SEA-FISHERIES LABORATORY. 253 



the odds against are 20 to 1, and on D over 1,000 to 1. The 

 samples of Winter 1914 compared with those of 1921 agree on 

 all three characters, with odds of more than 1,000 to 1 against 

 the differences being due to random sampling. 



The trawled herrings from the Smalls (Table IV) — October 

 1913 compared with October 1914 — show somewhat conflicting- 

 results. D and V suggest that the differences are due to 

 causes other than random sampling, but l.cp.l. shows much 

 lower odds against — about 10 to 1 — whereas in the other two 

 cases the odds are over 1,000 to 1 against. 



It is difficult to explain these anomalies. We consider 

 the most reliable character to be l.cp.l. from the measurer's 

 point of view and would regard it with more confidence perhaps 

 than any other character. It is clearly defined and does not 

 suffer from the effects of distortion due to bad condition, 

 freezing or softening, as might Z>, 7, or A, and in nearly all 

 cases P, when calculated on this character is low. In the one 

 case only — that of the Smalls, October 1913 and October 1914, 

 does this character give any value of P which indicates any 

 approach to a " good fit." So that the bulk of the evidence 

 inclines to the theory that there is no racial homogeneity in the 

 samples compared. 



It seems pretty clear that the herrings inhabiting the 

 Irish Sea are a mixture of sub-races, or genotypes, and 

 that one of these predominates at one time, and another at 

 another time. For the chances that the differences observed are 

 due solely to errors of random sampling are, very often, far too 

 small to allow of any other conclusion. Further complications 

 are the choice of the diagnostic character, or whether we should 

 not, preferably, choose a combination of characters, or whether 

 some other character than those studied must be sought ? 

 Evidently the question of races is not so simple a one as has been 

 thought and some further investigations, as to what characters 

 are truly germinal ones and what are environmental only, must 



