334 M. W. Siemens on the Question of 



to which this might be attributed, viz. variations of temperature 

 of the copper wire with which the tubes were compared, to 3° C, 

 and of the tubes themselves to 2° C. Besides this, Sabine has 

 pointed out in his paper that these differences were ascribable 

 principally to errors in the measuring- apparatus, being quantita- 

 tively plus or minus according as the balance-wire was bisected on 

 one side or the other of the middle point. And even were the 

 differences apparent in the first determination not to be satis- 

 factorily explained, they would nevertheless say nothing against 

 the system, because the tubes chosen were not so cylindrical as 

 they might have been — the object of this determination not being 

 so much to establish normal standards, as to prove that the pro- 

 posed method was capable of doing so. For practical purposes, 

 at that date, an exactness of OS per cent, was considered amply 

 sufficient ; and indeed Dr. Matthiessen was himself contented 

 with determinations of the conducting-powers of the metals 

 which differed many per cent. 



The little standards which I have distributed were all adjusted 

 to within 0*05 per cent, of the mean value of the third repro- 

 duction, from the description of which it will be evident that, 

 with normal weights and measures, the mercury unit is repro- 

 ducible within the same limit of exactness. 



I do not think that Dr. Matthiessen should have compared 

 the results of his determination with mine, as he has departed, 

 in several material points, from the method which I laid down. 



The correction-formula for conicalness employed by Dr. Mat- 

 thiessen regards the tubes as series of different cylinders, and 

 not as cones, as in my formula. The coefficient is, however, 

 very small, and would not account for the difference between 

 the determinations, although it would be in the same direction. 

 A more important deviation from the modus operandi which I 

 followed, is in filling the tubes by dipping them into a trough 

 of mercury, and lifting them out again by pressing the ends 

 between two fingers. The soft skin of the points of the fingers 

 would naturally be pressed into the openings, whereby the 

 weighed contents of the tubes would be too small, and the 

 calculated resistance therefore too large. 



It is also barely possible that Dr. Matthiessen has omitted 

 to invert the tubes in the bridge, and to take means of the read- 

 ings in each direction. 



At any rate, the difference 0*8 per cent, exists between his 

 determination of the mercury unit and mine ; and if it is not 

 explained by these points of difference in our manipulation, they 

 serve to show, at least, how little care Dr. Matthiessen has 

 taken to follow in the footsteps of the method I had pointed 

 out. On this account I do not think that the result of his de- 



