Mr. Marsh's Thermo-Electric Apparatus. 327 



the flame of the lamp in this situation of it, will give rise to a 

 slow rotation in either direction according to the circumstances 

 of the case. The same principles will enable us to explain 

 the increase of acceleration produced by two magnets with 

 their opposite poles applied as in fig. 7 and fig. 10, and the 

 cause of the non-action of a central magnet, except as in the 

 apparatus of Pi'ofessor Gumming, where one of the branches 

 is carried further from the centre; in which case a slight ten- 

 dency to rotation is exhibited equal to the difference of the 

 two opposite forces. 



We see thus the marked difference between the electro- 

 magnetic rotations produced by the application of the lamp, 

 as in the cases above, and those produced by the galvanic 

 machine ; in the latter case it is essential to have the magnet 

 central, whereas in this the magnet must be exterior to pro- 

 duce the desired effect; and the reason is obvious (refer- 

 ring for example to Experiment X, Essay on Magnetic Attrac- 

 tions), for here, and in all similar cases, the fluid being trans- 

 mitted from the centre passes down the several branches in 

 the same direction, and is therefore acted upon by the central 

 magnet all in one sense ; whereas in these, the fluid being 

 ascending in one branch, and descending in the other, the 

 forces on one side counterbalance those on the other, and the 

 machine remains quiescent. 



It has been objected against the hypothesis I have advanced 

 of a tangential force varying inversely as the square of the 

 distance, that if such were the case, the cylinder alluded to 

 above (in Experiment X of my Essay) ought to revolve by the 

 application of an exterior magnet ; whereas it is almost en- 

 tirely insensible to its action. But this objection must fall, if 

 the nature of the forces on the periphery of a circle from an 

 exterior point be properly considered; for let p (fig. 15) be 

 an exterior point, and pa, pb, tangents to the circle abd, 

 then the part of the circumference between a and b will be 

 acted upon in one direction, and all the other part of the cir- 

 cumference in the opposite one; and although the line of ac- 

 tion is more considerable in the latter case, the intensity is 

 greater in the former, and the difference between the two is 

 not sufficient to cause the rotation. In other words, the centre 

 of attraction of the periphery of a circle estimated from a 

 point without, falls so near to its centre, that the effect to pro- 

 duce rotation is too weak to render itself sensible. 



P.S. — It may be proper to observe, that for the conveni- 

 ence of making the drawings I have represented the magnet 

 as standing upright ; but the experiments were generally made 

 with powerful magnets placed horizontallv. 



LXVII. On 



