106 THE FUR SEALS OF THE PRIBILOF ISLANDS. 



It is evident that between these two seasons the decrease in the herd was more 

 strongly marked than between any other two seasons. The reason tor this is of 

 course plain and will be referred to again in its appropriate connection. In a word, 

 the resumption of pelagic sealing in 1894, nearly doubling as it did the draft of the 

 preceding year on the herd, naturally showed itself very strongly on the rookeries 

 in 1895. 



PHOTOGRAPHS OF ABANDONED TERRITORY. 



There is one way in which the photographs of successive seasons show definite 

 results, and this is in the recording of the absolute abandonment of breeding ground. 

 Thus on the flat at the head of the "slide'' on Ardiguen rookery there were 78 cows 

 in the season of 1890 and none whatever in 1897. This fact is clearly recorded in the 

 photographs of the two seasons. Photographs of the large breeding masses on Keef 

 rookery, Tolstoi, and Vostochni, which are calculated to show most plainly the effects 

 of shrinkage, give clear evidence of the fact even between two successive seasons. 

 This evidence might not, however, in view of the daily fluctuations in rookery popula- 

 tion, be so clear if it were not corroborated by more definite proof. 



THEIR LIMITATIONS. 



There are, on the whole, many reasons why photographs are at best unsatisfactory 

 guides to the actual condition of the rookeries from year to year. In the first place, it 

 is difficult to take them on exactly the same dates on account of adverse weather 

 conditions, and to be of value for comparison between two successive seasons they 

 should be so taken. Again, the period during which photographs of any value can be 

 taken is short. It falls within a few days before or after the loth of .July, which was 

 found in the season of 1897 to be the maximum date of rookery population. But 

 between this maximum and the population of the 8th of July there had been an 

 increase of 20 per cent, while on the other hand from the maximum of the 20th of the 

 same month there was a decline in population of 38 per cent. 



THEIR RELATION TO THE DAILY COUNTS. 



To take a concrete example: The population of the Amphitheater of Kitovi, as 

 counted at its maximum on July 15, showed 703 breeding cows present. On the 14th 

 its population was 556, a difference between the two days of 20 per cent. Photographs 

 for these two days of this rookery in the breeding season of 1897 would have indicated 

 20 per cent of difference, if they indicated anything. Suppose similar conditions for 

 the year 1896, and that a photograph taken on the 14th of .July in one year is to be 

 compared with one taken on the same date of the next, or vice versa. Such a com- 

 parison would clearly be misleading. The result would be more striking it the 

 comparison was made between a photograph in one season for the l."ith and one in 

 another season for the 20th. If we continue the comparison we find that by the .'.1st 

 of July our population of breeding cows has declined 40 per cent from its maximum. 



Here, however, comes in another element of confusion in the use of photographs. 

 The pups have been growing in the meantime and are becoming more and more 

 conspicuous. They are always at least twice as numerous as the cows, and in a 

 distance photograph they can not be readily distinguished from their mothers. It 

 therefore happens that a photograph taken on the 31st of July for the Amphitheater 



