Mapping the World of an Ancient Greek Historian: The HESTIA Project

HESTIA (the Herodotus Encoded Space-
Text-Imaging Archive
) is an interdisciplinary
project, sponsored by the AHRC and involving
the collaboration of academics from Classics,
Geography and Archaeological Computing, that
aims to enrich contemporary discussions of
space by developing an innovative methodology
to the study of an ancient narrative, Herodotus’
Histories
. Using the latest ICT, it investigates
the ways in which space is represented in the
Histories
, and develops visual tools to capture
the ‘deep’ topological structures of the text,
extending beyond the usual two-dimensional
Cartesian maps of the ancient world. In addition
to exploring the network culture that Herodotus
represents, one of its stated outcomes is to
introduce Herodotus’ world to new audiences
via the internet. This paper will set out in
more detail that methodology, paying particular
attention to the decisions that we have made
and the problems that we have encountered,
in the hope that our project can contribute not
only to offering a more complex picture of space
in Herodotus but also to establishing a basis
for future digital projects across the humanities
which deal with large text-based corpora.
For the purposes of a twenty minute
presentation, we address three key areas of
interest:
1.
To provide the background to the data capture
and digital mark-up of the
Histories
. Our
project differs from many by utilizing a
digital resource already in the public domain:
the text of Herodotus freely available from
Perseus (
http://www.perseus.tufts.edu/hop
per/
). Though the capture of the digital text
from Perseus (version P4) gave our project
a welcome initial boost, a number of issues
have had to be overcome including procedural
conversion, which involves handing back a P5
text to Perseus.
2.
To sketch out the the type, structure and
categorization of our spatial database. A
PostgreSQL database was chosen because
its PostGIS extension provides excellent
functionality for spatial data and is widely
supported by other applications: one key
principle of HESTIA has been to use open
source software in order to maximize its
potential dissemination and reusability of its
data. By storing information about references,
locations and the text in the database, it
has been possible to provide it to both a
Desktop GIS system and Webmapping server
simultaneously (see figure 1).
3.
To present a sample set of results of the
maps that we have been able to generate
using the geo-referenced database. While
sections 1 and 2 will be of particular concern
to anyone wishing to understand how one
may interrogate spatial data using the digital
resources available, this last stage holds the
greatest interest for the non ICT expert since
it demonstrates the use to which data in
this form can be employed: hence the main
focus of this paper will be on explaining the
five kinds of map that we have been able to
generate:
i.
Geographical Information System (GIS)
maps
. The most basic maps that are
generated simply represent a ‘flat’ image of
the spatial data: that is to say, they mark
all the places that Herodotus mentions over
the course of his work with a single point,
thereby providing a snapshot of the huge
scope of his enquiry. In this way one is
able to gain an overview of the places
mentioned in Herodotus and divide them
according to three different kinds of spatial
category: settlement, territory and physical
feature (see figure 2). A variation on this

2
basic model depicts places according to the
number of times they are mentioned (see
figure 3).
ii.
GoogleEarth
. In order to start
experimenting with public dissemination it
was decided to expose the PostGIS data as
KML: a mark-up format that can be read by
a variety of mapping applications including
GoogleEarth. With this ‘Herodotus Earth’
application, users will be able to construct
‘mashups’ of visual and textual data. So,
for example, since all places are linked to
entries in the database, when one clicks
on a particular location in GoogleEarth, it
will be possible to bring up a dialog box
containing Herodotus’ text (in both English
and Greek) for that particular location for
every occasion when it is mentioned in the
narrative (see figure 4).
iii.
TimeMap
. Whilst it is possible to visualise
narrative change using graphs, and static
differences using GIS, it is more difficult
to visualize spatial changes throughout
the narrative; GIS does not have useful
functionality in this regard except for
the ability to turn layers on and off, a
process which becomes impractical beyond
book level. The most likely candidate to
provide this kind of functionality is an Open
Source JavaScript project called TimeMap,
developed by Nick Rabinowitz, which draws
on several other technologies in order
to allow data plotted on GoogleMaps to
appear and disappear as a timeline is
moved. In collaboration with the project’s
IT consultant Leif Isaksen, Nick Rabinowitz
has adapted his schema in order to
represent the book structure of Herodotus’
narrative in a similar way (see figure 5).
iv.
Database-generated network maps
. Since
the GIS maps outlined in i. have little to say
per se regarding Herodotus’ organization of
space, a key next step has been to explore
rapidly-generated networks based on the
simple co-presence of terms within sections
of the text. The purpose of producing
networks of this kind is to start exploring
the connections that Herodotus himself
makes between places, seeing how strongly
the narrative is bound to geographical
regions, and flagging up potential links
between particular locations (see figure
6). Figure 7 illustrates one such simple
network, that for “territories” across the
entire
Histories
. It shows a series of links
connecting Greece to other areas within the
Mediterranean world: but the territory that
has the strongest connections in this basic
network culture is Egypt. While surprising,
it does make sense on reflection, since for
a better part of one book Herodotus uses
Egypt as the touchstone against which other
cultures, including Persia and his own,
Greece, are compared. It is as a tool of
comparison, then, that Egypt appears to be
the centre of Herodotus’ network picture of
the Mediterranean. Figure 8 complements
this picture by presenting the networks
of physical features, which envelop the
comparison between Greece and Egypt.
v.
Manual network maps
. The automated
maps outlined in iv. rely on ‘counting’ the
number of times two or more places are
connected to each other: they have little
to say about the kind of connection being
drawn. We end our presentation, then, with
a brief comparison to text-based qualitative
analysis, which attempts to categorize
relationships according to fundamental
geographical concepts of movement or
transformation, based on the close reading
of one book (5). Our different approaches
are intended to complement, challenge
and inform each other with a view also
to suggest ways by which the automated
process may be extended, such as by
adopting text-mining procedures.
In sum, this paper aims to meet three outcomes:
1.
To outline a methodology for dealing with
digital data that may be transferred, adapted
and improved upon in other fields of digital
humanities.
2.
To demonstrate the value of digital projects
within the humanities for helping to achieve
‘impact’ by bringing the world of a fifth-
century BC Greek historian into everyone’s
home.
3.
To show the potential for the digital
manipulation of data in posing new kinds of
research questions.