Electromagnetic Effect of Convection-Currents. 459 



This is a fair example of the way the deflexions run. As a 

 further illustration of this take #17 and #18 ; these two are 

 identical in arrangement, but the direction of rotation is in 

 one case got by crossing the belts from the countershaft to 

 the disks and leaving the main bolt straight; in the other the 

 main belt is crossed while the auxiliary belts are straight. 

 The deflexions are the same. This, too^ shows that the 

 difl"erence cannot be due to any effect of the countershaft. 

 The cause of this has not yet been explained. The work is 

 to be continued with this and also with new apparatus, made 

 like the Berlin apparatus, but with the disk much larger, 30 

 centim. in diameter ; at least double the speed then obtained 

 will be used. This ought to give deflexions on reversal of 

 1*5 to 1*7 ceutim. 



The values of V do not agree so well as might be looked 

 for ; but when, in addition to the numerous difficulties already 

 mentioned, the smallness of the deflexion is considered, and 

 the possibility of the needle being affected by currents or 

 magnets in other portions of the laboratory, so far away as 

 not to be guarded against, and Avhich might well be changed 

 between the time of taking the observation and the deter- 

 mination of the needle-constant, and, finally, that a disturb- 

 ing cause of some kind is still undoubtedly present, the 

 agreement is seen to be as good as could justly be expected. 



Physical Laboratory, 



Johns Hopkins University, 

 April 22, 1889. 



Note, added April 29. 



There seems to be a misunderstanding in certain quarters 

 as to the nature of the deflexion obtained in Prof. Rowland^s 

 first experiment. The paper reads : — " The swing of the 

 needle on reversing the electrification was about 10 to 15 mm., 

 and therefore the point of equilibrium Avas altered 5 to 

 7 "5 mm." This has been construed to mean that the deflexion 

 was merely a throw, and that no continuous deflexion was 

 obtained. This is entirely erroneous; there was always a 

 continuous deflexion. The throw was read merely because 

 the needle was always more or less unsteady, and better 

 results could be got by seizing a favourable moment when 

 the needle was quiet and reading the throw, than by attempt- 

 ing to take the successive elongations, or waiting for the 

 needle to come to rest. In the experiment described above 

 the needle was very steady and no such trouble was 



