especially by not completing a line of ditto marks down a page. One 
of the best banding jobs ever done on the redhead duck lost mich of its 
value because the bander gave neither the age nor the sex of the birds 
he banded in the summer of 1932. 
In some cases the schedules now on file at the Patuxent Refuge 
represent third-hand (rather than second-hand) information. This condi- 
tion is the product of an office operation called "reducing the files." 
Where two schedule pages for the same species are less than one-half 
filled out, clerks have occasionally copied the data from the second 
page onto the first. Many of these workers (apparently in the WPA) 
failed to transfer the year of banding as well. "File reduction" has 
not only made for such errors but it has been frequently characterized 
by failure to copy age and sex. I was unable to locate a sizeable 
fraction of the original schedules thus reduced. Because the schedules 
are always second-hand sources of data subject to typically human errors 
of transcription, it is regrettable that the Fish and Wildlife Service 
has never required its cooperators to turn over to the government their 
original records when their interest in the program has terminated. (This 
would at least permit the safekeeping of primary data by inexpensive 
microfilming. ) 

Reports by the public.--It is obviously rather difficult to 
test the accuracy of recovery data sent by the general public to the 
Fish and Wildlife Service. The amount of information supplied by lay- 
men varies considerably. I noticed that about 12 per cent (in 1281 
cases) failed to mention the places of recovery, over 10 per cent 
(in 1445) failed to specify the date of recovery, and that at least 
33 per cent (in 373) were accompanied by enclosure of the band. In 
the fall of 1947, duck hunters were enclosing bands in about one-half 
of the letters received at Patuxent. 
Two tests of the accuracy of numbers reported to it have been 
used by the Fish and Wildlife Service with varying degrees of consistency: 
(1) In some cases the species and sex cited in a recovery letter can be 
compared with the bander's data as to species and sex. In one series 
of 1445 Mallard cards, I found that there had been no follow-up in 12 
cases (0.8 per cent) where bander and hunter differed as to sex. In 
this sam series, some hunters differed with the bander's report as 
to the species involved. The significance of the latter discrepancy is, 
of course, debatable. (2) In other cases, the Service has used a 
screening test of geographic implausibility. A New York—banded greater 
scaup shot on the Pacific coast would fall into this class. (One 
actually did.) A yellow palm warbler recovered in Newfoundland in 
November, 52 days after it was banded on Cape Cod (Lincoln 1939: 137) 
would be another example. Reports like these are usually followed up 
by requests that the band be shipped to the Service. 
Among 375 letter writers who enclosed bands, I noted that 25 
mentioned the band number correctly in their letters, 25 wrote the num 
ber without giving the serial prefix (0, 41, 42, etc.), and one mentioned 
19 
