20 



and pups draw back to the higher gromuls, oftentimes leaving entirely bare the 

 space originally ocenpied. The general lorni of the new arrangement corresponds 

 roughly to that of the old, and remains constant during September and October. 

 Eeef rookery, which was ])itrticnlarly observed in this regard, showed on September 

 !S the cows and imps drawn back so thiit the line of Townsend's crosses,* none of 

 which had been reached in the breeding season, ibrmed the center line of the mass. 

 Though all the cows and pups of this rookery were driven otf at least three rimes in 

 the last week of September, they resumed their old position so exactly, that the eye 

 could detect no ditlerence iu the arrangement. This position, ]iractically unchanged, 

 they occni)ied on the date of the \iisi observation, October 22. It would doubtlcsa 

 only be broken uji with the final dejjarture of the cows and pups iu November. 

 This backward movenu-nt etiectually obscures any line of demarcation that might be 

 formed in the breeding season, and there is nothing to guide tlie e,\ e in determining 

 the difference in location, if any, between the rookeiies of one year and of the next. 

 For this reason, faithiul as the work of the observer and maj) jnaker may be, it can 

 not be made exact. 



HALF TIIK cows ONLY PUKSEXT AT ANY OXK TIME. 



Furthennore, it is now known that definite and relatively constant as are tbe 

 boundaries of the rookeries at the height of the season, the numl)er of cows then 

 found is not the numlier of breeding seals. At no time in the ye.ir are more than 

 about one-half the breeding seals ]irescnt on the rookeries at once. This is shown 

 by the faet that the number of ]iu]is on any rookery is nearly double the greatest 

 number of females counted on it at any one time. The yearlings and virgin two-year- 

 olds do not come to the islands until after the rookeries have passed their jieriod of 

 greatest compactness, and none were present at the time of the counts made in 1895. 



COUNT OF PUPS. 



Until this year it ha<l been regarded as impossible to test these counts and esti- 

 mates by an actual count of the pups. The finding of 6.049 ])ups on Kitovi rook- 

 ery, while the maximum nundjer of cows in the height of the season was but 3,152, 

 shows that the number of cows arriving late is much greater than had been sup- 

 posed. Examination of other rookeries sliows the necessity of adding 75 per cent to 

 the count of females to include these absentees. 



ESTIMATES FOR 1895. 



The investigators for 1895 have estimated the number of breeding cows on the 

 I'ookeries at any one time in the height of the season as 70,42.3 for the two islands. 

 This estimate is not far from correct and is probably slightly below the truth. There 

 are two reasons for thinking that it is an underestimate. First, the count was made 

 a little early, when "the rookeries may noth.ave quite reacheil their breeding height." 

 At Lagoon rookery, for example, all the harems counted were visible from the boat 

 in Village Cove, l^ater in the season we found some twenty or more harems on the 

 inner side of the ridge of rocks and out of sight from the waters of the Cove. These 

 were doubtless overlooked in 1895. Second, the estimate that the massed rookeries, 

 as Tolstoi, Vostochni, are but twice as densely i)opulated as Lukauiu and Kitovi, is 

 probably incorrect. A count of dead bodies on Polovina killing grounds shows an 

 average of 13^ square feet to each body. On Ardiguen rookery an isolated harem 

 of 33 cows was noted, and its l>oundarie8 marked from a distance by stcnies which 

 circumscribed it. When the seals were absent at the time of the count of starved 

 pu])S, in September, this space was carefully measnied, and showed an average of 8 

 square feet for each seal. Where seals are massed on rookeries, the si)ace occupied 

 by each seal is more nearly 12 than 23 scpiare feet. The estimate, therefore, of 23 

 square feet is too high, and consequently the number of seals based upon it is too 

 low for the massed portions of the rookeries, even if 46 square feet, which, as a mat- 

 ter of fact, is doubtless also too low, be taken for the other estimate. 



Accepting the figures of last vear, 70,423 cows on the rookeries would mean an 

 aggregate of 123,240 breeding cows. To this number nuist be added 25,000 to 40.000 

 virgin two-year-olds and as many yearlings to form an estimate of the actual num- 

 ber of cows for 1895. That the figures given by us for 1896 are slightly higher than 

 those for 1895 does not mean an increase in numbers since 1895, but simply an increase 

 in the data on which an estimate may be made. Messrs. True and Townsend, for 

 example, count 2,640 cows on Kitovi rookery. This count is the most important 



* The white crosses were painted in 1895 on rocks marking the backward extension 

 of the large masses of seals on the various rookeries at aliout the middle of July. 



