I. AVOID FRUITLESS EXPERIMENTS 25 



better pr-epared to expedite them. It seems inci-easiugly likely that 

 future research will requii'e more co-opei-ation between individual 

 scientists rather than less. 



Some experiments by their very nature impose demands on time 

 which exceed the human life span. Where investigations of this kind 

 must end with the hfe of the experimenter, the scientific returns may 

 not be commensurate with the investment in time and energy. It is 

 therefore partly meaningless to inaugurate such long range experi- 

 mental programs unless facilities can be assured in advance for their 

 continuance into the indefinite future. Examples of this type of in- 

 vestigation occur in the field of experimental genetics where life cycles 

 are long, as for elephants, turtles, and certain species of trees. In the 

 same classification are many of the observational problems in the 

 field of human genetics, which suffers under the great handicap of not 

 being an experimental science. Since it is conceivable that it may 

 become an experimental science in the future, perhaps many problems 

 in the field should await that time. 



G. SATISFY IMPORTANT TECHNICAL DEMANDS 



1. Design Experiments 



In the words of Fisher, who has wTitten an excellent book on the 

 design of experiments (7) : "It is possible, and indeed it is all too fre- 

 quent, for an experiment to be so conducted that no valid estimate of 

 error is availal^le. In such a case the experiment cannot be said, 

 strictly, to be capable of proving anything. Perhaps it should not, in 

 this case, be called an experiment at all, but be added merely to the 

 body of experience on which, for lack of anything better, we may have 

 to base our opinions." 



To make the best use of statistical methods, it is not sufficient to 

 attempt an analysis subsequent to the recording of experimental ob- 

 servations, but the experiment must be planned from the beginning 

 with the demands of the statistical analysis in mind. As a conse- 

 quence numerous economies often can be effected in the conduct of 

 experiments without an accompanying decrease in the reliability or 

 the significance of the results. An example of this type of application 

 is the determination of the most efficient number of organisms to 

 serve as controls and for the various parts of an experiment in order to 

 achieve comparable accuracy throughout. Moreover, it is possible 

 that a careful analysis of the schedule for an investigation wdth respect 



