Database of Greek inscriptions ‘Hellenisation of Ancient Thrace’: a final report on...
Transcript of Database of Greek inscriptions ‘Hellenisation of Ancient Thrace’: a final report on...
67
10. Database of Greek inscriptions ‘Hellenisation of Ancient Thrace’: a final
report on investigations during 2013-2014
Petra Janouchová
With Pls. 30–32
Abstract: The article discusses the purpose, design and development of an epigraphical database
created for a dissertation project with the title "Hellenisation of Ancient Thrace". The database
has been populated with epigraphic data from a geographic area broadly conceived as falling
within Ancient Thrace in a manner that crosses national and linguistic boundaries, and with an
emphasis on the spatio-temporal attributes of individual inscriptions. The database makes
possible the quantified analysis of more than 4000 inscriptions and their attributes, and is
available for use and reuse by other scholars in epigraphy, philology, archaeology and other
disciplines.
Keywords: Greek epigraphy; Ancient Thrace; hellenisation; epigraphical database; identity
Project rationale
Epigraphy offers a unique view into the life in Ancient Thrace, informing the political
organisation of cities, political agreements and treaties, existence of road communications and
documentation of migration processes within the society. Epigraphy complements archaeology
and historiography in providing an insight into the past and has its place in interdisciplinary
research (BODEL 2001, 1). A database of published Greek-language inscriptions from Ancient
Thrace thus contributes to and facilitates a historiographical and archaeological study of the area.
The need for a database with a comprehensive and structured dataset arose from by my
dissertation project ‘Hellenisation of Ancient Thrace based on epigraphical evidence’, conducted
at Institute for Greek and Latin Studies at the Faculty of Arts, Charles University in Prague. The
dissertation aims to reevaluate the process of hellenisation of Ancient Thrace using the post-
colonial approach to material evidence from a broadly defined area of Ancient Thrace (e. g.
TSETSKHLADZE 2006, xxiii–lxxxii; DANOV 1976, 348–368; MIHAILOV 1979, 267; ISAAC 1986,
136, 291–292; BOUZEK 2005, 6). The dissertation proposal stipulated the development of a
database of all published and accessible inscriptions, which was designed and populated with
data over the course of 2013 and 2014.1 The resulting 'Hellenisation of Ancient Thrace'
(henceforth HAT) database represents the empirical basis for my dissertation. As every database,
the design of the structures and concepts used in the HAT database is driven by the needs and
aims of my dissertation research.
The HAT database aggregates all major published corpora of Greek inscriptions from
1 The project was supported by the Grant Agency of the Charles University in Prague, project GA UK No. 546813.
I would like to thank my collaborators Barbora Weissová, Markéta Kobierská and Jan Ctibor for their help with the
project and the data entry, and to Adéla Sobotková for her useful suggestions and instructive comments to the HAT
project from its inception and also for the introduction to the FAIMS project.
68
Thrace and represents the best digital source of structured epigraphic data from the region. The
corpora that form the bulk of the database include (with their abbreviated form): MIHAILOV 1956
– 1997 (IGBulg 1–5); LOUKOPOULOU et al. 2005 (IAegThr); KRAUSS 1980 (ISestos); SAYAR
1998 (IPerinthos); LAJTAR 2000 (IByzantion). The structured and normalized nature of database
records allows for their quantification, spatial and temporal analysis, and makes possible the
statistical assessment of aspects traditionally associated with the concept of 'hellenisation'. The
aspects associated with 'hellenisation' include the spatio-temporal occurrence of personal names,
collective group affiliations, identity markers as well as terminology for political and social
organisation or professed religious association. Besides the raw data extracted from the editions
of corpora, the database contains additional attributes with my own qualitative interpretation of
the primary data.
In the beginning of January 2015, after two years of database development, the database
contained 4282 inscriptions, from ten different epigraphical sources, spanning across three
modern countries.
