In 1907 it had been established by Marr and Geerts, that margalitic soils with a content of less than 0.03 % P205 soluble in cold 25 % HC1 or 0.01 % soluble in 2 % citric acid were P deficient. An attempt was therefore made to explain why some soils with P well above this limit responded to superphosphate.Assuming that the availability of Ca3(PO4)2 depended on its specific surface area, fractions by size were isolated from the soils by Mohr's method of mechanical analysis, and P was estimated in the fractions. P content of the sand fraction was high and decreased with decreasing particle size to a value below the established limits. This contrasted with the rule for soils in temperate climates. Neubauer tests showed that P in the clay fractions was 100 % available, but that of the silt and sand fractions only for 30 %. The distribution of P over the size fractions was especially unfavourable for the margalites rich in P and explained their favourable reaction to P fertilizer.