We have devised a computer-assisted graphical method for correlating chemical analyses in suites of related igneous rocks. The method provides a direct and empirical means of sample identification using all of the reported chemistry. In a study of basalt of the Columbia River plateau, the method has been used for (1) checking field identifications of rocks, (2) spotting analytical errors in analyses of samples of known chemical type, (3) identifying repetitions of chemistry in thick stratigraphic sections, and (4) matching analyses of dikes to those of flows that the dikes may have fed.

Program OXVAR creates a set of oxide-oxide variation diagrams on a CALCOMP plotter; any oxide may be used as the base (for example, MgO, SiO₂). For each oxide-oxide variation diagram, program CTSET creates the smallest convex polygon that encloses all the data points as well as a larger polygon that takes into account the error in determination of each data point. A set of nested polygons, one for each oxide, then represents a chemical type. Program CTYPE identifies unknown analyses by comparing them with all defined chemical types. If all the elements or oxides of an analysis plot within the defined polygons for a given chemical type, then an identification is printed. If at least one element of an analysis falls outside the set of polygons defined for each chemical type, then “unidentified” is printed, and qualitative information on the amount of deviation from the defined chemical types is given.