A REFINED TECHNIQUE FOR MEASURING CRYSTAL SIZE DISTRIBUTIONS IN THIN-SECTION

Log-normal size distributions are of the form n=n(o)e(-L/alpha), where n=number density, L=crystal length, and alpha is a constant. A method for measuring three-dimensional lee-normal crystal or grain size dist ributions (CSDs) in thin section has been deduced from computer experi ments, in which 2D sections were cut through assemblages of 3D solids. The size ranges and distributions studied were appropriate for igneou s microphenocryst to megacryst populations. Conversion from 2D to 3D i s based on an exact correction for spheres of uniform diameter. Cumula te numbers of polygons with length greater than or equal to L (N-2D) a re converted to N-3D by the equation: ln(N-3D=ln(N-2D/[L . S])+ln(gamm a)-beta/[L . S] The number density is then obtained as n=-dN/dL. The p arameters S and gamma correct the measured lengths and n(o) (n(o)=numb er density at L=0) respectively, and are functions of crystal shape. T he parameter beta is a weak function of the degree of spatial orientat ion of the crystals. Highly symmetrical shapes such as cubes, octahedr a, and elongated prisms can be accurately measured when randomly orien ted; however, rectangular solids with a not equal b not equal c cannot be accurately measured because they produce bimodal length distributi ons in cross section. Strongly oriented textures (trachytic or lineate d) can be accurately measured regardless of crystal shape. New CSD dat a from alkaline rocks and a kimberlite give examples of CSDs modified by megacryst retention, xenocryst addition, phenocryst accumulation, a nd groundmass nucleation.