VARIATION IN BEES. 



FRANK E. LUTZ. 



In the study of evolution, there is nothing more important 

 than the investigation of variations, since the whole doctrine rests 

 upon the premise that organisms do vary. There was a time 

 when it was sufficient in such an investigation to take a series of 

 specimens and from the general looks of things postulate theor- 

 ies. But the world has become more critical now and demands 

 that when a statement is made concerning some phenomenon, 

 exact data accompany the statement. Hence, the statistical 

 study of variation which attempts to exactly measure the varia- 

 tions and correlations of different organs and to set them down 

 in figures which "cannot lie." And here we cannot allow the 

 other proverb which says that figures will prove anything, for 

 figures truthfully handled can only prove the truth. But there 

 is, on the other hand, great danger that, having collected a set 

 of measurements, we make a show of accuracy that will lead us 

 and others astray by reason of careless or insufficient analysis. 

 Such work is most troublesome because of its seeming exactness 

 and the difficulty of detecting errors. 



Messrs. Casteel and Phillips, in the December (1903) number 

 of this BULLETIN, have taken up a very interesting and vitally im- 

 portant problem. The comparative variability of the drone and 

 worker bees hits, in a way, at the very root of the variation 

 question. Accordingly, while we lament with the authors the 

 smallness of their series, it seems well worth while to consider a 

 few points about the paper. 



In the first place, we have to disagree with the statement that 

 if the variability is "due to chance," it is "not in accordance 

 with any law," for it is well known, and has been for years, that 

 nothing is more bound by law or more expressible in mathe- 

 matical formulae than "chance." However, we will heartily 

 agree with them that the "true test of the relative variability" 

 is the " descent in numbers of individuals " in the different classes 

 as they are removed from the mean ; but we w r onder greatly 



217 



