The "sample" is the commercial kill. Since only 3— ear-old 
A-tags in 1950 and only 4-year-olds could be sb a in eae om 
necessary to estimate the number of 3-year-olds in the 1950 kill and the 
number of 4-year-olds in the 1951 kill. The sample is a composite ot these 
two totals. A further adjustment must be made for the fact that in 1950 
and 1951 the sealers were instructed to kill f. ay 
to table 19. or tags see the headnote 

Figure 8. Pup tagging. A seal pup is held on the 
table while a metal tag is clamped to the left foreflipper 
and a perforation is made in the web of the left hindflipper. 
Pups awaiting tagging are restrained by a portable corral. 
Polovina rookery 24 September 1947 (KWK 47-35-11). 
The most important complication is caused by the homing instinct 
of the fur seal. This is well illustrated by table 29. Since varying 
numbers of tags were placed on the different rookeries, the sample is not 
necessarily random with respect to tag recoveries, and the Petersen 
formula will be invalid. This problem of heterogeneity in tagging and 
sampling operations has been discussed by Schaefer (1951). A method of 
estimating the population size under these circumstances has been given 
by Chapman and Junge (1953, MS). The method has been applied to the A-tag 
returns to yield the following results. Calculations and discussions are 
given in appendix F, and a brief summary here. 
Estimated number of pups alive on St. Paul Island by hauling- 
ground groups, September-October 1947.--Caiculations were made for the five 
rookery groups: Northeast Point, the Polovinas, Reef-Gorbatch, Tolstoi, 
