One of my main goals is to make not only my results, but also my data and my methods available to the widest possible audience. I believe that having full access not only to the results of a research, but also to the sources, the methods, and the tools that have been used is crucial: everyone should not only be able to discuss the outcome of my study, but should also have the possibility of picking up and re-elaborating all of my data according to one’s specific needs, in order to carry on the research as freely as possible.
These considerations have shaped my decisions regarding the software and methods I have used. In particular, I have decided to work with open source, widely available, and possibly easy to use software. This decision is not only in agreement with my sharing philosophy, but it also has the advantage of allowing everyone to have easy and free access to my data, tools, and techniques. The flexible architectures, the freely available plugins, and the thriving online communities of open source software are very important aspects, in this respect.
Finally, there is the problem of compatibility. Technology evolves very quickly, and it is very difficult to foresee which format will still be available and which one will have disappeared in ten or fifteen years from now. To rely on opensource software or licenced but freeware software supported by large communities online should give a slightly higher probability of continuity, or at least a slightly higher probability of backwards compatibility that should at least allow to access and then update old data formats.
Here is a list of the software I have used:
Raster graphics editor.
Licence: GNU General Public License v3+
I used Gimp as editor for most of the figures of the book. It is freely distributed, and it allows to edit a wide range of image formats.
Professional vector graphics editor.
Licence: GNU General Public License v3+
I used Inkscape to create and edit vector images. Inkscape was also useful in the preliminary elaboration of the plans of Dunand’s excavations used in the 3D modelling of the Obelisk Temple (§4.4).
Graph digitiser software.
Licence: proprietary, distributed freeware
I used GraphClick 3.0.3 to retrieve the coordinates of objects and architectural features from the plans of Dunand’s excavations.
Geographic Information System.
Licence: GNU General Public License
I used QGIS to elaborate satellite data and the geographical maps, as well as to calculate and generate the data used the heatmaps in §3.3, §3.4.
3D computer graphics software.
Licence: GNU General Public License v2+
I used Blender to elaborate satellite data and to generate 3D geographical models, such as the images of the Lebanese coast used in §5.4.
SketchUp Make 2014
Raster graphics editor.
Licence: proprietary, distributed freeware
I used Sketchup to create the 3D models of the Obelisk Temple (§4.4), and to elaborate the terminus sub quo, the terminus super quem, and the approximate stratigraphic sections derived from them (Appendix C). SketchUp is not an opensource software, but its basic version is freeware. Moreover, Sketchup projects can be shared online in virtual libraries and can be integrated into other packets, such as GoogleEarth, which could be useful in possible future developments of my project.
“Data mining” software framework.
Licence: GNU Affero General Public License v4+
I used ELKI in the preliminary analysis and elaboration of archaeological data.
GNU Octave 3.4.0
Software primarily intended for numerical computations.
Licence: GNU General Public License
I used GNU Octave in the preliminary analysis and elaboration of archaeological data.
In order to freely share and distribute not only the results, but also the sources of my research, I have decided to make all my data freely available online.
To make it easily accessible, I decided to use GitHub, a free opensource cloud service which already host thousands of projects and datasets. In addition to freely storing the data indeterminately, Github offers also the possibility to download the files it hosts. This means that new analyses and subprojects could easily be created and linked with my original repository if anyone wishes to do so.
The address of the GitHub repository associated with this book is the following: https://github.com/Kilani-DPhil/Byblos
1 GitHub Repository – Content
The following sets of data are available in the GitHub repository.
As explained in §2.3, one of the results of my research was the development of a database collecting the objects found by Dunand in his campaigns of 1926–1932 and 1933–1938, and published in his catalogue volumes (§2.1.3). The resulting data, collected in .csv files, are available in the GitHub repository. The table presents the catalogue number, the findspot coordinates, a transcription of Dunand’s entry automatically obtained with OCR software, as well as various other information about the objects themselves (such as their category, material, provenance etc). A readme file explaining how to read the .csv table is also available in the github repository. A working, searchable version of the database is also available online. It is currently hosted on Heroku, a git-based provider using a PostgreSQL database engine. The address is the following: http://kilani.herokuapp.com/byblos_db/. Note however that the location of this database could change in the future. The updated address of the database, however, is available in a specific readme file in the main github repository.
3D models – Obelisk Temple
Files of the 3D models of the Obelisk Temple discussed in the book (in particular §4.4). Both the files of the 3D models (.skp format) and the raw sets of coordinates used to generate them (.csv format) are available.
