JoHS. Schmidt 115 



My own interpretation of the offspring analyses, which is supported 

 by the results arrived at in 1903 by W. Johannsen in his studies on bean 

 communities, is roughly as follows. It was found possible by population 

 analyses to subdivide Zoarces viviparus into numerous local " races." 

 The individual offspring analyses showed that the smallest unit hitherto 

 considered, the " race," may be resolved into still smaller elements, ex- 

 pressed by the means of the offspring samples. These smaller elements 

 may differ widely one from another, and likewise from the average of 

 the population (" race "), but when added together they reproduce the 

 picture of the race itself as the latter is expressed in the results of the 

 population analyses. 



As already mentioned above, it is a generally accepted view among 

 fishery biologists that racial differences are due to the differing external 

 conditions under which the various races live. The salinity of the water 

 is here as a rule considered of great importance, and as this factor is also 

 the best elucidated up to date, I have endeavoured to ascertain how far 

 any agreement could be found to exist between the salinity and the 

 average values characterising the races of Zoarces. A few examples 

 may be quoted. Of the populations we have analysed, that of St. 57 

 in the North Sea (Anstruther, Scotland) is the one living in saltest 

 water, St. 52 in the Gulf of Bothnia in the Baltic (Hudiksvall, Sweden) 

 that with lowest salinity. The difference between these two places is 

 very great, the salinity at the Scottish station being about 34 °/^^, and 

 in the Gulf of Bothnia about 5 °/„^ only. The population at St. 57 thus 

 lives in a salt solution more than six times the strength of the water at 

 St. 52. The graphs in Fig. 9 show the values for number of vertebrae 

 in these two populations. It will be noticed that there is no difference 

 to speak of between the two graphs, either as regards course or average 

 values, the latter being about 1160 and about 116-4 respectively. This 

 example t hus seems to suggest that the s alinit y is not of any real im - 

 portance. 



Another example is illustrated in Fig. 10. The two populations 

 here shown are from Roskilde Fjord, Sealand, Denmark, and Kjelds 

 Nor, Langeland, Denmark. At these two places the salinity is very 

 nearly the same and rather low, viz. about 12°/^^. But it is at once 

 noticeable that the values for number of vertebrae exhibit an enormous 

 difference as between the two populations, amounting to no less than 

 about 11-5 mean. So great is the dissimilarity indeed that the two 

 populations have not a single common variate. The picture presented 



