480 EVOLUTION, GENETICS, AND EUGENICS 



tricians and these show peculiarities in embryonic membranes similar 

 to those of armadillo quadruplets. Thus it will be seen that the human 

 twinning situation rests largely on the armadillo quadruplet situation. 



As to the causes of human one-egg twinning, we have no positive 

 knowledge. It seems probable, however, that the cause is somewhat 

 similar to that described for the armadillo. Some early hitch in the 

 normal correlation of embryonic and maternal physiological relations, 

 involving a delay in placentation, is believed to be responsible for 

 twinning in man. Technical studies of tubal pregnancies strongly 

 support this view; but lack of space forbids the introduction of this 

 testimony. 



In concluding the discussion of the causes of human twins, it must 

 be admitted that as yet we do not know enough to enable us to control 

 or to permit the production of twins. So we can give no practical 

 advice to those who desire to have twins or to avoid them. 



II. THE TWIN METHOD OF STUDYING HEREDITY IN MAN 



In recent years there has grown up a new method of studying 

 heredity in man by making use of twins, and especially of identical 

 twins. A number of European workers, notably Siemens and von 

 Verscheur, have written extensively on this method and have con- 

 tributed many new facts about the heredity of pathological conditions 

 and normal peculiarities in man. 



The validity of the method rests on the ability of experienced 

 students of twins to diagnose twins accurately as to whether they are 

 identical (monozygotic) or fraternal (dizygotic). Both Siemens and 

 von Verscheur are confident of their ability to do this, and I fully agree 

 with them, having had no real difficulties in making this diagnosis in 

 considerably more than a hundred cases. 



Assuming that the two types of twins can be accurately separated 

 by experts, how can twins be used in determining what characters are 

 hereditary? The working hypothesis is that those characters that are 

 the same in all cases of identical twins, but are usually different in 

 fraternal twins, are hereditary; and that those characters that are 

 different in identical twins, or that occur in one such twin and not in 

 the other, are non-hereditary. Now there is one weakness of this 

 hypothesis, namely, that some undoubtedly hereditary characters ac- 

 tually do express themselves differently in the two individuals of a 

 monozygotic pair. This is not surprising, for often in a single individ- 

 ual a hereditary character may express itself on one side of the body 



