenceable map. Beulé’s map (1861: pl. I)
cannot be georeferenced at all, since it contains numerous distortions and does not have a
consistent scale. Delattre’s map (1893a, pl.
XI; drawn by Bonnet-Labranche) also provides no datums, and
even though it does not show features found by Beulé (despite the fact that they are immediately
adjacent to it), it can be indirectly (if poorly) georeferenced relative to some of the
substructural walls of the southwest temple in
183
The biggest shortcoming of the maps of the French Mission is their feature symbology, which is often more
ambiguous than their predecessors, making it difficult—and sometimes impossible—to distinguish between extant,
reconstructed, and hypothesized features.
Beulé‘s (1861, pl. I) map is very
inaccurate (see e.g., n. 131). Delattre’s
map (1893a, pl. XI), compared to
my dataset (which is derived mostly from the Byrsa volumes) is probably accurate to within
3-4 m. Lapeyre's maps (1934,
pl. 1, 2), along with his discussion, selectively emphasize
certain features to support his interpretation of wall D (Byrsa I, 106-113)—see discussion in
Section 5.1.1.1 above, esp. n. 53.
As I explored in Chapter 6, problems with ambiguous symbology can also be found in the
profiles of the Byrsa volumes, which often “project” (i.e., unfold) features from multiple different
planes onto the plane of the illustration. This process of projection can create discontinuities that are not
always shown in the profile (e.g.,
In a project of the magnitude of the Byrsa volumes, some
inconsistencies and errors are bound to crop up, but the sheer quantity and scale of the problems with the maps
and profiles, combined with the disregard for them in subsequent publications, suggests to me
The simple fact is that since the beginning of the discipline, many,
if not most, archaeologists of the Mediterranean have been unable (or unwilling) to draw their own maps and
profiles, relying instead on architects, cartographers, or artists for their technical illustrations.
185
While this division of labor tends to produce illustrations that are clean, professional, and often quite
beautiful, it also creates a multitude of problems, not simply logistical (how does the archaeologist coordinate
the illustrators with the excavators?), or ethical (how should the illustrator be credited?),The rise of digital recording and drawing techniques
has left this traditional division of labor largely unchanged, though the scheduling of that labor
has been weighted more heavily towards the post-processing phase.
186
but also, as I have shown through numerous examples in Chapters 5 and 6,
epistemological. In the context of the Byrsa Hill, this division is partly why later scholars of the
site, when reproducing or citing problematic maps and profiles from the French Mission, have tended not to point
out or engage with those problems.An apt comparison for the question of credit may be
found in the film industry. The odds of a technical illustrator for an archaeological publication
getting partial authorship are about as bad as those of a visual effects artist for a movie getting
included in the opening credits. This runs completely contrary to authorship norms in many natural
science fields, where the creation of visual data often requires that one receive partial
authorship.
187
More broadly, this division is one of the main reasons why
archaeologists very commonly do their analyses
and write their reports using maps and profiles that are different
from maps they publish. As long as this
practice continues, publications will continue to be produced that do
not actually report the data on which their results are based. 7.1.2 Distance Between Different Archaeologists
Given that the
Byrsa Hill has been the subject of excavations for over
160 years, it is striking, but not surprising, how disjointed and insular the efforts of various researchers
have been. Delattre, in his published work, largely avoided the places where Beulé had excavated,
188
and the same is mostly true for Lapeyre. Consequently, as I outlined above in Chapter 3, the excavation history of the Byrsa is a patchwork, with
excavations conducted all over by different researchers with different motivations, frameworks, and approaches.
Some
notable exceptions to this trend include
Delattre's (1893b; 1894) excavation of the amphora wall in the southeast corner of
the complex, part of which was also found by Beulé (1861, 39), and Delattre's
excavations slightly west of the eastern perimeter, which was first excavated by (1861, 66-74). I have been unable to
locate the publications corresponding to Delattre's latter excavations—two articles are listed in
Ennabli (2020, 89), but one of them (“RT, 1901, p. 280”) seems to be incorrect, and the other
(“BSAF, 1901, 131”) uses an acronym that is not defined in the text.
The prevailing attitude seems to have been that in places where
excavations had already occurred, there could not be any more remains of interest. This attitude is to some
extent understandable. Firstly, each of the major campaigns was separated from the others by 20-30 years, and
during those interludes features which had been uncovered were deteriorated due to a combination of spoliation
and erosion. Secondly, the fixation of all of the excavators on tombs and inscriptions, combined with their
penchant for large, open-air excavation methods, all but ensured that no material of interest to them would
remain in a given area after a single campaign there.
On the other hand, this attitude was also an outgrowth of the colonial
explorer mentality that has characterized most of European and North American archaeological research in North
Africa—and many other places around the world—since the beginning of the discipline. Very often, the “discovery”
and excavation of a feature bestows some sort of
189
and in the tendency of each campaign to omit features found by other excavators in their own maps.For a brief overview of some of these systems, see
Section 2.2.
190
The result of this mentality is that, with a few exceptions, for more a century, excavations on the Byrsa largely functioned as isolated scholarly domains.
Bonnet-Labranche, who drew the map published in (1893a, pl.
XI), did not bother to include south perimeter apses found by Beulé, even though they
were probably less than 10 m from the apses excavated by Delattre. Thouverey, who drew the maps
published in (1934, pl. 1, 2), did not bother to include
outlines of walls N,
S, O, and other perimeter structures along the southwest and east perimeter, even
though he certainly knew about them and they were still visible (1934, 344).
The excavators of the French Mission of the 1970s and 80s were the
first to attempt to unify these disparate domains into a collection of more-or-less coherent datasets, which
they presented in individually-authored site reports. To some extent, they were successful: this thesis is a
testament to that success, since the overwhelming majority of my data
192
Other features are basically not described at all, either due to a simple lack of prioritization, as
happened with the south slope retaining wall (see Section 5.1.1.7) and wall
G (see Section 5.1.1.2), or due to a misunderstanding, as with the western
C piles.193
As noted in Section 5.1.2.2, Carrié and Sanviti (in Byrsa I, 132) refer the reader to
Deneauve and Villedieu's report (also in
Byrsa I) for discussion on the relationship between the western C piles and wall
N. This is an empty reference, since Deneauve and Villedieu basically do not discuss the
relationship.
These problems are exacerbated by the absence of an inventory, index,
and precise ID system for extant architectural features in the Byrsa volumes.
194
The investigations of Roman architecture in these volumes are characterized by a search for extant
sections of previously recognized, pre-defined features. Granular identification is eschewed for small,
recognizable names such as “wall D,” and “pile C1.”
While this system generally works well for a
small set of features that are obvious and well-defined—a description
which admittedly applies to many of the
Roman remains on site—it has some notable shortcomings. First, it does
not discriminate between sections of a feature, which creates
ambiguities for very large or
partially destroyed features (also applicable to most of the Roman
remains) and makes it difficult to tell
whether a feature under discussion has been actually found, or was
only hypothesized or reconstructed. Second,
their ID system is not On this problem in Byrsa III, see n. 152 above. Note that this lack of precision does not
extend to the remains of Punic pottery, which are given unique, context-specific IDs, and tombs,
which are also given unique IDs and mapped and photographed exhaustively.
