114 



ALASKA FISHERIES AND FUR INDUSTRIES IN 1917. 



Average Harems in 1915, 1916, and 1917 on Rookeries on which the Puis were 



Counted in 1917. 



The significant portions of these data are the totals as compared 

 with the totals for the entire herds for each island sho^vn as follows: 



Thus it is seen that in 1915 the average harem for the rookeries 

 counted in 1917, 48.80, almost exactly coincided with the average 

 for the herd, 48.13. The diff'erence was 0.67 over. In 1916 the 

 difference was 2.17 in the other direction. It would seem this in- 

 dicates with sufficient clearness that if the average harem for tlie 

 herd in 1917 is placed at 25.92, which was found on 1 1 rookeries, it wiU 

 be within 2 of being correct. As the total number of harems was 4,850 

 it makes the possible error in the total number of breeding cows 

 9,700. The maximum error coefficient in this enumeration there- 

 fore becomes 7.71 per cent. It is regretted that this is so high, but 

 there are no known means whereby it can be reduced when a com- 

 plete pup count is not practicable. Upon consideration of all other 

 data this is not as discouraging as it might seem at first glance. 



The further the subject progresses the greater appears the neces- 

 sity of assuming that the average harem for the entire herd is near 

 that of the rookeries upon which counts were made. It can not 

 vary more than two from this. Upon the completion of all calcula- 

 tions and when all assumptions had been made for rookeries not 

 counted, the average harem for the herd was raised from 25.92 to 

 26.39, as shown in table on page 109. The principal influence bringing 

 about, this change was the fact that the average for the herd in 1916 

 was 2.17 greater than the average on the same rookeries coimted in 



