CARCINOGENESIS FROM INDEPENDENT EVENTS 167 



gressive pathologic change involving endarteritis of fibrosis, or whatever it may be, or 

 even the changes of ageing which simulate such conditions, in which case one cannot 

 eliminate latent period if this is indeed a secondary event in the situations being considered. 

 This applies to the whole question of linearity and threshold also. If you consider that a 

 heterogeneous population, such as the human population, is made up of individuals with 

 a wide range of dose thresholds for anj^ particular late radiation effect, all the way from 

 almost no threshold to a very high threshold, then it's easy to see why one might show 

 that a small dose of radiation would produce an increase in incidence in some individuals 

 of such a population. This would mean an increase in absolute incidence as far as the 

 population is concerned, by a small dose of radiation. It's also not surprising when we 

 think that experiments usually use relatively homogeneous groups of young adult 

 animals, that we get curvilinear relationships because the range of dose thresholds here 

 is very narrow and similar, although not identical. In other words, there's a difference 

 in amount of non-radiation induced causes for each of these non-specific events which 

 determines the threshold of an individual. When you said the other day that the thres- 

 hold dose problem was theoretical then I certainly agree. 



DANiELLi: If you do come up with an answer that requires two events there must be 

 some biological explanation and, however dubious the data that you're working on may 

 be, nevertheless an explanation is required. 



MOLE: I would agree with you that it badly needs an explanation in biological terms. It is 

 very easy to think of chromosomal changes in view of what we learned yesterday in a 

 review of chromosomal changes in neoplastic cells. Perhaps it's too easy to make that 

 jump, and I'd be very reluctant to do it just now. The precision of the data, the number of 

 bone tumours that you find depends entirely on how intensively you look. All I've done 

 is to assume the same intensity of observation throughout one experiment. If, however, 

 3'ou do compare the two experiments with repeated doses and with single doses, which 

 were done manj' years apart, you come up with the same probability of biological event 

 per radioactive particle, within less than a factor of two. The data do seem to form a 

 consistent whole, in spite of the fact that experiments were done over a long time, and that 

 there were differences in the survival times, in animal hygiene and tliis kind of thing. 

 But the actual errors I wouldn't have any idea about at all. 



rotblat: I would hke to ask Dr. Mole about this application of the square ratio between 

 incidence and dose, to the two events wliich we discussed yesterday. If you have a square 

 ratio between incidence and dose, doesn't this imply that the two events must come very 

 close after each other, ^^■hen you have given the dose? 



mole: Not necessarily, no; because what I have done you see is to subtract 150 days from 

 the time of the occurrence of death. This is what I have called 6 or development time 

 which includes not only the time from the beginning of the tumour to the point at which 

 it kills the animal, but also any other time required for an interval between the two events. 

 I am trying to get still further and see in fact if I can get any sort of idea of what the 

 intervals between the two events should be. But I don't think with relatively long-lived 

 radioactive materials, like *^Sr and '"Sr one can get anywhere. 



muller: Did I understand you to say tliat you had also tried a fit with a one hit and a 

 three hit event, but that they didn't go as weU? 



mole: I didn't actually say that, but they certainly don't fit a one hit idea, because 

 Brues tried that and he had to put in a "latent period" (which has an uncertam meaning) 

 which varied with the dose. I think that Finkel herself might like to feel that the three 

 hit event might fit some of the data better. 

 muller: Didn't you put in a "latent period"? 

 mole: I put in this fixed development time, which. . . . 



