Error rates in buccal-dental microwear quantification using scanning electron microscopy.

Dental microwear, usually analyzed using scanning electron microscopy (SEM) techniques, is a good indicator of the abrasive potential of past human population diets. Scanning electron microscopy secondary electrons provide excellent images of dental enamel relief for characterizing striation density, average length, and orientation. However, methodological standardization is required for interobserver comparisons since semiautomatic counting procedures are still used for micrograph characterization. The analysis of normally distributed variables allows the characterization of small interpopulation differences. However, the interobserver error rates associated with SEM experience and the degree of expertise in measuring striations are critical to population dietary interpretation. The interobserver comparisons made here clearly indicate that the precision of SEM buccal microwear measurements depends heavily on variable definition and the researcher's expertise. Moreover, error rates are not the only concern for dental microwear research. Low error rates do not guarantee that all researchers are measuring the same magnitudes of the variables considered. The results obtained show that researchers tend to maintain high intrapopulation homogeneity and low measurement error rates, whereas significant interobserver differences appear. Such differences are due to a differential interpretation of SEM microwear features and variable definitions that require detailed and precise agreement among researchers. The substitution of semiautomatic with fully automated procedures will completely avoid interobserver error rate differences.