FUR-SEAL CENSUS, PRIBILOF ISLANDS, 1917. 115 



1917. Had the average for the herd been placed 2 greater than that 

 found on the 11 rookeries counted, however, it would have placed 

 the total number of cows at 135,000, an increase which observations 

 do not indicate as having taken place. 



In deciding upon the average harems for those rookeries not 

 counted which, together with those counted, would make the grand 

 average 26.39, due consideration was given to make that estimate in 

 every case conservative. 



If the maximum variation allowable, 2.17, had been split in two 

 and the average for the herd been placed at 27.08, it would have 

 indicated that more cows came back than we now figure; that is, 

 the loss was not 50 per cent during the first three years of the seals' 

 life or the loss due to old age was not 10 per cent of the breeding cow 

 class. Either of these conditions may be true and either makes the 

 count as recorded herein farther on the side of safety; that is, it 

 is much more apt to be under the actual figures than ov^er, and this 

 is as it should be. No absolute law can be established for losses at 

 sea, and the percentages are undoubtedly subject to more or less 

 variation from year to year, due to causes of which we at present 

 have no knowledge. 



3. PERCENTAGES OF DEAD PUPS AND PERCENTAGES OF INCREASE. 



Before saying definitely what the total number of pups on a 

 rookery which is not counted shall be in a season when other rook- 

 eries are counted, consideration should be given to tw^o other factors 

 than the average harem and the areas. These are the percentages of 

 dead pups and increases. Either of these might form the basis of a 

 census, especially in a year when harem counts for some reason are 

 not made. 



To analyze the values of these we must inquire into the causes of 

 the conditions which are found. As shown in a preceding para- 

 graph, page 102, the percentage of dead pups found on the rookeries 

 was a fairly constant factor during the time when there was a large 

 average harem. But with the increase in the number of bulls 

 there has likewise been an increase in the number of dead pups 

 found at the same seasons of different years. It is perfectly 

 natural to record these as cause and effect. It is not disputed 

 that 90 per cent of all dead pups found now^adays have been kiUed 

 by the bulls fighting and charging over the breeding areas in the 

 early part of the season. At the time of the count, however, few 

 are in a fit state of preservation for autopsy. It is definitely known 

 that some buUs kill more cows and pups than others do, and no law 

 governs the return of these bulls to any particular rookery. The 

 topography on some areas makes the death rate there higher than 

 on others. 



Therefore, while there is certainly a relation existing between the 

 percentage of the dead pups to the total number of pups and the in- 

 crease of bulls, it hardly seems necessary to resort to this method for 

 census estimation, except in emergency. When the minimum aver- 

 age harem exists there is no doubt the percentage of dead pups from 

 trampling is practically constant and the total number of pups could 

 be arrived at very closely by the simple expedient of counting those 

 dead after the breeding season. In future years it will be well to 



