86 



THE BOTANICAL MAGAZINE. 



[Vol. XXVI. No. 304. 



will show the correlation table in Fukure fruits. These fruits 

 were collected 1911 in A habitat. 



As the table shows, the variation of seeds will give a clear 

 skew curve, which is almost one-sided. This fact is probably 

 dependent on its poor fertility. Thus we see 23 non-seeded 

 fruits among 674, that is about SA^/o. 



As regards the correlation-coefficient, I have obtained only 

 0.076 ±0.038. That is to say, they are correlated each other 

 in a ver3^ slight degree. Glancing at the correlation-table, we 

 see a peculiar negative correlation^^ in 10-13- vesicled fruits. 

 Namely, the more the number of vesicles increases above 11, the 

 more that of the seeds decreases. 



Next I will show the table, obtained in IJnshumikan 1911 

 in A habitat. 



Although the number of variates is verv small in my obser- 

 vations, it is nevertheless clear, that the seeds of Unshumikan 

 show a " halbe Galton-Kurve." Thus we see 38 non-seeded 

 fruits among the total of 52 fruits; that is 73 o/^ of these fruits 

 are not fertile. 



Here the manner in which the seeded fruits were produced 

 is not clear. It may perhaps be due to the fertilization of other 

 mandarins as the late Prof. Ikeda pointed out. At any rate it 

 seems to be true that poor fertility shows a half Galton-curve. 



') Camp. Vernon: Variation in animals and plants, 1903. p. 73. 



Johannsen: Elemente dcr exakten Erblichkeitslehre, 1909. p. 248. 



