Lifting Power of the Electro- Magnet. 301 



deserving full confidence, and that it is necessary even with tliefeeblefit tranf<ference 

 of electricity. This especially applies to those methods in which a current is 

 divided between two conductors, and its respective quantities in them are esti- 

 mated from their relative resistances, previously determined. That relation 

 involves the temperature of each, and varies with the current. In many re- 

 spects this rheostat was a great improvement on that which I previously used, 

 these probable errors being 0M6 and 0\28 ; but I soon found it could not in- 

 variably be trusted. Occasionally a film of water would adhere so obstinately to 

 the platinum, that its contact with the mercury did not occur till two inches 

 below the surface of the latter ; and this state would continue for several days. 

 A little solution of potassa lessened this tendency, but made the water too good 

 a conductor ; I therefore abandoned the mercury in that part of the instrument, 

 and made the contact by a spring clip of platinimi. This change enables me to 

 use a wire of palladium instead of platinum ; the former resisting twice as much 

 with the same section, and, what is more important, varying its resistance ten 

 times less by a given change of temperature ; being, in this respect, the lowest of 

 all the metals which I have examined. These alterations have improved the ac- 

 curacy of the rheostat, its probable error being now only 0'.06. The wire is 

 ^jij- diameter, and its range 15 inches, read to O'.Ol, by a vernier.* If greater 

 resistance be required, 19 equivalents of the same wire, also immersed in water, 

 can be added to the circuit. I wish I could give some more definite statement 

 of this wire's resistance than is contained in the mention of its diameter, for that 

 alone is not suflicient. Platinum wire I find, even when drawn in a gemmed 

 hole, and heated white hot after its passages, resists unequally in different parts 

 of the same piece, — much more may different specimens be expected to differ. 

 A tolerable approximation to it, however, is given by the fact, that if we use 

 the current unit just described, the intensity (or the electro-motive force of the 

 contact theory) of a Groves' cell, determined by the tangent rheometer, = 47'282 

 inches of this wire.f 



Another measure (which I hope may ultimately prove an accurate one) is 

 afforded by the electrolytic intensity of water (the imaginary polarization of 



* Equal to 970 inches of J^ copper wire. 



t Mean of the last 20 I observed, the greatest being 48'675, the least 45'345. 



2 H 2 



