168 EXPERIMENTS TO TEST 



other strips / and g, in such a manner that they c^an be 

 freely turned round, so that each can be turned at will 

 either to h or tyi. Under ordinary circumstances the 

 paper /, as in the figure, was turned to the larvse ; but 

 whenever any ant, excepting the marked one, came, I 

 turned the papers, so that / led to m and g to h. The 

 result was striking, and I give the observation in full 

 in the Appendix. In all, 17 ants came, every one of 

 whom took the wrong turn and went to m. 



Although the observations above recorded seem to 

 Fig. .5. me almost conclusive, still I varied the 



experiments once more (see fig. 5), 

 making the connexion between the 

 board B and the glass containing the 

 larvse by three separate but similar 



A 



6 



A strips of paper, cZ, e, and/, as shown in 



^ the figure. Whenever, however, a 



// "->•., strange ant came, I took up the strip / 



' ' " and rubbed my finger over it two or 



three times so as to remove any scent, and then re- 

 placed it. As soon as the stranger had reached the 

 paper e, 1 took up the strip cZ, and placed it so as to 

 connect e with the empty glass m,. Thus I escaped the 

 necessity of changing the paper /, and yet had a scented 

 bridge between e and m. The details, as before, are 

 given in the Appendix. 



In this experiment the bridge over which the 

 marked ant passed to the larvae was left in its place, 

 the scent, however, being removed or obscured by the 



