Repeatability of measurements of small mammalian fossils with an industrial measuring microscope

Publisher Website
Google Scholar
Abstract
Five small (<2.5 mm), morphologically complex mammalian teeth were measured repeatedly (10 measurers, 2 dimensions per tooth, 1 daily measuring session repeated over 10 days, to accuracy of 0.01 mm) using a “Shopscope” measuring microscope. The experiment was designed to show, within and among measurers: (1) the ranges of variation that may be expected; and (2) the primary components of observed variation. Variations in repeated measurements consistently were very low (for lumped samples, N = 100: 80% of time SEX was 0.002 mm or less; mean of SD = 0.019 mm), even though specimens were reoriented before each trial and several participants had no prior experience with the instrument. A randomized, complete block design (with repeated measures on blocks) analysis of variance showed that, in 8 of the 10 required length and width measurements, the component of intrameasurer variation explained a larger part of total variance than did the component of intermeasurer variation. Assuming that unambiguous instructions for measurements are provided, very close agreement in estimates of biological variation can be expected from one measurer to another with use of an instrument such as the Shopscope. Closeness of repeated measurements to the same quantity (“precision”) is high, and the instrument itself is inherently accurate. Most “error” within any individual measurement (i.e., difference from mean) probably reflects minor variations in specimen orientions and/or individuals' choices of measurement end-points. We plead for researchers to publish routinely definitions of measurements used.

This publication has 0 references indexed in Scilit: