360 Mr. 11. Campbell on the Uniformity 



unifonnitj' usually exhibitecl by classes of phpenoinena — such as 

 not only births, deaths, marriages, but such things as those ap- 

 parently depending upon mere caprice — such as the number of 

 letters annually sent without directions through the Post-office, 

 or those which appear likely to depend on the most capricious 

 and irregular causes, such as murders and suicides ; and he pro- 

 ceeds to draw certain moral conclusions from the fact of this 

 uniformity, namely, the existence of certain moral laws by which 

 a section of the community, delinite in number, is always im- 

 pelled to such acts. But before entering upon a discussion of 

 the legitimacy of such results, there appears to me always a 

 previous question to be solved, namely, whether the observed 

 degree of unifoi-mity is remarkable or not. For supposing the 

 observed uniformity to be not more than that which might be 

 expected from events, the occurrence of which to individuals was 

 conceived of as perfectly fortuitous, the whole argument would 

 resolve itself into a pure metaphysical question, from which it 

 would be hopeless to expect any practical issue. 



I propose, therefore, the following problem : — To find some 

 test by which we may ascertain whether a certain observed degree 

 of uniformity, or the reverse, in tables which give the numbers 

 yearly occurring of a certain class of phsenomena is to be looked 

 upon as remarkable or not ; and I propose the following method 

 for finding such a test. Starting from the supposition that we 

 know nothing except the total niimber of such phpenomena which 

 have occurred during a certain number of years, let us try to 

 find the degree of uniformity with which we should expect the 

 phsenomena to be distributed through the difi'erent years. If the 

 uniformity observed in the real tables is much greater than this, 

 we may fairly conclude that there is some cause of this uniformity 

 which we might hope to discover ; and further, tiiat if there is 

 any known cause which might tend to produce such uniformity, 

 Ave may fairly ascribe such uniformity, in part at least, to such 

 cause ; and if, on the other hand, we found the figures in the 

 real tables to vary much more widely than we should expect from 

 the knowledge of the mere fact from which we started, we might 

 fairly say that the number in any year which presented such a 

 remarkable deviation is most likely assignable to some disturbing 

 cause acting in that particular year. 



Such positive result will be matter of probability ; but of this 

 negative result we may be certain, that if the uniformity thus 

 arrived at is very much the same as that of the real tables, we 

 shall not be justified in drawing any moral inference from that 

 uniformity alone ; for it would be shown that such uniformity is 

 only what is to be expected if we knovv' nothing except the total 

 sum of the tables. 



