460 Prof. Donkin on certain Questions relating to 



being much smaller than the latter, it is a priori mnch more 

 likel)'^ that some irregular event will happen than sotne regular 

 one. But it is immediately seen that this explanation is unsatis- 

 factory, because the division into parcels is perfectly arbitraiy. 

 We might just as well divide the events into a small and a large 

 parcel on any other principle, and then be surprised at the oc- 

 currence of an event belonging to the smaller parcel. In fact, 

 if this account of the matter were admitted, it would merely 

 prove that the surprise felt on such occasions results from a fal- 

 lacy, and that a well-instructed person ought not to be surprised 

 at all ; a conclusion repugnant to experience and common sense. 

 Nor is the difficulty removed by the supposition, that in the case 

 of a regular event we involuntarily consider it as having been 

 predicted, or at least singled out and contemplated beforehand. 

 For the probability of an event is in no way altered by the fact 

 of its having been predicted ; so that if it happen after being 

 predicted, there is no more reason for surjjrise, so far as a priori 

 probabilities only are concerned, than there would have been 

 without the prediction. 



6. I conceive that the true answer to this question is, that in 

 such cases we compare, not the a priori probabilities of different 

 events considered as accidental, but the a priori probabilities 

 of the same event considered on the one hand as resulting 

 from accident, and on the other from design, or some other 

 simple cause. A symmetrical phfenomenon suggests a simple 

 cause, because such causes coumionly produce such plijenomena ; 

 an irregular pha?nomenon suggests a complex combination of 

 causes, for a similar reason. When we see a symmetrical phge- 

 nomenon, withoTit knowing how it was produced, we therefore 

 immediately refer it to some simple cause, with a conviction of 

 which the intensity depends upon a comparison between the 

 facilities with which the phnenomenon might have been produced 

 by such a cause or by other supposable complex causes. [It was 

 shown in the fonner paper that there is no difficulty of principle 

 in estimating the force of such conclusions mathematically.] If, 

 then, we see a symmetrical phsenomenon, and know, or have 

 reason to Ijelieve, that it was really produced by accident, we are 

 surprised, because we see an effect resulting from the less likely 

 of two causes capable of producing it. From this follows of course, 

 what is obviously true in fact, that our surprise depends entirely 

 upon our knowledge of the causes which might have been sup- 

 posed to produce tlie phajuomenon, and of the relative facilities 

 with which they might have produced it. If a person who can 

 read were to draw letters out of a bag and produce his own 

 name, he would be very much surprised ; whereas, if he could 

 not read, he would not be suiiirised at all. The u priori pro- 



