Hollow Spheres of Metal by Heat. 101 



The direction of motion appears to me to be simply dependent 

 upon the direction «f the first impulse. Another mode of ex- 

 planation is thus : — Let A signify a point or locality upon the 

 advancing side of the moving ball, which will, by the motion 

 of the latter, be brought into contact with one of the heated 

 rails. The sources of A's gaining heat are by radiation from 

 the heated copper, and by conduction from the hotter parts of 

 the ball; and its means of losing heat are by radiation into 

 the atmosphere, and by conduction to colder parts of the ball; 

 in addition to these influences, its temperature is slightly 

 affected by upward currents of heated air from the ring. As 

 this point or locality A is moving downwards towards the rail, 

 its temperature is gradually augmented, and goes on increasing 

 not only until it touches the rail and is precisely under the centre 

 of the ball, but probably also until it has passed that position a 

 minute distance, because its rapidity of loss of heat by con- 

 duction is less upon the receding than upon the advancing side 

 of the ball, the advancing side having been more completely 

 cooled by radiation. If the temperature of the locality A is at 

 its maximum at some small distance behind the point of contact 

 of the ball with the rail, the maximum of expansion must also 

 be a little behind that point, and thus constitute a propelling 

 power, as in the previous explanation. 



As nearly as I have been able to ascertain, the following appear 

 to be the most favourable conditions to a successful result : — 1. 

 The rails perfectly equidistant and horizontal. 2. The ball as 

 spherical as possible, and of uniform thickness, or at least with 

 its centre of gravity in the centre. 3. The maximum of tem- 

 perature in the rails and the minimum in the ball. 4. The 

 maximum of heat-conductivity in the rails and the minimum in 

 the sphere. 5. The maximum of expansibility in the metal of 

 the ball and the minimum in that of the rails. 6. A sufficient 

 degree of thinness in the metallic shell to ensure quickness of 

 local expansion. 7. A sphere of large diameter; and 8. The 

 surfaces of the rails and ball as smooth as possible. The same 

 conditions, with modifications of 3, 4, and 5, would probably be 

 also the most favourable ones in the experiments of rotation by 

 electric heat, already referred to. 



The apparatus has been presented to the Iloyal Institution, 



and may be seen there by those who are interested in the matter. 



In conclusion, I have to thank Messrs. Griffiths and Co., of 



Birmingham, for kindly granting me the unrestricted use of 



their enamelling furnace for the experiments. 



Birmingliain, June 14, 1859. 