3D models – Stratigraphy
Files of the 3D renditions of the terminussubquo and terminussuperquem plans discussed in Appendix C. Both the files of the 3D models (.skp format) and the raw sets of coordinates used to generate them (.csv format) are available.
Files of the approximate vertical stratigraphies along the x axes of Dunand’s excavation grid for the campaigns 1933–1938. Presented also in Appendix C.
3D model of the Lebanese coast
3D model of the Lebanese coast obtained from satellite data and used in figures 5.3 and 5.4.
Maps and plans
QGIS files for the various maps and plans used in the book, including the heatmaps presented in §3.3 and §3.4.
Data about specific categories of objects discussed in the book. In particular, .csv tables with data about Mycenaean pottery (§3.2), scarabs (§3.3), spindle whorls (§3.4), and loom weights (§3.4) are available.
Any image created by the author and used in the book.
1 Definition of the Theoretical Framework
Without stratigraphic data, it is challenging to identify the Bronze Age layers of Byblos. The structure of the site is an obstacle: not only is there the complex and often uneven stratigraphic development typical of any urban area, it is also clear that many sectors were disturbed or destroyed by natural phenomena (erosion) or by later (especially Graeco-Roman and medieval) activities. Dunand himself observed that “à Byblos les couches hellénistiques et romaines sont souvent superposées directement aux couches du Moyen Empire” (Dunand 1939, 64; see also p. 9). Still, mentions of Late Bronze Age material and architectural remains are scattered through his publications, suggesting that Late Bronze Age layers did survive here and there. It is thus important to determine where these layers could have been. While it is clear that it is not possible to reconstruct the stratigraphy of the whole site, Dunand (1954, 3–5) suggested a method for making a partial reconstruction possible (see above §2.1.2, §2.1.3, §2.1.4).
Dunand’s ideas are a good starting point and can be further developed. In particular, on the basis of the principles of stratigraphy one can assume that an object found above another one is later in date than the latter, unless some disruption has occurred. Therefore, the lowest object of a given period can be taken as a reference point, a “terminus sub quo,” under which the uncontaminated layers of the previous periods could be located – “could be” rather than “are,” because layers of previous periods might have been contaminated by later material or might be absent, either because they never existed or because they were destroyed. If uncontaminated layers of previous periods do exist, then they must be located below that point. If instead there has been a disruption, one may find a later object below an earlier layer. In this case, the object will still be a valid “terminus sub quo,” but one that will mark the possible limits of the disruption itself.
So for example, if in a given Column (see §2.3 for the terminology used here) there is Greek material in levées 1, 3, and 4, it can be assumed that the Late Bronze Age layer was either located somewhere below levée 4 or, if it was above those levels, it was contaminated or destroyed by later activities. Obviously, such a limit is only a possible stratigraphic indication. If the Late Bronze Age layers never existed or were destroyed, this limit may lie directly on earlier strata, dating for instance to the Middle or Early Bronze Age.
Another aspect to take into consideration is that not all the columns of the excavations have yielded material that can be used to define such a terminus sub quo. This problem can be managed by using those Excavation Units that did yield dating objects as reference points for a triangulation that creates a virtual plane encompassing all the relevant area. This plane approximates the limit of the terminus sub quo for the Late Bronze Age layers also in those columns that did not yield any diagnostic dating artefact. In this way the gaps in the information represented by these empty columns will be filled in by an average value represented by the plane and inferred from the values of the terminus sub quo in those columns around it that have yielded objects. This approach is illustrated graphically in fig. C.1.
After having defined the position of the plane corresponding to the terminus sub quo, it is possible to obtain a series of vertical sections of the site along the x or y lines of the excavation grid. These sections enable one to visualise where the limits of the layers so defined are located in relation to the levées. They can therefore be considered as an approximate guideline for the stratigraphy of the site, although a rough and loose one, with a resolution of only one Excavation Unit (fig. C.2).
It is clear that if the columns and Excavation Units with diagnostic objects are close to each other, the resulting approximation of the layer in the adjacent and intermediate empty columns will have a relatively high degree of reliability. If, however, the Excavation Units with dating objects used as reference points are some distance away, the resulting plane is potentially less reliable and precise (fig. C.3).
Due to the limitations of the data available, the resolution of such a reconstructed stratigraphic model is relatively low. Since the location of the objects is given only by Excavation Units, the resolution of the reconstructed stratigraphy cannot be higher than that of the unit. This means that although it is possible to identify the Excavation Unit containing the limit of the layer, it is not possible to define where and how this limit passes within that Unit. According to Lauffray,1 Dunand did use a more precise system of coordinates to record the absolute location of every object discovered. These data could be used to refine the stratigraphy, but if they have survived, they have never been published.
