Table 1.--Methods of Constructing a Composite Life Table -- I 
Unchecked hunters! reports of adult mallards banded 1926-3 on the 
Pacific coast and shot in some subsequent year are here grouped 
together to show the results of two different methods of calcula- 
ting annual mortality rate. When such rates would involve less 
than 100 alive at the start of an age interval, they were not 
calculated in this report. , 


Time-specific Life— 
Original Data Dynamic Life-table Analysis table Analysis 
x dx Ix . ay ity dy aS” 
Age Interval No. Alive Number Mort. Rate Alive Calcu~ Mortality 
(in Years) Shot at Deaths per Year at lated Rate 
Start Start Deaths per Year 
x41 to x42 845 1411 845 59.9 845 561 66 
X+2 to x+3 284 566 28h, 50 28h, 160 56 
X+3 to x+ 12h 282 12) bh 12h 52 42 
x+y to x45 72 158 72 446 72 43 - 
x4+5 to x+6 29 86 29 = 29 0 = 
x46 to x+7 29 57 29 - 29 17 us 
X+7 to x+8 12 28 12 = 12 5 - 
x48 to x49 7 16 #xF = 7 2 - 
X+9 to x+10 5 9 5 - 5 2 33 
x+10 to x+11 3 hy 3 - 3 2 - 
X#11 to x+12 1 1 1 - 1 1 = 

*The mean mortality rate per annum, this is always the most statistically 
reliable mortality rate in an avian life table. 
In this table I have used some Fish and Wildlife Service re- 
ports taken from hunters! letters as an illustration of the two methods 
of analyzing a composite life table. Most of these birds were banded 
during the hunting season by A. J. Butler at Chilliwack, B.C., and by 
personnel of the Malheur National Wildlife Refuge at Voltage, Ore. 
First-year recoveries were not included in this table. These birds 
were unaged at the time of banding; their ages in the table therefore 
run X+1, x+#2, etc. | 
Lack's method has the disadvantage of requiring a number of 
years to elapse before the life table can include the results of a 
given banding operation. The method makes for small sample size 
because recent banding work will not give data on the full life span 
of many species. This can be circumvented by expressing deaths for 
each age interval as a percentage of the number of banded birds 
available. Such a device tends to reduce sampling errors in the 
early age intervals. It is illustrated in table 2. The percentage 
results are, of course, expressed as birds dead or alive per 100 
banded, and in succeeding chapters I will frequently convert such 
statistics to the number per 1000 banded birds available for study. 
10 
