MULTIVARIATE STATISTICAL ANALYSIS OF CLINOPYROXENE COMPOSITIONS FROM MAFIC AND ULTRAMAFIC XENOLITHS IN VOLCANIC ROCKS

Abstract

Multivariate statistical comparison of clinopyroxene compositions in various types of mafic and ultramafic xenoliths in volcanic rocks has shown that the compositional differences are adequately described by two principal factors that are linear combinations of the major chemical constituents of the minerals. Within the framework of our statistical model, the analytical expressions of these new factors are P 1 = −0⋅02 SiO 2 + 0⋅09 TiO 2 + 0⋅64 Al 2 O 3 + 0⋅59 Cr 2 O 3 + 1⋅35 FeO + 9⋅65 MnO−0⋅50 MgO + 0⋅29 CaO−2⋅26 Na 2 O + 8⋅0 P 2 =−0⋅93 TiO 2 + 0⋅07 Al 2 O 3 + l⋅23 Cr 2 O 3 −0⋅46 FeO + 1⋅74 MnO + 0⋅36 MgO + 0⋅10 CaO−1⋅66 Na 2 O + 8⋅0. A classification diagram in P 1 − P 2 space is presented in which three linear zones corresponding to the compositions of clinopyroxenes under the conditions of garnet, spinel, and plagioclase depth facies are delimited. Fields inside these zones are identified which correspond to the compositions of pyroxenes in different types of mafic and ultramafic xenoliths representing different temperature equilibria. The proposed diagram has several advantages over binary and ternary diagrams: the positions of data points are determined by the whole chemical composition of the mineral, and the effect of each component can be assessed from such factors as the rock composition, temperature, and pressure. This P 1 − P 2 classification diagram allows identification of depth facies of the particular mafic and ultramafic xenoliths or intrusions based only on their clinopyroxene compositions even if the corresponding depth mineral-indicators (plagioclase, Al-spinel, or garnet) are lacking