Archaeometry 49, 4 (2007) 749–764 doi: 10.1111/j.1475-4754.2007.00333.x
*Received 15 August 2005; accepted 10 August 2006
†Corresponding author: email mpanagop@central.ntua.gr
© University of Oxford, 2007
Blackwell Publishing Ltd Oxford, UK ARCH Archaeometry 0003-813X © University of Oxford, 2007 XXX Original Articles Identifying hands on ancient athenian inscriptions S. V. Tracy et al. *Received 15 August 2005; accepted 10 August 2006
IDENTIFYING HANDS ON ANCIENT ATHENIAN
INSCRIPTIONS: FIRST STEPS TOWARDS A DIGITAL
APPROACH*
S. V. TRACY
American School of Classical Studies at Athens, 54 Souidas Street, GD-106 76, Athens, Greece
C. PAPAODYSSEUS,† P. ROUSSOPOULOS, M. PANAGOPOULOS,
D. FRAGOULIS, D. DAFI and TH. PANAGOPOULOS
National Technical University of Athens, School of Computer and Electrical Engineering,
Zografou 157 73, Greece
In this paper, a novel methodology is introduced for the identification of the workmen
(hands) that carved ancient inscriptions. This methodology employs specific geometric
characteristics of each letter and computes the mean value and variance of these
characteristics for each one of the available inscriptions separately. Subsequently, we define
original decision thresholds that make use of the statistical distribution of the difference of
these values in order to attribute an inscription to a given hand. The inscriptions of the
hands under consideration have been properly processed and all information extracted, both
visual and statistical, is stored in a suitable database. Application of this methodology to
nine Athenian inscriptions, some of which contain very similar letters, offered correct, clear-cut
hand identification.
KEYWORDS: AUTOMATED LETTER-CUTTER IDENTIFICATION, AUTOMATED INSCRIBER
IDENTIFICATION, HANDWRITER IDENTIFICATION, AUTOMATED CLASSIFICATION OF
HANDWRITING STYLE, STATISTICAL HANDWRITING RECOGNITION
1 INTRODUCTION—A BRIEF DESCRIPTION OF THE PROBLEM AND THE APPLIED
METHODOLOGY
What follows is a first methodological article exploring the possibility of using digital technology
to aid the recognition of hands on ancient Athenian inscriptions. In Tracy (1990, 1995, 2003),
it has been established that it is actually possible to reliably identify the hands of individual
ancient letter-cutters. Nevertheless, employing mathematics, digital image processing and
pattern recognition for identifying the hand that formed an inscription may make the process
automated and essentially more objective. In fact, as things are at the present time, identifying
writers of ancient inscriptions incorporates a significant amount of subjectivity.
The first step towards the development of the methodology introduced in this paper was
to set up a test case. Professor S. V. Tracy, as the epigraphist—that is, the specialist in
inscriptions among the authors—provided the rest of the team with photographs of stones cut
by very prolific workmen who were inscribing decrees in Athens during the second half of the
third century bc. In particular, Professor Tracy offered high-quality digital images of the
following inscriptions with reference numbers: IG II
2
336, Agora XV 240, Agora XVI 208, IG