ERROR RATES OF SURFACE ROUGHNESS PARAMETERS USING OPTICAL MICROSCOPY AND LEICA STEREOEXPLORER SOFTWARE

Our laboratory employs the computer program Leica StereoExplorer (Leica Microsystems Schweiz, AG, CH-9435 Heerburugg, Switzerland) to study a variety of anthropologic, paleontologic, and geologic specimens including lithic artifacts, bone surfaces, scavenging marks, pottery shards, and rock samples. 3D surface topography models and analyses generated in the Leica StereoExplorer software are created by capturing stereo image pairs from a conventional stereomicroscope. Because the error and reproducibility of our results remains unmeasured, the reliability and comparability of our data is unknown. Here, we assess the intra- and interobserver error by taking repeated measurements of a known surface roughness (Ra) by one or multiple workers. We documented an overall intraobserver error rate of ~1.5% and an interobserver error rate also of approximately 1.5%. Because both intra-and interobserver errors are likely influenced by variables associated with the data collection process (lighting, sample orientation, magnification, etc.), we further examined the relative importance of each variable. We found that changing the orientation of the light source induced ~3% error and different sample orientations gave ~23% error. Data of the same object at different magnifications yielded an average error of ~33% with more error at higher magnifications. By changing the size of objects (of known values) and keeping the magnification constant, we found that we had an overall error rate of 14.5% with more error associated with smaller objects. Our results indicate that inconsistent data is generated by changing any of the data collection variables; however, as the intra-observer error rates indicate, if the data collection process is consistent, error is around 2%. Furthermore, our results indicate that magnification and object size are important considerations when measuring an object. There is likely an optimal balance between magnification and object size to achieve the best results and discovering this relationship will be the next step in our reproducibility study. In light of our reasonably low inter- and intrarobserver error rates, we provide examples of how 3D surface topographic models and roughness analyses can be used to answer interesting archeological questions and to quantify taphonomic variables.