I20 THE BIOLOGY OF TWINS 



farm possessed four functional nipples and a practically- 

 pure line of sheep with four miUc-bearing nipples was 

 estabUshed.- Subsequently many five- and six-nippled 

 sheep appeared and occasionally seven- and eight-nippled 

 ones were produced. The result was due, not to the 

 selection of fluctuating variations, but to the appearance 

 of mutations and the selection of these suddenly appear- 

 ing new types as the parents of succeeding generations. 

 The attempt to establish a correlation between 

 supernumerary nipples and twin-bearing was dis- 

 appointing. In a subsequent paper Bell says: 



At first it appeared that four-nippled ewes were less fertile 

 than ordinary sheep, for they had a smaller proportion of twins; 

 but this turned out to be due to the fact that the process of 

 selection had necessarily resulted at first in a flock composed 

 mainly of young ewes, and young sheep rarely have twins. After 

 the four-nippled ewes had grown to full maturity they were found 

 to be as fertile in this respect as ordinary sheep, if not more so. 



The four-nippled stock proved a failure in so far as twinning 

 was concerned, so in 1909 the flock was cut down to six-nippled 

 ewes alone. There were indications in i9r2 that the six-nippled 

 stock win ultimately turn out to be twin-bearers, as a rule, when 

 they become fully mature. 



Whether or not this expectation was reaUzed I am 

 not in a position to say. On the whole, it does not 

 appear that the production of a twinning race of sheep 

 by selecting for supernumerary nipples is a success, for 

 at best only a very general correlation between the twin- 

 bearing tendency and the tendency to extra nipples 

 ejasts. 



A much more promising method of producing a 

 twin-bearing race of sheep would be to breed exclusively 

 from twin individuals irrespective of, or in correlation 



