252 



SCIENCE 



[X. S. Vol. XXVIII. Xo. 712 



gives in the second filial generation (not in 

 the first) four varieties, viz., agouti, chocolate, 

 black and cinnamon. We could only have 

 shaken our heads and looked wise (or skep- 

 tical). 



Then we had no explanation to offer for 

 such occurrences other than the " instability 

 of color characters under domestication," the 

 " effects of inbreeding," " maternal impres- 

 sions." Seriovis consideration would have been 

 given to the proximity of cages containing 

 both black and cinnamon-agouti mice. 



Now we have a simple, rational explanation, 

 which any one can put to the test. We are 

 able to predict the production of new varieties, 

 and to produce them. 



We must not, of course, in our exuberance, 

 conclude that the powerS of the hybridizer 

 know no limits. The result under considera- 

 tion consists, after all, only in the making of 

 new combinations of unit characters, but it 

 is much to know that these units exist and 

 that all conceivable combinations of them are 

 ordinarily capable of production. This valu- 

 able knowledge we owe to the discoverer and 

 to the rediscoverers of Mendel's law. 



W. E. Castle 



Zoological Laboeatort, 

 Habvabd Uktversity, 

 July 15, 1998 



THE ORIGIN OF VARIETIES IN DOMESTICATED 

 SPECIES 



The great diversity of varietal forms, or 

 races, amongst domesticated animals and 

 plants, as compared with the corresponding 

 ^.i[d species has always been a subject of 

 remark, and it has generally been assumed 

 that domestication, involving as it does rad- 

 ical changes in environment, induces varia- 

 tion. The considerations mentioned below 

 indicate that we may have given too much 

 prominence to the effect of domestication in 

 inducing variation. It wiU be seen that, at 

 least in many cases, domestication merely 

 gives opportunity for the segregation and de- 

 velopment of variations which may have ex- 

 isted practically unnoticed in the wild species. 



The writer has previously shown that the 

 poll character in cattle is a dominant Men- 



delian character, the dominance being some- 

 what variable, but the heterozygotes always 

 being distinguishable from the extracted 

 reeessives. Let us suppose that a number of 

 polled and horned cattle be allowed to inter- 

 breed freely. Let the number of the oiiginal 

 polled cattle be a, and the horned cattle i, 

 both types being equally divided between the 

 sexes. The total number of cattle is a-\-i. 

 The chance that in any mating the male shall 

 be polled is a/a-\-i, the chance that it is 

 homed is i/a -j- i. The chance that the fe- 

 male is of a particular type is the same. The 

 following table (table I.) shows the proba- 

 bility of each of the various types of matings, 

 and the corresponding probability of progeny 

 of each type. Since the denominators of all 

 the probability fractions in this table are the 

 same, and since we are concerned only with 

 the ratios between the types, only the numera- 

 tors of the fractions are used. 



TABLE 1. 



Ratios of Types of Progeny descended from 

 polled (a) and horned (i) cattle, under con- 

 ditions of random mating. [A = polled char- 

 acter, a its aisence.) 



Generation F^. 



Type numbers I II III 



Type formulae AA Aa aa 



Ratios of types a — t 



Generation F-^. 



Proba- 

 Matings bility of 

 M. F. mating 

 IX I aXa a- 



IXlII aXi ai 



IlIX I iXa ai 



IIIXIII iXi 



a' 2ai 



Generation F,. 

 IX I a-Xa' a* 



IX II a'X^ai a'i aH 



IXlII a-Xi' aH' 



II X I 2aiXa' a'i a'i 



II X II 2aiXZai aH' 2a=i" 



II XIII 2aiXi' a »= 



IIIX I i'Xa' a'f 



IIIX II rX2ai ai' 



IIIXIII f=Xt= 



Probability of progeny of each type 



a'(a+i)'^ 2ai(a-\-i)' i"( 

 a' 2a i 



a+i)' 



