Multivariate Analysis of Geochemical Data on Tektites

A fundamental problem in the origin of tektites is the estimation of the composition and nature of the parent material from which they were derived. Of equal importance is the nature and complexity of the system of processes that led to their formation. Some information relevant to these problems is thought to be contained in the estimates of correlation among various chemical constituents in tektite specimens. These correlations, however, are subject to restraints imposed on the data by the fact that the constituents within each specimen sum to 100 per cent. The restraints affect correlations among both major and minor constituents. As an alternative, correlations among the ratios of each constituent to SiO₂ have been examined. These correlations are independent of the effects of closure, and correlation coefficients of zero may be taken as indications of total unrelatedness. By use of ratio correlations the samples are compared on an equal SiO₂, rather than the traditional equal total weight, basis. Applying the multivariate statistical methods of factor analysis to the ratio correlations, a model of tektite formation is developed which can account for most of the covariance relationships among major and minor constituents in twenty-one tektite (bediasite) specimens from east-central Texas. According to the model, the principal factors controlling the tektite compositions have been the amount of meteoritic material incorporated into the principal parent material and the nearness of the site of formation to a point of meteoritic impact.