tabulate those alive at the start of each year from then up to the 
present. The amount of mortality each year would obviously be primary 
in nature, so that the annual mortality rates would be accurate and 
reliable, but only for this particular cohort. This type of table 
is seldom used in the vital statistics of human populations, because 
many of the data are essentially historical and because public health 
officials are concerned more with mortality statistics of the present. 
Dynamic life tables have, however, been utilized in survival studies 
of birds like the song sparrow (Nice 1937) and pheasant (Leopold, 
Sperry, Feeney, and Catenheusen 1943; Buss 1946), which. were banded 
and, with varying success, retrapped alive in subsequent years. Deevey 
(1947) has termed such tabulations "horizontal life tables." 
In time-specific life tables (termed “vertical tables" by 
Deevey) all the primary data are compiled from a single period of 
observation. An entire population may be available for study, as 
in a ten-year census of the United States (Glover 1921; Hill 1936), 
or when all the fish of a single lake are poisoned at the same time 
(Eschmeyer 1939;G. W. Bennett 195). The age composition of bird 
populations has somewhat similarly been observed at specific times 
by study of banded samples. Michener and Michener (1933) have reported 
on the age distribution of house finches retrapped during a single 
month; Austin (1942) has shown the ages of common terns when they were 
retrapped on their nests; and Kortlandt (192) has observed banded 
European cormorants by means of a telescope at their nesting colonies. 
The registered mortality data that now supplement census figures for 
human populations have no counterpart in the time-specific studies 
of wild animals. The mortality often cannot be readily inferred, 
Since the age distribution of the living at a specific time is a 
function not only of the number of young produced in the past but 
also of mortality in previous age intervals. Mortality rates are 
therefore very difficult to compute in time-specific analyses of 
natural populations. 0. L. Austin (1938, 1942, 1945, 1947a) has 
utilized this approach in his studies of the population dynamics of 
Cape Cod Sterninae. A comparison of his conclusions with those 
arising from a dynamic life table analysis is particularly instruc- 
tive, but its presentation is postponed to a later section on the 
Laridae. 

Composite life tables have occasionally been constructed 
from miscellaneous collections of mortality data that do not permit 
either dynamic or time-specific definition. Life expectancies (é,) 
have, for instance, been roughly computed for ancient Egyptians by 
Karl Pearson (19025 working with the age at, death recorded on 141 
mummy cases. In such a sample the individuals have been born in 
different years and have died in different years. Strictly speaking, 
the sample is neither a dynamic one nor a time-specific one. Although 
Dublin and Lotka (1935) state that "correct age specific rates cannot, 
of course, be computed on such a basis," I believe that this criticism 
really holds for a time-specific treatment of such data. The 11 
persons can be studied as though they all were born at the same time. 
This assumption at once makes the population a hypothetical one, but 
