DISCUSSION 



convenient way of investigating the point you have in mind and might give results 

 which are easier to interpret. 



Dr. Watson : At the beginning of this lecture you referred to the question raised by 

 Sir Macfarlane Burnet concerning the linearity of dose response relations. Dr. Gray 

 gave three examples, of which the first was a virus. I think it is true to say that not all 

 viruses are inactivated in the way Dr. Gray indicated. The second was induction 

 of the lysogenic state in bacteria, and the third was the induction of chromosome 

 structural damage. Now, there is a common feature in all these — they are organized 

 structures — and it would be unfair to argue from them to the general case of any 

 cells. Is that correct? 



Dr. Gray: I have not discussed which forms of radiobiological damage are or are 

 not likely to show a linear dose response relation. My purpose was to indicate condi- 

 tions under which an extrapolation can usefully be made if it had already been 

 demonstrated experimentally that the dose response relation was linear down to the 

 lowest doses investigated. Have I answered your question or not ? 



Dr. Watson: That is perfectly clear. I was just going to point out that in fact these 

 three are special types of structure as opposed to an ordinary cell, and that there is, 

 in fact, no evidence of any kind which would suggest linear relations within an ordin- 

 ary cell — a complete cell. 



Dr. Gray: The chromosome structural damage and the induction of lysogeny to 

 which I referred were both initiated in cells, namely, microspores and bacteria 

 respectively. These examples of what I beheve to be established linear dose relation- 

 ships came to mind. I do not generalize to other types of cell, though it is my belief 

 that these examples are not in fact unique. 



Dr. Loutit: I should like to ask Dr. Gray in this connection whether Homer has 

 nodded or not. On the board he warns us, very rightly, against too facile interpre- 

 tation of linear log graphs, and then as one of his examples of biological response to a 

 single physical event he showed Marcovich's data of which the ordinate was in logs. 

 He was unable to show the abscissa, but that is also in logs. 



Dr. Gray: I am sorry about that. 



Dr. Loutit: In fact, my statistician has replotted this on a linear scale and the points 

 are highly significantly not linear. 



Dr. Gray: Is the relation not linear at the 1 r level? 



Dr. Loutit: It is not linear at the low dose level. There is a definite curvature like 

 a two-headed curve just at the point where we are interested in it. 



Dr. Gray: Well, this one is evidently not a very good example. I am sorry. I will 

 withdraw this example pending further examination of Marcovich's data. 



Note added later: Marcovich presented his data as a log/log plot. This is a satis- 

 factory way of testing for linearity since if strict proportionality exists between two 

 variables, the log/log plot will be a straight line. It is not quite as satisfactory as a 

 simple plot of one variable against the other because lack of strict proportionality 

 between the variables may leave the log/log plot linear, and the lack of proportion- 

 ality must be detected by the fact that the straight line is not at 45° to the axes. 

 The log/log plot is not open to the objection which I raised against a linear log plot, 

 which inevitably presents the appearance of a threshold when the variables are 

 linearly related. 



I have re-examined Marcovich's data, and cannot find, from a statistical examin- 

 ation of the average of the results obtained in all experiments, any indication of 

 departure from proportionality between the dose and the number of bacteria in 

 which the lysogenic transformation was induced. 



299 



