538 liADIATloN ItloLOCiY 



no case did a tumor develop as the result of a single dose. Change in the 

 sehedule of doses, for example, by diseontinuing them after a given time, 

 markedly delays the appearance of tumors and alters the shape of the 

 eurves, as is seen in Figs. 14-8 and '.). 



Particular attention was paid to the reproducibility of the doses of 

 ultraviolet radiation, since uniformity was necessary if the experiments 

 were to have any (juantitative significance.^ The dose was customarily 

 measured in units of 10" ergs/cm- of radiation of wave lengths 0.313 m 

 and shorter delivered l\y the particular type of intermediate arc used. 

 The number of these units per dose is indicated in the following discussion 

 by the symbol D. Experiments designed to test reciprocity showed that 

 td is independent of intensity, at least dowMi to a level well below that at 

 which the data used in the following analysis were obtained. 



THEORETICAL" 



Any theoretical treatment of carcinogenesis meets a difficulty that is 

 seemingly insurmountable at the present time, namely, that a cancer is 

 detectable only after it has reached a relatively advanced stage. An iso- 

 lated cancer cell has the same general morphological characteristics as the 

 normal cells of the tissue from which it arises. A cancer is recognized by 

 the arrangement and behavior of these cells, and there must be present 

 something like 1000 to 10,000 cancer cells before an observer can be sure 

 that a cancer exists. The same is true of any chemical measurement, 

 since changes are determinable only when a large number of cancer cells 

 are already present. 



Confronted with a cancer at such a relatively advanced stage, the 

 observer may ask whether it has sprung into being suddenly or has grown 

 for a considerable time before it was first detected. This ciuestion cannot 

 be answered directly in the present stage of knowledge. Lacking any 

 present means for detecting the original cancer cell, the observer can only 

 extrapolate back from what he finds after the time of appearance of the 

 tumor. Such extrapolation is certainly justified, but any reasoning based 

 thereon must be considered as tentative and likely to be in error. When 

 data are available such as those which have been obtained with ultra- 

 violet-induced tumors-" highly reproducible and exact within the limits of 

 experimental error — the experimenter feels compelled to attempt to use it 

 to learn something of the events prior to the appearance of the tumor if 

 there is any chance that this can be done. 



' The original paper (Blum et al., 1941) should be consulted for such details as a 

 description of the method of exposure of the animals and measurement of the dose. 



'^ This analysis was recently presented in a paper which should be consulted for all 

 details (Blum, lySOb). An extension of this analysis is now in course of preparation 

 for publication. 



