Page:Popular Science Monthly Volume 54.djvu/683

herd. The counts of harems, and especially of cows present, are much inferior in value. The latter counts, however, point in the same direction. The harems on all the rookeries were counted in both seasons. In 1896 there were 4,932; in 1897 there were 4,418, a decrease of 10.41 per cent. The cows actually present on certain rookeries at the height of the season were counted in both seasons. Where 10,198 were found in 1896, 7,307 were found in 1897, a decrease of 28.34 per cent.

The important element in these special counts, undertaken with a view to determining the relative condition of the breeding herd for the two seasons, is the count of pups. All other classes of rookery population fluctuate from day to day, but the pups remain constantly on shore and near to the place of birth for the first six weeks of their lives, and it is merely a matter of patience and skill in counting them. Such a count on any rookery is an absolute record of the number of breeding females which has visited it for the season in question.

The minimum figure of nine per cent adopted by Professor Thompson is based upon a recount of a single rookery made by himself under conditions less favorable for accuracy than in the case of the official counts, which give the larger figure of twelve per cent, and which were made jointly by representatives of both commissions.

8. It is not easy to apply the various counts in the form of a general average to all the rookeries of the islands. We recognize that a notable decrease has been suffered by the herd during the twelvemonth 1896 to 1897, without attempting, save by setting the above numbers on record, to ascribe to the decrease more precise figures.

This is a rather extreme statement of the uncertainty which may be assumed to attach to these figures. The problem is not an easy one at best and its factors are complex. This should always be borne in mind, but not to the extent of doubting the value of the figures. The areas counted were large enough to be fairly typical. The counts were carefully done, and are accurate enough for all practical purposes. The probable error for the 15,000 more or less pups counted would not exceed 500. But as the counting was done in exactly the same manner and by the same persons for the two seasons, such errors as may exist are common to both counts and the relative conditions are unaltered. The figure of twelve per cent, moreover, must be taken as in itself a minimum, since it is the result of a number of individual counts varying in accuracy; and all in a sense underestimates, inasmuch as more animals are always overlooked among the rocks than are counted twice.

But the exact percentage of decrease is immaterial. That it has been a “notable” decrease is sufficient, and this is unquestioned. It may be noted in passing that this unequivocal decrease occurs in two seasons during which there was perfect enforcement of the regulations of the Paris award.