182 



THE AMERICAN NATURALIST [Vol. XLII 



them upon the most flax-sick areas. But whether these 

 resistant strains have been selected through crossing or 

 by this gradation method from bulk seed or from indi- 

 viduals taken from a known pure strain, I have found 

 that abrupt heightening of disease conditions of too great 

 violence may undo the whole work. 



It holds as well for wheat when speaking of rust at- 

 tacks. Under uniform conditions of rust infection, all 

 wheats arise rapidly to a stage of marked resistance to 

 general uredospore infection whether caused by the type 

 Puccinia graminus or P. rugigo-vera, which resistance 

 seems to be characteristic for each variety concerned, 

 but may then fall a ready prey to sudden attacks in- 

 troduced by properly conducted aecidial infection. All 

 this points to the role played by the irritations of environ- 

 ment, which either govern the appearance of mutations 

 or produce other changes which are very worthy of the 

 breeders' and croppers' attention; and allows one to as- 

 cribe much more merit to methods of mass selection and 

 breeding in cereals than the DeVriesian doctrine of con- 

 stancy of elementary species will allow one to assume. 



No phase in this argument touches upon the ultimate 

 causes of disease resistance or immunity. But the facts 

 do point quite clearly to the probable influence of chemi- 

 cal agencies, perhaps toxines, arising from the direct 

 existence of fungus attacks upon the hosts. In my mind, 

 there is not the slightest doubt but such attacks originate 

 heritable resistance, in much the same sense as Mac- 

 Dougal's chemical injections upon ovaries are supposed 

 to have originated new types. If later experiments prove 

 MacDougal's observations to be well founded, the results 

 will be of far reaching importance. If these suppositions 

 that fungus attacks upon the host may induce fungus- 

 resistant qualities in the progeny from the matured ova- 

 ries, are correct, no doubt the unit characters, so called 

 (whether simple and definite in number or whether they 

 may be considered as composed of countless and variable 

 elements of the cromatin structures) may be effected, or 



