LIFE HISTORY OF LAKE HERRING OF LAKE HURON 



283 



Table 1. — Comparison of duplicate scale counts with the original for various numbers of herring taken 

 October 29, 1921, and November 1, 1922, at Bay City, Mich. 



The duplicate scale count for 113 herring of the 1921 collection averaged only 

 0.42 less than the original, while that for 63 herring of the 1922 collection averaged 

 1.02 more than the original. As was expected, on account of the large number of 

 scales missing in the lateral line the personal error in the counts for the latter col- 

 lection exceeded that involved in the counts for the former. 



Table 4 seems now definitely to answer our inquiry. The range (0.10 to 0.99) 

 in the differences as well as the mean (0.55) of the differences between the grand 

 averages of the scale number of the two se.xes (p. 282) may be accounted for very 

 well by the personal factor involved in the scale count; but the range (1.27 to 2.44) 

 in the differences and the mean (2.09) of these differences of the two size groups 

 (p. 282) can not be so accounted for, as they greatly exceed the personal errors involved. 

 Neither is it probable that random sampling can account for these large differences, 

 inasmuch as the greatest difference obtained between any two scale counts of one 

 and the same series, as given in Table 4, amounted to 0.60 (80.29 to 79.69), which 

 means that the greatest difference due to random sampling amounted to approxi- 

 mately 1.20. The consistent difference between the average scale number of small 

 and large fish of an age group is apparently, then, significant; the large herring have 

 the greater number of scales in the lateral line. 



The greater number of scales in the large fish may be accounted for in one of 

 two ways: (1) The scale number can be determined more accurately for large fish 

 than for the small, due especially to the larger size of the lateral-line scales at the 

 caudal extremity in the former, or (2) the large fish of an age group were always 

 large in their year class and consequently needed and always had more scales than the 

 small individuals. 



The data already accumulated for the 4-year herring collected in 1921 and 

 referred to above (p. 282) may be employed to determine the status of the first propo- 

 sition. The fish were divided into two size groups, and the averages of the duplicate 

 scale counts of each size group were compared with those of the original counts. 

 Sixty-two herring 231 milhmeters or less in length were found to have an average 

 of 80.08 in the first scale enumeration and 80.02 in the second, a difference of 0.06. 

 Fifty-one individuals 232 millimeters or more in length were found to have an average 

 of 81.39 in the first count and 80.63 in the second, a discrepancy of 0.76. These 

 data indicate that scale enumeration is as accurate for the small as for the large fish 

 considered. 



Table 5 has been constructed to test the postulate that the fish with the larger 

 number of scales were always large. Each age group of the two collections employed 



