FUR-SEAL CENSUS, PEIBILOF ISLANDS, 1917. Ill 



would indicate over 100,000 in existence, a figure differing not greatly 

 from tlie one derived later in this report from the birth rates. (See 

 census summary, p. 123.) 



This table contains actual counts in some cases. Here it was 

 found there were 25 per cent as many bulls, old and young, as bache- 

 lors. This would make 5,580 of the former and 16,743 of the latter. 

 It was manifestly impossible to determine the proportion of old and 

 young buUs with any degree of accuracy worthy of notice. 



THE DATA ON ROOKERIES WHERE COUNTING OF PUPS WAS NOT DONE 



It is admitted by aU that an actual count of all living and dead 

 pups is the best known means of arriving at a close estimation of all 

 classes of seals in the herd. But when the number has increased to 

 such an extent that complete pup counts become impracticable other 

 means must be devised. Several methods may be considered. 



1. AREA COMPUTATION. 



In the early days of sealing, when the herd was very large as com- 

 pared with its present size, this was the only method considered in 

 arriving at a census. Thus Charles Bryant, the first agent of the 

 Treasury Department on the Pribilofs, used it, and so far as known 

 originated it. He was followed by H. W. Elliott in 1872-74 and 

 again in 1890. His work was exhaustive, but the results were such 

 that they received severe criticism. So bitter was the fight waged 

 that the actual good in the area method of computation was lost 

 sight of and was scarcely again considered until 1917. 



The basis of any computation by this method must of necessity 

 be a unit of area for eacli seal. A close approximation of the total 

 breeding areas of any and all rookeries can be obtained by plotting 

 them on charts. Then the division of the area by the unit gives the 

 total number of cows, buUs, and pups in the space. The accuracy 

 of the method depends upon (a) the skiU of the obseri^er in sketch- 

 map work and (h) the correctness of the seal unit. Regarding the 

 first, little can be said. The writer would consider Elliott's areas as 

 given in his several reports very accurate indeed. His skiU in 

 delineation is well recognized. In 1915 and 1916 the writer plotted 

 the breeding areas on the large scale Coast and Geodetic Survey 

 charts of the rookeries at the height of the breeding season. Little 

 time, however, could be given to this phase of the work, and it was 

 attempted solely to give a general idea of the location of the breeding 

 masses. Greater accuracy was not desired. 



In 1917 A. C. Reynolds was detailed to assist in the seal-census 

 work, and his training as a civil engineer especially fitted him for 

 this phase of the work. His results are entirely satisfactory, and, 

 coupled with other information related to census work, these give, 

 in the writer's opinion, the most comprehensive view of area calcu- 

 lations which has so far been possible. 



The unit of area assumed by EUiott to be occupied by each seal 

 was 2 square feet. He made allowances for the difference in size of 

 the bulls, cows, and pups, and the fact that only about one-half the 

 cows are present at any one time on the rookeries. But he did not 



