830 



SCIENCE. 



[N. S. Vol. II. No. 51. 



tie extra care of the passengers. The fourth 

 class, the trespassers, seem, as usual, to 

 have looked out for themselves and to have 

 come out with a little less than their share 

 of damage as shown by the previous year's 

 experience. 



The numbers for the year 1893 are inter- 

 esting. This included a period of excessive 

 passenger traffic, under conditions likely to 

 considerably increase the total number of 

 casualties. It was really, however, little 

 different from and somewhat less than for 

 the year 1892. Nor is the variation in ratio 

 of distribution great, the several percentages 

 agreeing well with the mean of the whole 

 period. 



But the most curious and interesting re- 

 sult brought out by the investigation is the 

 constancy of ratio of distribution of injuries 

 among various classes, such as death, loss of 

 limb, loss of finger, fracture, etc. It is dif- 

 ficult to estimate the a priori probability of 

 any one of these occurrences, and the facts 

 here cited furnish a remarkable illustration 

 of the operation of the principle or Law of 

 the Long Run, as defined above. Indeed, 

 in numbers relatively so small it is ex- 

 tremely surprising to find so many in- 

 stances of persistency. General results as 

 to character of injury and withoiit refer- 

 ence to occupation are shown in diagrams 

 6, 7, 8, 9 and 10, and also numerically in 

 Table B. We are here treating a total of 

 about 15,000 casualties, an average of some- 

 thing less than 4,000 per annum. It would 

 hardly have been expected in advance that 

 during these four years there would be a 

 nearly constant percentage of the total 

 number of injured who would lose a limb, 

 or that year after year almost exactly the 

 same proportion would lose a finger or toe, 

 or that the ratio of fractures and disloca- 

 tions to the whole would be still more per- 

 sistent, and this notwithstanding the fact 

 that the total number of casualties would 

 vary more than thirty per cent. Even 



when the analysis is carried much further 

 there appears striking evidence of the same 

 uniformity of distribution, although natu- 

 rally there might be much less of it. In 

 illustration of this I may cite the following : 

 AVhen the class of employes alone is con- 

 sidered and their injuries classified into the 

 eight groups above referred to, in the two 

 groups which contain the number of those 

 who have suffered from loss of limb, or loss 

 of finger or toe, injuries in which there is 

 nttle chance of mistake in diagnosis, we 

 find: 



Thus while the actual number of casual- 

 ties varied considerably, the proportionate 

 distribution remained extremely constant, 

 particularly in the case of the loss of finger 

 or toe. Although agreement in results 

 where the numbers involved are not large 

 must, itself, be regarded as fortuitous, it is 

 interesting to note that in the class of tres- 

 passers, composed, it may be assumed, very 

 largely of ' tramps,' of whom little regu- 

 larity of any kind might be expected, and 

 of whom about 450 are annually injured in 

 one way or another, the ratio of loss of 

 limb was in 1891, 10.8%; in 1892, 9.7%; 

 in 1893, 9.4%, and in 1894, 9.8%. Of the 

 same uncertain class, it is curious that in 

 1891, 8 sufi'ered the loss of a finger or toe ; 

 in 1892, 9 suffered in the same way ; in 

 1893, 8; but in 1894 this was reduced to 4. 



Examples might be multiplied to almost 

 anjr extent, but it is believed that enough 

 has been shown to establish the existence, 

 in this instance, of the principle under con- 

 sideration. Indeed, so strong is the evi- 



