December 20, 1895.] 



SCIENCE. 



831 



dence that we may feel quite justified in 

 declaring that some error has crept into the 

 classification of tlie injuries included in the 

 three groups, sprains, cuts and bruises, and 

 miscellaneous, as shown in the table for the 

 year 1894. In short, it is more probable 

 that error exists, either clerical, or arising 

 from unusual professional carelessness in 

 diagnosis, than that percentages of distribu- 

 tion, which have persisted so regularly dur- 

 ing the three preceding years, should sud- 

 denly change to the extent shown in the 

 table. 



It is quite likely that the several accident 

 insurance companies of the country have 

 accumulated material relating to fortuitous 

 events much more extensive than the above, 

 which would yield equally interesting re- 

 sults if subjected to analysis. 



There is one point to which it seems 

 worth while to invite especial attention, 

 namely, the confusion which often exists as 

 to the inherent improbability of certain 

 events. Such events are those which, for 

 reasons entirely independent of the proba- 

 bility of their occurrence, have a particular 

 interest. As an illustration, I may refer to 

 the chance of the appearance of a particu- 

 lar hand at whist. Two or three yeai'S ago 

 those interested in games with cards were 

 greatly excited by the alleged occurrence of 

 an event in the Boston & Albany railroad 

 station in Boston. It was nothing less 

 than that during the progress of a game of 

 whist played by three railroad conduc- 

 tors and a mail agent, while waiting for 

 the hour of departure of their trains, 

 on taking up the cards after a deal 

 each man found himself in possession 

 of the whole thirteen cards of one suit. 

 The a priori probability of such an event is 

 all but infinitely small, and it was thought 

 to be necessary to fortify the account pub- 

 lished with affidavits of all the players and 

 also of one or two gentlemen who happened 

 to be watching the game. It probably oc- 



curred to few who read this account that 

 the chances against any other particular 

 distribution of the cards were just as great 

 as against this, and that the result of every 

 deal of the cards is just as remarkable as 

 this and as little likely ever to occur again 

 in the lifetime of the players. Indeed, any 

 event of life, when considered in connection 

 with contemporaneous and related events, 

 in all their ramifications, will be found to 

 have a priori chances so overwhelmingly 

 against it that it seems impossible that it' 

 ever should happen. An ' accidental ' death, 

 for example, is an event generally unlikely, 

 but in any specific case enough collateral 

 circumstance and related fact can always 

 be found to render the a priori probability 

 of the combination nearly infinitely small. 

 The chances of any man whom you may 

 name meeting his death by falling from the 

 third-story back window of the house be- 

 longing to his grandmother on his mother's 

 side, and impaling himself on the point of a 

 cotton umbrella accidentally left wide open 

 in the garden below by the man servant of 

 a gentleman named AVitherspoon, tempor- 

 arily stopping at the nearest inn, to whom 

 he had loaned it on the day before at 2 P. 

 M., in the lull of a thunderstorm which 

 came from the north, are indefinitely small; 

 yet I have been told that a man actually 

 lost his life in just that way, and it is easy 

 to see that the exact repetition of the sim- 

 plest event in life, with all of its accompany- 

 ing conditions and relations, M^ould be just 

 as incredible as this. 



T. C. Mendenhall. 

 WoECESTEE Polytechnic Institute. 



HOBTICULTUBE AT CORNELL. 

 In response to a request from the editor 

 of Science, a brief outline of the purposes 

 and methods of the work in horticulture at 

 Cornell University is here given. This is 

 the more willingly given because no fuU 

 statement has been made of the capabilities 



