12 B. Benjamin 



pass through those ages within a short period of term. The 

 population observed is therefore a combination of a large 

 number of generations born over a very long period, as long 

 indeed as the extent of the table. It does not follow that those 

 who passed through age 40 in 1950-52 will experience at 60 

 (in 1970-72) the mortality rates given at age 60 by English 

 Life-Table No. 11. 



For actuaries the Life-Table is an experience which they 

 know will never be reproduced but is nevertheless a model 

 which serves to guide estimation of the future expectations. 



It should therefore be borne in mind that the senescent 

 deaths are of people born long before those whose deaths fall 

 in the "anticipated" category, so that it is not strictly correct 

 to compute the ratio which one group bears to the other as in 

 the "attainment" proportions referred to above. Similarly 

 the peak age of senescent deaths refers to generations born 70 

 years or so ago and does not indicate the natural lifespan of 

 those who are now in their infancy. However, the indications 

 of secular trend are acceptable. 



Strictly we should calculate "generation life-tables", i.e. 

 tables each based on the observed mortality of a single genera- 

 tion of births (for practical purposes those born in a narrow 

 interval of years, say five) but the recorded mortality of this 

 country does not permit the calculation of complete tables for 

 generations separated by more than ten to 20 years. 



Accuracy of age at death 



In all the kinds of analysis of death rates or of the curve of 

 deaths which are undertaken by actuaries and have been 

 considered above, use is made of a life-table model rather than 

 the actual deaths of a particular year in order to base the 

 analysis on a population with a fixed birth entry, thereby 

 avoiding the irregularities in the run of deaths from one age 

 to another which affect the published deaths statistics of 

 any one period and arise from birth fluctuations or other 



