FUR-SEAL CENSUS, PEIBILOF ISLANDS, 1917. 113 



The accuracy of computing complete censuses from areas is not 

 as OTeat as from actual pup counts, and in 1917 it was not believed 

 to be as reliable for computing the numbers on the rookeries not 

 counted as the average harem based on a partial pup count with the 

 average harem on the same rookeries which were counted this year. 

 We know better the number of animals on a rookery by comparison 

 with it m former years than we would know by assuming that the 

 unit is the same for it as for some other which was counted. It so 

 happens, however, that the two methods come out with more exact- 

 ness than was at first anticipated. Rookeries not counted but esti- 

 mated upon the basis of the average harem show a seal unit in close 

 approximation to that found on rookeries which were counted. 



The area method and the seal unit should be of increasing value 

 as the herd grows from year to year until finally it supersedes all 

 other methods. But as long as possible every known method of 

 census work should be run in conjunction with every other, because 

 at best any seal census can not be more than a close approximation 

 to the truth. 



2. THE AVERAGE HAREM ON ROOKERIES WHERE PUPS WERE NOT 

 COUNTED. 



While it was suspected on the islands that the increase in cows 

 was not up to normal even before any pups were counted, owing to 

 lack of expansion of certain breeding areas, it was well known after 

 the first few rookeries had been counted. The percentage of increase 

 of the pups, the number of dead pups, the expansion of the breeding 

 areas, and the average harem all pointed to the same general result 

 and indicated early about what the total number of breeding cows 

 would be. It so happened, however, that Lukanin and Zapadni, the 

 first rookeries counted, ran low-average harems this year, while last 

 year they were normal. Conclusions drawn from those alone would 

 have been fallacious, as subsequent counting proved, but after it 

 was known that the percentage of increase in pups was approxi- 

 mately 9, it was evident about what the total number of breeding 

 cows would be. This left the necessity, however, of assuming an 

 arbitrary figure for the average harem on those rookeries which could 

 not be counted. 



Realizing the importance of determining the average number of 

 cows to each bull on these rookeries as closely as possible, every 



Shase of the sealing industry since 1912 was carefully considered. 

 TaturaUy, one of the most instructive helps in this was the table 

 of average harems for those years, as shown. When the known 

 figures were placed in the proper columns, figures for the unknown 

 were immediately suggested; but this was not enough. The table 

 which follows was computed. It was run back to 1915 only, be- 

 cause previous to that the scarcity of bulls made the average harem 

 a conjectural and inconstant quantity. 



99805°— 19 21 



