294 



SCIENCE 



[N. S. Vol. XLIII. No. 1105 



cancer rate may be dominant, while in 

 some others the opposite tendency predomi- 

 nates, and in a few an intermediate result 

 is obtained. Our experiments established 

 for the first time the cancer rate for a num- 

 ber of different strains ; each strain was fol- 

 lowed through several successive genera- 

 tions and in each generation a large num- 

 ber of animals were observed. The re- 

 sultant figures for the various generations 

 of the same strain were usually in fairly 

 close agreement. Animals belonging to 

 such strains were used for hybridization 

 experiments. The results of hybridization 

 experiments which we obtained do not seem 

 to be compatible with the view recently ex- 

 pressed that the tendency to cancer is a re- 

 cessive character and that all results can 

 be explained on such a basis.^ 



Of a different character is a problem in 

 heredity first studied by E. E. Tyzzer. It 

 is well known that some strains of mice 

 are a favorable soil for a certain trans- 

 plantable tumor, while other strains are 

 not. In crossing a favorable and an un- 

 favorable strain Tyzzer found conditions 

 apparently incompatible with Mendelian 

 principles. We obtained likewise in sub- 

 sequent experiments with M. S. Fleisher, 

 results similar to those of Tyzzer, and we 

 suggested that the results might be ex- 

 plained by assuming the presence of multi- 

 ple factors. The same interpretation may 

 apply to the heredity of autochthonous 

 tumors to which we referred above and in 

 which also simple Mendelian proportions 

 do not seem to exist. 



These studies of the cancer incidence in 

 various strains of mice and the methods 

 used therein have, however, a much wider 

 significance. On the basis of a thorough 

 knowledge of the cancer incidence in cer- 

 tain families, and on this basis alone, will it 



2 Maud Slye, Interstate Medical Journal, XXII., 

 July, 1915, p. 692. 



be possible to analyze certain other factors in 

 the etiology of tumors, and the understand- 

 ing of these latter factors, as well as of 

 heredity, will perhaps ultimately provide 

 us with a rational basis for the prevention 

 of cancer. Without a thorough knowledge 

 of heredity, conclusive results as to the sig- 

 nificance of other factors could not be ex- 

 pected. Acting on this principle, we found 

 that castration in sexually mature mice at 

 the age of three to eight months reduces 

 the cancer rate in a very pronounced way. 

 Prevention of pregnancy, while it still has 

 some effect in reducing the cancer rate, as 

 we found several years ago, has very much 

 less significance than castration. 



These results and some additional ones 

 to be mentioned shortly permit us to classify 

 the causes of tumors into two main divi- 

 sions, internal and external ones. Hered- 

 ity belongs to the former class. The point 

 of attack of these hereditary factors we do 

 not yet know. In some cases, they may per- 

 haps stand in relation to some other in- 

 ternal factors, which are in all probability 

 of significance in certain cases. I refer to 

 the spontaneous parthenogenetic develop- 

 ment of the egg within the ovary and else- 

 where in mammals, a process which, ac- 

 cording to our findings in the guinea pig, 

 is not a rare occurrence, and may even 

 normally proceed to the formation of the 

 anlage of the central nervous system. To 

 this class of factors may also belong devel- 

 opmental errors which were already sus- 

 pected by Cohnheim and which as we know 

 may appear as inheritable mutations in 

 various groups of animals. 



The external factors may be further 

 divided into chemical and mechanical, and 

 both may be derived either from within 

 the body or from the outside world. As 

 an example of a chemical factor originating 

 within the body, we may cite the great im- 

 portance of the internal secretion of the 