The research areas
The spatial definition of Ancient Thrace is ambiguous as the geographic boundaries were vague
already in contemporary Greek historiographical sources, such as Herodotus or Thucydides. The
territory was generally outlined by the Strymon River (Struma) in the West (Thuc. 2. 97), the
Aegean coast in the South, the Bosporus in the Southeast (Thuc. 2. 97), the Black Sea coast in the
East (Thuc. 2. 96–97), and the Stara Planina mountain range and Danube River in the North
(Istros; Hdt. 5.10). Translating this to modern political boundaries, the region of Ancient Thrace
is located on the territory of modern-day states of Bulgaria, European part of Turkey and north-
eastern Greece.
The modern state division does not respect the extent of Ancient Thrace and as such, it
often puts a restraint on the epigraphical corpora. They are published on a national scale, which
means that inscriptions are often excluded from national-scale research. What is more, the
corpora are often published in languages of the particular state, which further poses an additional
barrier. The HAT database combines sources from all available sources from Greek, Bulgarian
and Turkish Thrace (Pl. 30), and provides them all with English as the unified output language to
facilitate an accessibility. Similar idea of trans-national and trans-linguistic collection of primary
data has already been undertaken by Dumont and Homolle in 1892, but without any academic
followers until now (DUMONT 1892, 307–581).
The HAT database aims to cover the entire area of Ancient Thrace, though some parts are
yet to be covered, as can be seen on Pl. 30. These existing geographic gaps between the research
areas are caused by the absence of published material and not by the lack of epigraphical
evidence per se. This is mainly the case in the area between Hadrianopolis and Byzantion, which
has not been systematically published, and most likely inscriptions from this part of Thrace are
hidden in local museums and inventories. However, the overall number of corpora and
inscriptions provide a sufficient statistical sample with fair coverage over 130 000 sq. km.
Therefore, the statistical relevance of the data is not ideal, but represents well the state of
available epigraphic data as of 2014.
69
Methodology and technical challenges
The basic purpose of creating any database is to be able to handle large amount of data
effectively (STONES – MATTHEW 2005, 2–42). There are various systems enabling data
management together with basic analysis of stored information, each with its own advantages and
shortcomings depending on its user. For the purpose of this project, I have chosen online data
management software ‘Heurist’, developed specifically for Humanities researchers by Ian
Johnson and the Arts eResearch unit at the University of Sydney.2 My main reason for choosing
Heurist was the possibility to have several people editing the database at the same time from
different locations via a web browser. Using a web browser for data entry, Heurist is platform-
agnostic and places no constraints on the types of computers and operating systems used. All in
all Heurist fitted my requirements, was simple to use and offered the service free of charge for
academic purposes.
One of the main advantages of Heurist and similar databases is the ability to handle
spatial data and to visualise it with Google maps, and export it to other GIS software. As GIS
software I use QGIS, version 2.0 Dufour. For combined statistical analysis I use software R,
version 3.1.2., both of them are also open source and free of charge for any researcher. The focus
on spatial information was one of the goals of the HAT database in order to analyse the spatio-
temporal relations in Ancient Thrace, the socio-economic development and nuances in migration
among the population.
Unfortunately, as the corpora have been created over many years by many people, the
spatial data have not been provided consistently. Thus, I had to create a methodological approach
how to deal with often incomplete or missing spatial information, or uncertainty concerning the
find spot of each inscription. Every database record, representing an inscription, has a set of
geographical attributes, providing the information of the findspot with assigned geographical
coordinates, further specified by the ‘Position accuracy Index’. The ‘Position accuracy Index’
presents the radius of a circular buffer in which the inscription was most likely found, in other
words, how far from the geolocated point is the area with the most probable findspot of the
inscription (with values up to one, five, 20, and more than 20 km radius). The ancient names are
recorded (if known) together with the modern name of the location. Any available information
about archaeological context, or possible reuse on an inscription are also important as they affect
the precision of the provenance (reused inscriptions are likely to be transferred to the nearest
village, church and the proximity of archaeological site indicates possible provenance of
inscription).
