Abstract

In the present study multivariate statistical methods emerging from log-ratio methodology are applied to an extensive major element dataset of Cenozoic volcanic rocks of Hungary. For an easy comparison, several conventional data plots are also given. The revealed compositional geometry shows a good agreement with geological models based on geoscientific methods which do not fulfil a definite statistical approach. Subcompositional patterns within the two major groups (alkaline basalts and calc-alkaline rocks), and their good separation, known from earlier interpretations of stratigraphical, petrographical, geochemical and other data, are statistically confirmed. The linear compositional patterns disclosed here for these two rock series provide additional details helping to describe quantitatively and, hence, elucidate the nature of petrogenetic processes that affected the Cenozoic volcanites in Hungary.