Dr. Matthiessen on the so-called Mercury Unit. 369 



Copies of a unit may be made, I maintain, to any accuracy 

 required : all the B. A. units agree together, when issued, to 

 within O01 per cent., of course at the temperature stated 

 on them : they are all kept a month after making ; if during that 

 time they alter in the least, they are taken to pieces and remade ; 

 out of fifty already made, only two have had to be altered, the 

 cause of failure being in both cases faulty soldering. 



This statement, together with the above fact of the four dif- 

 ferent copies of the 1864 mercury unit agreeing so well together, 

 prove that copies of a unit can be, and are, made to agree 

 together much more accurately than it can be reproduced even 

 when great care is taken. 



If the reproduction of a unit be so easy a matter, how 

 are we to explain the large differences observed in the copies of 

 the different issues of the mercury unit. The simple truth is 

 that the reproduction of any unit, where great accuracy is 

 required, is exceedingly difficult. Take, for instance, the repro- 

 duction of the normal pound by Professor Miller of Cambridge : 

 one would think, at first sight, nothing easier ; but when we 

 enter into the minor details, we find how difficult the problem 

 was to solve. 



It would have been better if, instead of reproducing the unit 

 from time to time, Messrs. Siemens had done what the Com- 

 mittee on Electrical Standards appointed by the British Associ- 

 ation have done — namely, first determined the value of their unit 

 as accurately as possible, and then made copies of it in several 

 different materials. It is not possible that all these copies will 

 alter to a like extent by age, and it is very probable that most 

 of them will not alter at all : they will, of course, be tested from 

 time to time; and at the Kew Observatory, where they will be 

 deposited, arrangements will be shortly made for comparing also, 

 from time to time, at a small cost, the coils issued by the Com- 

 mittee, thus offering a ready means of ascertaining whether the 

 coils issued are constant or not. The coils issued by the Com- 

 mittee are copies of the B. A. unit, and at the temperature 

 marked on them are correct to within 0*01 per cent. By 

 using these coils, the results of different observers become 

 comparable. The Committee of Electrical Standards do not 

 wish to imply that the resistance of their unit is exactly equal 

 to 10 7 f~^ Weber's electromagnetic units, but only a close 

 approximation to it. If in a few years' time new and more ac- 

 curate determinations be made, then, supposing a correction 

 required, the copies of the B. A. unit (1864) being all exactly 

 equal to one another, will have a small value to be added to or 

 subtracted from their resistance, and all the comparisons made 

 with them corrected by the same coefficient. 