Inscriptions often help to date an archaeological site and together with coins are used as
the most precise date determining factor. Thus, recording the date is one of the most important
features every epigraphical database should have. But similarly as in the case of spatial data, the
data provided by various corpora are of different quality and levels of completeness. Moreover,
some inscriptions can be dated to one year’s precision, but mostly they are dated to centuries or
even to periods. This heterogeneity of data is very typical; it nevertheless requires a tailored
methodological solution which enables further analysis. The HAT database stores a date in three
different formats: the first format stores an absolute date as an interval between the two years,
where any historical period is transformed into this two year interval, e. g. the Classical period is
marked by years -479 and -338, 2nd
c. AD by years 100 and 199, for the detailed guideline see the
Tab. 1. The second format stores the date as numeric value representing the century to which a
2 Version 3 available at http://heurist.sydney.edu.au/h3 (accessed 01/01/2015).
70
particular inscription belongs, e. g. date 2nd
–3rd
c. AD is stored as two and three value, 4th
c. BC
as -4 value. The second format is a reduction of the first format for the purpose of later analysis
in GIS. The third format uses a coded value, representing a relative date, mostly applicable only
to polarize Roman and not-Roman date on the basis of used nomenclature and the letter shapes
(the relative date has not been determined previously in the case of IGBulg at all). The third
format mostly consists of my personal interpretations, as the dating criteria by letter shapes and
occurrence of personal names have been omitted by some editors. By using the relative date, I am
trying to overcome the fact that large part of inscriptions remained undated.
An overview of the database contents
The total number of inscriptions is 4282 from ten epigraphical corpora, covering the area of
130 000 sq. km in parts of modern Bulgaria, Greece and Turkey. Tab. 2 shows their division with
the basic data, such as the area in sq. km, number of inscriptions and the density/ 100 sq. km. The
coverage of all studied areas is displayed on Pl. 30, together with the locations (find spots of
inscriptions). The total number of locations is 650 (represented by dots), covering known ancient
cities, smaller sites, random finds or inscriptions coming from secondary context, see Pl. 30. The
largest clusters of inscriptions come from the area of Philippopolis in central Bulgaria, Serdica in
West Bulgaria, Odessos on the Black Sea coast, Mesambria Pontica, Apollonia Pontica,
Byzantion, Perinthos in Propontis and from Maroneia on the Aegean coast, as can be seen on Pl.
31, with the largest densities of over 200 and 300 inscriptions respectively.
When compared to the system of Roman roads, the majority of inscriptions cluster not
only around major sites, but also along main communication routes. These roads had been fully
documented by the Roman period, but their existence is the previous periods, is more than likely
and in some cases mentioned already by Greek historiographers of the Classical period (DANOV
1970, 135–144; ISAAC 1986, 143–144). For example, Herodotus’ description of the Persian
march through Thrace mentions the road from the Propontic region to Apollonia Pontica as
taking four days (4. 90–92). Thucydides (2. 97) explains that a healthy man could walk through
the territory of the Thracian tribe, the Odrysai, in 11 days, going straight from Abdera to the
Danube (Istros) River. Appian (BC 4.11.87) mentions the only known route from Asia to Europe
at the time leading through the territory of the Thracian tribe, the Sapaioi around Abdera and
Maroneia. Another important connection of the Thracian inland with both coasts was provided by
the river transport network (BOUZEK 1996, 221–222; ISAAC 1986, 2–4, 143–144). Ancient
Tonzos (Tundzha), Hebros (Maritsa), Strymon (Struma) and Nestos (Mesta) were most probably
navigable during Antiquity, or at least facilitated transport of some kind in their proximity. It is,
therefore, not surprising that the majority of inscriptions was localised in the immediate vicinity
of these rivers.
As to the date of inscriptions, in 1746 cases the original corpus provided an absolute date,
which represents 40 % of the total number (Fig. 1). The dated inscriptions span from the 7th
c.
BC to 7th
c. AD, with the majority (42 %) dated to the 2nd
-3rd
c. AD. Unfortunately, the
quantification of the date was influenced by different approaches of editors: some corpora
provided date for more than 80 % of the dated material (IAegThr (98 %, IByzantion 86 %,
IPerinthos 81 %). On the contrary, the IGBulg 2 provided dates with 19 % of inscriptions, and the
IGBulg 3,1 only 7 %. The average availability of dated inscriptions from Bulgaria is on average
very low (less than 20 %). Due to these limitations, a new dating strategy was applied: the
occurrence of Roman nomenclature and the letter shapes were applied as secondary dating
criteria. This way, I was able to determine the date at least relatively as Roman where the criteria
71
have been met. This way I was able to assign a relative date to 836 inscriptions (20 % of total).
