BEHAVIOR OF THE COMMON ROACH. 349 



FG). When in use the maze is supported, in a horizontal posi- 

 tion, by glass pillars which rest in wide pans of water (Fig. 21. 

 These pillars were made by inserting glass stirring rods in the 

 corks of wide-mouth bottles. When thus supported, the maze 

 was about eight inches above the surface of the water which 

 extended beyond it, in all directions, to a distance of eight to 

 twenty inches. To facilitate the taking of accurate notes the 

 parts of the stage were labeled as indicated in Fig. i . 



FIG. 2. Photo showing the maze and the upper portions of its support as 

 arranged for use. 



These experiments were conducted, in the summer time, in 

 an out-of-doors insectary, the whole north wall of which, except 

 a narrow strip for a door, was constructed of wire netting. The 

 three other walls were without either windows or doors, hence 

 this north opening was the only source of light. Except in the 

 few special cases mentioned in the body of this article, the maze 

 was always arranged with the side /I on the north and parallel 

 to that side of the house. 



As originally planned the experiments were to test the ability 

 of the roach to learn to go by the shortest route from the portion 

 of the maze marked "/" to a dark box placed at some definite 

 place on the maze. A few preliminary experiments demonstrated 

 that the box did not make a satisfactory goal. Some insects on 

 reaching the box would enter it and remain therein; others would 



