212 ANIMAL BEHAVIOUR 



and guards against its occurrence. Beavers cut a canal right 

 through an island in a big river — a task not practically justi- 

 fied till it is completed. It is strictly impossible to 'prove 

 that these animals really put two and two together as we 

 do in perceptual inference, but no less generous interpreta- 

 tion seems adequate. 



When Dr. G. T. Romanes's chimpanzee was asked for a 

 number of straws up to five, it used to pick up the required 

 number and present them with the ends exposed between 

 finger and thumb. When it was right it got its reward. Some- 

 times, however, if asked for four straws, it would gather three 

 to save time and double one of them so that four ends showed. 

 When a reward was refused on such occasions, it would 

 straighten out the doubled straw, pick up another one, and 

 present the required number. In a case of this sort we are 

 inclined to admit intelligence, for it was rather subtle and 

 novel, and we know that the chimpanzee has a highly de- 

 veloped brain. 



But pass to one of Miss Drzewina's experiments with 

 hermit-crabs. She removed them from their borrowed shells 

 and gave them similar shells which had been plastered up. 

 The hermit-crabs spent a long time trying to get into these 

 closed shells. Eventually^ however, they gave it up as hope- 

 less, as of course it was. When some shells of the same 

 sort, but empty, were put into the aquarium the hermit- 

 crabs would not look at them. The established association 

 was too strong. Yet when some other shells of a different 

 shape were introduced, the hermit-crabs tried them at once. 

 The question is whether this also was intelligent behaviour, 

 or whether it illustrated what we do not understand, a profit- 

 ing by experience on a lower than an intellectual level, such 

 as must form the basis of the very effective agency of the 