195
See e.g., the many miscellaneous substructural walls
discussed in Section 5.1.1.7. The lack of IDs for these features
in the Byrsa volumes made this section especially difficult to write.
7.1.3 Distance Between the Archaeologist and the Editor
Many of the problems in the previous two categories are reinforced
by
constraints imposed by the systems of archaeological publication. A book-length primary site publication cannot
be constructed around a systematically extensible feature ID system if the publisher is not willing to provide
the editing resources required for a large and robust cross-referencing system.
196
Similarly, such a publication cannot include a comprehensive dataset if the publisher is not willing to
print a large number of high-quality maps, profiles, and many pages of find inventories. While these
restrictions are to some degree out of aesthetic or rhetorical concerns, the primary motivation is economic. For
essentially all archaeological publishers, the first concern must be whether a product will be financially
feasible. When a book’s expected sales are numbered in the hundreds, rather than the thousands, publishers are
often unwilling to pay the costs required for overcoming these problems, because their very existence is
dependent on how well their products suit the market. Though this requirement has long been a major barrier
to densely-referenced publications, this thesis demonstrates that freely available tools can largely
eliminate this barrier for archaeologists who are willing to learn new document-preparation systems.
In addition to the
economic considerations, there are professional
factors at play. For over a century, excavations on the Byrsa Hill were almost exclusively conducted by public
officials and priests who were curious, well-connected, and largely untrained. These were amateurs—in the
original sense of the word—and their goal was not to compose comprehensive writeups of their excavations, but to
communicate their most exciting finds to their wider, and also largely amateur, audience.
197
After Beulé’s singular, self-funded expedition and monograph chapter (1861, 1-84),
the next century of publications on the site was dominated by small,
bite-sized communications (e.g., Delattre 1893a; 1893b; 1894; 1924b) with some very uncommon longer articles
(e.g., 1934) or thesis sections (e.g., in 1901). The systems of
archaeological publciation in these decades
prioritized small, self-contained preliminary reports, discouraging the collation and organization of data into
unified frameworks or archives. Of course, this relationship was complex (excavators had many different priorities
aside from
publishing), and recursive (the publications that excavators read probably somewhat influenced how they wanted
to excavate). Yet, the fact is that no publication longer than Beulé’s monograph chapter was produced until (1955), and no wider-ranging, more comprehensive publication was produced until Byrsa I (published 1979).
It almost goes without saying that the vast majority of the data not
published by Beulé, Delattre, Lapeyre, and other excavators has been lost.
198
The result of this is that for at I am not sure how much archival work has been done to
confirm this. (2005,
67 n. 251) indicates that the archives of the Pères
Blancs have been searched fruitlessly for maps of the area disturbed by the construction of
the cathedral. It is less clear if descendants of Beulé, Delattre, Lapeyre, and other early
excavators have been contacted to check for the existence of excavation archives among family
heirlooms.
199
On the lack of a centralized archive, see n. 2. Serge Lancel passed away in 2005 ( 2005), and Jean Deneauve passed away in 2017 ( 2018). I am not sure about Jean-Michel Carrié, Nicole Sanviti,
Jean-Paul Thuillier, and Françoise Villedieu, who are all co-authors on Byrsa I. I am also not sure
about Gérard Robine and Philipe de Carbonnières, who drew most of the maps and figures for the
Byrsa volumes. According to Academia Europaea, Pierre Gros remains on faculty as a
professor emeritus at Université de Provence (https://www.ae-info.org/ae/Member/Gros_Pierre), but
this school no longer exists, having merged in 2012 to become Université Aix-Marseille. According to
the Centre Camille Julian, Jean-Paul Morel is also a professor emeritus at the Université de
Provence (https://ccj.cnrs.fr/spip.php?article48).
7.2 Reforming the Publication of Archaeological Research
As I discussed in Chapter 1, these
problems of publication are endemic, not just in the archaeology of Carthage, but in Mediterranean archaeology
more generally. Although they have, in some circles, occasionally provoked genuine inquiry (e.g., 1975; 1983; 1992) and introspection (e.g., 2004; 2010; 2013), the problems have
proven largely intractable, and according to some (e.g., 1996; 1996b; 1996), they have grown
significantly worse. The lack of
I believe that this predicament constitutes an existential threat to
the field, because it threatens our academic integrity, it is incredibly wasteful for researchers of legacy
data, and it harms the mental and financial health of the multitude of archaeologists who have not been trained
to efficiently and sustainably collect, analyze, and publish data at the scale of an entire site. The threat to
our academic integrity has already been pointed out, e.g., by (1996, 103-104), who noted that absent or incomplete publications create inequities in
data accessibility. This makes it very difficult, and sometimes impossible, for other researchers to critique
the excavators' results or propose alternate interpretations of their published data, since the excavators are
privy to meaningful information that has not yet been published.
200
When additional publications are not forthcoming, because the site director has moved on to another
project or (more often) passed away, they often take with them an intimate knowledge of the organization and
contents of their datasets, leaving behind scant documentation of these topics for future researchers and making
the task of consuming, synthesizing, and arranging the remaining data for publication far more expensive than it
already was. Sometimes, the cause for such languishing is not that the site director left or passed away, but
that they simply cannot manage the data themselves. Perhaps the dataset has become too sprawling and variegated
for them to effectively understand ( 1996, 42; 1996, 71); or maybe they have run out of funding, their In Herzog’s assessment (1996, 104), this makes the field of archaeology an “imprecise science,”
but I feel more strongly: this makes the field of archaeology not a science at all.
“The archaeologists who fail to publish are not, for the most part, bad people. They are not disinterested or intentionally neglectful or lazy. They fail to publish, for one reason or another, because they can't. They simply cannot do it. It is a mountain they cannot climb whether for legitimate or psychological or other reasons. In short, they need help. And it should be the task of the profession to find some means to give them that help.”
This crisis requires a combination of solutions. First, we must
rethink and redefine what it means to properly publish archaeological research. It has been clear for quite some
time that the monograph is not well suited to the task of publishing data at the scale of a site. It is terribly
tedious and inefficient to typeset thousands of pages of documentation, drawings, and spreadsheets into a
manuscript, and it is terribly expensive to print it. The rise of digital publication technologies has offered
an unprecedented opportunity to rethink the publication norms of our field, enabling researchers to forego the
costs of printing and to use more flexible formats than manuscripts. These formats include self-published online
databases (e.g., 2019; 2020),
interactive site maps (e.g. 2016; 2020), and
linked-data e-books (e.g., 2016). Unfortunately, such technologies
remain significantly under-used, in no small part because they tend to be under-represented in Review, Tenure,
and Promotion (RPT) requirements. There needs to be a concerted effort to redefine RPT requirements to encourage
such publications.
