became practical. This would seem to include mortality in the nest 
but Paynter (1949) subsequently pointed out that bands found on dead 
nestlings were removed and placed on live birds "so that, as nearly 
as possible, every banding record represents a fledged bird." His 
life tables based on banding data thus begin "sometime in early 
August" (Paynter 1949). The actual difference in starting dates 
thus appears to be a small one. 
(bo) Banding dates.--Marshall (1947, p. 192) used birds banded 
through 1939. Since his paper was published in April 197, his returns 
must have been those available as of September 1, 196 at the very 
latest. Birds banded as late as 1939 could, therefore, contribute 
to no more than the first 7 age groups in his life table. The 
mortality for at least the last 10 age groups (7-8 to 16-17 inclu- 
sive) is thus not completely represented. This error tends to raise 
the apparent mortality rates for the early years somewhat slightly 
and for the later years somewhat importantly. Paynter sought to 
avoid this error by dividing his recovery series by the number of 
years for which reports were available. 
The validity of Paynter's adjustment in 1947 depends on 
the assumption that practically the same number of birds are banded 
each year. Quite to the contrary, the work of the Bowdoin Scientific 
Station did vary widely from one year to another. The extensive 
banding operations early in the Bowdoin program thus left an unduly 
large number of older birds in Paynter's recovery series. This is 
clearly the reason why his mortality rates as published in 199 tend 
to be higher than those he published in 197. 
The banding dates in Paynter's 1949 paper are confined to 
the year 1936 and the recoveries include only those received "up to 
June 30, 1946." Paynter calls this an ll-year study. In a sense 
this is so, but only 10 full yearly age intervals are thus possible 
for life-table compilation, and one wonders how he obtained 11 
(table 7, Paynter 1949, p. 162). This of course introduces only 
minor errors in the calculations. It is curious that Farner (195), 
Marshall (1947), and Paynter (1949) have all committed this over- 
sight, but that none of the Europeans have, 
(c) Type of life table.--When I arrived at the Patuxent 
Research Refuge, rubber bands still enclosed the mortality reports 
that Marshall quite evidently had used. Except in one instance, I 
found that he and I did not differ importantly in interpreting what 
constituted a valid record of a bird reported to be found dead. 
For birds banded from 1923 to 1933, however, 2.5 per cent of the 
cards had not been completed and placed on file; most of these 
were from age 9-11 to 15-16, and all could be pardonably overlooked 
by a visiting investigator. Adult mortality rates calculated by 
Marshall thus tended to be too high. Marshall used only reports 
of birds found dead, and upon these he constructed a dynamic life 
table following Lack's method exclusively. 
Paynter's first life table for herring gulls is based 
upon the results of an extremely interesting analysis of his 
93 