In addition to the definition of this terminus sub quo for the Late Bronze Age layers, it is also possible to try to define a terminus super quem, that is, the upper limit of the layers of the periods preceding the Late Bronze Age. The principle is the same, but reversed: the terminus super quem can be defined using the highest attestation of material of previous periods.
This terminus super quem, however, is much less indicative than the terminus sub quo, since whereas only direct vertical contamination can introduce a later object into an earlier layer, many factors can cause an earlier object to be found in a later layer. In addition, if a later-object-into-earlier-layer contamination is usually limited in space to the area affected by the disturbance, an earlier-object-into-later-layer contamination can also involve significant horizontal displacements: for example earth (and associated objects) dug out from a pit could be deposited very far from the pit itself. Similarly, an object that has been circulating for decades or even centuries could end up in chronologically and spatially unrelated contexts. As a consequence, the presence of an object in a certain place does not necessarily imply that the layers below it belong to the same or to an earlier period, nor that the area has been disturbed: the object could have been brought to the surface elsewhere and could have been moved there at a later time.
These are serious limitations, and therefore the terminus super quem should be seen only as a very approximate guide. With some adjustment, however, these limitations can be contained and the terminus super quem can be fine tuned in order to obtain a relatively valid stratigraphic indicator. In particular, identifying coherent ensembles of objects, rather than isolated objects, improves the reliability of a terminus super quem. It is clear that one isolated Middle Bronze Age pot is not indicative of a Middle Bronze Age layer, but a group of Middle Bronze Age vessels within a small vertical or horizontal range of Excavation Units would point to the presence of a stratigraphic ensemble that might reflect a Middle Bronze Age layer. Focusing on intact, fragile objects is another productive strategy, so that ceramics are a very good choice. Vessels break easily, either before deposition or as a consequence of later disturbance, and their sherds can easily get out of context. Sherds are thus poor indicators for the terminus super quem, but vessels that are intact or broken but complete, can indicate deposition and little or no disturbance.
2 Definition of the Terminus Sub Quo and Terminus Super Quem
As mentioned above, the terminus sub quo and the terminus super quem for the Late Bronze Age layers can be defined only for the section of the site excavated between 1933 and 1938, which was published in enough detail for such an approach to be viable. The grid of the earlier excavations was too wide and too irregular, while the material excavated later has never been published. In terms of material, the area excavated between 1933 and 1938 is also the most productive for such an approach: the area excavated before 1933 is relatively small, with limited quantities of objects,2 while the campaigns after 1938 focused mainly on the walls or on areas outside them and, according to the published reports, they uncovered mainly, if not exclusively, remains from the Iron Age or the Persian, Greek, and Roman periods (See §2.1.3).
In order to define the terminus sub quo, abundant and widespread objects that postdate the Bronze Age are needed, so that many reference points across the surface under analysis will be available. Pottery and potsherds are ideal, but two problems have to be considered. First, only a small part of the pottery found was listed in Dunand’s catalogue (1939, 9). Second, even the published sherds are not described or assessed in detail, and therefore too often cannot be dated. Some later studies treat particular categories of pottery from Byblos, but these mostly focus on pre-Late Bronze Age material and are not useful in the definition of the terminus sub quo (compare Williams 1975; Saghieh 1983; Thalmann 2008). There are two notable exceptions. The first is a group of Iron Age vessels identified and studied by Grace Homsy (2003). Homsy was able to study some of the pottery excavated by Dunand that is now kept in the Museum in Byblos. Her work is not exhaustive, as the vessels she had access to represent only a fraction of the pottery excavated,3 but she succeeded in identifying 120 Iron Age vessels of various types. However, their usefulness for the present study is limited since, of the 120 vessels she lists, only 20 came from the area of the 1933–1938 excavations, with one of those being a surface find whose original place of deposition is not known.
Handles of amphorae and other vessels bearing stamp impressions with Greek inscriptions and Greek iconography can also be used here, as they are all later than the Late Bronze Age. Dunand published them in his catalogue and, in view of the close attention that he paid to all kinds of inscribed objects, the handles that he listed are likely to represent a good part, perhaps the totality, of those found in the excavations (see Dunand 1954, x). Given their abundance and their ubiquity on the site, they are very good stratigraphic indicators.
