A Simple Grid Deviation Technique of Study of the Areal Composition Variations in Granitic Bodies

A method of preparing grid deviation maps for the representation of the areal composition variations in granitic bodies is described. A comparative study of this method and the trend surface method by using the data of forty-five modal analyses of Bahalda hornblende granodiorite indicates greater effectiveness of the former in several respects. THE point count method of modal analysis has made possible the assessment of the areal composition variations in bodies of granitic rocks with reasonable speed and accuracy (cf. Chayes, 1949, 1952 ; Fairbaira, 1951 ; Whitten, 1957 ; Saha, 1959). But attempts to represent such variability have not been altogether satisfactory so far, because of (a) rather small scale variations exhibited by such bodies, (b) lack of adequately spaced exposures of rocks, and also (c) the subjective element involved in contouring irregularly spaced, highly variable sets of data. The technique of fitting least-square trend surfaces, based on the solution of equations involving non-orthogonal polynomials, as introduced by Whitten (1959, 1961), removes effectively the subjective element in contouring, but it has been criticized on several other grounds (Chayes, 1961 ; Saha, 1963). In the course of detailed statistical studies of the areal composition variations in parts of the composite mass of Singhbhum granite, the writer has used a simple technique of preparing grid deviation maps on the basis of the modal data. This technique is illustrated here by reference to the grid deviation maps of a 15 square miles area of the Bahalda granodiorite (Saha, 1954 ; Saha and Chakraborti, 1963)— one of the constituent units of the Singhbhum granite—on the basis of forty-five careful modal analyses. A square grid pattern (one mile to a side) was superposed on the geological map containing the data points such that (a) each of the squares contained at least one datum point and (b) there was no undue concentration of the points in any particular grid square, In all, nineteen grid squares covered the entire sampled area (Text-fig. 1). Next, the average values of K feldspar percentage, quartz percentage and colour index within each grid were computed, these averages being assumed to be located at the centre of gravity of the data points in each grid. Using these grid averages, the average values of the three variables for the entire sampled area were computed, half weightage being given to those marginal areas that cover GEOL. MAG. VOL. 101 NO. 2 10 F ig . I. ® +2 .7 -3 .0 -/