Group Discussion 291 



mean by lifespan. If one could establish, once and for all, the normal 

 lifespan for a given species, then the observed mortality curve would 

 tell us the deviation due to environmental conditions, such as being 

 kept in a zoo, or hunted. Then we might introduce another index to 

 take these conditions into account. We could take the 50 per cent 

 survival time from the actual mortality curve and the ratio of this 

 to the normal lifespan might serve as an index of, say, longevity or 

 senescence. These two quantities, the normal lifespan and the 50 

 per cent survival time, might be the proper parameters to use. 



Comfort: The limit, as I say, is not always advantageous, because 

 of the very great divergence in the reports of maximum age in 

 animals. My own feeling is that once you can get a family of curves 

 like this, or a rough indication of the family, you could perhaps take 

 a point 10 per cent back from the limit. 



Roiblat: If you take the mortality curve, then depending on the 

 condition in which the animals are kept we would obtain different 

 times for the 50 per cent survival but the same limit. 



Comfort : I am thinking of cases where you have not got curves as 

 good as that. If you do have such a family of curves, there is no 

 difficulty. The trouble is to know what to do with the sort of thing 

 that Sacher was describing, where he wants to compare a whole 

 range of mammals ; merely to get a rough correlation it is necessary 

 to give some sort of figure which one can compare. 



Perks: The difficulty with the modal value is that it is influenced 

 to a certain extent by the infantile mortality, and by mortality from 

 accidents and infections. If you reduced those mortalities, then, as a 

 mere piece of arithmetic, the mode is advanced and the curve of 

 death becomes much steeper, by the mere fact that more lives 

 survive to the ages at which the rates of mortality are high. I have 

 been thinking about this question of what would be a useful measure 

 of lifespan. Some sort of technique such as actuaries use, which we 

 call multiple decrement technique, might be used, at least in theory. 

 It is rather laborious, but if you could separate from the death curve 

 all those deaths which have nothing to do with lifespan — accidental, 

 predatory, infantile, and anticipatory deaths, — then there is a 

 technique for getting a residual life-table, a hypothetical life-table, 

 that is concerned only with those causes of death which affect 

 biological lifespan. This would have all the disturbances taken out of 

 it, and of course you would get a much later and taller death curve. 



Comfort: The trouble with that is to know which causes are age- 

 dependent. 



Rockstein: I think we are being unrealistic about this. I would 

 suggest that mean longevity seems to be a thing that you could 



10* 



