A STATISTICAL STUDY OF NITROGEN FIXATION BY CLOVER PLANTS

Abstract

Clover plants inoculated with Rhizobium trijolii were grown on agar in cotton-plugged bottles and on sand in open jars in the greenhouse and the following measurements taken: mgm. of N fixed; number, type, and location of nodules; length of stems and roots; and number of leaves. From these data, a statistical analysis was made, including the following tests and calculations: homogeneity of data on mgm. of N fixed and on number of nodules per plant; distribution curves of N fixed and number of nodules; simple and multiple correlation coefficients among variables capable of quantitative estimation; and association tests among those variables admitting of only qualitative specification. Statistical methods used are detailed. Distribution curves of variables examined are essentially normal except that of number of nodules in the sand series. This departure appears to be due to inaccuracies in counting nodules rather than to any gross abnormalities in type of distribution. A mathematical demonstration is given showing that normal statistics can be applied with confidence to the populations investigated (mg. N fixed and number of nodules), if a few plants are combined for a measurement. The simple correlation coefficients indicate that no 2 variables measured are very highly correlated with each other. In the sand series the greatest correlation was observed between number of nodules and mgm. of N fixed (0.63), and length of stem and mgm. of N fixed (0.63). In agar, the highest correlation coefficient was between length of stem and mgm. of N fixed (0.52) ; the correlation between number of nodules and mgm. of N fixed was only 0.39.