1252 records, which he broke down into 67); "natural recoveries" (birds 
found dead presumably from natural causes) and 578 “artificial recover- 
ies" (mortality due to man, birds captured alive, vague records). 
Three statistical objections to his interpretations of this breakdown 
have already been mentioned under "Internal Comparisons" in Chapter III. 
It is merely necessary to add here that 803 reports of birds dead, 311 
reports of birds captured, caught, found ill, or observed, and 138 
others of a vague category (band recovered~-11, no information--23, 
"found"—=58, etc.) were subsequently used as a mortality series. 
Although Paynter may well be justified in assuming that some of these 
311 birds died the year they were captured, this may not be true of 
all of them (such as 5) caught by fishermen and 25 "captured and re- 
leased"). Captured birds make up 25 per cent of his sample. It is 
interesting to note that under similar circumstances Dutch investi- 
gators would regard all 1252 records as birds known to be alive at 
the start of the year in which they were recovered, and that they 
therefore would analyze them as a survival series in a time-specific 
table. Paynter's decision to use these records as mortality data in 
a dynamic life table, instead of a time-specific one, at least pro~ 
duces minor statistical distortions that snould be clarified and 
justified. 
In the hope of shedding some further light on Marshall's 
results, I constructed a life table based on birds banded from 1925 
to 1930 inclusive (table 40). Tnese reports were not checked against 
banders! schedules. 
The 60 per cent first-year mortality rate here is identical 
to the 60 per cent found by Marshall, but the mean mortality rate 
for later years (27 per cent) is lower than his data (35 per cent) 
indicate. The difference in survival curves is clearly evident in 
figure 10 where--after age 10-11 in the 1925-30 curve-——the apparent 
mortality rate seems to be increasing. If this was a band-loss ef- 
fect and the adult mortality rate was constant at all ages, a 
straight line running from age l=2 or 2-3 could be extended through 
the point at age 10-11 and intercept the 1 per cent line close to 
age 23-2. Under these circumstances, the inference is that, among 
every 500 birds (50 in table 0) alive on the September 1 following 
their hatching, one individual in every 500 presumably would live 
to this advanced age. This correction for band loss suggests an 
adult mortal.ty rate of 25 per cent per year. In Europe where the 
banding program is much older, herring gulls have been known to 
sae at least 25 years of age in the wild (Schitz 1935; Paludan 
1951). 
If all herring gulls breed at age 3-, the 1) adults in 
table 40 would need to raise 2.) young per pair to September 1 to 
keep the population stable. With a band-loss correction as indicated 
above, this productivity would have to be about 2 young per pair. 
Preliminary nesting data secured by Paynter (199) indicated a mean 
productivity of less than one fledgling per nesting pair in the 
heavily crowded Kent Island (N.B.) colony in 19h7. The life-table 
results obtained here cannot, therefore, be accepted at this time. 
9h 
