94 THE FUR SEALS OF THE PRIlilLOF ISLANDS. 



tliougli the actual population has been diminished by the more frequent and pro- 

 longed absence of the cows. The number of bulls therefore found for those two 

 rookeries on July 31 can not be accepted as representing the number of families in the 

 breeding season and must be reduced. 



CORRECTIONS ON ST. PAUL. 



The need for revision in the census of St. Paul Island is not so great; but our 

 experience during the past season shows that in all probability the harems on Zapadni 

 and Tolstoi rookeries were slightly overestimated and those on the main part of 

 Polovina underestimated. 



Zapadni rookery occupies a long gradual slope back of the more abrupt bowlder 

 beach. This slope was filled with savage bulls, making close approach in 1890 impos- 

 sible. The count of harems was made from a boi»t offshore and was not corrected by 

 a count from behind. From this point of view of the water the whole field could not 

 be iuspected, and the figures of last year were felt to be merely a rough approximation. 

 The estimate for those not clearly seeu was certainly too large. This year the rookery 

 had shrunk so largely that observation points from the rear, commanding the entire 

 breeding ground, could be reached. In like manner the estimate of Tolstoi for 1896 

 was made under difticulties, which were more satisfactorily oveicome in 1897. 



On Polovina the conditions were similar to those on Zapadni. The main part of 

 the rookery lies on a gradual even slope, behind which the idle bulls were cotlected in 

 such numbers as to prevent near approach, while there was no opportunity to inspect 

 the rookery from the water. This year the conditions were so changed as to permit 

 of much closer inspection, and although the mass of breeding seals had plainly 

 shrunk at every point, practically the same number of harems were counted, making 

 an addition of harems necessarj'. 



SIVFTCH ROCK. 



A third correction must be made in the estimate of Sivutch Kock. The original 

 estimate of this rookery was made after an inspection of the rock with a field glass 

 from the plateau of the Keef, which showed 27 bulls in charge of harems on the 

 northern face. At the time of the count of dead pups in August a landing was 

 eft'ected and an attempt made to estimate the number of harems, as the rookery 

 seemed plainly greater than our original estimate allowed for. Sixty-seven bulls 

 occupying harems were then found, though it was plain that the rookery was wholly 

 disorganized. This season Sivutch Rock was inspected from all sides in a boat at 

 the height of the season and an adequate notion of the extent of the rookery obtained. 

 Harems were found on the southern side and in places where their presence was not 

 suspected in 1896. The total number counted in 1897 was 102. It is therefore 

 necessary to largely increase the original estimate for this rookery. 



In balancing these under and over estimates, however, they practically neutralize 

 each other and leave the total for St. Paul for 1896 about what it was. For St. George 

 Island also the difference resulting from the revision is not great, being only about 600 

 cows. 



THE IMPORTANT CORRECTION IN THE CENSUS OF 1896. 



But the chief error in the census of 1896, and this applies to both islands, lies in 

 the percentage of correction to be added for absent cows. This has already been 



