Sex-limited inlieritance in Lychnis dioica L. 287 



of the "hereditary niechauism"; others go to the opposite extreme and 

 maiutain that they are simply a shorthand description of the results of 

 genetic experiments. It appears to me that an intermediate position 

 is the proper one, and I believe that this is the attitude taken, con- 

 sciously or unconsciously, by most of those who use genetic formulae. 

 A formulation must represent with approximate correctness the experi- 

 mental results, — otherwise it can have no value; but a formula is 

 both something less and something more than such a description. It 

 constitutes not only an abbreviated statement of results already secured, 

 but at the same time it involves a prophecy as to future results. Both 

 of these functions of genetic formulae are imperfectly performed; instead 

 of saying that they describe the results of genetic experiments, it would 

 be more correct to say that they approximate those results so closely 

 that they may be advantageously used as standards of comparison, and 

 that, although they may not correctly represent in any case the actual 

 genotypic apparatus, they do symholize an internal mechanism which 

 could give rise to approximately the observed genetic results. 



Predictions as to the outcome of genetic experiments, based upon 

 this hj'pothetical internal mechanism, fail with a frequency sufficient to 

 add zest to the work of every geneticist. To those who are following 

 genetic investigations on a fairly extensive scale, these constant sur- 

 prises are a sufficient safeguard against giving the genetic formula too 

 important a significance as a picture of the actual constitution of the 

 genotype. The limitations of genetic formulae in this respect become 

 even more convincing when it is seen that each genetic situation, in- 

 cluding both empirical results and prophetic inferences, can be equally 

 well expressed by several different sets of formulae. On several occa- 

 sions I have pointed out such alternative formulations for given sets 

 of genetic facts (Shull 1909, 1910a, b, 1911), and the literature abounds 

 in examples of the application of different formulae by different writers, 

 to the same or essentially the same situation. 



As the literature increases in volume it becomes more important 

 that genetic terminology be simplified as far as this may be done con- 

 sistently with the necessities of a correct statement of the ideas intended 

 to be conveyed. Several suggestions have already been made toward 

 simplification (Lang 1910, Morgan 1913 a, b. Castle 1913, Emerson 

 1913), but the more radical of these schemes, even if generally adopted, 

 would lead to confusion rather than to simplification. No scheme which 

 would overturn or change beyond recognition, the formulae which are 

 now in use, and which would consequently disconnect future genetic 



