1 42 Mr TREVELYAN'S Experiments on the 



When the heated bar is laid upon the leaden block, the ca- 

 loric passes into these prominences ; and as their power of con- 

 duction is not great, it does not rapidly diffuse itself through 

 the rest of the mass ; of course, they instantly expand and elon- 

 gate, and by that sudden elongation they give an impulse to the 

 incumbent bar. Soon, however, the caloric moves into the ad- 

 joining mass and the prominences contract, and at the same time 

 come into a state ready to admit a renewed accession of caloric 

 from the bar ; they receive that caloric ; again expand, and give 

 a second impulse to the bar. This goes on incessantly; and 

 though the first impulse be infinitely small, and altogether in- 

 adequate to produce any sensible movement of the bar, yet, by 

 incessant repetition, an accumulation of effect takes place, and 

 the movements gradually reach a magnitude sufficient to become 

 easily discernible. 



As soon as the bar and block arrive within a certain limit of 

 difference of temperature, the impulses become feebler and 

 feebler ; and at length the bar comes to rest. 



It has been mentioned, that the smoother the bar is, so much 

 greater is the effect. I conceive that this smoothness operates 

 by increasing the celerity with which the surface possessed of it 

 communicates the caloric to the projecting points of the block, 

 and thereby the elongation which gives the impulse to the bar 

 is increased both in quickness and extent. 



It is obvious, that had the bar any considerable degree of as- 

 perity, the points of contact between the two metals would be 

 fewer, and the passage of the caloric between them more 

 tardy. 



When the surface of the block is highly polished, the expe- 

 riment does not succeed ; no tremors occur. This result pro- 

 ceeds probably from the circumstance, that the caloric enters 

 into every part of the surface of the block equally, and is more 

 quickly diffused through the mass, and hence there are none of 



