Three orthogonal planes (two vertical and one horizontal with respect to the recognizable bedding structure) of a core sample of apparently graded graywacke sandstone have been petrographically analyzed in order to determine the characteristics of the lithologic gradients in the specimen.

Compositional and textural properties were sampled in thin section according to an orthogonal grid pattern on each of the three faces.

Variability among rows and columns (plus interaction) with respect to each petrographic variable was tested by analysis of variance using a two-cross classification. A larger number of properties show significant variability among rows (in a vertical direction) in the two vertical planes of the specimen than in the horizontal plane.

Quality control models with confidence limits which expose graphically the trends displayed by individual constituents indicate that feldspar proportion, mica proportion, and quartz grain size show significant gradients in the vertical direction perpendicular to the bedding. Quartz proportion oscillates in a non-systematic fashion, and quartz grain shape shows no trend in the vertical planes of the specimen. The petrographic variability observed in the horizontal plane of the bedding is generally trendless. Independence of the individual gradients is apparent, reflecting the influence of explicit processes on the spatial distribution of the petrographic variables in the rock.

Partial trend surfaces show the trends in quartz grain size to be dominant, according to the comparative amounts of variance accounted for by each fitted surface. Again, the major portions of variability are explained by the surfaces in the planes perpendicular to the bedding. The textural homogeneity of the rock in the horizontal (bedding) plane is reflected in the minor trends defined by the surfaces in this direction.

These results indicate that only apparent gradients in the lithologic properties of the rock are detectable by these methods. To assess accurately the "real" gradients, rotational transformations, which would bring the faces into complete orthogonality with the trend, and criteria (such as maximum variance explained) should be employed.

End_of_Article - Last_Page 337------------