ESTIMATION OF AVERAGE DOMINANCE OF GENES 515 



It should not be necessary to comment on the role of statistics in the de- 

 vising and evaluation of schemes for investigating the inheritance of quanti- 

 tative characters. If the importance of statistics in this area of research has 

 not been adequately demonstrated by the foregoing, general statements 

 could hardly be expected to be convincing. The point to be emphasized is 

 that more theoretical investigation of experimental technics in quantitative 

 inheritance is badly needed. For example, insofar as the three experiments 

 considered here are concerned, more information is needed on the biases re- 

 sulting from various sorts of epistasis. It is possible that such biases are 

 greatest in Experiment III and would detract from its apparent superiority. 

 It is also possible that the biases from epistasis differ between the experi- 

 ments and that the differences vary with type of epistasis. In that event, 

 comparison of results from two or more of the experiments could conceivably 

 contribute to our knowledge of epistasis. 



Investigation of the power of a variety of technics used in quantitative 

 genetic research also would be fruitful. The intent is not to imply that there 

 are no such procedures for which the power is known within satisfactory 

 limits, but only to point out that there are some for which this is not the 

 case. For example, mention has been made of the use of parent, Fi, F2, and 

 backcross means for investigating ep)istasis, but to the authors' knowledge 

 there is nothing in the literature concerning amount of data required to in- 

 sure that the chances of erroneous conclusions from such a study would be 

 small. 



Equally important is continued search for useful technics and procedures. 

 It is entirely possible that approaches may thereby be discovered which are 

 more efficient than any presently known. As a case in point, at the time the 

 work described by Robinson et at. (1949) was planned we had not thought of 

 the procedure designated here as Experiment III which, so far as we know, 

 has not been previously described as a technic for investigation of dominance. 

 Judging from findings of the preceding section, the same amount of work 

 using the latter procedure would have provided considerably more precise 

 estimates. 



While attention herein has been devoted to estimation of average level of 

 dominance, the experiments described provide other information as well. The 

 data collected can be used also for estimation of additive genetic variance, 

 variance due to dominance deviations, and the genetic and phenotypic co- 

 variances and correlations of pairs of characters. 



LITERATURE 



No attempt has been made to cite all of the various publications that in 

 one way or another were stimulatory to the above discussion, since a careful 

 attempt to assign credit where due would have made the manuscript consid- 

 erably longer. Most interested readers will be familiar with relevant litera- 



