ULTRAVIOLET RADIATION AND CANCER 555 



data for reasons that have already been discussed. Even if an action 

 spectrum could be obtained experimentally for the mouse, this would not 

 apply quantitatively to human skin because of the difference in the trans- 

 mission of the ultraviolet radiation. Direct measurements of the inci- 

 dence of carcinogenic ultraviolet wave lengths have not been made at a 

 sufficient number of points on the earth to give a complete picture of this 

 distribution in detail, and it would be extremely laborious to do so. Such 

 measurements would, moreover, be of uncertain value in the solution of 

 the present problem, because their interpretation would depend on the 

 missing information regarding wave-length dependence of carcinogenesis 

 in human skin. Some general ideas can be obtained, however, from data 

 that are available. 



For the purpose, data calculated by O'Brien (1943) for antirachitic 

 action (bg,sed on the absorption spectrum of provitamin D) , which has the 

 same long-wave-length limit as carcinogenesis, are used here. There is no 

 reason to believe that the action spectrum for carcinogenesis in man is 

 related to that for antirachitic action, and the action spectrum for the 

 erythema of sunburn might be thought more closely representative. 

 But there are inherent objections to using an accepted erythemal spec- 

 trum for this purpose, as was pointed out in Chap. 13; in any case 

 O'Brien's calculations indicate that about the same relative values would 

 be obtained if the erythemal spectrum was used in place of the antirachitic 

 spectrum. Such estimates must obviously be very rough, at best being 

 influenced by numerous factors which cannot be readily taken into 

 account, for example, cloudiness and dust. However, they provide as 

 satisfactory an index as is now available. 



The variation of cancer incidence and carcinogenic radiation with lati- 

 tude is indicated in Fig. 14-11. In order to have a relative basis of com- 

 parison, the value for the lowest latitude, 32°, has been taken as 100 per 

 cent for each variable. When the data are plotted in this way, the inci- 

 dence of cutaneous cancer shows about the same magnitude of change 

 with latitude for both sexes (curves III and IV). Two curves for annual 

 incidence for carcinogenic radiation are plotted, one for 2.0 mm of ozone 

 in the atmosphere (curve I) and one for 2.8 mm of ozone (curve II). 

 Neither of these curves shows as great variation with latitude as do the 

 cancer incidence curves. Ozone, which is the principal limiting factor 

 with regard to the shorter wave lengths of sunlight, varies with latitude, 

 and the two curves represent approximate extremes for the latitudes 

 covered by the cancer incidence data, 2.8 mm representing the northern- 

 most and 2.0 mm the southernmost condition. Thus the true range of 

 variation with latitude should be greater than is indicated by either curve 

 alone. In curve V, the annual ultraviolet radiation corresponding to 

 2.0 mm of ozone is used for latitude 32° and that for 2.8 mm of ozone is 

 used for latitude 41°. The resulting curve which should represent more 



