Modelling the possible archaeological past(s): Agent-based modelling of Harappan seal use and survival

Abstract

The Harappan civilization, also known as the Indus Valley Civilization, flourished in South Asia in the middle of the third millennium BCE. The early twentieth century excavations at Mohenjo-daro and Harappa are still the largest excavations ever carried out at Harappan sites. Among the millions of artefacts excavated from hundreds of sites, the most remarkable, best known, and most extensively analysed are the ones that bear peculiar signs, which have been widely proposed as elements of a writing system. Within this set of artefacts, the most characteristic item is the Harappan (or Indus) seal. These tiny carved and inscribed objects, which have been recovered from more than 80 sites, are at the centre of a significant portion of archaeological discourse in Harappan archaeology. The majority of this scholarship revolves around three conjectures: (i) the signs or sign sequences which appear on these objects indicate linguistic components and are part of a script with fixed set of signs, (ii) the ‘seals' were used as amulets or as objects of economic or administrative activities, and (iii) the primary use of seals is for creating the impressions which form `sealings'. However, the small number of actual sealings found in the excavation of Harappan sites seems to challenge this last conjecture, suggesting that perhaps seals found at Harappan sites were not always intended primarily for administrative activities that would be characterised by the creation of sealings. This paper approaches the question of seal function by using agent-based modelling (ABM) to model the quantities of seals and their impressions in Harappan culture. Amongst multiple possibilities, the model seeks to suggest a hypothetical baseline for the number of sealings one might expect to find if all excavated seals were in fact used to create an impression and thus to produce sealings. The objective is to see if and under which conditions the model results in the quantities of artefacts that are currently available as archaeological evidence. If none of the model outputs are equivalent to the presently excavated material, does it mean that the data needs to be interpreted differently? On the other hand, in order to reach the presently available quantities from the excavations, which parameters could have contributed the most? The results of the experiment may indicate a need for reappraisal or re-interpretation of this data and the systems they manifest.