The number is only to increase, as I will consult visual documentation when available. At
present, only about 60 % of all corpora provided visual documentation of any kind (photos,
drawings, squeezes) and it is more than likely that some of the inscriptions with missing visual
documentation could be relatively dated on the basis of letter shapes. In this way, the percentage
of dated inscriptions increased from 7 to 50 % in the case of IGBulg 3,1. Similar increase was
recorded also in the rest of IGBulg 2, 3,2 and 4. Nevertheless, even after application of the
relative dating method, 40 % of texts remain undated.
The spatial distribution of dated inscriptions (Pl. 32) confirms that the first inscriptions in
Thrace come from the Greek colonies on the Black Sea coast, the Propontic and the Aegean
coast, already in the Archaic and Classical period, with only occasional occurrences inland.
During the Hellenistic and Early Roman period inscriptions started to appear around major roads
connecting inland regions with the coasts. This trend only increased under Roman rule, as the
major roads such as the Via Diagonalis (Singidunum – Serdica – Philippopolis – Hadrianopolis –
Byzantion) were used frequently. More detailed study of spatio-temporal distribution will be an
inherent part of my dissertation.
The typology of recorded inscriptions reveals the following pattern: the majority of
inscriptions are private texts (77 %), official texts are less common (15 %) and the rest is
indeterminable due to its fragmentation (8 %). The official inscriptions are mostly honorary
decrees (45 %), milestones (20 %) lists and religious regulations (17 %) and other documents
issued by political authorities (18 %). The private texts are from approximately one half
dedications to the deities (52 %) and from approximately one half funerary inscriptions (47 %).
The remaining 1 % is mostly the ownership marks. The dedicatory inscriptions are more
dispersed in the landscape than the funerary inscriptions. This may be due to the existence and
availability of local sanctuaries, often located in the mountains or other remote areas.
The majority of inscriptions consists of only 1-2 lines of text (46 %), or 3-5 lines (29 %)
but yet the short inscriptions provide valuable information about the people involved in the
inscribing of the texts (dedicants, the deceased or their families etc.). The study of personal
names and personal statements is very useful for describing the structure of the society and
underlying processes. Every individual is identified by personal name, kinship, geographic origin,
in some sense also ethnicity, and personal achievements. The HAT database contains records of
4700 individuals, over 200 collective group affiliations and over 200 geographic names often
pointing to the origin of individual persons. However, the detailed study of the society of Ancient
Thrace is out of the scope of the current, but will form an integral part of my dissertation.
Future of the project
The records in the HAT database will be made available for public viewing and download after
the completion of my dissertation. Moreover, I have made the structure (data schema and forms)
of the HAT database available to other researchers by converting it to a mobile application on the
FAIMS Mobile Platform3 as of January 2015. The HAT - Inscription Recording Module is freely
available for download from the FAIMS Mobile Platform Demo Library of modules. All Android
4.1+ users can download this module after they have downloaded the FAIMS app 2.0 version
from Google Play. Please see the instructions for trialing this app on the FAIMS User Guide.
They can use this module to record epigraphic inscriptions and assess the overall design of
3 www.fedarch.org (accessed 01/01/2015).
72
database as well as the utility of its mobile version. Besides my dissertation, the HAT database is
being used in several archaeological projects in Thrace, such as the related Tundzha Regional
Archaeological Project4, the Yambol regional study
5 and the Stroyno excavation project
6 as a
comprehensive source of comparative material. The HAT database aims to be a collaborative
interpretative tool for such regional-based studies.7
Bibliography
BODEL, J. 2001: Epigraphy and the ancient historian. In: BODEL, J. (ed.): Epigraphic Evidence:
Ancient History from Inscriptions. London, New York, 1–56.
