306 
BULLETIN OP THE BUREAU OF FISHERIES 
3-inch-net samples failed to be consistently higher than the average lengths determined 
for the same age groups from samples that were taken in smaller meshes. 
From the above examples it may be concluded that selection by numbers is 
much sharper than selection by length. Apparently nets of different mesh size, 
which in their operation upon a population take samples that disagree only slightly 
as to average length, may take their samples with unequal degreees of facility, with 
the result that while their catches agree closely in average length they^may differ 
markedly as to the actual number of fish taken. Table 78 (catch ob2%-, 2/-, and 3- 
inch nets) shows such a situation. If small differences in the sizes of mesh are ac- 
companied by large differences in the number of fish captured from a certain size 
range, then small differences in the average sizes of different groups of fish may be 
expected to produce large differences in their numerical abundance in the catch of a 
particular size mesh. 
From the above conclusions it can be seen that a series of nets that may take quite 
reliable samples of particular age groups, as average length is concerned, may at the 
same time fail to capture the members of these age groups in numbers corresponding 
to their actual relative abundance in the population of which they are part. Conse- 
quently gill-net samples must be employed with extreme caution in the study of the 
relative abundance of age groups and year classes. 
GENERAL CONCLUSIONS 
If the collections from the four lakes are considered as a whole it may be stated 
that the action of a gill net of specified mesh is predictable only with reference to the 
specific nature of the population to be sampled. The action of a net of specified mesh 
depends first upon the range of length and abundance of the fish within the population 
and second upon those morphological characteristics that determine in what manner 
the fish is held captive. A demonstration of the reliability of a sample obtained by 
use of gill nets from one population does not indicate that the same gear will obtain 
an equally reliable sample from a second, different population. 
On the question of the use of gill-net samples for growth studies it was demon- 
strated that the analysis of growth data with respect to size of mesh of the gear in 
which samples are taken aids in the detection of age groups whose appearance in the 
sample is by reason of selection not representative. These age groups should be elimi- 
nated completely from the data. Such selected age groups were detected in three of 
the four populations considered here. It was demonstrated further that if these 
selected groups are eliminated the remaining growth data can be considered accurate 
and trustworthy within very narrow limits. No such high degree of reliability can 
be claimed for gill-net samples in the study of the relative abundance of the different 
age groups. 
SUMMARY 
1 . This study of the growth of the cisco was based on the determination of age of 
3,882 specimens and the calculation of growth from scale measurements of 3,694 
specimens. The data for Trout, Muskellunge, Silver, and Clear Lakes are presented 
in the general paper while the smaller samples from Allequash and Tomahawk Lakes 
are treated separately in an appendix. 
2. It was assumed that Van Oosten’s (1929) demonstration of the validity of the 
scale method in Leucichthys artedi can be considered to apply to the four cisco popu- 
