All remains will be physically distributed and manually refitted over worktables. I am going to focus on bones larger than 3 cm except in the case of burning or anthropic modifications. Both mechanical (connection of two or more fragments of the same skeletal portion) and anatomical refits (union of different skeletal parts of the same individual by grouping contiguous anatomical elements of the same or opposite side) will be undertaken. For the former, particular attention will be dedicated to fresh fractures, as they are related to nutritional activities. The refitting protocol will follow the methodology of Fernández-Laso adopted during my PhD: 1) grouping and refitting of bones divided per unit, according to accumulation areas, first by square and then between squares of the area and the accumulation areas (differentiation by taxon and weight); 2) observation of refits between all accumulation areas of the three units, to verify similarities/differences and possible recurring patterns. Considering bone preservation, the refit programme can undergo variations depending on such cases. The necessary time to carry out refits is based on my PhD experience: for 6,000 fragments, 5 months of work (8 hours per day) were required. This valuation has been useful to choosing the number of remains to analyse during ANDSU: I will select all identified bones (taxa), adding those with anthropic traces or with relevant degrees of combustion. This will avoid unnecessary waste of time according to objectives and research questions. A consolidation carried out with UHU patafix glue pad sticks (removable) will allow the study of fracture patterns. All recovered information is going to be collected in a relational database management system (MySQL).
Data processing and spatial analysis interpretation
Archaeological data will be stored on a geodatabase to carry out spatial analysis employing the GIS and R-programming language. The plotted fragments with related zooarchaeological and taphonomic information, 3D Cartesian coordinate, orientation and slop (to check the incidence of post-depositional movements) are going to be georeferenced and incorporated into QGIS. Three-dimensional data recording enables the development of different types of maps that are useful to summarise bone dispersion. This information will be processed with multivariate statistic techniques: k-mean cluster analysis and hierarchical classification methods to identify the occurrence of clusters and their contents as evidence for spatial structure; discriminant analysis to measure the differences between individual groups of objects; point pattern analysis to identify spatial interaction between object assemblages or subsets and assessing the occurrence of influencing external factors.
3D models and virtual refits
Is dedicated to a 3D scan of a sample of bones. A high-precision 3D scanner, manufactured by Polygon Technology (Darmstadt, Germany) will be used. The light pattern is recorded by two digital cameras and results in a complete model of the artefact, which is digitally stored as a triangulated point-cloud model. The accuracy is about 0.2 mm per pixel. On the basis of the previous literature, each bone requires between 15 and 45 minutes to be scanned and post-processed (including two cluster scanning, registration and triangulation – all following the scanner’s protocol) depending on its degree of complexity. I will concentrate only on refits (150-200 samples) to obtain a procedure for bone junction comparison which is unique in a specimen. Nevertheless, a margin of error for excess or defect is taken into consideration. As refitting is the primary method for increasing the number of identifiable specimens but it is time-consuming and vulnerable to the researcher’s experience, I wish to develop a computational tool for future bone refits analysis. To test the feasibility of the method I carried out a pilot study on 20 bone remains of various sizes. Each scan took approximately 1/1,30h as I am a novice with 3D tools. Supplementary scans were produced when very irregular shapes hid portions of the surface during the first shot round. Breuckmann’s SmartScan structured light scanner currently available at IPHES has been used. Despite good results (shape, colours and geometric features) they are not sufficient for the research but they attested to the feasibility of this method. 3D models are rapidly developing due to their excellent way of recording and, above all, providing novel insights.