BOUZEK, J. 1996: Pistiros as a river harbour: Sea and river transport in antiquity. In: BOUZEK, J. –
DOMARADZKI, M. – ARCHIBALD, Z. H. (eds.): Pistiros I: Excavation and Studies. Praha,
221–222.
BOUZEK, J. 2005: Urbanisation in Thrace. In: BOUZEK J. – DOMARADZKA, L. (eds.): The Culture
of Thracians and their Neighbours. Oxford, 1–7.
DANOV, C. 1976: Altthrakien. Berlin, New York.
DUMONT, A. 1892: Inscriptions et monuments figurés de la Thrace. In: DUMONT, A. – HOMOLLE,
T. Mélanges d'archéologie et d'épigraphie. Paris, 307–581.
ISAAC, B. 1986: The Greek Settlements in Thrace until the Macedonian Conquest. Leiden.
KRAUSS, J. 1980: Die Inschriften von Sestos und der Thrakischen Chersonesos. Bonn.
LAJTAR, A. 2000: Die Inschriften von Byzantion I. Bonn.
LOUKOPOULOU, L. D. – PARISSAKI, M. G. – PSOMA, S. – ZOURNATZI, A. 2005: Inscriptiones
Antiquae Partis Thraciae Quae ad Ora Maris Aegaei Site Est: Praefecture Xanthes,
Rhodopes et Hebri. Athens.
MIHAILOV, G. 1956 – 1997: Inscriptiones Graecae in Bulgaria Repertae, vol. 1–5. Sofia.
MIHAILOV, G. 1979: Documents épigraphiques de côte bulgare de la mer Noire. In: PIPPIDI, D. M.
(ed.): Actes du VIIe congrès international d’épigraphie grecque et latine, Constantza, 9-
15 Septembre 1977. Paris, 263–271.
SAYAR, M. H. 1998: Perinthos-Herakleia (Marmara Ereğlisi) und Umgebung. Geschichte,
Testimonien, griechische und lateinische Inschriften. Vienna.
STONES, R., MATTHEW, N. 2005: Beginning Databases with PostgreSQL. Berkeley.
TSETZKHLADZE, G. 2006: Revisiting Ancient Greek Colonisation. In: TSETSKHLADZE. G. (ed.):
Greek Colonisation: An Account of Greek colonies and Other Settlements Overseas, Vol.
1. Leiden, Boston: xxiii–lxxxii.
4 Info available at www.tundzha.org (accessed 01/01/2015). The project is conducted by Adéla Sobotková and
Shawn A. Ross at the Macquarie University, Sydney. 5 Conducted by Barbora Weissová at the Institute of Classical Archaeology, Charles University.
6 Conducted by Petra Tušlová and Barbora Weissová at the Institute of Classical Archaeology, Charles University.
7 For any questions or access to the database, please contact me via email [email protected].
73
The term used in corpus StartYR EndYR
half of the 3rd c. BC -260 -240
first half of the 3rd c. BC -300 -250
second half of the 3rd c. BC -250 -200
last third of the 2nd c. AD 270 299
end of the 2nd c. BC -215 -201
beginning of the 2nd c. BC -300 -285
after mid-2nd c. BC -250 -230
last years of 2nc c. BC -210 -201
Archaic period -700 -479
Classic period -479 -338
Hellenistic period -338 -30
Early-hellenistic -338 -280
Mid-hellenistic -280 -150
Late-hellenistic -150 -30
Roman -30 476
Imperial -27 284
Principate -27 284
Dominate 284 476
ante anno 212 -27 212
post anno 212 212 476
ante ABC -700 ABC
post XYZ XYZ 476
Tab. 1. The translation of corpus provided date to numeric values used in the HAT database.
Tab. 2. HAT study areas and density of inscriptions.
Name Area sq. km Inscriptions Density per 100 sq. km
IAegThrace 9704 501 5
IByzantion 2499 452 18
IPerinthos 1248 266 21
ISestos 1039 72 7
IGBulg 1 7667 658 9
IGBulg 2 50161 413 1
IGBulg 3,1 15072 653 4
IGBulg 3,2 23137 319 1
IGBulg 4 19490 442 2
IGBulg 5* not applicable 506 not applicable
*covering the same area as IGBulg 1-4