274 BULLETIN OF THE BUREAU OP FISHERIES 



lateral line and just posterior to the stipra-claviclc of the shoulder girdle. Ordinary 

 straight calipers were used for both measurements. To find the point on the body 

 proper for the Hi measurement the point of the caliper is inserted as far as possible 

 under the soft, posteriorly directed flap of the supraclavicle. This point of the body 

 is usually more dorsal than the one chosen on the suboperculum for H and is more 

 constant in position. The H head measurements were made by Doctor Koelz for 

 taxonomic purposes. The 11^ head measurements were made later by the writer, 

 as they represent more nearly the true head lengths as included in the length measure- 

 ments of the body. Hi measurements are more often parallel with the long axis of 

 the body than the H measurements. The head lengths were not corrected for 

 shrinkage. The distance between the caliper points was measured on a steel milli- 

 meter tape. T and Ti represent the length of the body proper, excludmg the head 

 and tail. They were obtained by substracting the head from the body length {K), 

 as follows : T= K- H; T, = K- Hi. 



The weights of all fish were recorded to the nearest J^ ounce and were taken 

 with a sealed "Chatillon improved circular spring balance." It is assumed that the 

 readings of the balance were accurate. After it had been used for this work it was 

 damaged, unfortunately, before its error could be determined. The herring collected 

 at Bay City, Mich., in 1921, 1922, and 1924 were weighed on shore while fresh. In 

 order to determine whether the preservation of a herring in formalin and alcohol 

 materially alters its weight the herring collected and weighed November 1, 1922, 

 were weighed again individually in March, 1923, after having been hardened in 

 formalin and transferred to alcohol. Before weighing the j)reserved fish they were 

 piled in a tray to allow the excess alcohol to drip off. The 499 herring averaged 5.17 

 ounces before preservation and 4.96 ounces after preservation, a loss in average 

 weight of 0.21 ounce. As the weights were read fairly accurately to only ]4, ounce 

 the error involved in the weight of each individual should not have exceeded ±0.125 

 ounce, and average weights obtained by reweighing identical material should differ 

 by less than ±0.125 ounce. The above averages obtained by weighing before and 

 after preservation differ by 0.21 ounce, or about 4 per cent. There seems, therefore, 

 to have been some loss of weight in preservation. Jarvi (1920) found that specimens 

 of Coregonus alhula increased in weight after preservation in formalin. The average 

 increase for the sLx specimens employed amounted to 0.5 gram. 



SCALE MEASUREMENTS AND COUNTS 



All scales were removed from the left side of the body and whenever possible 

 from the area situated midway between the dorsal fin and the lateral line. This area 

 was chosen after a careful examination had shown that its scales were less variable 

 in shape and size, when compared one with another, than those of other parts of 

 the body. To compare the body-scale ratio of different fish corresponding scales 

 were thought to be possibly essential in order to eliminate the errors due to the vari- 

 ability of scales from different body regions. (See p. 311.) In the marine herring 

 Lea (1910) located corresponding scales by means of their position on a definite 

 myomere of a definite scale row. He found that, the caudal extremity excepted, 

 only one scale is superimposed upon a myomere. In the coregonids the scales do 

 not follow the myomeres so closely. I found, by enumerating both the myomeres 