Second, there must be a curricular shift toward data management and
digital publication. Far too many archaeologists, particularly in the Mediterranean, never complete coursework
on data management. Instead, they learn the data recording and organizational strategies of their advisor, or
perhaps their colleagues, on the fly. Without formal training, it can be difficult to create a data management
plan that is comprehensive (describing all data collected within the project), easy for other researchers to
understand, flexible enough to accommodate the addition of new types of data, and systematically extensible to a
sufficiently large number of loci. Similarly important (and under-used) is coursework on digital
publication. All of the digital publishing formats mentioned in the previous paragraph require a significant
amount of training to create, and if researchers do not have at least some experience with them, they are
unlikely to seek out such training on their own. Moreover, it is often very difficult for an archaeologist to
effectively use such technologies if she did not factor them into her original data management plan.
Third, we must change the mentality—still all too common in our
profession—that the publication of site reports represents a necessary evil and the “last stage of research”
(pace 1996, 88). Complete and open
communication is essential to scientific inquiry and academic community. If it is prohibitively difficult to
communicate one’s data completely and openly, then something has gone wrong. Doing archaeological research
should be predicated on participating in the archaeological community—in other words, excavation should be
predicated on publication, not the other way around. The reticence of many project directors toward full
publication belies the nature of the threat posed by publication delays: the greatest risk of a project director
moving on or passing away is not that their interpretations will be lost. The greatest risk is that their
knowledge and understanding of
In this thesis, I have demonstrated that it is possible to restructure
old archaeological data in a sustainable and systematically extensible way.
201
There are undoubtedly flaws and gaps in my work, but I was still able to resolve many of the problems
inherent in prior publications. I have also demonstrated that the restructuring of old data can give rise to new
interpretations and questions about a site, some of which were not possible within the old organizational
framework. In the final section of this chapter, I will discuss some of those interpretations and questions for
the Byrsa Hill. Two computational tools were essential in this work.
The first tool was QGIS, a geospatial mapping program. As I have shown many times in the preceding
chapters, previously published maps of the Byrsa have too many inconsistencies for any one of them
to be completely reliable, and GIS software made it generally straightforward to identify and
resolve those inconsistencies. Another essential tool was LyX, a document preparation program. The
simple truth is that this thesis could not have been written in Microsoft Word. There are too many
figures, tables, bibliographic links and cross references for Word to reliably handle, and the
proprietary formatting of .docx files ensures that it is all but impossible to fix references if
they break or retrieve data if the file becomes corrupted. By using LyX, an open-source
What-You-See-Is-What-You-Mean word processor built on the LaTeX document preparation system, I had
access to extremely robust cross-reference and hyper-reference systems, all while writing plain
text-formatted files.
7.3 What can we do with Restructured Data?
The restructured dataset has enabled new interpretations about the
design and development of the Byrsa complex, particularly with regards to the substructural walls and perimeter
structures in and along the complex’s southern side. Some of these interpretations require additional proof,
highlighting certain areas of the site and certain questions for further investigation.
7.3.1 New Organization Enables New Interpretations
The most
significant new interpretation that I have proposed above is
that the Phase 1 substructural systems and perimeter structures and of the Byrsa complex were constructed in a
modular, piecewise fashion. This flexible approach allowed construction crews to adapt more easily to the local
availability of stone rubble from the Punic destruction layer, while still following the original design of the
complex. The approach is evident in wall D’s variable and somewhat crooked construction, both of which
were obscured by the reductive symbology used to depict it in the maps of the Byrsa volumes. The
approach is also evident in the vastly different dimensions and construction of the apses lining the
southwestern, southern, and eastern apses. The apses of Series 1 and 2 were significantly wider and had much
thinner walls than the apses of Series 3.
202
The apses of Series 3 themselves exhibited a remarkable shift in dimensions, growing significantly wider
from west to east, and they seem to have been significantly longer than the apses of Series 4, although I
believe that this last discrepancy may have been at least partially made up for by I have also proposed several smaller interpretations that still want
for additional proof. First, I showed that the late dating of the B
piles that was proposed by the original excavators seems
questionable, and I suggested some reasons why a Phase 1 dating may
make more sense. Second, I proposed an order and rationale for the
development of the substructural walls of the Phase 2 southwest
temple, and showed how the interaction between these walls and the B piles and C piles was
probably more complicated than previously acknowledged. Third, I reconstructed the approximate extents of the
amphora wall in the southeast corner of the complex, and showed that it may have functioned in concert with the
southeast staircase nearby. Finally, after pointing out that the previously proposed dimensions of the Phase 1
eastern apses have no basis in material evidence, and in fact contradict the construction strategies used in the
southwest corner of the complex, I hypothesized that the Phase 1 dimensions may have been identical to the Phase
2 ones.
7.3.2 New Interpretations Prompt New Questions
These preliminary interpretations and hypotheses highlight the
attractiveness of certain areas of the site for further investigation, and prompt new questions about the
complex’s design and development. A fuller excavation (and publication) of the substructural walls of the Phase
2 eastern basilica would probably provide helpful constraints on the dimensions and design of the building’s
Phase 1 predecessor, particularly in the southwest corner, where the building’s substructures met wall D’
(built in Phase 1). A new field survey
The complex’s design
and development could be greatly clarified by
making a serious investigation of the structural requirements imposed by topography of the Punic destruction
layer and the volumes of earth and debris found above it. In the past, excavators have tended to make vague
statements about the role of the apses, the substructural walls, and the foundation piles in resisting outward
pressure from the Augustan embankment. These statements are very shallow, from an engineering standpoint. What
sorts of pressures did these structures need to resist? How large were they, and in what directions? How did
these pressures change over time, as the soil naturally shifted, compacted, and creeped? How did these changes
influence the decisions which governed the construction of Phase 2 structures? These are necessary questions,
but they cannot be answered without a comprehensive dataset describing the topography of the Punic destruction
layer and the vertical and horizontal extents of all of the Roman and Punic features. As my preliminary work in
Chapter 6 seems to indicate, such a dataset may well be
obtainable at least for the southern part of the complex.
Such a dataset could also reveal how the Phase 1 Roman construction
crews navigated the site’s Punic remains. By integrating the horizontal and vertical extents of both Roman and
Punic features, we could determine how the various Roman substructures cut, abutted, and included Punic walls,
cisterns, tombs, and other features. This would offer insight into what sorts of Punic features the Phase 1
Roman construction crews would have seen during their work, and what they, or their superiors, decided to do
about them. Which Punic walls were removed to make way for Roman ones, and which ones were merely included? How
many tombs did the construction crews encounter, and how many did they fully or partially destroy? How might the
presence or discovery of tombs have influenced the construction decisions that followed—e.g., as in the
southwest temple? These, too, are necessary questions, which cannot be answered without restructuring the
current architectural dataset for the Byrsa complex.
