522 THE ROYAL SOCIETY OF CANADA 



at the age of 20, the corresponding number was only 134. On this 

 basis, if e.g. the elder daughter was 30 and the younger 20, the indef- 

 inite probability in favor of the 1st conclusion would be not merely a 

 balance of probability but a ratio of 6,359:134, approximately 47/48 

 or 47:1. In the 2nd conclusion the probability depends approxi- 

 mately on the chances that the younger daughter is married, as op- 

 posed to the chances that the mother was twice married. According 

 to the statistics of five principal countries summarized in the same 

 volume of Brockhaus, p. 927, an average of 5,888-6 persons (or about 

 6,000) out of every 10,000 of marriageable age were either married, 

 widowed or divorced. This would make the probability of the daugh- 

 ter's marriage approximately 60/100. In the same edition of Brock- 

 haus, vol. V, p. 747, the averages of every 100 women marrying in six 

 principal countries are given as 90-97 spinsters, and 9-03 widows and 

 divorcées. This would make an approximate probability of 9/100 

 that Mrs. Jones was twice married, and the chances for and against 

 the 2nd conclusion respectively would be 60/100 and 9/100, yielding a 

 probability of 60/69 or 60:9 in favor of the conclusion. 



In the 3rd conclusion the principal alternatives are children by 

 successive husbands or full and legal adoption of one of the daughters 

 into another family. While the binding out especially of orphan 

 children, under their own names, as semi-servants until of age, is not 

 unusual, permanent adoption with change of name, particularly while 

 the mother is living, is comparatively rare. On the other hand, 

 mothers not infrequently have children by successive husbands; 

 therefore the probability of this conclusion is conservatively estimated 

 at 5/6 or 5:1. In the 4th conclusion the probability is fixed by the 

 instances where sisters have different surnames by complete adoption 

 of one of them into another family, as opposed to those instances 

 where such difference in surnames arises from any other causes (bar- 

 ring those excluded by the evidence here available). Instances of 

 such adoption, especially when both father and mother are living, are 

 rare, but the opposing instances are much rarer; hence the probability 

 of this conclusion has been estimated at 10/11 or 10:1. 



Case 9. 



Case 4 illustrated the feature that scientific results should be 

 stated ordinarily in positive, and not in negative, form because per- 

 sons receiving information wish generally to know what a thing is, 

 not what it is not. An exception to this rule is where positive infor- 

 mation on one point of interest to the reader raises in his mind a ques- 

 tion on a related point, concerning which he would like an answer 

 either positive or negative. In such circumstances even a negative 



