Siemens Mercury Unit of Electrical Resistance. 299 



larities were met with. In physical measurements, however, it is 

 always dangerous to assume that large errors of experiment can 

 only be of accidental origin and are eliminated by a sufficiently 

 great number of observations. 



Indeed, if we now look at the final published results *, they 

 seem to justify some slight degree of hesitation in accepting them. 

 These mean values still exhibit differences amongst themselves 

 which rise as high as 1*4 per cent. We find further that the 

 slow spins give a mean result which differs by about 0*5 per cent, 

 from that of the rapid spins. In like manner the publication 

 of a few observations belonging to one and the same day (p. 175) 

 makes it possible to form some opinion. Four results are given, 

 which differ amongst themselves by as much as 2*3 per cent., 

 although each of these numbers is obtained by a series of ob- 

 servations extending over about a quarter of an hour and inclu- 

 ding about one hundred scale-readings, from which, if need were, 

 the least-accordant numbers had already been struck out. An 

 unprejudiced reader must always have his attention arrested by 

 differences as great as these. 



But the discrepancies, amounting in some cases to as much 

 as 8*5 per cent., according as the inductor was turned towards 

 the left or towards the right, are to me totally unintelligible. 

 According to a statement by Mr. Jenkin (Poggend. Annalen, 

 vol. cxxvi. p. 387), the explanation of this circumstance lies in 

 the fact that "the thread, to which the magnet was hung, 

 exerted a slight influence in one direction." One is inclined to 

 conclude from this that there was a permanent torsion of the 

 thread one way, whereby the deflection to the two sides would 

 of course be made unequal. But in order to explain such differ- 

 ences as occur here, the torsion would have to be great enough 

 to deflect the magnetic axis of the steel sphere many degrees 

 from the magnetic meridian. It is hardly admissible to suppose 

 that such an oversight was made in suspending the magnet. 

 Should it, however, have occurred, it seems to me that the series 

 of observations affected by it ought to be rejected; for such a 

 want of symmetry would not be willingly allowed even in ordi- 

 nary working, and must be regarded as specially dangerous when 

 the spherical form and feeble magnetization of the magnet are 

 considered. For to say that the magnetic axis of the magnet, 

 on the permanence of which every thing ultimately depends, was 

 really permanent within minutes of arc, when it did not coincide 

 with the direction of the magnetic directive force, would be a 

 hazardous assertion. 



* Brit. Assoc. Rep. 1864, p. 350 ; Poggend. Annalen, vol. cxxvi. p. 386. 



