From all the results above mentioned, the following summaries may be deduced. 1. The variation in the number of pulp-vesicles in the four races of Citrus Aurantiun L. Subsp. nobilis MAK. var. Tachibana MAK. shows a monomodal curve and the mode is held on 9, 10 or 11. The variation in the seeds of poorly fertilized mandarins gives a strongly skew curve, even a half Galton-curve, while well-fertilized ones, show a symmetrical curve. On the point of number of seeds and vesicles, Fukuremikan and Uushumikan are very nearly allied. Kishumikan is slightly distinguishable from the above two by the number of vesicles, while in the same point Kojimikan is clearly different from other mandarins. The difference of the four kinds of mandarins is more pronounced in the number of seeds than in that of vesicles. 2. As far as my observations go, the local variation of vesicles is not so great as to cause its mode to change its position and it is very like the annual variation in the same tree. It may therefore be conceded that the number of vesicles is of an almost stationary character. 3. The greater average weight in the fruit of the upper part of the Fukuremikan tree is not always correlative with the greater average number of vesicles. The correlation-table between the weight of the fruit and the number of vesicles clearly shows that they are not correlated in a remarkable degree, and so the above result is obtained. 4. The correlation between the number of seeds and of vesicles can be well observed in well-fertilized mandarins, but in poorly-fertilized ones it is not great; for instance, in Fukuremikan a negative correlation may be observed in part of the correlation-table and hence the correlation-coefficient is very small.