EXPLANATION OF FOKM AND COLOUEING. 221 



This method, apart from its general ease and sureness and its capacity for 

 being combined with the offering, where necessary, o£ large series, itself 

 incidentally keeps the experimenter in touch with the state of the animal's 

 appetite and possesses the further inestimable advantage of practically 

 eliminating the personal factor. There is no absolute need to mention the 

 animal's manner : the bird's acceptance of one insect immediately after the 

 refusal of another is usually enough to indicate their relative value in its 

 estimation. Thus where an animal, hungry after a fast or empty after voiding 

 a pellet, eats Z but (after, say, a grasshopper) refuses a second Z yet eats a Y, 

 refuses a Y but eats an X, refuses an X but eats a W, and so on up to A, 

 a second or third individual of which species he rejects or refuses, while 

 treating in the same way an insect that he has been found to eat habitually 

 with eagerness up to actual repletion-point, there can be no difference of 

 opinion amongst even a hundred spectators of the experiment. All, whatever 

 their bias and however differently they might have construed the bird's 

 manner of acceptance or rejection, are likely to agree, on the whole, that A 

 (in spite of the bird's final repugnance to it) is probably a great favourite and 

 Z (in spite of the eagerness of the first acceptance) very much the reverse, 

 and that the bird's preferences, in so far as they can be judged from that 

 single experiment, are probably in alphabetical order. Should these results 

 have been repeated in subsequent experiments, perhaps in very many 

 subsequent experiments and at varying intervals of time, and sufficiently 

 checked by changes in the order of offering, the probability of their truly 

 representing the bird's preferences will have been increased to something- 

 like a certainty. A complicating factor that I have already referred to — the 

 stimulation of the appetite by a very welcome offering or its inhibition by a 

 very unwelcome one — will, of course, have to be watched for, but I have 

 found the latter, at any rate, of less importance here than in the case of my 

 carnivorous mammals. 



It is on results obtained by the above relatively mechanical method that I 

 have placed my main reliance. At the same time, as will be seen, I have not 

 hesitated, in numerous instances, to base deductions on the indications that 

 are so often afforded by an animal's manner. This method is less reliable 

 than the other in that it affords an opportunity for the entry of differences of 

 opinion and unconscious personal bias, but, used with due caution (as I have, 

 I hope, used it) and especially with a knowledge of the state of the animal's 

 appetite (for without this it is useless), it becomes after practice far too 

 valuable a source of evidence to be entirely neglected. 



I would go further, and say that without a knowledge of the animal's state 

 of appetite (best gauged by what of known value the bird will or will not at 

 the moment accept) it is not only useless — it may be hopelessly misleading. 

 A bird will frequently reject with, apparently, some disgust an object 

 that if slightly hungrier he would have eaten with apparent relish ; he will 



X7* 



