480 Journal of Agricultural Research voi. iv, no. 6 



of the present paper is to submit an analysis of data with a view to testing 

 the foundations of such a belief when we are dealing with a class of pure- 

 bred sheep. 



To indicate birth as a single, in twins, or in triplets, we use the symbols 

 "i " for single, "2" for twin, and "3" for triplet. 



SOURCE OF DATA 



The source of all our data is the American Shropshire Sheep Record. 

 We have taken individuals with numbers from 325502 to 344869,^ and 

 have looked up their parents and grandparents with respect to the state 

 of birth in singles, twins, and triplets. 



All cases are omitted where either parent is imported, for the reason 

 that the English records do not show whether an animal is born single, 

 in twins, or in triplets. 



For each of the offspring above mentioned, with American-born 

 parents, we have made a card, showing whether this animal is born in our 

 symbolism as a i , 2 , or 3 and showing in which of the states i , 2 , or 3 

 its parents and nonimported grandparents are born. 



DISCRIMINATION IN FAVOR OF OR AGAINST RECORDING TWINS 



In beginning this investigation we made some inquiry concerning 

 possible discrimination in favor of or against the recording of twins, and 

 found no reason to believe that there existed such discrimination. In 

 the Journal of the Royal Agricultural Society of England,^ we find that 

 46.84 per cent of Shropshire ewes involved in the data there analyzed 

 have twins. This would mean that nearly 64 per cent of the lambs born 

 are twins. But the percentage of lambs born as twins and triplets that 

 we have found in fairly large classes of oflFspring does not seem to exceed 

 43, which is very different from 64 per cent. The difference seems to mean 

 that either Shropshires in America are less fertile than in England or 

 there is discrimination against twins in the matter of recording. This 

 does not mean that the discrimination is made directly against twins, 

 but probably in an indirect manner for some such reason as the better 

 development of singles when the selections are made. However, for the 

 main purpose of our problem, we are concerned with the elimination of 

 discriminations where one would record twins or singles because certain 

 ancestors are twins or singles — that is, we are concerned with the elimina- 

 tion of the kind of discrimination that would give an affirmative answer 

 to the question, Do breeders tend to record an increased or decreased 

 proportion of twins on account of the fact that certain parts of the 

 ancestry consist of twins or of singles? Such discrimination is doubtless 

 much less likely than a more general sort of discrimination that would 

 lead to the recording of a larger or smaller proportion of twins than we 

 find in a random sample. 



^American Shropshire Sheep Record, v. 25, p. 1-1314. 1912. ^ Heape, Walter. Op. cit., p. 23s. 



