COMPARATIVE PSYCHOLOGY 27 



accept. At the same time, much that can not be re- 

 garded as wholly reliable may prove suggestive and 

 serve as the starting-point of investigations. But there 

 is no reason why many points now bearing the character 

 of uncertainty and indefiniteness might not be sub- 

 mitted to the test of experiment. Doubtless not a few 

 supposed facts would vanish into thin air if subjected 

 to such examination. However, I must at the same 

 time state that a careful perusal of the accounts of the 

 experiments of even the most skilful investigators by 

 this method, with its clearly defined but artificially 

 arranged conditions, has convinced me that such do not 

 wholly meet the case. They bear with them the danger 

 of fallacy against which one must constantly be on the 

 watch. It must always be considered that the great 

 question is, not how an animal's mind may act, valuable 

 as that may be, but how it normally does act ; that is to 

 say, what are the natural psychic processes of the class of 

 animals under investigation? The same caution, in 

 drawing conclusions, must be observed in the allied 

 science of physiology, one in which the conditions can 

 be much more accurately regulated. Plainly, it will be 

 desirable to keep our facts very sharply apart from our 

 explanations. The science of psychology is a very 

 youthful one, that of comparative psychology still more 

 so ; and, at the present stage of the science, any one 

 who contributes a single fact will be a real friend to 

 their progress. We must endeavour to secure a large 

 number of correspondents who will furnish accurate 

 accounts of phenomena in this realm, of which they 

 have been themselves the observers. We must place 

 all material coming at second-hand by itself, not as 

 worthless, but as calling for special scrutiny. But so 

 long as we have facts only, we have no science ; such, 

 indeed, are as the wood and stone for the building, and, 

 unless worked up into scientific form, may prove an 



