ABSTRACT The search for relations among sedimentary properties is a basic procedure in the interpretation and reconstruction of sedimentary environments. These relations are commonly expressed as the correlation coefficient, r, between pairs of variates plotted on scatter diagrams. The observed correlations are closely related to trends or gradients (i.e., systematic patterns of areal variation) in the data. The coefficients of trend surfaces can be used to predict the correlation between two mapped variables, and by expressing the correlation in terms of its covariance and variances, the observed relations can be partitioned into systematic and residual portions. A shift of emphasis for consideration of correlation coefficients (which are not additive) to covariance sums of crossproducts, which are, affords additional opportunities for analysing the contributions of two or more sources of covariation in the total observed correlation between sedimentary attributes. In this approach trend surface analysis plays an important role in partitioning the crossproduct sum. A statistical model is presented for this purpose, and some sources of covariation in beach foreshore properties are examined in part by simulation studies of areal patterns of variation and the population densities of mapped variables.