No. 552] SHORTER ARTICLES AND DISCUSSION 



745 



chances are about 995:5 or 199:1 against the validity of the 

 first hypothesis and only 69:31, or about 2:1, against the 

 second. 



Hence, P = .0185 or the chances are about 980:20 against such 

 discrepancies being chance deviations from the theory. Thus 

 either the theory must be discarded or reasons for the discrep- 

 ancies found. 



A conspicuous advantage of this method of Pearson is that in 

 its application the deviation of observation from theory for 

 each class and the amount which this discrepancy contributes 

 to x 2 are under the worker's eye. 



If used with the caution that should be exercised in the draw- 

 ing of any conclusion from probable errors, 9 I believe that this 

 well-known criterion of goodness of fit will prove most useful 

 to Mendelians. 



J. Arthur Harris 



9 Some biologists apparently seem to feel that the calculation of a statis- 

 tical " probable error" covers all the biological sins which may be com- 

 mitted in the collection or manipulation of their data. 



