Furthermore, supposing the group of these observational points as a 

 random sample taken out from the normal population of two variables, 

 the rejection-ellipse(l) has been drawn at the level of significance of 

 '0-01. By means of the rejection-ellipses differences of the type of corre- 

 lation of the two indices have been tested between the Pacific and the 

 Japan Sea coasts, and further it has been also tested whether Soya, 

 Iturup Island, and Hukuoka belong to another type or not — that is, 

 points fallen outside of the ellipse are distinguished from the type at the 

 level of significance of 0-01. Thus it has been observed that the values 

 ■of Soya and Iturup Island belong neither to the. type of the Pacific 

 Coast nor to that of the Japan Sea, that Hukuoka is distinguished from 

 the type of the Pacific Coast but not from the Japan Sea, and that between 

 both types of the Pacific and the Japan Sea there exists no significant 

 difference. Then the correlation between the distributions of the lati- 

 tudinal values in the flat fishes and of the coefficients of closeness in 

 mollusca probably shows no significant difference as well in the Pacific 

 Coast as in that of the Japan Sea ; in other words, the mode of distri- 

 bution of species in the flat fishes are parallel to that in Mollusca, but 

 in the localities of Soya and Iturup Island the type of fauna is distinguished 

 from that in which the distribution of species gradually changes with 

 the latitude, mixing the northern and southern fauna in the Japanese 

 coasts having the centre at lat. 35° N. — that is, at the latitudes of these 

 localities the fauna contains scarcely southern forms ; therefore, it may 

 be estimated that the northern limit of distribution of this type is lat. 

 43-44° N. and oppositely the southern limit is lat. 26-27° N. 



It has been said biogeographically that the Japanese coast extends 

 over the Boreal Pacific Province and the Tropical Indopacific Provirxe, 

 but their boundary is not so clear. This fact is natural from the view- 

 point of coefficient of closeness, and rather the boundary must be made 

 by the statistical conception. From this point of view the northern 

 limit discussed above as that of the type of the main Japanese coasts 

 may be just equivalent to the boundary between both Provinces. 



Nomura and Hatai(2) (1936) have divided Japanese seas into seven 

 provinces from the point of distribution mainly in Mollusca and Brachio- 

 poda. When northern and southern limits of distribution of species 

 exist in a definite range it is possible to find out several species 

 characterizing each province, but the distributions of many species 

 deviate little by little from the definite range and extend over the 

 imaginary biogeographical provinces, and therefore it is unnatural to 

 set up a province only from the characteristic species in that range. 

 Accordingly the biogeographical division is subject to an expediential 

 boundary. The division can be expressed rather strictly in terms of the 

 coefficient of closeness than by the border-lines. In order to discuss the 

 biogeographical distribution it is rather necessary to take account of the 

 gradual decrease in the number of common species from locality to 

 locality or with the distance expressed in terms of the latitude, by means 

 of the coefficient of closeness. 



(1) Masuyama, M. (1943) : loc. cit. 



(2) Nomura, Sitihei and Hatai, Kotora (1936) : A Note on the Zoological 

 Provinces in the Japanese Seas. Bull. Biogeogr. Soc. Japan, 6 (21), 207-214. 



322 



