THE AGE INCIDENCE OF CANCER 



of persons of those ages, the incidence rises so steeply with 

 advancing age that a logarithmic scale is required. This 

 gives a curve with a gentle upward convexity. If now the 

 age scale is also plotted logarithmically the curves become 

 a straight line (Fisher and Hollomon, 1951; Armitage and 

 Doll, 1954; Metcalf, 1955). There are a number of deviations 

 from this regularity but all the exceptions are 'reasonable' 



30 



40 



50 60 



Fig. 10. To show log-log relationship between cancer mortality and age. 

 Lung cancer mortality at ages of cohorts born around the year shown 

 (males, Australia). Ordinates, mortality per 10* at ages; abscissae, 

 age in years; both expressed logarithmically. 



ones. Cancer of the lung, for instance, shows approximate 

 straight lines if we work with cohorts instead of total popula- 

 tions — an indication of progressive change in the last fifty 

 years in some social factor heavily correlated with the in- 

 cidence of the disease. Like most people, I believe that the 

 factor has been the increase in cigarette smoking. Special 

 features associated with changing functional activity at the 

 menopause lead to a slowing down of the increase after the 

 age of 45 of cancer of breast, uterus and ovary. 



189 



