430 THE BORDERLAND OF SCIENCE. 



dies), 6 when it does happen, almost always happens 

 with A, and very rarely without it. That is to say, 

 such is the phenomenon asserted ; and all who rationally 

 refer it to casualty affirm that B is happening very often 

 as well as A, but that it is not thought worthy of being 

 recorded except when A is simultaneous.' I must ven- 

 ture to express my dissent from this statement; it 

 seems to me incredible that any person would, as De 

 Morgan asserts, rationally affirm that spectral appear- 

 ances are 'very often' seen. 'In talking of this 

 subject,' he proceeds, 6 it is necessary to put out of the 

 question all who play fast and loose with their secret 

 convictions : these had better give as a reason, when 

 they feel internal pressure for explanation, that there is 

 no weathercock at Kilve : this would do for all cases. 

 But persons of real inquiry will see that, first, experi- 

 ence does not bear out the asserted frequency of the 

 spectre, without the alleged coincidence of death ; and 

 secondly that if the crowd of purely casual spectres 

 were so great that it is no wonder that now and then v 

 the person should have died at or near the moment, we 

 ought to expect a much larger proportion of cases in 

 which the spectre should come at the moment of the 

 death of one or another of all the cluster who are 

 closely connected with the original of the spectre.' 

 (This is not very distinct : any wrong spectre, with or 

 without close connection with any particular moribund, 

 would seem to serve De Morgan's purpose in this argu- 

 ment equally well. He appears to insist, however, on the 

 fact undoubtedly such that if spectres were com- 



