428 



Till': Joi'K.NAi. OF I ll•:KI•;l)lT^■ 



other types could be found rei^resentin^ 

 still other sta^^es in the evolutionary 

 develo]jment or possessing a different 

 assortment of factors. If a sufficient 

 number of these intermediate types 

 were available, the forms would in- 

 crease until the differences became so 

 small that, although the different stages 

 might still behave as Mendelian units, 

 the fact could not be definitely demon- 

 strated. 



THE L.\W OF SEGREGATION. 



Looked at from this standpoint, the 

 law of segregation may be said to hold 

 good. The difficulties and inconsis- 

 tencies in its application have resulted 

 from the assiunj^tion that visible dif- 

 ferences are "unit characters" repre- 

 sented by .single determiners. According 

 to the inter]jretation here advanced, 

 visible characters are to be looked ujjon 

 as complexes of an indefinite number of 

 heritable elements which arc so small 

 as to be immeasurable and which are 

 more or less tenaciously united. On the 

 degree of coherence, or tenacity with 

 which the elements are bound together, 

 de])ends the a])])arent regularity of the 

 Mendelian behavior. If the elements 

 remain inseparably associated, the in- 

 heritance follows the simjjle mono- 

 hybrid ratio. If the elements separate 

 into two or three groups, di- or tri- 

 hybrid ratios result. If the association 

 of elements is still more completely dis- 

 solved, the inheritance is blended and 

 api^arently non-Mendelian. On Ijreak- 

 ing u]) the characters, smaller groujjs of 

 slightly coherent elements may fonn, 

 and these groups themselves ma\- not 

 be entirely independent. It is thus 

 possible to account for cases in which 

 indications of both mono- and di- 

 hybrid ratios occur with res])ect to the 

 same visible character. With characters 

 that arc strictly alternative in their 

 nature, dissolution results in aberrant 

 or comjjlieated ratios. Where inter- 

 mediate stages of the characters can 

 come into cx])rcssion, dissolution results 

 in indistinct classes and ai)i)arently 

 imperfect segregation. 



The difference between the waxy and 

 homy endos])erm of maize affords a 

 good exam])le of a character whose 

 expression is dcfinitelx' alternative'. In- 



termediate forms are completeh' absent, 

 the nearest aiJi)roach to an intermediate 

 being the extremely rare occurrence of 

 seeds in which Ijoth homy and waxy 

 endosperm are present as a mosaic. 

 The inheritance of endosperm texture 

 in horny x waxy hybrids behaves as a 

 sim])le Mendelian character, and this 

 Ix'havior was thought to be perfectly 

 regular until the use of larger numbers of 

 individuals made it ]X)ssible to determine 

 the ratio with greater accuracy. It 

 was then found that instead of the mean 

 ])ercentage of wax\' seeds being the 

 expected 25, it was in reality only 

 about 23. 



If the coherence between the various 

 genetic variations that result in changing 

 waxy to horny endos]jerm was absolute 

 the Mendelian behavior of the characters 

 would be regular, but if one or more of 

 the elements was occasionalh- disasso- 

 ciated from its fellows there would 

 result a variation from the expected 25% 

 of recessives. 



The change from waxy to homy may 

 be likened to a chemical change that 

 accompanies the transition from an 

 acid to an alkaline reaction, or the 

 change from a solid to a liquid state that 

 follows a rise in temperature. The in- 

 crease in alkalinity or the rise in tem- 

 perature may be slow and continuous, 

 but the resulting change in appearance 

 is abrui)t. 



MANY CASES REPORTED. 



l'2xamples of coherence and the factor- 

 ing or subdivision of characters are 

 constantly lacing reported. Castle has 

 shown that some Mendelian characters 

 are amenable to change by selection. 

 The latest results in this line are those 

 of Castle and Phillips. In a scries of 

 carefully planned ex])eriments, it is 

 shown that the hooded character in 

 rats, which exhibits ])erfectly blended 

 inheritance between different stages or 

 degrees of the character, may still 

 behave as a unit, giving monohybrid 

 ratios when crossed with ft)nns not 

 showing the character. 



After changing the mean of the 

 hooded character in both directions l)y 

 selection, individuals were crossed with 

 the wild stock which is not hooded. 

 The wild fonn is dominant and the 