These questions touch on a deeper theme underlying the foundation of
the Roman colony at Carthage, a theme that has been largely neglected by modern scholars of this site: the role
of reconstruction and erasure in shaping public and private narratives about the past and present. Still bearing
the physical and psychological scars of the recent violent upheavals at the end of the Republic, the Romans who
labored on the colony’s foundation literally took apart the remains of their country’s greatest geopolitical
rival—and deeply despised nemesis—stone by stone, and then they re-made the city in Rome’s own ideal image. What
did this project mean to the laborers, surveyors, architects, artisans, and soldiers who undertook it? What did
it mean to their families and children? How did it influence their perception of the new political order? These
are questions that can and should be asked of this site. By collating, digitizing, restructuring, and
re-publishing archaeological
References
Aksoy and Bayar
2016, "Archaeological Ethnography of the Battle of Aslıhanlar
(29–30 August 1922): A Case Study of Public Archaeology, Visual Storytelling, and Interactive
Map Design", European Journal of Archaeology
19, 1 (2016),
pp. 68--94.
Alperin et al.
2021, "The Value of Data and Other Non-Traditional Scholarly Outputs in Academic
Review, Promotion, and Tenure in Canada and the United States", in Open Handbook of Linguistic Data Management (Cambridge, MA: MIT Press, 2021).
Altekamp and
Khechen 2013, "Third Carthage: Struggles and Contestations Over
Archaeological Space", Archaeologies 9, 3 (2013), pp.
470--505.
Architecture
2018, "The Destruction Of Yazidi Cultural Heritage",
Forensic Architecture (2018).
Audollent
1901, Carthage romaine, 146 avant Jésus-Christ-698 après
Jésus-Christ (Paris: L'Université de
Paris, 1901).
Babelon 1896, Carthage (Paris: E. Leroux, 1896).
Bates 2019, "The
Published Archaeobotanical Data from the Indus Civilisation, South Asia, c.3200–1500BC",
Journal of Open Archaeology Data 7,
0 (2019), pp. 5.
Benichou-Safar
1976, "Carte Des Nécropoles Puniques de Carthage",
Karthago XVII (1976), pp. 5--35.
Bernbeck 2010, "Heritage Politics: Learning from Mullah Omar?", in Controlling the Past, Owning the Future: The Political Uses of
Archaeology in the Middle East (Tucson: University of Arizona Press, 2010),
pp. 27--54.
Bernbeck and
Pollock 2016, "`Grabe, Wo Du Stehst!' An Archaeology of
Perpetrators", in Archaeology and Capitalism : From Ethics
to Politics 1st edition (New
York: Routledge, 2016), pp. 217--233.
Beulé 1861, Fouilles à Carthage aux frais et sous la direction de M. Beulé
(Paris: Imprimerie Impériale,
1861).
Boivin 1997, "Insidious or Just Boring? An Examination of Academic Writing in
Archaeology", Archaeological Review from
Cambridge 14, 2
(1997), pp. 105--125.
Callot 1986, "La
région nord du palais royal d'Ugarit", Comptes rendus des séances de
l'Académie des Inscriptions et
Belles-Lettres 130, 4 (1986), pp. 735--755.
Campbell 2007, "Beulé, Ernest-Charles-Auguste", in The
Grove Encyclopedia of Classical Art and Architecture (Oxford: Oxford University Press,
2007).
Carver et al.
1992,
"Archaeological Publication, Archives and Collections: Towards a National
Policy", British Archaeological News 7 (1992).
Cronin and La Barre
2004, "Mickey Mouse and Milton: Book Publishing in the
Humanities", Learned Publishing 17, 2 (2004), pp.
85--98.
de Chateaubriand
1859, Itinéraire de Paris à Jérusalem et de Jérusalem
à Paris (Paris: Garnier,
1859).
Cunliffe 1983, "The Publication of Archaeological Excavations: Report of a Joint Working Party
of the Council for British Archaeology and the Department of the Environment", Department of the Environment (1983).
Curl 2006, A Dictionary of Architecture and Landscape Architecture Second (Oxford University Press,
2006).
Davin 1930, "Etude sur
la cadastration de la Colonia Julia Carthago", Revue
Tunisienne 2 (1930), pp. 73--85.
Delattre 1891, "Tombeaux puniques de Carthage: Nécropole de la colline de Saint-Louis",
Revue Archéologique XVII
(1891), pp. 52--69.
Delattre 1893a, "Fouilles archéologiques dans le flanc sud-ouest de la colline de Saint-Louis
en 1892", Bulletin archéologique du Comité des travaux
historiques et scientifiques (1893), pp. 94--123.
Delattre
1893b, "Un mur à amphores romaines découvert à l'angle sud
de la colline de Byrsa (Carthage)", Comptes rendus des
séances de l'Académie des Inscriptions et Belles-Lettres 37,
3 (1893), pp. 152--155.
Delattre 1894, "Le mur à amphores de la colline Saint-Louis à Carthage", Bulletin archéologique du Comité des travaux historiques et
scientifiques (1894), pp. 89--119.
Deneauve 1983, "Le
tracé monumental de Byrsa à l'époque romaine: État des recherches",
Cahiers des Études Anciennes 16
(1983), pp. 89--106.
Deneauve 1990, "Le
centre monumental de Carthage: Un ensemble cultuel sur la colline de
Byrsa", in Histoire et Archéologie de l'Áfrique du Nord:
Carthage et son territoire dans l'Antiquité: Actes du IVe colloque international sur
l'histoire et l'archéologie de l'Afrique du nord vol. I,
(Paris: Comité des Travaux Historiques et
Scientifiques, 1990), pp. 143--155.
Dever 1996, "The
Importance of Research Design", in Archaeology's Publication
Problem vol. 1, (Washington,
DC: Biblical Archaeology Society, 1996),
pp. 37--48.
Duval 2005, "In
memoriam Serge Lancel (1928-2005)", Revue d'Etudes Augustiniennes et
Patristiques 51, 2 (2005), pp. I-V.
Effros 2018, Incidental Archaeologists: French Officers and the Rediscovery of Roman
North Africa (Ithaca, NY: Cornell University Press, 2018).
Ennabli 2020, Carthage: "Les Travaux et les Jours": Recherches et Découvertes
1831-2016 (Paris: CNRS
Éditions, 2020).
Estabrook and
Warner 2003, "The Book as the Gold Standard for Tenure and
Promotion in the Humanistic Disciplines: Report for the Committee on Institutional
Cooperation", Committee on Institutional Cooperation
(2003).
Faniel et al.
2013, "The
Challenges of Digging Data: A Study of Context in Archaeological Data
Reuse", in Proceedings of the 13th ACM/IEEE-CS Joint
Conference on Digital Libraries (New York, NY,
USA: Association for Computing Machinery, 2013), pp. 295--304.
Ferron and Pinard
1955, "Les Fouilles de Byrsa: 1953-1954", in Cahiers de Byrsa (Paris:
Imprimerie Nationale, 1955), pp.
31-81 and pl. I-LXXVI.
Ferron and Pinard
1960, "Les Fouilles de Byrsa (Suite)", in Cahiers de Byrsa (Paris:
Imprimerie Nationale, 1960), pp.
