The Bonetrade: Studying the Online Trade in Human Remains with Machine Learning and Neural Networks

There is a thriving online trade in anatomical, ethnographic and archaeological human remains that makes ready use of new social media such as Instagram, Facebook, Etsy, and until recently, eBay. The “fetishization” of the ‘exotic’ dead that underpins this trade by its very nature transforms pieces of the body into material culture: curios, commodities or objets d’art. This practice has deep Colonial-era roots, but today’s e-commerce and social media platforms have only expanded collectors’ reach and made participation open to anyone with interest and spare finances. The sheer volume of materials being produced, shared, and sold can be overwhelming for a small team to study. The market moves so fast.

Can we teach machines to identify from photographs alone patterns in the ‘visual rhetoric’ that signal materials for sale? Can ‘licit’ materials be discerned from ‘illicit’? Are there geographical patterns? Can we trace materials back to a source?

This website is the public face of our research. Here you will find updates, code, research compendia, papers, presentations, and other elements of our work over the next five years. We intend to produce a body of data and of code that will enable other researchers to repurpose our research into other allied fields (such as the more familiar trade in archaeological antiquities, and in the markets for other ‘grey market’ and black market commodities).

Open Research

All our code and data relating to this project are publicly available. All our publications and reports are accompanied by a compendium of code and data files that are deposited in a trustworthy data repository and referenced via DOI. We use GitHub to publicly host our code in development, at We archive code and data files at the Open Science Foundation.


Graham S. and Huffer D. 2018. The Bonetrade Project Website - Zenodo. DOI: 10.5281/zenodo.1245091

Graham S. and Huffer D. 2018. The Bonetrade Project Source Text Files. Zenodo. DOI: 10.5281/zenodo.1241500