In the Markham Valley of Papua New Guinea, multivariate graphical displays of serological, anthropometric, and dermatoglyphic population structures are compared pursuant to the hypothesis that fingerprints have a slower velocity of evolutionary change and, therefore, are preferable biological markers for prehistorical reconstructions. Samples from nine villages, which represent three geographical and linguistic populations, are plotted in two dimensions using appropriate multivariate techniques for maximally portraying between sample variability. Both the serological and morphometric displays are found lacking in close clusters, which demonstrates a lack of congruence with the shared languages and, presumably, the ethnohistorical origins of the three populations. These discrepancies between biology and prehistory appear to reflect recent environmental and stochastic perturbations. On the other hand, fingerprint displays conform closely to language affiliations, relatively undisturbed by environmental variation and genetic drift.The relative congruence between ethnohistorical affiliation and fingerprint diversity is further corroborated by comparing the three measures of population structure with geographical distances, using partial rank correlations. In terms of explained variance, the level of association with fingerprints is more than twice than with anthropometry, and 13 times greater than with serology. Whereas metric and serological data provide distorted portrayals of known biohistorical relationships among the study populations, fingerprint data mirror these relationships. The theoretical foundations and consequences of this observation are discussed with respect to the broader question of polygenic versus monogenic biological markers.In a previous paper on the prehistorical implications of fingerprint data in Papua New Guinea (Froehlich and Giles, 19801, we argued that multivariate population comparisons with fingerprints support hypotheses of migration and gene flow along likely geographical pathways, corroborated by linguistic data and oral traditions. In other words, fingerprints provide a population structure which is compatible with processural models derived from language, geography, and ethnohistorical studies.Previous studies of anthropometric and serological variation have rarely portrayed known or suspected prehistorical relationships in Melanesia, seeming rather to reflect environmental differences or recent microevolutionary events. Thus, there is an implicit conclusion from successful prehistorical reconstructions t h a t fingerprints may be more phylogenetically stable than other biological