Variation between samples. - The individ- 

 ual samples of a year -class taken in one area 

 during one year were tested for evidence of 

 significant variation between means. Only one 

 series of samples, out of 64, showed significant 

 heterogeneity between the means of samples. 

 Inasmuch as at least one divergent series could 

 be expected among such a large number of series 

 due to chance if sampling from a homogeneous 

 population, this instance was not accepted as 

 evidence of heterogeneity . Rather, it was as- 

 sumed that each series of samples was drawn 

 from a homogeneous population in respect to 

 vertebral number . 



Variation between age groups . - -Fi sh of 

 the same area and year -class but taken in differ- 

 ent seasons were compared to determine whether 

 they could have been drawn from one population 

 in respect to vertebral number. Out of a total 

 of 52 series involving two or more age groups of 

 the same year-class, only two were found to ex- 

 hibit significant differences between means . The 

 two cases could represent freakish samples 

 drawn by chance from a homogeneous population, 

 although Tester offers alternative explanations: 

 He concluded, however, that he would accept the 

 view that different ages of a year -class in an 

 area were drawn from one population with re - 

 spect to vertebral number . 



Variation between year classes. --I am 

 reproducing table IV from Tester's paper. The 

 table presents in concise form the basic informa- 

 tion on mean vertebral count according to year- 

 class and area . From an inspection of the table 

 it can be seen that significant differences occur 

 between the means of year classes in each area . 



annual variation in mean water temperature in 

 March at William Head at the southern end of 

 Vancouver Island. Using augmented data Tester 

 found a highly significant negative correlation 

 between variation in March water temperatures 

 at William Head and variation in the mean 

 vertebral count of west coast of Vancouver Island 

 year-classes over the period 1927 to 1938. Hence, 

 the correlation previously found for Barkley 

 Sound was found to be general along the whole of 

 the west coast of the island. 



Variation between areas . --The unweighted 

 border mean vertebral counts for areas (refer to 

 table 5) show a slight graded increase from the 

 southern area to the northern area. An analysis 

 of variance was made using two criteria of 

 classification; year-classes and areas. The re- 

 sults show a highly significant heterogeneity 

 between the means of year -classes (which was 

 anticipated) and a highly significant heterogeneity 

 between the means for areas. According to 

 Tester, "This shows that, with due allowance for 

 variation between year -classes, more than one 

 population in respect to vertebral number is 

 present among the five areas along the west coast 

 of Vancouver Island." (Ibid :415 .) In this analysis 

 the mean square for "interaction was also found 

 to be highly significant. Two explanations are 

 advanced as to why the mean counts for each year- 

 class and area failed to vary in the same manner 

 as the border means, one presuming local aberra- 

 tions in temperature conditions during the time 

 of early development and the other presuming 

 partial mixture or interchange between the runs 

 to the five areas. No choice is made of these 

 alternatives, but Tester warns that "the general 

 conclusion that each west coast of Vancouver 



Island area (with the 

 possible exception of 

 one) has an essentially 

 discrete population 

 should be accepted with 

 reservation as to the 

 possibility that mixture, 

 more extensive than 

 'limited' may take place 

 occasionally" (ibid: 4 16) . 



Tester had shown in a previous paper (Tes- 

 ter, 1938) that variation in mean vertebral counts 

 of successive year -classes (1927 to 1935) at 

 Barkley Sound was inversely correlated with 



Tester then tested the reliability of the 

 foregoing with results obtained from tagging and 

 tag recovery experiments. To quote Tester, 



56 



