year-classes* are represented in several calendar years, the year-to-year variation in recov- 
ery data resulting from changes in shooting pressure has been minimized for each year-class 
but that for 7-8, table 1. 
After analyzing figures obtained from banded mallards that return to the traps in 
subsequent years, we have concluded that these data do not allow for a valid interpretation of 
population losses because of the small size of the sample and because of yearly variations in 
rate of return of ducks to the same flyway resting or feeding grounds. Our band returns show 
that, although in each year there is some degree of return by mallards to the same flyway rest 
stops, the degree of return varies from year to year. 
Because band recoveries from shot ducks representing the extensive areas over 
which ducks fly are not only many times as numerous as records of live recaptures but are not 
subject to yearly variations in the return to a single spot on the flyway, these band recoveries 
were used in this study to measure population losses in the mallard, black duck, and blue- 
winged teal. 
YEAR-OF-BANDING RECOVERY CORRECTIONS.--In order to use recovery data ob- 
tained the same year that mallards and black ducks were banded, we found it necessary to com- 
pensate for the fact that banding was done south of the breeding grounds during the hunting sea- 
son rather than on the breeding grounds immediately preceding the hunting season. To correct 
for the chronological and geographical differences involving band recoveries in the season of 
banding (part of hunting season and south half of flyway) and band recoveries in subsequent sea- 
sons (entire hunting season and entire flyway), we used a simple ratio that involved the number 
of days banded ducks were actually available for shooting and the number of days these banded 
ducks would have been available if all had been banded before the beginning of the season. Cor- 
rections were made separately for each calendar year and for each age and sex group. 
The first step in making the corrections in year-of-banding data involved finding the 
number of ducks banded in each autumn banding period and then locating for each period the 
mean date of banding activities. This mean date assumes that all ducks were banded on a sin- 
gle day a definite length of time after banding was begun and before it was ended. The next 
step was to determine the number of days from the mean banding date to the end of banding. 
The daily rate of mallard kill was found to be approximately the same before and af- 
ter the mean banding date. This rate of kill was determined by tabulating returns through the 
hunting season for mallards banded at least one year previously. Very few mallards were 
killed in September, more in October, and most in November; then the kill declined through 
December. So that the data would more nearly resemble a normal curve, the number killed in 
September was added to the number for October, and October 1 was assumed in calculations to 
be the first day of banding. 
* The ducks in a year-class are those killed in the year of banding or in a specified number 
of years after being banded and are indicated in table 1 by the columns headed O-1 corrected, 
1-2, 2-3, etc. Ducks in year-class 1-2 are those killed during the period between 1 and 2 
years after being banded. 
