286 



BULLETIN OF THE BUREAU OF FISHERIES 



Table 6 shows that the average scale number (80.54) of the 2-year fish is in general 

 not strikingly different from the averages of the older fish. 



A comparison of the differences between the means of these year classes with 

 those due to personal errors (Table 4) or to size of the fish (Table 3) leads us to the 

 same conclusion as that obtained by the application of statistical methods; namely, 

 that the number of scales in the lateral line remains constant with the year classes 

 studied and consequently with age. 



If, now, the average scale number remains constant with the year classes, we 

 should e.xpect to find insignificant differences between the averages in Table 6 of the 

 same age groups taken in different years. We find that the differences between the 

 averages of the 1921 and 1923 fish, which vary from 0.80 to 1.22 (average = 1.05), 

 are unquestionably insignificant, but that the differences between the averages of the 

 1922 and 1923 fish, which vary from 2.66 to 4.35 (average = 3.36), are much greater. 



Table 7. — Frequency distribution of scales in the lateral line of Bay City herring 



Table 7, showing the frecjuency distribution of scales in the lateral line, likewise 

 indicates that the 1921 and 1923 material belong to the same population, whereas 

 the 1922 fish probably belong to another, in so far as scale number is concerned. I am 

 inclined to beheve, in spite of the fact that the difference between original and dupli- 

 cate scale counts diminished with an increase in the number of variates (Table 4), 

 that the averages of the 1922 fish were low because, as stated on page 282, a large per- 

 centage of these individuals had lost many of their lateral-line scales, and scale pock- 

 ets are overlooked more easily in counting than scales. Tliis belief is supported by 

 the fact that the 1922 averages are persistently low, irrespective of the year class 

 involved. The large differences between the scale averages of tlie 1922 and 1923 age 

 groups, as given above, therefore should not be considered as evidences for the 

 variability of the number of lateral-line scales with the year classes. 



Whether the 1922 herring are actually different or whether the difference is due 

 to inaccurate scale count, it is at least clear that the 1922 collection can not be com- 

 pared in its scale number with the others on the assumption that all the collections 

 form a unit.^ In studying the scale averages of the different age groups belonging to 

 the same year class, only the 3-year fish of 1921 and the 5-year fish of 1923 may be 

 compared with confidence. The sixth age group of the 1923 collection includes too 

 few specimens to permit valid comparision with the fourth age group of the 1921 

 collection. 



! If such an assumption be held, then it is dear the data of Table 6 show that neither a decrease nor an increase in scale number 

 With age occurs, as the averages of all year classes decrease in the 1922 and increase in the 1923 age groups. 



