When the numbers of birds found dead or reported shot are 
expressed as year-by-year percentages of the cohort originally banded, 
the recovery series so constructed is every bit as real a mortality 
series (d,) as the original raw data. The chief advantage of the con- 
version is that it permits recently banded birds to appear in the 
calculations for those age intervals where they were available for 
recovery. In composite life tables, the dynamic and time-specific 
methods of analysis stili apply—and with the same variables discussed 
in this chapter, 
Raw mortality data can similarly be expressed as the number 
recovered per 10,000 or 1000 banded and available for recovery in a 
given year or age interval. In an interesting life table for herring 
gulis, Paynter (1947) has attempted to utilize cohorts of recently 
banded birds by dividing the recoveries for each age by the number 
of years of recovery data available. This short cut is justified 
only in those rare cases where approximately the same numbers of 
birds are banded each year, 
Summary 
Avian life tables invariably lack primary data on either 
the number of birds surviving or dying. Even the field data gathered 
for one of these categories by banding samples may not always be 
typical of a species' population. Major sources of bias in life 
tables may be noticed when the results are checked by a productivity 
analysis, or when standard statistical tests are used to compare re- 
sults obtained by two or more methods of recovery. When mortality 
rates are constant from one age to another, the overall mortality 
rate can be ascertained even though only one type of mortality pro- 
vides all the band recovery data. Under such circumstances the re- 
sults are not impaired by consistent crippling losses encountered 
(but not reported) by hunters or by the amalgamation of survival and 
mortality records into a single survival series in a time-specific 
analysis. When a mortality factor varies in degree from one year to 
another, the dynamic method of life-table analysis gives a better 
approximation of the overall mortality rate than the time~specific 
method does, especially if the banding recoveries precede or coincide 
with the operation of this factor (such as hunting) on the population. 
An alleged short cut for survival rate, obtained by dividing the 
deaths reported in one year by the deaths reported in the preceding 
year, is found to be satistically hazardous and is termed survival 
index in this paper. Expressing raw survival or mortality data as 
a percentage of the number of birds banded and available for study 
is recommended as a means of utilizing recent banding work in the 
construction of a life table. | 
These problems reduce themselves to these questions: Can 
banded birds reported as dead be properly regarded as raw material 
for a survival series? Are they examples of the living or examples 
of the dead? TI conclude that in large randomized samples gathered 
43 