77-170 and pl. I-XCV.
Freed 2001, "Bibliograph of Publications by Alfred-Louis Delattre (1850-1932)",
Centre d'Etudes et de Documentation Archéologique de Carthage
(Carthage) 20 (2001), pp. 3--60.
Freed and Moore
1996, "New Observations on the Earliest Roman Amphoras from
Carthage: Delattre's First Amphora Wall", Centre d'Etudes et
de Documentation Archéologique de Carthage (Carthage) 20
(1996), pp. 19--28.
Frere 1975, "Principles of Publication in Rescue Archaeology: Report by a Working Party of
the Ancient Monuments Board for England", Committee for
Rescue Archaeology (1975).
Gros 1982, "Le forum
de la haute ville dans la Carthage romaine d'après les textes et
l'archéologie", Comptes rendus des séances de l'Académie des
Inscriptions et Belles-Lettres 126, 3 (1982), pp. 636--658.
Gros 1985, Byrsa
III: Rapport sur les campagnes de fouilles de 1977 à 1980: la
basilique orientale et ses abords vol. 41, (Rome: École Française de Rome,
1985).
Gros 1990, "Le
premier urbanisme de la Colonia Julia Carthago", in L'Afrique dans
l'Occident romain: Ier siècle av. J.-C. - IVe siècle
ap. J.-C.: Actes du colloque de Rome (3-5 décembre 1987) (Rome: École Française de Rome,
1990), pp. 547--573.
Gros 1992, "Colline
de Byrsa: Les Vestiges Romains", in Pour sauver Carthage.
Exploration et conservation de la cité punique,
romaine et byzantine (Paris: UNESCO/INAA, 1992), pp. 99--104.
Gros 1997, "Les
Bâtiments Administratifs de La Carthage Romaine. Problèmes
d'identification et de Localisation", Mitteilungen des
Deutschen Archäologischen Instituts Römische Abteilung 104
(1997), pp. 341--350.
Gros 2018, "In
memoriam : Jean Deneauve (1922-2017)", Antiquités africaines.
L'Afrique du Nord de la protohistoire à la
conquête arabe (2018), pp. 5--7.
Gros and Deneauve
1980, "Hypothèses sur le centre monumental de la Carthage
romaine, d'après les recherches récentes sur la colline de Byrsa", Comptes rendus des séances de l'Académie des Inscriptions et
Belles-Lettres 124, 2 (1980), pp. 299--332.
Gros and Deneauve
1996, "La Carthage romaine restituée", Archéologia (1996), pp. 54--61.
Gutron 2010, L'archéologie en Tunisie (XIXe-XXe siècles): jeux généalogiques sur
l'Antiquité (Paris and Tunis: Karthala and IRMC, 2010).
Herzog 1996, "With
Time, We're Getting Worse", in Archaeology's Publication
Problem vol. 1, (Washington,
DC: Biblical Archaeology Society, 1996),
pp. 87--110.
Hodder 1989, "Writing
Archaeology: Site Reports in Context", Antiquity 63, 239 (1989),
pp. 268--274.
Hurst 1994, Excavations at Carthage: The British Mission vol. II, 1, (Sheffield: Oxford University Press, 1994).
Isakhan and Meskell
2019, "UNESCO's Project to `Revive the Spirit of Mosul':
Iraqi and Syrian Opinion on Heritage Reconstruction after the Islamic State", International Journal of Heritage Studies 25, 11 (2019), pp.
1189--1204.
Jacobs 1996, "The
On-Line Digmaster Database", in Archaeology's Publication
Problem vol. 1, (Washington,
DC: Biblical Archaeology Society, 1996),
pp. 71--78.
Ladjimi-Sebaï
2005, La colline de Byrsa à l'époque romaine: étude
épigraphique et état de la question vol. 26,
(Paris: C.E.A.M., Institut d'Art et
d'Archéologie, 2005).
Lancel 1977, "Le
secteur A", Antiquités
africaines 11, 1
(1977), pp. 19--49.
Lancel 1995, Carthage: A History (Oxford:
Blackwell, 1995).
Lancel et al. 1979, Byrsa I: Rapports préliminaires des fouilles
(1974-1976) vol. 41, (Rome:
École Française de Rome,
1979).
Lancel, Morel, and Thuillier
1982, Byrsa II: Rapports
préliminaires sur les fouilles 1977-1978: Niveaux et vestiges puniques (Rome: École française de Rome,
1982).
Lapeyre 1934, "L'Enceinte Punique de Byrsa d'après Les Dernières Fouilles de La Colline
Saint-Louis de Carthage", Revue Africaine 75 (1934), pp. 336--353.
Lin 2003, Lading of the Late Bronze Age Ship at Uluburun (College
Station: Texas A&M
University, 2003).
Manière, Crépy, and
Redon 2020, "Geospatial Data from the ``Building
a Model to Reconstruct the Hellenistic and Roman Road Networks of the Eastern Desert of Egypt, a
Semi-Empirical Approach Based on Modern Travelers' Itineraries'' Paper", Journal of Open Archaeology Data 8, 1 (2020), pp. 7.
Margueron
1995, "La Palais Royal d'Oufgarit: Premiers Résultats
d'une Analyse Systématique", in Le Pays d'Ougarit Autour
de 1200 Av. J.-C. Histoire et Archéologie: Actes Du Colloque International, Paris, 28
Juin-1e Juillet 1993 (Paris: Éditions Recherche sur les Civilisations, 1995), pp. 183--202.
Matoïan 2008, "Les
Objets Du Palais Royal d'Ougarit: Un État de La Question", in
Le Mobilier Du Palais Royal d'Ougarit: Maison de l'Orient et de La
Méditerranée (Paris: Éditions
Recherche sur les Civilisations, 2008), pp. 17--71.
Mazar 1996, "Final
Reports: What They Should Include", in Archaeology's Publication
Problem vol. 1, (Washington,
DC: Biblical Archaeology Society, 1996),
pp. 23--36.
Meskell 2018, A
Future in Ruins: UNESCO, World Heritage, and the Dream of
Peace (Oxford, New York: Oxford
University Press, 2018).
Meskell 2020, "Sites
of Violence: Terrorism, Tourism, and Heritage in the Archaeological
Present", in Embedding Ethics 1st edition (London: Routledge, 2020), pp. 123--146.
Morel 2011a, "Les
fouilles de Byrsa (secteur B) à Carthage : un bilan", Comptes rendus
des séances de l'Académie des Inscriptions et
Belles-Lettres 155, 1 (2011), pp. 325--363.
Morel 2011b, "Mission archéologique de Carthage-Byrsa", Les
nouvelles de l'archéologie (2011), pp. 39--43.
Morgan 2013, "Publishing Archaeological Research: Comparative Perspectives on Common
Concerns", Mélanges de la Casa de Velázquez 43, 43-2 (2013),
pp. 279--288.
Opitz 2018, "Publishing Archaeological Excavations at the Digital Turn", Journal of Field Archaeology 43, sup1 (2018), pp. S68-S82.
