REVISED CLASSIFICATION OF SOME IRON-METEORITES BY USE OF STATISTICAL PROCEDURES

Au, As, Ru, Pd, Rh, Os, Ir and Pt were quantified in 120 iron meteorites using graphite furnace atomic absorption spectrometry and instrumental neutron activation analysis. The data were combined with literature values for Ga, Ge and Ni in order to devise a statistical group classification system based on principal components and discriminant analyses, while at the same time preserving the integrity of the existing classification system as far as possible. If Ga and Ge are excluded, 72.9% of the total variance of the system was accounted for by abundances of Ir, Os, Pt and Ru. Very little of the variance was accounted for by Ni and Pd. If Ge and Ga are included, they become the most important diagnostic elements and the order of importance for group classification is then Ga > Ge > Ir > Os > Ru > Pt. The discriminant analysis of the abundance data involved use of the D2 statistic, initially over all groups excluding anomalous irons. Where group changes were inferred, meteorites were transferred to their new groups and the calculations repeated. The next stage involved adding the anomalous irons to the pool of data to establish whether any of these could be assigned to a group. The final stage of classification involved examining the morphological and mineralogical properties of nine meteorites whose change of group had been inferred by the statistical procedures. On this basis, reassignment of four meteorites was rejected and the remaining five accepted. The latter included Cranbourne, Jaralito, Mundrabilla and Nantan, formerly low-Ni IIICD meteorites, that we now recommend should be returned to group IAB from which they had previously been removed. We also recommend that Denver City (formerly ungrouped) be reassigned to group IIIF Anom. Our statistical procedure confirms the validity of previous classifications not involving statistical discrimination.