528 fV ill I am Morton Barrows 



It is a source of continual surprise to see how accurately and 

 quickly many flies will find food. Not only do the flies find 

 food easily and certainly during flight, but they can also find 

 it successfully when creeping. To test this the following experi- 

 ments were tried. A small piece of fermenting banana was placed 

 on a glass plate one inch square and the glass plate with the 

 banana was put in the center of a square sheet of paper ruled into 

 25 squares each one an inch on a side (Fig. 4, a to h). The glass 

 cylinder used in the first experiments (p. 5161 was then placed over 

 this paper in such a position that the four corners of the paper 

 just touched the lower edge of the cylinder. To prevent air cur- 

 rents from driving the odorous particles away, the chamber was 

 closed by a glass plate above. Hungry flies were admitted singly 

 at the bottom of one side of the chamber. They flew as usual 

 toward the light side and upward, but in the course of a minute 

 started to creep down the glass toward the bottom of the chamber. 

 As they came on the paper their course was carefully plotted on a 

 duplicate sheet of ruled paper. The courses given in the diagrams 

 (Fig. 4) show the characteristic paths traversed by twelve flies. 



These paths show that inmost cases the flies took the most direct 

 route in order to reach the food, i.e., they did not merely run upon 

 it by chance. The paths are usuallv so direct that it appears as 

 if the flies found the food by sight, but that this is not so, is shown 

 by subsequent experiments, in which after the removal of the 

 antennae the flies seldom found the food, though their eyes were 

 intact. 



It is clear that both in flight and in creeping the movements 

 toward food are at first irregular and afterward more accurately 

 directed, so that the fly eventually takes an almost straight course 

 to the food. The beginnings of the courses in flight and in creep- 

 ing are such as to suggest the trial and error method of response. 

 This view is supported by what is sometimes seen in the zigzag 

 course taken in the return of flies against the current of odor in the 

 trough in the experiment described on p. 524. The conclusions of 

 the courses however in both flight and creeping are so accurately 

 directed toward the food that trial and error can play no part in 

 explaining this condition; one is forced on the contrary to assume 



