432 THE SCIENTIFIC MONTHLY 



degree of manipulative skill and to these should be assigned the type 

 of man who is naturally handy with his hands. Conversely there are 

 problems of a mathematical nature that would be exceedingly irksome 

 to the preceding type, whereas the theoretical type could handle them 

 efficiently. It is the problem, then, of the research director to so arrange 

 the work that each man of his staff will be working to the greatest 

 advantage. 



Undoubtedly the most important quality that any experimenter can 

 have is persistence. There are many problems that require months and 

 perhaps years of hard, discouraging work before the first ray of success 

 can be discerned, and in such cases it requires the utmost faith in the 

 ultimate result to enable one to keep up his interest. It is only by the 

 careful, systematic elimination of each obstacle as it comes up and the 

 direction of the work continually into new and more promising chan- 

 nels that will make successful what to the less determined experimenter 

 would be a failure. 



Perhaps it will not be out of place to say a few words about the 

 presentation of results. In the first place, the data should be given, 

 wherever possible, in graphical form. There are some instances where 

 the accuracy of the results is greater than can be represented by a plot, 

 in which case the data must be also given in tabular form, but in most 

 cases a curve is sufficient and the tables may be omitted. When plotting 

 curves, no points should be used except those directly computed from 

 the experimental results, as the practice of some very reputable labora- 

 tories of taking points from a faired curve as the basis of plotting is 

 very misleading as to the regularity of the results. On the other hand, 

 no experimental curve should be shown without including the actual 

 points, otherwise the results can not help but be regarded with sus- 

 picion. It is always better to present a few well checked results than 

 a multitude of irregular ones. 



One of the secrets of experimentation is to know when to stop, for 

 it is a natural tendency to carry the work further than the value of the 

 additional results will warrant, and it is inefficient to allow a nearly 

 finished piece of research to drag along. There is of course, no definite 

 point where a piece of work can be considered finished, and often times 

 one feels that he is in a position to efficiently commence the work only 

 when the allotted time or money is exhausted. For this reason one 

 of the most important functions of an investigation should be the pav- 

 ing of the way for more extended work. Therefore every report should 

 contain an account of the difficulties encountered, and most important 

 the recommendations of the experimenter for the conduction of further 

 research, for every difficulty, and every failure, should be made of 

 value by preventing others from encountering the same obstacles. 



It is urged that every investigation that produces results of interest 



