452 APPENDIX 



Loophole experiments have been described by Claparede, who 

 is, on the whole, our best and most impartial witness. Krall, 

 in Claparede's presence, wrote a problem on the board and retired 

 along with Claparede from the stable. The horse was left alone 

 and his proceedings watched through a loophole from outside. 

 The first sum so given was V 6 14656, and the answer, given in 

 some seconds, was 28, which is right. The next was ^4879681, 

 for which Muhamed, after 30 seconds, gave 117 and then 144^ 

 The right solution, which is 47, Muhamed did not find. For the 

 next fourth root, where the right answer was 56, he gave the 

 following attempts : 43, 73, 267, 34, 74, and 84, and then, 

 being set another sum, he rapped out 56, which would have been 

 right in the previous case. The results of this experiment are 

 curious and I think instructive. The only sum which the horse 

 does right is the first. It looks as though this answer might 

 have been somehow conveyed to the horse, or that possibly he had 

 the figure or the appropriate rappings for the figure 28 in his 

 memory previously. We shall find some corroboration for this 

 suggestion in other answers given by the horse. After the first 

 one he fails, but in the same curious way he seems to have the 

 right answer to sum No. 3 in his mind, for it comes out, though 

 on the wrong occasion. It is clear that Claparede's record, taken 

 alone, is unsatisfactory for the defenders of Muhamed, and with 

 regard to the loophole experiments in general Dr. Maday points 

 out l that, so far as our descriptions go, the use of audible signs 

 is not altogether excluded. In several cases the door is partly 

 open, for instance, and Krall habitually talks to the horse from 

 outside. In 21 out of the 22 recorded cases Krall himself wrote up 

 the problem. Four of these cases were so-called Kuvert Versuche, 

 that is to say, the sum set was put inside an envelope, which Krall 

 opened immediately before writing it up on the board, and in 

 these four the proportion of false answers to the right was large. 

 One was not solved at all, but here again, as in Claparede's case, 

 the answer was given when the next sum was set. But in these 

 experiments again an awkward incident occurred. The horse 

 gave 33 as a cube root where the answer should have been 23. 

 Krall called out " False," and told the groom to give the horse a 

 stroke of the whip, whereupon, after another error, the horse 

 gave 23. This occurred, says Dr. Buttell Reepen, notwithstanding 

 that he had not told Krall the right solution. Krall's explanation 

 is that, in many cases, he knew a false answer from the manner of 

 pawing, but it must be admitted that the incident throws con- 

 siderable doubt upon the value of experiments in which Krall is 

 supposed to be ignorant of the answer to be given. 



1 Op. tit. p. 297. 



