THE RUSSIAN FUR-SEAL ISLANDS. 105 



in wliieli I fouud, or thought 1 found, diflereuces, but in the main I agreed, with one 

 notable exception, however, viz, the estimation of the number of iseal.s on theroolieries. 

 Of course, his estimate rehited only to the Pribyh)f group, and as I liiiew the latter 

 only from his description, I felt bound not to criticise him. But I became sure of 

 this: His methods and results did not apply to the Commander Islands. Elliott's 

 method was to ascertain the area of the rookeries in sipiare feet and then multiply this 

 with an average figure calculated from the number of seals, large and small, counted 

 on a certain piece of ground. But I found insurmountable obstacles. In the first 

 place, the method recpiired not only a very detailed and accurate topograi)hical survey 

 on a large scale, of each rookery, but the calculation of the area presented an exceed- 

 ingly difficult problem. Xo two pieces of ground are alike. In some the beach is 

 smooth and the seals are lying close; others are covered with smaller or larger rocks 

 and stones, where the seals lie scattered as a matter of necessity. In other places, 

 again, there are open spaees or thin spaces. Then, again, the outlying rocks and reefs 

 defy close calcnlation as to number and area. On Copper Island small herds of seals 

 would be found in corners and coves, on ledges of clitts,and under overhanging rocks, 

 sometimes entirely out of sight and most times beyond compntation. I iound that 

 every factor of the calculation would have to be estimated averages, and that these 

 averages in their turn had to be founded upon estimated items; in short, that the 

 whole calculation would have to be a product of guesses multiplied by gnesses. As 

 we have to deal witli large figures, it is evident that a mistake in the estimated factors 

 must result in disastrously great mistakes in the total number. 



Suppose, for instance, that I had '-estimated" the area covered by the seals on 

 both islands to be i,000,0()0 s([uare feet. If I "'estimated" the average ground covered 

 by a seal (mother, pup, and bachelor) on the rookeries to be 2 square feet, I would obtain 

 a total of 2,000,000 seals on the Commander Islands. But, on the other hand, if I 

 guessed that on the average a seal, large and small, on the rookery occupies 5 square 

 feet — and this would possibly have been more nearly correct — I would get only a 

 total of 800.000 seals, large and small. According to this method, various persons 

 might estimate the number of seals on Xorth Eookery, Bering Island, from 20,000 

 to 120,000, and yet it might be impossil)le to convince any of them that they were 

 mistaken. 



A numeration of tlie seals being utterly valueless unless accurate, or at least 

 approximately accurate, I naturally regarded such an estimate of the number of seals 

 on the rookeries not only as useless, but as downright pernicious. Actual counting 

 being impracticable, and an individual judgment of the number being about as value, 

 less as the above method of calculation, unless aci[uired by a very long practice, I 

 gave uj) all attempts at presenting figures. 



When, after twelve years. I again visited these rookeries the same question con- 

 fronted me. In one place, where I had au unusually good opportunity, I tried to make 

 an estimate of the average ai-ea occupied by a seal on that particular rookery. On 

 -luly l(i, watching the seals before me on Kishotchnoye Itookery, Bering Island, 1 wrote 

 in my notebook as follows : 



Here is a harem right iu front of me, 1 sikatch, 16 matki, and about as many pups. They ai'e 

 lyiug as close together as ahout the average, and they easily cover a piece of ground 20 by 20 feet, 

 40(1 aiiuare feet, or more than 11 snuare feet per animal, pups ard all. Ten square feet per animal for 

 this rookery is, therefore, I think, a fair estimate. 



