the Conductivity of Liquids. 333 



a specific gravity that a drop of it would just sink in the 

 liquid used ; and when the magnets were revolving at a great 

 speed a series of drops were delivered at different distances 

 from the centre. They all fell in spiral lines, each one re- 

 volving more quickly than the one outside it. Then, on re- 

 versing the engine and repeating the experiment the spirals 

 were found to be going the other way, but, in each case, 

 in the same direction as the magnets. The average speed 

 of the liquid (30 p. c. H 2 S0 4 , 70 p. c. H 2 0) was about one 

 turn in ten minutes. It is now possible to see to what extent 

 movement of liquid in the cell affects the result. As soon as 

 the rotation of the liquid has become constant, the force 

 urging it forward is equal to that due to friction retarding it ; 

 the torsion of the wire therefore is an exact measure of the 

 force on the moving liquid. But the force is directly propor- 

 tional to the relative speed, and not to the actual speed, of the 

 magnets ; and we have seen that the liquid does not revolve 

 more than once for 20,000 turns of the magnets ; and there- 

 fore the error made on the supposition that the liquid is at 

 rest is not more than the 20,000th part of the whole result — 

 a quantity altogether inappreciable, for neither the speed nor 

 the torsion can be measured with such accuracy. 



The apparatus described was finished and the experiments 

 begun on April 5, 1880; but it was immediately found that 

 the behaviour of the wire was such that no results of any 

 value could be obtained ; for the zero on the scale (that is, the 

 position without torsion) was constantly changing, so much 

 so that sometimes without any apparent cause the scale would 

 move ten divisions in a few hours. As 11 '4 divisions corre- 

 spond to an angular movement of one degree, and as the 

 greatest torsion ever observed caused a movement of only 45 

 divisions, it seemed that some other wire would have to be 

 used. But, again, steel was the only metal that could be 

 used, as a weaker metal must have been much thicker, and 

 the torsion of a thicker wire would have been too great. 

 Weights were hung on the wire and twisted several times 

 round and then left, to see if the wire would improve by such 

 treatment, but without effect. Then some other wire was 

 sent for, and the machine left for six weeks. During this 

 time the old wire became a little rusty; and when the 

 experiments were to have been continued with the new 

 wire, the old wire was tried once more and was found greatly 

 improved, possibly from the removal of the hard skin by rust. 

 During the course of one experiment, lasting over an hour, 

 the zero had not shifted the tenth part of a division. The 

 next day the same acid was examined again ; and the two 



Phil Maq. S. 5. Vol. 10. No. 63. Nov. 1880. 2 B 



