LEUKAEMOGENESIS: QUANTITATIVE ASPECTS AND CO-FACTORS 11 



handicap in epidemiological studies. Second, there is good evidence that the 

 leukacmogenic effect of radiation does not depend simply on dose but that the 

 circumstances of tlie irradiation as well as the reactivity of the irradiated 

 individual are also quantitively important. Third, "small" doses of radia- 

 tion of the magnitude encountered in diagnostic radiology have a real, though 

 low, probabihty of causing leukaemia, at least for whole-body irradiation in 

 utero. Nevertheless this does not necessarily disprove the contention that 

 serious tissue damage is a necessary factor in radiation-induced leukaemia. 

 The doses involved may damage some tissues quite markedly (Table I) and 

 no investigation seems to have been made of the haematological damage in 

 foetuses exposed to diagnostic irradiation. Lastly it still remains true that 

 the exposure dose-rate was high in all the situations in which radiation is 

 known to have caused leukaemia in man. 



KINDS OF MECHANISMS FOE LEUKAEMOGENESIS BY EADIATION 



One basic kind of hypothesis is that the late somatic effects of radiation 

 are due to some direct cellular effect of radiation, such as somatic mutation. 

 Another basic kind of hypothesis is that cancers, including leukaemias, follow 

 gross radiation-induced disturbances in physiology or tissue structure. It seems 

 to me a serious mistake to put these hypotheses in opposition to each other. 

 They are not logical opposites and both could easily be true: tissue damage 

 could well facilitate the expression of direct cellular effects by changing the 

 cellular environment or encouraging cell division. 



It is easy to waste a great deal of time in wholly unnecessary argument 

 about tlie mechanism of cancer or leukaemia. Surely we should recognize that 

 multiple correlation is a characteristically biological phenomenon (Huxley, 

 1958), as we would aU accept if we discussed the mammary tumour agent of 

 the mouse and the dependence of tumour manifestation on genetic constitu- 

 tion and hormonal influence. The nature of scientific proof entails that 

 experimental conditions can be arranged in which the tumour yield depends 

 critically on one factor only, but since such an experiment can be done for 

 each of the three factors separately it is not possible to single out one of them 

 as the aetiological factor except in some private, arbitrary sense."f" I think we 

 should avoid arguing about whether one hypothesis or the other is true but 

 instead try to define their relative importance and fields of action. 



It is probably a fair summary to say that the somatic mutation hypothesis 

 of carcinogenesis appeals especially to those who have no first-hand acquain- 

 tance with cancer or leukaemia, a category which includes its originator 

 Boveri. Its advantages are that many kinds of mutation can indeed be 



t The word " co-factors " in my title is similarly to be deplored. 



