Pith location tool and wood diameter estimation: Validity and limits tested on seven taxa to approach the length of the missing radius on archaeological wood and charcoal fragments

Abstract The study of timber wood and wood charcoal fragments from archaeological sites (aka anthracology) constitutes a relevant archaeobotanical field of research for both landscape reconstruction and the study of past people-woodlands interactions. Regarding this second research field, variables other than taxa are known to be a key to the social organization of woodland management. In this sense, wood diameter constitutes a core factor of fire management, fuel provisioning and both firewood and timber procurement. These wood uses are most commonly represented in the archaeological record by charcoal fragments (both dispersed in the sediment and/or concentrated in fire structures) due to the fact the wood experiences both mass loss and fragmentation during carbonization. So the original form of the wood used (trunk, branch, twig) is no longer recognizable. Different pith-location tools (PLT) have been proposed previously in order to virtually locate the charcoal fragment in relation to the central part of the stem or trunk (pith) where the used wood originally came from. Among them, PLTs based on trigonometry are proven to be the most reliable, but have not yet been extensively tested on referential datasets in order to establish reliable analysis of the accuracy of the measurement of the missing radius, margins of error and correction factors. In this study we present an experimental referential dataset for 7 different taxa, both angiosperms and gymnosperms. The first aim was to move forward on the establishment of the trigonometric tool by testing if it is also suitable and valid for all the woody species producing tree rings. The second purpose was to provide a ready-to-use tool to estimate the missing radius with an interval of confidence. Lastly we also tested the effect of the carbonization on two taxa. According to the results obtained, a measuring protocol, correction factors and guidelines to interpret the subsequent results are established. In addition, an R function is now available to estimate the real radius from the calculated one with PLTs.