ULTRAVIOLET RADIATION AND CANCER 



537 



that the points scatter about a smooth S-shaped curve, which is the inte- 

 gral of a normal distribution. For different doses or different schedules 

 of exposure, the curve changes its position along the time axis, but the 

 shape and slope are not changed. In this figure, data from different 

 experimental series involving different dosages and schedules are brought 

 together to a common point at 50 per cent incidence of tumors. The 



100 I- 



20 



2 7 



2 1 2.2 2 3 2.4 2.5 2 6 



TUMOR DEVELOPMENT TIME Ud). log days 



Fig. 14-2. Curves, for two experiments, of percentage incidence plotted against log td. 

 Curve II might represent a lower dose D or a longer interval between doses i than does 

 curve I. Note that the shape and slope are the same for both experiments although 

 the development time is longer for curve II than for I. 



100 r- 



100 200 300 



TUMOR DEVELOPMENT imEit^), days 



Fig. 14-3. The same percentage incidence curves represented in Fig. 14-2 but plotted 

 against days instead of log days. 



drawn curve represents a standard deviation of 0.081 log days, repre- 

 senting large variance in spite of all the precautions that were taken, 

 although less than has Vjeen found for other types of carcinogenesis. The 

 nice fit to a probability function makes it possible to assess the signifi- 

 cance of the data statistically ; and the uniform distribution of the points 

 around the curve indicates the reliability of the data and the constancy 

 of the shape and the slope of the curve. The success of the logarithmic 

 plotting is indicated by comparing Fig. 14-2, in which the log of td is 

 plotted, with Fig. 14-3, in which td is plotted directly. 



To induce tumors it was necessary to repeat the dose many times. In 



