You are here

Computational Methods for the Study of Graeco-Egyptian Magical Gems: A Case Study in the Anguipede

Walter Shandruck

            The study of ancient magical gems has in recent years been greatly augmented by efforts to publish some of the largest collections since Bonner (1950) and Delatte and Derchain (1964). The publication of the British Museum collection (Michel 2001), various Italian holdings (Mastrocinque 2003, 2007) and now the re-editing and publication of an expanded Paris collection (Mastrocinque 2013 forthcoming) are but the latest examples. Such abundance of data has also brought with it the challenge of applying traditional research techniques to such a large body of data. Michel's seminal Die Magischen Gemmen: Zu Biltern und Zauberformeln auf geschnittenen Steinen der Antike und NeuZeit (Berlin: 2004) whose analysis involved about 2,600 gems from numerous collections and thus benefited from the decades long process of catalogue publications provides a clear example of the traditional method by which magical gems have been studied, which relies on a taxonomic approach informed primarily by iconography. Richard Gordon characterized Michel's monograph as favoring a "learned but uncritical iconographic-museographic approach" (Gordon: 716). The method’s shortcoming lies principally in the underlying assumption that iconography offers the best organizational criterion to understand the significance and specific uses for these amulets.

            Recent investigations have rightly suggested a corrective by pointing out the importance of other gem characteristics, such as material and color (Faraone 2011, Mastrocinque 2011). This is not to suggest that Michel does not recognize the importance of material and color—indeed, she makes frequent reference to the typical materials and colors that, according to her, characterize different gem groups. The failure is in treating material and color as largely dependent variables used to describe categories such as "Regeneration und göttlicher Schutz"—the title of the second chapter—which have already been established according to iconographic criteria. Stated differently, the iconographic assumption prevents an objective classification and assumes that other gem characteristics, such as material, color and inscription, can not provide equally strong (or stronger) taxonomic criteria.

            This presentation will demonstrate the utility and insights to be gained from applying computational methods, already widely in use in the social sciences, to arrive at a more objective taxonomy for ancient magical gems equally informed by all gem attributes (color, iconography, material and text). The primary analytical approach will be through Network Theory in order to map the various correlations between different gem attributes in one-to-one and one-to-many relationships. A preliminary new taxonomy for magical gems will be suggested by noting which nodes in the network are the most central, with the greatest number of vertices.

            The same computational tools will then be applied in a specific case study concerning the so-called anguipede gems, distinguished by the presence of a peculiar rooster-headed snaked-legged divinity. These anguipede gems form a significant subset of all magical gems and are all the more fascinating for the apparent absence of this particular figure from the magical papyri. The analysis will (1) hope to demonstrate why Nagy (159-172) is likely correct in concluding that the anguipede is best construed as a visual "epithet" of the divine name Iaô ( = Yahweh) and (2) evaluate how various hypotheses about the relationship of color and material measure up against the quantitative results for these specific gems. By applying network theory to both a general (magical gem taxonomy) and specific (anguipede gems) problem this presentation hopes to demonstrate the utility of computational techniques for a broad range of datasets.

Session/Panel Title

Ancient Amulets: Language and Artifact

Session/Paper Number


© 2020, Society for Classical Studies Privacy Policy