Opitz, Mogetta, and
Terrenato 2016Opitz, Rachel,
Mogetta, Marcello, and Terrenato, Nicola, eds., A
Mid-Republican House from Gabii (Ann Arbor, MI:
University of Michigan Press, 2016).
Picard 1951, "Vestiges d'un Édifice Punique à Carthage", Karthago III (1951), pp. 119--126.
Pollock 2010, "Decolonizing Archaeology: Political Economy and Archaeological Practice in the
Middle East", in Controlling the Past, Owning the Future:
The Political Uses of Archaeology in the Middle East (Tucson: University of Arizona
Press, 2010), pp. 196--216.
Pollock 2016, "Archaeology and Contemporary Warfare", Annual Review
of Anthropology 45, 1 (2016), pp. 215--231.
Pulak 1998, "The
Uluburun Shipwreck: An Overview", International Journal of Nautical
Archaeology 27, 3 (1998), pp. 188--224.
Rakob 2000, "The
Making of Augustan Carthage", in Romanization and the City:
Creation, Transformations, and Failures:
Proceedings of a Conference Held at the American Academy in Rome to Celebrate the 50th
Anniversary of the Excavations at Cosa, 14-16 May, 1998 (Portsmouth: Journal of Roman
Archaeology, 2000).
Richards 2004, "Online Archives", Internet
Archaeology (2004).
Ronayne 2006, "Archaeology against Cultural Destruction: The Case of the Ilisu Dam in the
Kurdish Region of Turkey", Public Archaeology
5, 4 (2006),
pp. 223--236.
Ronayne 2016, "The
Culture of Caring and Its Destruction in the Middle East: Women's Work,
Water, War and Archaeology", in Archaeology and
Capitalism: From Ethics to Politics 1st edition
(New York: Routledge, 2016), pp. 247--265.
Saumagne
1924a,
"Colonia Iulia Karthago", Bulletin
archéologique du Comité des travaux historiques et scientifiques (1924), pp. 131--140.
Saumagne 1924b, "Notes de topographie carthaginoise: La colline de Saint-Louis",
Bulletin archéologique du Comité des travaux historiques et
scientifiques (1924), pp. 177--193.
Schimanski and
Alperin 2018, "The Evaluation of Scholarship in
Academic Promotion and Tenure Processes: Past, Present, and Future", F1000Research 7 (2018), pp. 1605.
Shanks 1996aShanks, Hershel, ed., Archaeology's Publication Problem vol. 1,
(Washington, DC: Biblical Archaeology
Society, 1996).
Shanks 1996b, "Getting Assistance in Writing Reports", in Archaeology's Publication Problem vol. 1, (Washington, DC: Biblical Archaeology Society, 1996),
pp. 49--54.
Shanks and Magness
1999Shanks, Hershel and Magness,
Jodi, eds., Archaeology's Publication Problem vol.
2, (Washington, DC: Biblical Archaeology Society, 1999).
Shaw 1738, Travels, or Observations Relating to Seveval Parts of Barbary and the
Levant (Oxford, 1738).
Styrenius et al.
1986, "The Swedish
Carthage Excavations: Preliminary Geodetic and Cartographic Studies", Bulletin of the Museum of Mediterranean and Near Eastern
Antiquities 21 (1986), pp.
81--91.
Townsend et al.
2020, "Digital
Archaeology and the Living Cherokee Landscape", International Journal
of Historical Archaeology 24, 4 (2020), pp. 969--988.
Appendix A Correcting Distortion in Figures from Persée
A.1 Summary of the Problem
One of the most important
repositories of scholarship on ancient
Carthage is Persée, an open-access digital library that contains an enormous selection of books, conference
proceedings, and journal articles spanning over 200 years of French archaeological research. Particularly
useful is Persée’s selection of books and articles on the Byrsa Hill, which include Byrsa I and
III (curiously, not Byrsa II), and several key studies in Comptes Rendus des Séances
de l'Académie des Inscriptions et Belles-Lettres (CRAIBL) and Cahiers des Études
Anciennes (CEA), including (1983) and
(1980). All of these publications have been manually scanned, de-skewed, and
optimized by Persée, and are available for download as OCR-enabled PDFs.
A significant number of figures in these PDFs are visibly
distorted. Three examples of distorted images from the Persée PDFs (
203
In all three examples, it In the interest of concision and clarity, all
images will be referred to by a three-part ID, which indicates the publication from which the
image was taken, the identity of the scanner, and the page number from the
Figure A.1: Three example pages
from Persée-scanned PDFs that exhibit distortion. Perfect squares are shown in red.
is clear that the reference grids are not square.
These images are not distorted in
their original paper
publications. In Figure A.2 are shown the same images as in
Figure 1, but scanned instead by me (
This image distortion is a major problem, because the distortion
is not slight. As will be shown below in Figures 5-7, features in depicted in the Persée PDFs diverge wildly
from their personally-scanned counterparts. These distortions are present in multiple publications (Byrsa I, Byrsa III, and 1980), across multiple years (1979, 1985, 1980), and in multiple formats
(book and journal article). This suggests that the
Figure A.2: Three pages from
personally-scanned PDFs that do not exhibit distortion. Perfect squares are shown in red.
distortions present in these PDFs may be found in many
other documents in the Persée system.
In the following sections, I diagnose the cause of the problem and
demonstrate a solution.
A.2 Diagnosing the Cause of the Distortion
A close examination of the three
examples discussed above suggests
a common cause for the distortion: all of the examples are composed of non-square pixels. Figure A.3 shows a close-up perspective on each of the three example pages from
Persée, with perfect squares shown in red for comparison. In all three examples, it is clear that the pixels
themselves are not square. Pixel aspect ratio (AR), calculated as the ratio of pixel width (w) to pixel
height (h), is calculated for each page in Figure A.3. These AR values vary
significantly. In
Figure A.3: Close-up view of the
three example pages shown in Figure A.1. Perfect squares
are shown in red. In all three pages, the pixels are not square.
Although square pixels are generally the
norm, digital images may
be encoded with different pixel aspect ratios when created in certain contexts and for certain purposes.
Even so, the pixel aspect ratios calculated from Figure A.3 are very
different from one another, and they do not match any of the standard non-square values used in common image
and video formats. Moreover, a quick survey of other pages in (1980), Byrsa I, and Byrsa III shows that every
page is composed of pixels with a different aspect ratio. These factors suggest that the different
aspect ratios may have been produced accidentally, perhaps during the post-processing of the original,
unoptimized scan files.
The next step was to export the three Persée PDF pages as JPEG
image files, so that key features in them could be traced and overlaid onto the pages I personally scanned.
By examining the sizes of the gaps between the two sets of features, it is possible to determine if the
mismatch is due only to a difference in pixel aspect ratio, or if there are additional sources of
distortion.
When the three PDF pages are exported as JPEG files, the
non-square pixel aspect
Figure A.4: Metadata for JPEG
files created from the three example PDF pages. Note that for
ratios are retained. The metadata of the resulting files are
presented in Figure A.4.
