It seems to me that the annual kill of waterfowl can also 
be reliably ascertained if we ignore the actual extent of the two 
unknowns, unreported hands and crippling losses, and obtain in future 
studies of banding statis ticsthe precise relation of seasonal recovery 
rates to overall annual mortality rates. A hypothetical case will 
illustrate the kind of results that can be expected from such an 
analysis, 
In figure 21 I have plotted annual mortality rates for 
certain adult mallard populations (on the ordinate) against certain 
first-season recovery rates that were available to me at the end 
of this study. A regression line expressing the interrelationship 
of these two rates has been drawn on the graph from A to B. In no 
case did I have actual recovery rates taken from the same samples 
in which mortality rates were also available. These data are there- 
fore crude ones, but they do bring out the nature of the statistics 
that other investigators can obtain with this technique and more 
precise facts. The line AB represents the total overall mortality 
rate per year. It has been drawn in here purely by visual inspec- 
tion. This line mans that a seasonal recovery rate of about 20 
per cent would occur when the overall mortality rate was 70 per 
cent. Similarly with a first-season recovery rate of zero the over- 
all mortality rate would be about 20 per cent per year. The slope 
of this line probably varies slightly when the data are segregated 
from different banding stations. Point A can be recognized as an 
upper limit of nonhunting mortality occurring in the population. 
If a line from point A is next extended parallel to the abscissa, 
this will intersect the ordinate on the right at D. Line AD can 
be said to represent that fraction of the povulation that is annually 
lost through nonhunting factors. It is likely that nonhunting 
mortality does not have the inflexibility implied in this diagram. 
It does, however, give us for the first time a working statistic 
in our exploration of population of this species. 
A third step is to construct line AC because recovery 
rates up to 20 per cent are included in this diagram. C can be 
set as 20 percentage points above D. AC can now be regarded as a 
recovery rate-line over and above the fraction of the population 
lost from nonhunting causes. The area ACD therefore represents 
mortality that is normally reported to the Fish and Wildlife 
Service by the banding program. 
The area ABC includes mortality involved in crippled and 
unretrieved birds and such dead banded birds.as are retrieved but 
never reported to the government. 
The area ABD represents the total fraction of the popula- 
tion dying because of hunting factors. It should be borne in mind 
that the mortality rates used to prepare this graph represent that 
percentage of the total population dying between September 1 and 
August 31. The various areas in this graph therefore represent frac- 
tions of the total population dying for the reasons here assigned. 
155 
