164 



EXPERIMENTS TO TEST 



Fig. 3. 



similar results. The details were given in the Appendix 

 to the earlier editions, but have now been omitted to 

 make way for other matter. 



I then tried the follovring experiment : — 

 In figure 3, A is the ants' nest, o the door of 

 the nest, m is the section of a pole on which the 

 whole apparatus is supported. B is a 

 board 2 feet long ; C, D, E, and F are shps 

 of glass connected with the board B by 

 narrow strips of paper G, H, i. k is a 

 movable strip of paper, 1^ inch long, 

 connecting the glass f with the strip e ; 

 and L is another movable strip of paper, 

 as nearly as possible similar, connecting 

 H and I. On each of the slips of glass c 

 and F I put several hundred larvse of L. 

 Jiavus. The object of the larvae on c was 

 to ascertain whether, under such circum- 

 ll*^ stances, other ants would find the larvae acci- 

 dentally ; and I may say at once that none 

 * did so. I then put an ant (a), whom I 

 had imprisoned overnight, to the larvse on F. She 

 took one, and, knowing her way, went straight home 

 over the bridge K and down the strip H. Now it 

 is obvious that by always causing the marked ant 

 (a) to cross the bridge K on a particular piece of 

 paper, and if at other times the papers k and L were 

 reversed, I should be able to ascertain whether other 

 ants who came to the larvae had had the direction 

 and position explained to them j or whether, having only 