Notably, the JPEG files for
In the resulting JPEGs of
In each Figure, the trace was overlaid in such a way that one
point from it overlapped perfectly with its corresponding point in the image below. I then measured the
distances between this shared point of optimal overlap and five features. I calculated the gaps between
these two sets of distances, and plotted them against the distances for my personal scan. In each case, the
resulting plot was completely linear (R² > 0.99), indicating that the mismatch was caused by a different
linear scaling factor. This limits the potential
Figure A.5: Comparison of
Figure A.6: Comparison of
Figure A.7: Comparison of
causes of the mismatch to differences in pixel aspect ratio
and image resizing, and rules out non-linear causes such as excessive page curling. In each case, the slope
of the plot is essentially equal to the difference between the Persée aspect ratio and 1, which shows that
the gaps are caused entirely by a difference in aspect ratio.
A.3 Demonstrating a Solution
Because pixel aspect ratio is
solely to blame for the mismatch,
the solution to the problem is simple: edit the metadata of the Persée JPEGs and set the pixel aspect ratio
to 1. Image metadata is stored in EXIF format, and can be edited in a variety of ways, including via image
editing software (e.g., Adobe Photoshop), EXIF editing software (e.g., EXIF Pilot), and python scripting
(e.g., with the piexif package). Adjusting the image metadata in this way yields an almost perfect overlap,
as shown in Figure A.8.
Figure A.8: Comparison of
features in corrected Persée pages (red) and their personally-scanned counterparts. After the image
metadata of the former are edited to set pixel aspect ratio to 1, their features overlap perfectly
with the latter.
Appendix B Reproduced Figures
This appendix presents additional reproduced figures that did not
fit in the main text of the thesis. These figures are included here for the sake of transparency and
reproducibility, since my results cannot be reproduced if my initial dataset is not clear. All figures in
this appendix are also included as higher-resolution image files in the supplementary files accompanying
this thesis online.
All figures are presented with
either a citation or Figure ID. The
Figure ID system is explained in Section 2.1.
Figure B.1:
Figure B.2:
Figure B.3:
Figure B.4:
Figure B.5:
Figure B.6:
Figure B.7:
Figure B.8:
Figure B.9:
Figure B.10:
Figure B.11:
Figure B.12:
Figure B.13:
Figure B.14:
Figure B.15:
Figure B.16:
Figure B.17:
Figure B.19:
Figure B.20:
Appendix C Supplementary Tables
C.1 Feature Metadata
C.1.1 Axes of Reconstruction
Table C.1:
Metadata for axes of reconstruction.
| Axis ID | Description |
|---|---|
|
|
western façade of the corner room in |
|
|
line
of the reconstructed façade of the first 4 apses in the southern alignment in
|
|
|
east
façade of the basilica in |
|
|
north
façade of the basilica in |
|
|
line
between the NW corner of |
|
|
line
from the W side of |
|
|
line
extending eastward from the NE corner of |
|
|
line
extending eastward from the SE corner of |
|
|
line
extending eastward from the NE corner of |
|
|
line
extending eastward from the NE corner of |
|
|
line
extending eastward from the SE corner of |
|
|
line
from SW corner of |
|
|
line
along E side of |
|
|
central axis of |
|
|
line
extending 26 m parallel to |
|
|
line
from the center of |
|
|
line
between the center of |
|
|
east
façade of |
|
|
south
façade of basilica in |
|
|
west
façade of the basilica in |
|
|
|
C.1.2 Walls (MU)
Table C.2:
Metadata for walls.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed same dimensions as |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed same dimensions as |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed width about 1.5 m; assumed E and W sides
perpendicular to |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed width about 1.3 m; assumed E and W sides
perpendicular to |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed width about 1.1 m; assumed E and W sides
perpendicular to |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed width about 1.0 m; assumed E and W sides
perpendicular to |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed width about 1.1 m; assumed E and W sides
perpendicular to |
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 |
|
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 1 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 |
assumed dimensions similar to W part of |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed S side continued the curve of
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed S side continued the curve of
|
|
|
Roman | Phase 1, Phase 2 | rubble joined with lime, lined with opus reticulatum |
assumed curve of |
|
|
Roman | Phase 1, Phase 2 | rubble linked with lime |
assumed curve of |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed S side continued the curve of
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed S side continued the curve of
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 2 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 1 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 1 |
|
|
|
|
Roman | Phase 1 | opus quadratum (base); brick-like stonework (impost) |
|
|
|
Roman | Phase 1, Phase 2 | opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
Assume same size and orientation as |
|
|
Roman | Phase 1 | opus quadratum |
Assume same size and orientation as |
|
|
Roman | Phase 1 | opus quadratum |
Assume same size and orientation as |
|
|
Roman | Phase 1 | opus quadratum |
Assume same size and orientation as |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus reticulatum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed length such that apse opening was
proportional to that reconstructed in |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed length such that apse opening was
proportional to that reconstructed in |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed length such that apse opening was
proportional to that reconstructed in |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed length such that apse opening was
proportional to that reconstructed in |
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
assumed length such that apse opening was
proportional to that reconstructed in |
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed same dimensions as described by Beulé 1861:
60; assumed opening width of 2.28 m, proportional to openings of southwest
apses; assumed contiguous with façade walls of adjacent corner room; assumed S
side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed contiguous with façade wall of adjacent corner room; assumed E side
flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed contiguous with façade wall of adjacent corner room; assumed E side
flush with |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed thickness identical to |
|
|
|
Roman | Phase 1, Phase 2 |
assumed thickness identical to |
|
|
|
Roman | Phase 1, Phase 2 |
assumed thickness identical to |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed thickness assumed opening width of 2.28 m, proportional to openings of
southwest apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed thickness assumed opening width of 2.28 m, proportional to openings of
southwest apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed thickness assumed opening width of 2.28 m, proportional to openings of
southwest apses; assumed S side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed contiguous with façade wall of adjacent corner room; assumed E side
flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed E side flush with |
|
|
|
Roman | Phase 1, Phase 2 |
assumed wall width of 1.0 m following Beulé 1861: 60;
assumed opening width of 2.28 m, proportional to openings of southwest apses;
assumed contiguous with façade wall of adjacent corner room; assumed E side
flush with |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | assumed wall width of 1.0 m following Beulé 1861: 60; assumed opening width of 2.28 m, proportional to openings of southwest apses; assumed contiguous with façade wall of adjacent corner room | |
