In geology, as in physical science in general, it is sometimes possible to approach complex problems by arranging observations or experiments so that selected variables are held constant. In many cases, however, such a simplification is either difficult to achieve, or unrewarding by reason of uncertainties and artificialities entailed in attempts to duplicate natural conditions in the laboratory. Advent of high-speed digital computers has made it possible to treat such problems by various multivariate statistical techniques. One of these techniques, factor analysis, can be applied to problems involving as many as 200 variables recorded quantitatively or qualitatively on any number of samples. By treating such a data array simultaneously, relationships that might otherwise be missed can be identified and objectively evaluated. One of the most useful features of this system is its ability to specify the number of independent casual influences at work in a given problem and to assess the relative importance of each variable at each locality. Applications to modern sediment data and ancient sedimentary rocks indicate that the scheme is a useful supplement to usual rational procedures of data analysis.

End_of_Article - Last_Page 269------------