particularly true after predators or scavengers have worked over the 
the corpses. This would potentially bias the resulting mortality 
reported. 
Lack's (1946b) suggestion that January 1 should be used as 
the date from which to calculate the survival of birds on the basis of 
' banding recoveries has been heartily seconded by Farner (197) but is 
not followed in the present report. The use of January 1 will, of 
course, make more data available for the calculation of adult survival 
rates, but the determination of subadult survival should not be neglected. 
Lack (1946b) has demonstrated that in several species (European black~ 
bird, song-thrush, and lapwing) the mortality rate decreases from the 
bird's first August 1 to the following January 1. Subadult mortality 
or survival rates therefore have considerable significance in the popu- 
lation dynamics of some species, In the larger birds, which are particu- 
larly vulnerable to gunners, subadult mortality rates derived from band- 
ing tend, however, to be somewhat biased. When a life table derived 
from banding data for such species can start close to the time the young 
birds leave their parents, the investigator is in a particularly good 
position to check the bias in his results against an analysis of nesting 
productivity. As I will show in Chapter III, this is also in many cases 
an all-important check on potential band loss. 
Retrapping of adult birds 
An inmense amount of survival data on adult and unaged birds 
has been accumulated by the systematic retrapping of banded birds. The 
correct interpretation of these data depends on the banders themselves, 
as they are the only ones who can estimate annual variations in the 
intensity and efficiency of their own trapping activities. I therefore 
did not use survival data of this type when I encountered them in the 
centralized files of the Fish and Wildlife Service. 
The accuracy of adult survival rates based on retrapping 
apparently varies from species to species. Although the migrational 
homing of adult birds is a generally accepted phenomenon, Nice (1937, 
pp. 176-178) has already shown that it may even vary between sexes and 
that it is poorest among species with specialized nesting sites. Even 
during the nesting season, when bird populations are relatively station- 
ary, many adults that disappear from a study area are rather promptly 
replaced. These replacements must mean that the disappearance of an 
adult cannot always be safely taken as proof of its death, for the bird 
may have deserted and moved to a new location. Direct observation of 
this phenomenon during the nesting season has been reported by Green 
(1916), and circumstantial evidence has been gathered by various observers. 
Quantitative measurements of it are supplied for the song sparrow by 
Nice (1937, p. 173): In 6 years of observation, in which a total of 
565 adults were present at the start of the nesting season, 123 birds 
disappeared in the next two months. The 37 birds that soon replaced 
these 123 represented 7 per cent of the original population. I believe 
that this 7 per cent is a rough measure of observational error in the 
study-area technique that Nice used, and that it could well be added 
to the percentage annually retrapped in an attempt to approximate annual 
survival rate of adults. 
17 
