Nice (1937, pp. 180-189) has thoroughly reviewed the 
determining survival of young passerines on iatonsiaeis shied reeniee 
areas. Among house wrens (Kendeigh 193k), starlings (Kluijver 1935) 
and song sparrows (Nice 1937, p. 189), about 17 to 20 percent of the. 
young that leave the nests apparently mst survive to the next breeding 
season in order to keep the populations stable. These birds, Nice points 
out, are never all retrapped because only a substantial proportion of 
the survivors may return to the general vicinity. The manner in which 
the breeding density of these survivors decreases with distance from 
their birthplace is illustrated in figures 2 and 3. At least in 
migratory species having a reasonably wide variety of available 
habitats is which the young can nest for the first time, intensive 
observation and retrapping cannot give a reliable estimate of early 
survival. 
Among those species in which family ties persist for some time 
the observational handicaps are greatly reduced. Brood counts in , 
waterfowl and many gallinaceous birds can often be extended over a 
period of many weeks. In cranes, swans, and geese, these family ties 
seem to persist over an entire winter (Heinroth 1911; Stevenson 191); 
they are also prolonged in at least one of the titmice (Van Tyne 191,8). 
Under such conditions, successive observations of the size of broods 
throughout the summer can yield reliable data on survival. This ap- 
proach has been used on pheasants by Errington and Hamerstrom (1937), 
on blue-winged teal by L. J. Bennett (1938, pp. 73-74), and on redheads 
by J. Be Low (1945). Its limitations on diving ducks (in which hens 
frequently abandon their young) have been pointed out by Hochbaum 
(19h4h, pe 161). | 
There is a considerable difference of opinion as to the choice 
of starting date with which to compile mortality records of banded 
birds reported by the public. Paynter (1947). used July 1; Lack (194 3a, 
1943b, 1946a) used August 1; I have frequently used September 1; Farner 
(195) used November 1; Kraak, Rinkel, and Hoogerheide (190) used 
January 1; and Stuart (1948) March 1. It has now been shown that a 
starting date of August 1 in many passerine species does not exaggerate 
first-year mortality by giving the original banders a special opportunity 
to find dead fledglings (Lack 1946b). In a number of samples that I 
studied, August banding operations prevented me from starting the 
compilation at the first of that month. Both Paynter (1947) and Deevey 
(1947) seem to feel that life tables based on the banding of juveniles 
should start with the mean date on which the nestlings were banded. 
The choice of such a. date seems to me to introduce sampling variables 
of unknown influence. Where a scientific station is located at or 
near a nesting colony, total mortality in the nestlings might be accurately 
recorded by the investigators. Once the birds disperse from the area, 
reports by the public constitute an entirely different: type of sampling. 
I fail to see how the two types of data can be indiscriminately mixed and 
used to compute a first-year mortality rate. ‘Where there is no scientific 
station nearby, the dead bodies of small-sized, very young birds are less 
evident to a casual observer than those of larger, adult birds. This is 
15 
