THO UGHT-EEADING. 



319 



suggested that some code of signals may have been 

 arranged by which the proper answer was communi- 

 cated to those questioned. The following cases are 

 free from this objection : 



' ' It will be well to give/' writes Professor Barrett, 

 ' ' a group of results obtained when no member of the 

 family was aware of the selected object. Eleven times 

 running we chose a card at random, and on six of 

 these occasions one of the children named the selected 

 card (giving both suit and pips, or fully designating 

 the court card) correctly at the first trial ; twice the 

 card was named correctly on the second trial; and 

 three cases were failures. On none of these occasions 

 was it even remotely possible for the child to obtain 

 by any ordinary means a knowledge of the card se- 

 lected. Our own facial expression was the only index 

 open to her ; and even if we had not purposely looked 

 as neutral as possible, it is difficult to imagine how 

 we could have unconsciously carried, say, the two of 

 diamonds written on our foreheads. The outline of 

 results during the present investigation, which ex- 

 tended over six days, stands as follows : Altogether, 

 382 trials were made. In the case of letters of the 

 alphabet, of cards, and of numbers of two figures, the 

 chances against success on a first trial would naturally 

 be 25 to 1, 51 to 1, and 98 to 1, respectively; in the 

 case of surnames they would, of course, be indefinitely 

 greater. Cards were far most frequently employed, 

 and the odds in their case may be taken as a fair 

 medium sample ; according to which, out of the whole 



