PREY CAPTURE IN MANTIDS 



65 



exactly reached their goal. We shall see immediately that, though the latter 

 is not the case, there is a proportionality between [x and a, but of a very 

 astonishing form. In the example shown by Fig. 8, the fly is about 29 mm. 

 from the mantis. There are both continuous movements and jerks, the end- 

 points of the jerks being more strongly marked. The continuous move- 

 ments are weak and ineil^ective ; but the endpoints of the jerks are grouped 

 along two straight lines nearly equal in slope. The values lie on the lower 

 line only, if the fly is moved from left to right, and on the upper line only, 

 if it is moved from right to left. The distance between the two lines with 

 regard to abscissa is about 9° (see Fig. 9). Nine degrees at a distance of 



-^ Fly / Prothorax 



Fig. 9. The same experiment as Fig. 8 with head positions reached by jerks only. 

 The slope of the upper full line (regression coefficient) is 0.85 ± 0.03, that of the lower 

 full line 0.83± 0.01(6). The abscissa distance between these lines of about 9° cor- 

 responds to the angular width of the fly presented. Hence the upper and lower broken 

 lines indicate the positions which the head would have, if the right and left sides of the 

 fly were fixed exactly. Yet the differences in slope of full and broken lines are con- 

 siderable (15 and 17%, respectively), and would be expected due to chance with a 

 probability of less than 0.0005. 



29 mm. is about 4% mm. Four and a half mm. is about the mean horizontal 

 width of the flies presented. Thus the simplest explanation is to assume 

 that the animal tends to face the left side of the fly while the fly is going 

 from left to right, and the right side if it is going from right to left. But in 

 neither case was the side of the fly centered exactly. The slopes are 83 and 



