EXPLANATION OF FORM AND COLOURING. 225 



by my readers^ I should be grateful i£ they would kindly draw my attention 

 to them. The main method of deduction has been explained already : it is based 

 on the assumption that i£ B is refused but A then eaten, A is probably (other 

 things being equal) at that moment preferred to B. If A had been first eaten 

 and B then refused, the probability would have been that A was at any rate 

 not appreciably less liked than B. It would be unsafe to go further unless a 

 marked difference in manner justified it. 



Again, it will be noticed that up to a fairly late date in May, 1909, by 

 which time I was beginning to adopt more critical methods, the experiments 

 (as I now interpret themj were very barren of results. I have already 

 stated the reasons for this. In this relatively unproductive category come 

 all the experiments on Roller A and approximately the first fifteen on 

 Holler B. At the same time the further advance from even my later 

 experiments on the latter bird and on C to my latest experiments of all, 

 to be published in a future paper, was probably nearly as great, for it was 

 not until this latest period that I realized the effect on appetite of " the 

 preceding offering " and shaped my experiments accordingly. " Refused A 

 ate B, refused A ate G, refused A ate D " now takes the place of " refused 

 A, ate B, C and D," B's possible stimulatory influence making it not quite 

 safe to infer from the last formula that C and D as well as B were preferred 

 to A. Much evidence that at the time 1 had regarded as good is thus 

 rendered doubtful and I am refraining from using it. 



A.S for the discarded conclusions from my very early experiments, I freely 

 confess, for any who may still find preferences hard to believe in, that I too 

 at first started out, not, it is true, with a disbelief in " unpalatability " (for, 

 amongst other things, I had had the advantage of witnessing some of Mr. 

 MarshalFs experiments in 1899 on animals of my own), but at any rate with 

 much confidence, founded on what then seemed good evidence, that very 

 little unpalatability would be found to occur in our Satyrince^ NympJialince, 

 Pierince, or Lycsenidse. I confess further that it took quite a good deal of 

 evidence to convince me of the contrary, the rejecting animal being always, 

 I could not help for a time suspecting, " in a capricious mood " or " reckless,'"" 

 or perhaps " unsuitable " — at any rate, anything rather than right. I suppose 

 that nearly every investigator of such a subject is likely at first to pass 

 through a shorter or longer "refractory" stage of this kind before he shakes 

 off his first opinions and, throwing everything into the melting-pot, begins 

 to learn. 



It will be seen later that this was not the only idea of which my animals 

 soon disabused me, and, as a complete cure for too great confidence in any 

 particular view, I can thoroughly recommend a good long course of special 

 experimentation. 



For the explanation of the few abbreviations that, to save space and end- 

 less repetition, I have used in the course of this paper, I must refer the 



