life-table method would, in table 1h, yield exactly 50 per cent for 
all ten years except the last age interval. This method is, of . 
course, also affected by the discontinuity of whole numbers, but - 
the early mortality rates are not satistically dependent on later 
sp eae reports near the end of the mortality curve. 
In dynamic calculations of mortality rate, the Ai etortlon® 
caused by whole numbers at the end of a mortality curve 
may be formulated as follows: 
1 1 
dg = 100 (F - yaad 
where 
a. = net amount of distortion in the mortality rate 
expressed as a per cent, 
d. = number of deaths reported for a given year a, 
1, = number alive at start of that year, calculated by 
Lack's method, and 
m= the missing mortality values at the end of the 
mortality curve. 
Thus, in table 1) the distortion for year 5-6 is 
= 100 Ges sy)! 16 
2 100 (0.32258 - 0.31250)16 
4 1.61 
In actual work with bird=banding data, the small size of the 
sample available at the end of a mortality series makes for obvious 
sampling errors which mask the statistical creep described here and 
preclude the possibility of calculating mortality rates for specific 
age intervals near the end of the life table. Under such circun- 
stances, I have usually computed a mean annual mortality rate for these 
age intervals. As Part II will demonstrate, most of the samples are 
too small to yield valuable information on the later stages: of a 
bird's life. 
Adjusted survival series 
When the number of banded birds recovered in a given period 
is expressed as a percentage of the total number originally banded, . 
the result is called a recovery rate. The older banders, like Whittle 
and Whittle (1926), also called it a returning ratio--the value then 
referring to the percentage retrapped. Whittle (1929) was among the 
first to consider these annual percentages as an index of survival, 
but obvious trapping inefficiencies persuaded him and others against 
making refined calculations of annual mortality or survival rate. 
The best use of recovery rates appears to have been made by S. H. 
Low (1935), who used the trend in four years' retrapping data to con- 
clude that a banded sample of mourning doves would shrink at zero at 
about the ninth year. . 
h2 
