The data matrix of 192 classified IRD count data from H1-H6 and I1-I5 (excluding I6 because of too low IRD counts) by 22 IRD lithologies (https://doi.pangaea.de/10.1594/PANGAEA.971280) was row-normalized to 100% to give equal weight to all IRD assemblages independently of their specific IRD abundance, but not column-normalized in order to limit the influence of rare, statistically less relevant IRD lithologies. A Principal Component Analysis (PCA) of this data set reveales well interpretable PCA scores and disparate loadings for the first 5 PCA axes. PC1 reflects the contribution of Ooid-bearing Dolomite IRD, PC2 discriminates Muscovite-Biotite Granite IRD from other silicious IRD species, PC3 delineates Shale IRD vs. Hematite-stained Quartz IRD variability, PC4 quantifies Sucrosic Dolomite IRD and PC5 Microcline IRD percentage. We provide Heinrich layer and interlayer affiliations of all PCA scores, which some PC axes separate well, in particular PC1.

Corresponds to Figure 10.