FUR-SEAL FISHERIES OF ALASKA, 1909. 
37 
Contrast with 1908 . — Using the rookery of Ketovi as typical of 
conditions on the whole of St. Paul Island, we may contrast the 
estimate of cows given above with that made by the same method 
in 1908. 
In that year Ketovi contained 48 harems and 1,960 pups, affording 
an average harem of 40.8. As 1,062 harems were found on the 
entire island, the application of the average harem of 40.8 would 
produce a total of 43,329 breeding cows. On St. George, the typical 
rookery, North, had 94 harems and 4,123 pups, giving an average 
harem of 43.8. As there were 241 harems on the entire island, an 
estimate of the entire number of breeding cows on that island, 
based upon the average harem obtained as above, would give 10,555. 
For the two islands, therefore, in 1908, by the methods of estimation 
outlined above, there was in 1908 a total of 53,884 breeding cows. 
Since 45,786 cows were estimated to be present in 1909 and 53,884 
in 1908, it would appear that a decrease of 8,098 cows, or 15 per cent, 
has occurred during the year. 
DIFFERENT BASES OF ESTIMATION. 
This estimate of breeding cows, as has been stated, is based upon 
the size of an average harem obtained by counting the pups on only 
one rookery on each island. It may be granted that had another 
rookery on either island been selected as typical an entirely different 
result would have been obtained. This may be seen readily by a 
scrutiny of the average harems on the various rookeries on St. Paul 
on which pups were counted in 1909. As it is, the Ketovi rook- 
ery average harem of 34.1 represents a fair mean between the high- 
est and the lowest averages obtained by counting other rookeries, 
and might fairly be taken as typical of the entire island when con- 
sidering the size of the average harem. The fact that the average 
harem may be made to vary greatly by the selection of other rook- 
eries is cited, however, to show that the estimate made is open 
to criticism on the ground of uncertainty as to its accuracy. 
If, instead of constructing an estimate based upon an average 
harem and contrasting that estimate with one made in the same 
manner the previous year, we were to take the percentage of decrease 
shown by an actual count of pups on certain rookeries, it is believed 
that a more accurate idea would be obtained as to the changes occur- 
ring in the breeding cows. 
In observing the decrease in seal life it has appeared that certain 
rookeries, or rather portions of rookeries, either show a smaller rate 
of decrease than others or in some instances do not show any decrease 
at all, while others do. We may take it from this that certain locali- 
ties are focal points in the breeding areas, attracting the breeding 
