ML'LATIVK HKVERSIOXS IX fOTTDX. 13 



often be combined into a hybrid sentence without disturl)ing the 

 general grammaticid structure, much as "unit characters'' appear 

 to 1)6 substituted for each other in strictly Mendelian hybrids. In 

 other cases the words of two languages do not prove to be direct 

 etjuivalents, but recjuire diirerent grammatical relations. Sentences 

 can no longer be translated piecemeal, by individual words, but have 

 to be recast by whole phrases or clauses. Words derived from the 

 same language tend to keep together in the hybrid sentence, hi the 

 same way that characters of divei-se parental types hold together in 

 expression. Hybrids that gave intermediate or combined expression 

 of Egyptian and Upland charactere hf the early generations have 

 shown a distinct tendency toward more exclusive expression of 

 Upland charactei-s in later generations, even when selected for the 

 expression of Egyptian or intermediate characters. 



If the existence of determinant particles or character units is to be 

 assumed, it is more reasonable to suppose that the expression of the 

 characters is governed by positional relations among the particles 

 than by mere presence or absence of particles. The theory of posi- 

 tional relations of determinants was suggested by Mr. "Walter T. 

 Swingle, of this Department." It has the advantage of accommodat- 

 ing a wider range of facts than the Mendelian theory. The establish- 

 ment of definite positions among the particles w^ould account for con- 

 ditions of uniformity and for regular ISIendelian ratios of expression, 

 while mutative reversions and diversified hybrids can be ascribed to 

 disturbances of the positional relations. Thus the positional theory 

 admits the transmission of latent characters as a general condition of 

 inheritance, whereas Mendelian writers have treated latency as an 

 exceptional phenomenon recjuirmg to be explained by additional 

 theories. 



Mendelian combhiations of charactei-s do not promise to attain 

 great importance in cotton because of the general tendencies to rever- 

 sions and correlations of characters that interfere with stable combi- 

 nations between characters derived from different types. Reversions 

 transgress the M(nidelian program. They interfere with Men- 

 delian dominance in the first generation and with Mendelian combi- 

 nations and segregations of characters in the hxter generations. A 

 Mendelian combination of the naked seeds of one variety with the 

 abundant lint of another might be desirable, but naked-seeded hybrids 

 are prone to revert to sparse lint, so that the yield is not likely to be 

 maintained. Fuzzy-seeded types are preferred because of the greater 

 abundance of lint. 



" In a paper read at a meeting of the Society for Plant Morphology and Physiology, 

 entitled "Some Theories of Heredity and of the Origin of Species Considered in Rela- 

 tion to the Phenomenon of Hybridization." Abstract published in the Botanical 

 Gazette, vol. 2r,, no. Ill, 1898. 

 [Cir. .wj 



