The analysis which we have given of 1813 flowers is sufficient to show that the idea of a definite fixed number of sepals in the calyx, of petals in the corolla, of stamens in the andrcecium or of carpels in the gynaecium of cruciferous plants is not upheld by an examination of a large number of flowers of this species. In less than 1 per cent, of the flowers examined there was an increase* or decrease in the number of sepals in the calyx ;in less than 1 per cent, there was also an increase or
... lso an increase or decrease in the number of petals in the corolla, but in 2 per cent, there was an increase in the number of stamens in the androecium, while in 22 per cenfc. there was a decrease in the number. Since then the number of sepals, petals and stamens is not absolutely fixed for any of the organs it becomes necessary now to consider whether the number of members in one organ is related to the number in the others. As has already been pointed out we have not only to consider organs as a whole, but, in the case of the calyx and the andrcecium, the constituents of these organs, owing to the fact that these organs are each divided into two separate whorls which are inserted at different levels and are placed in directions at right angles to one another. Further, a special study has been made of the various positions in androecium to ascertain to what extent bilateral symmetry may be regarded as an inherent character of the flower under consideration. By this means also it seems that some definite information might be obtained with regard to the perplexing and, at present, hypothetical theory of chorisis, the reasons for the existence of which have been summarised on p. 219. The statistical part has been divided into two sections: (1) a study of the Means and Standard Deviations, aod (2) a study of the Correlation Coefficients.