334 H. A. Blair 



(e) could possibly apply only to the adult stage. It should be noted that this 

 curve is not convertible into a LDgo-life expectancy relation because, owing 

 to mortality among the animals, the sample at each succeeding age is different 

 from those going before. Those dying early have the shortest life expectancies 

 and presumably the lowest LDgo's, although this latter point cannot be proven 

 directly. 



In the legend to Fig. 2 are also given the days of hfe expectancy for 

 the adult data only. These are fairly linear but do not extrapolate to LD50 = 

 0, when Sq — S — 0, but to about 300 r. Presumably later points will diverge 



800 



£600 



3 400 



200 



100 300 500 700 



AGE IN DAYS 



Fig. 2. Median lethal dose in roentgens as a function of age in rats of the 

 Rochester strain (6). The animals at different ages are not directly comparable 

 because, for example, of a group selected at 100 days, only about two-thirds 

 survive to 500 days. The actual median survival times of control animals for the 

 groups irradiated at 5, 11 and 16 months, respectively, are 450, 375 and 330 days 

 after the time of irradiation. Therefore, life expectancy does not decrease as 

 rapidly as the age of selection increases. 



toward zero. Because it requires maintenance of animals for about three years 

 to obtain data for a point at the advanced ages, it may be some time before 

 the curves of Fig. 2 are well determined even in short-lived animals. However, 

 Grahn and associates (7) at Argonne National Laboratory have shown in mice 

 that the lethal dose as measured by repeated daily doses decreases rapidly from 

 middle age with an apparent tendency toward zero at old age. 



At the present time it is not possible to state the situation more clearly 

 than that the lethal threshold in the adult is some diminishing function of life 

 expectancy, not a linear function throughout as required by postulate (e). 

 This can be expressed also as 



LD50 = F(So - S) (7) 



and this as yet undetermined function should replace 5*0 — 5" in equation (6). 

 However, there is considerable indirect evidence which will be discussed later 

 that 



LD50 = k(So - S) (8) 



in fairly close approximation,/: being a constant for values of ^q — Sup to 20 per 

 cent of 5*0. It is important to establish the form of equation (7) in several 

 species so that estimates can be made of variation of LD50 with age in man. 

 It is not clear whether equation (6) fits chronic data because equation (7) is 

 sufficiently linear in the region in which most of the data lie (shortening of life 

 span by one-half or less) or because of some other compensatory factor. In 



