over many years (where adult mortality rates are reasonably constant), 
they may often be satisfactorily regarded as.a sample of the living 
and thus analyzed on a time-specific basis. Where. fluctuating mortality 
rates are encountered, especially in short-term investigations-and. - s 
small samples, such data must be used as mortality information in. a>:-- 
dynamic life-table computation. .The mortality rates thus derived. from 
banding cohorts. spread over a period of years should be reasonable - - 
approximations of average conditions affecting “theoretical” popula- -: 
tions (in the sense of Chapter II). These are the type examined in 
Part II that follows. When "real" populations are similarly studied, 
as when a single cohort is used to provide mortality rates for specific 
years, wide annual fluctuations in a single mortality factor may: take . 
place. This particular: condition will be encountered in Chapter: XIV | 
where the limitations of life-table calculations will be reconsidered. . 

Why, 