Coins can also help in the definition of the terminus sub quo, having been introduced to Phoenicia in the fifth century BCE, long after the Late Bronze Age. Like the stamped handles, coins are quite abundant, ubiquitous, and well documented by Dunand. Neither of these categories dates to the periods immediately after the Late Bronze Age, but they are the best available markers in the published material. This gap in time means that any possible Late Bronze Age layers could be a little deeper than the level indicated by the terminus sub quo. Since, however a gap corresponding to the Early and Middle Iron Age layers has been observed (Mazza 1994; Homsy 2003, 246), it is possible that a terminus defined in this way could be quite close to the limits of the Late Bronze Age layers.
For the definition of the terminus super quem, I use the Early and Middle Bronze Age vessels studied and classified by Saghieh-Beydoun (1983), Williams (1975), and Thalmann (2008). Although these vessels do not represent the totality of the pottery of these periods recorded by Dunand, they are numerous and widespread enough to cover the entirety of the area studied here. These vessels are generally complete,4 which increases their validity as stratigraphic indicators.
Among the objects that would ideally be used in defining the terminus sub quo and the terminus super quem, some have to be left out, either because they are clearly in a disturbed context or because their coordinates are flawed by errors in Dunand’s publication; these are listed in the online supplement (see Appendix B). This leaves ninty-five valid points of reference for the terminus sub quo and seventy-one for the terminus super quem. By using 3D modelling software (here Google Sketchup) it is possible to plot these points and to obtain a model of the excavation with the two termini represented as planes. In order to have a spatial frame of reference within which the planes of the termini can be plotted, I have defined the original topographical profile of the whole area excavated by Dunand. This topographical profile is obtained by plotting the coordinates of the highest objects found in each Column, that is, the first object found under the surface in each Column. The results of this process are displayed in §C.3.1 below.
3.1 Terminus Sub Quo and Terminus Super Quem
In figure C.4 the terminus sub quo plane is plotted together with the surface of the site, while in figure C.5 the surface is removed and the plane itself is visible. In figure C.6 the terminus super quem is plotted together with the terminus sub quo and with the surface. In figure C.7 the terminus super quem is plotted only with the surface, in figure C.8 only with the terminus sub quo, and finally in figure C.9 it is plotted alone. In the following figures the vertical dimension is magnified ten times in relation to the horizontal one. The general profile of the areas of the site that are being investigated is thus emphasised, which allows to better appreciate the vertical sections obtained.
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus sub quo (blue) together with the surface of the area excavated (white)
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus sub quo (blue)
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus sub quo (blue) together with the terminus super quem (yellow) and the surface of the excavated area (white)
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus super quem (yellow) together with the surface of the excavated area (white)
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus sub quo (blue) together with the terminus super quem (yellow)
Stratigraphic data – Terminus sub quo. a–d) 3D representation of the terminus super quem (yellow)
3.2 Approximate Stratigraphy
This 3D model can then be cut along the x and y coordinates of Dunand’s excavation grid to obtain vertical sections with representations of the termini. These vertical sections can be used as approximate stratigraphic guidelines for the site. The figures below show the approximate stratigraphies along the x axis, one for each x line of the excavation grid. The stratigraphies along the y axis can be inferred and built from them.
The Excavation Units located between the surface and the terminus sub quo are marked in blue, while those located below the terminus super quem are marked in yellow (dark blue/dark yellow = columns with diagnostic dating objects, light blue/light yellow columns without dating objects whose limit has been inferred from the projection of the planes). If present and not disturbed, the Late Bronze Age layers should be located below the blue Excavation Units, and possibly above the yellow ones. The vertical dimension is magnified ten times in respect to the horizontal one.
Lauffray 1995, 457. The existence of such a system can also be inferred from the precise location of the objects in the plans of Volume II and by the scattered mentions of specific coordinates in that volume (see e.g., Dunand 1954, 26, 272).
5756 objects found under Dunand between 1926 and 1933 (Dunand 1939) versus 12537 objects found between 1933 and 1938 (Dunand 1954).
The remaining part, which Homsy was unable to study, is in Beirut. It is impossible to estimate the size of this group and what percentage of the pottery found under Dunand it may represent. It is, however, clear that various vessels appearing in Dunand’s catalogue that could date to the Iron Age are absent from Homsy’s list (e.g., nos. 9864, 9931, 10378, 11199, and 13034, which are similar to 10377 and 14019, recognised by Homsy as dating to the Iron Age). This could be because she did not have access to them or she did not recognise them.
Although they could have been broken when discovered.