S62 EXPERIMENTS WITH 



7 inches liigh (in fact, a hat-box) for the flat paper, 

 cutting two small holes at / and g, so that the ants 

 passing from the nest to the food went through the box 

 entering at/ and coming out at g. The box ^vas fixed 



Fig. 25. 



at d, so that it might turn easily. I then, when they 

 had got to know their way, turned the box round as 

 soon as an ant had entered it, but in every case the 



Fig 26. 



ant tiumed round too, thus retaining her direction. I 



then varied the experiment as shown in Figs. 25 and 26. 



I replaced the white disk of paper, but put the 



food e at the middle of the board. When the ant had 