|
|
Roman | Phase 1, Phase 2 | assumed wall width of 1.0 m following Beulé 1861: 60; assumed opening width of 2.28 m, proportional to openings of southwest apses | |
|
|
Roman | Phase 1, Phase 2 | assumed wall width of 1.0 m following Beulé 1861: 60; assumed opening width of 2.28 m, proportional to openings of southwest apses | |
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 | opus quadratum |
|
|
|
Roman | Phase 1 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Roman | Phase 1, Phase 2 | assumed dimensions and shape described in Delattre (1894: 89-90) | |
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
assumed to extend to |
|
|
|
Roman | Phase 2 |
assumed to extend to |
|
|
|
Roman | Phase 2 |
assumed to extend east until |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | assumed smooth façade | |
|
|
Roman | Phase 1, Phase 2 | assumed smooth façade | |
|
|
Roman | Phase 1, Phase 2 |
assumed S side flush with |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | rubble linked with earth |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | rubble linked with earth |
|
|
|
Roman | Phase 1, Phase 2 | rubble linked with earth |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | rubble linked with mortar |
assumed to exist based on analogy with adjacent
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1, Phase 2 | large blocks (El Haouaria) linked with dovetail seals |
|
|
|
Roman | Phase 1 | rubble linked with earth |
|
|
|
Roman | Phase 1 | rubble linked with earth |
assumed analagous to |
|
|
Roman | Phase 2 | opus caementicium |
assumed analagous to |
|
|
Roman | Phase 1, Phase 2 |
assumed analagous to |
|
|
|
Roman | Phase 1, Phase 2 | rubble linked with mortar |
assumed analagous to |
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1 |
assumed E side abutting |
|
|
|
Roman | Phase 1, Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 |
|
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 2 | opus quadratum |
|
|
|
Roman | Phase 1, Phase 2 |
assumed constant width of 1.75 m, following Deneauve
1983: 105 fig. 8; assumed N and S sides perpendicular to |
|
|
|
Roman | Phase 1, Phase 2 |
assumed 58.94 m away from |
|
|
|
Roman | Phase 1, Phase 2 |
assumed constant width of 2.3 m, following Deneauve
1983: 105 fig. 8; assumed N and S sides perpendicular to |
|
|
|
Roman | Phase 1, Phase 2 |
assumed 10.1 m away from |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 1 |
|
|
|
|
Roman | Phase 1 |
|
|
|
|
Roman | Phase 1 |
|
C.1.3 Piles (PI)
Table C.3:
Metadata for foundation piles.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 |
assumed location halfway between |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
|
|
|
Roman | Phase 1 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1, Phase 2 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1, Phase 2 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1, Phase 2 |
assumed location along the sevenths between |
|
|
|
Roman | Phase 1, Phase 2 |
assumed original length and width of |
|
|
|
Roman | Phase 1, Phase 2 |
assumed original length and width of |
|
|
|
Roman | Phase 1, Phase 2 |
assumed original length and width of |
|
|
|
Roman | Phase 1, Phase 2 |
assumed original length and width of |
|
|
|
Roman | Phase 1, Phase 2 |
assumed original length and width of |
|
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stone rubble and mortar |
assumed length and width of 1.8 m; assumed spacing of
4.8 m; assumed alignment from |
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
|
|
|
Roman | Phase 2 | opus caementicium |
assumed original width about 1.8 m as shown in
|
|
|
Roman | Phase 2 | opus caementicium |
assumed original width and length about 1.8 m as
shown in |
|
|
Roman | Phase 2 | opus caementicium |
assumed length and width of 1.8 m; assumed spacing of
4.4 m; assumed alignment from |
|
|
Roman | Phase 2 | opus caementicium |
assumed length and width of 1.8 m; assumed spacing of
4.4 m; assumed alignment from |
|
|
Roman | Phase 2 | opus caementicium |
assumed length and width of 1.8 m; assumed spacing of
4.4 m; assumed alignment from |
|
|
Roman | Phase 2 | opus caementicium |
assumed length and width of 1.8 m; assumed spacing of
4.4 m; assumed alignment from |
|
|
Roman | Phase 2 | opus caementicium |
assumed length and width of 1.8 m; assumed spacing of
4.4 m; assumed alignment from |
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay | assumed original width and height about 1.65 m |
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay | assumed original width about 1.7 m |
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 | stones bound with clay |
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
C.1.4 Cisterns (CI)
Table C.4:
Metadata for cisterns.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
assumed central axis along |
|
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
C.1.5 Staircases (ES)
Table C.5:
Metadata for staircases.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
C.1.6 Tombs (TO)
Table C.6:
Metadata for tombs.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Punic |
|
||
|
|
Punic |
|
||
|
|
Punic |
|
C.1.7 Rudus (RU)
Table C.7:
Metadata for rudi.
| Feature ID | Period | Subperiod | Construction | Source |
|---|---|---|---|---|
|
|
Roman | Phase 1, Phase 2 |
|
|
|
|
Roman | Phase 1, Phase 2 |
|
C.2 Feature Locations
C.2.1 Walls (MU)
Table C.8:
Locations of walls.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
C.2.2 Piles (PI)
Table C.9:
Locations of foundation piles.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
C.2.3 Cisterns (CI)
Table C.10:
Locations of cisterns.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
C.2.4 Staircases (ES)
Table C.11:
Locations of staircases.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
C.2.5 Tombs (TO)
Table C.12:
Locations of tombs.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
|
|
|
C.2.6 Rudus (RU)
Table C.13:
Locations of rudi.
| Feature ID | Location |
|---|---|
|
|
|
|
|
|
Appendix D Corrective Notes for Previously Published Profiles and Maps
While determining the extents of the profiles, some
inconsistencies and ambiguities became evident. These are identified below, along with the solutions I used
to resolve them. In this appendix, I always use the term “trench” to refer to a unit of
excavation, unless otherwise specified. This is equivalent to the term “sondage” in the original
French.
D.1 Byrsa I
D.1.1 The extents of
profiles
The extents of
D.1.2 The extents of
profile
The extents of
D.1.3 The extents of
profile
In profile
The contexts and features in
204
It is surprising that this mistake made it into
two different maps of the sector. Perhaps
D.1.4 The extents of
profile
There is no label for profile
D.1.5 The scale given for
profile
The figure caption for profile
D.1.6 The scale given for
map
The figure caption for map
D.2 Byrsa II
D.2.1 The extents of
profile
The extents of profile
D.2.2 The extents of
profile
The extents of profile
D.2.3 The extents of
profile
The extents of profile
D.2.4 The extents of
profile
The extents of profile
D.2.5 The extents of
profile
The extents of
D.2.6 The extents of
profile
As with
D.2.7 The extents of
profiles
The extents of profiles
D.2.8 The grid cell label
for profile
The grid cell label of profile
D.2.9 The extents of
profile
The extents of profile
D.2.10 The grid cell label
for profile
The grid cell label for profile
D.2.11 The extents of
profile
The width of profile
It might seem that the scale bar for profile
The contradiction over extents can be resolved thus: in map
D.2.12 The extents of
profile
The extents of profile
D.2.13 The extents of
profiles
The extents of
D.2.14 The grid cell
labels for profiles
The grid cell labels for profiles
D.3 Byrsa III
D.3.1 The extents of
profile
The extents of profile
D.3.2 The extents of
profile
The extents of profile
D.3.3 The extents of
profile
The extents of profile
D.3.4 The extents of
profile
The extents of profile