100 Major- General Sabine on the Cosmical Features 



other the disturbance of the same element might differ widely in 

 amount, and might occasionally be even reversed in direction. 

 Not unfrequently also a disturbance showing itself at the same 

 instant at distant stations would affect one magnetic element at 

 one station and another element at another station, thus giving 

 increased probability to a surmise originally due to the sagacity 

 of M. Gauss, that " various forces might not improbably be con- 

 temporaneously in action, having possibly very different sources, 

 and proceeding independently of each other, the effects of these 

 various forces being intermixed in very different proportions at 

 various places of observation according to the directions and 

 distances of these from the sources from whence the perturba- 

 tions proceed." 



The problem of the origin and laws of the phenomena might 

 thus be deemed even more complex than had been at first ima- 

 gined. But in every case the solution required as the first step 

 the separation, at a particular station where observations had 

 been made hourly, of a sufficient body of the disturbed observa- 

 tions to enable their laws to be investigated, should such laws 

 exist. The difficulty which here presented itself was to decide 

 on some criterion by which a disturbed observation might be 

 distinguished from one which was not disturbed. It has been 

 already noticed that the fluctuations which characterize the dis- 

 turbances are frequently found to exceed the limits of any of the 

 known periodical variations. The magnitude of the discord- 

 ance of an observation, when compared with the mean or normal 

 position of the magnet in the same month and at the same hour, 

 was therefore assumed as a criterion of disturbance for a first 

 essay. Care was taken that the magnitude of disturbance, mea- 

 sured from the mean or normal position, should exceed the pro- 

 bable limit of irregularities which might be occasioned on parti- 

 cular days by the periodical variations of known amount, so that 

 the separated observations might, as far as possible, consist only 

 of those rendered discordant by the special class of phenomena 

 which it was desired to investigate; and the amount of discord- 

 ance thus taken as characterizing a disturbed observation was 

 made constant for the particular station, in order that the ag- 

 gregate amounts of disturbance shown by the disturbed observa- 

 tions in the several hours, days, months, and years at that sta- 

 tion might, as far as possible, be strictly comparable inter se. 



The success of the experiment thus conducted was immediate 

 and decisive. A single twelvemonth sufficed to manifest an order 

 and sequence in the ratios of the aggregate amounts of disturb- 

 ance at each of the twenty-four hours to the mean amount in the 

 twenty-four hours taken as unity, which placed beyond a doubt 

 the fact that, casual and irregular as the disturbances appeared 



