610 



THE AMERICAN NATURALIST 



[Vol. XLV 



later tests the blossoms were arranged in a line on the 

 same row of cotton. When the triangular arrangement 

 was followed, the points (a) and (b) were on plants in 

 the same row and about four feet apart. Point (c) was 

 situated on the next row back, equally distant, from (a) 

 and {b). The writer was concealed in the cotton directly 

 in front of these points so that each could be kept readily 

 under observation at all times (see Fig. 1). In each test 



the only changes made were concerned with the blossoms 

 at the points in question. One blossom at least in each 

 experiment served as a control. See the accompanying 

 triangular diagram. 



Experiment No. 1. — July 26. Time of observation, one 

 hour— from 9 to 10 a.m. 



At (a) a normal blossom was growing in situ as a 

 control. 



At (b) a normal blossom was growing in situ as a 

 control. 



At (c) petals only of a blossom were carelessly pinned 

 to a stem. 



An Elis plumipes once alighted on the petals of (c), 

 but immediately discovered the deception and flew away. 

 The species of Melissoflcs in no instance alighted. These 

 bees, it would seem, possess rather keen discerning 

 powers, since without alighting they quickly detect the 

 difference between a normal and a mutilated flower. In 

 many instances, however, they inspect very carefully 



