
sensitivity of the test decreases when low mortality rates are en- 
countered. Under conditions involving reasonably high rates of 
mortality, this test can probably be used to detect major bias in 
age distributions for species breeding at the end of one year. This 
really means that it is best used when one is wrking with small 
birds. In my own work, however, productivity analyses lost much of 
their potential effectiveness because so few facts are known about 
nigh productivity and the age at which the larger birds begin to 
ree e 
Internal comparisons.=«-Up to the present, separate life 
tables have not been published for birds found dead, those shot, 
those killed accidentally, and so on. Accidental mortality seems 
like a singularly interesting type of sampling, but it is either 
poorly represented or poorly identified in banding recoveries. 
Table 6.—Productivity Required to Keep a Population Level Stable 
The species governed by this table must breed at the end of one 
year; the young produced are those alive at the start of the life 
table, here taken as some date such as September 1. 
Adult Mortality First-year Mortality Rate 
70% 50% COZ LOZ 3 
Rate 0% 
per 
Year Productivity Required per Pair 
“60% LOO - ° “a 
50% 3-4 2.5 209 1.67 
LOZ 2el 2.0 1.6. 1.3 1.1) 
30% 2-0 1.5 1.2 1.0 0.86 
20% 1.3 1.0 0.8 0.67 0.57 

Paynter (1947) has attempted to test the age bias in 
herring gull recoveries by first dividing the reports into two 
groups: “natural recoveries" (those birds found dead presumably 
from natural causes) and "artificial recoveries" (those in which 
nan was the known agent of death or capture). The test then involved 
a comparison of the mean length of life of.the birds in each sample. 
Although birds aged O-1 comprised 46 per cent of the 67), "natural 
recoveries" and 60 per cent of the 578 "artificial recoveries," 
Paynter found an insignificant difference between the two means. He 
therefore concluded from this test that the two samples could be 
grouped together as the basis of an unbiased life table and that 
"those birds which died from human interference would in all proba- 
' bility have died at about the same time from natural causes." 
I believe that Paynter's commendable pioneer effort failed 
for at least two reasons: (1) The mean is a measure of central 
tendency in a frequency distribution; it is not a statistic specific 
for a given age. Differences in data on the first year of 
life or the last years of life are therefore masked when mean 
29 
