of the Approach and Recession of Bodies. 73 



the end of the balance furthest from the flame ; the colouring 

 matter will then be seen to descend perpendicularly, move 

 along the bottom towards the flame, and rise up the side of 

 the vessel next the flame, as in the last experiment. 



" These two experiments render it highly probable, I 

 think, that convection currents play an important part in the 

 production of the phenomena under discussion, so far at 

 least as they apply to air at the ordinary pressure. 



"If, as had been suggested, evaporation from the surface of 

 the pith was involved in these motions, it was conceived 

 that some confirmation of this might be obtained by a 

 comparison of the effects of heat on the same balance in 

 moist and dry air, other conditions being as nearly as 

 possible the same. 



" With this object in view a balance consisting of a 

 platinum wire carrying a pith ball at each end was sus- 

 pended horizontally within a round-bottomed glass flask. 

 The pith balls weighed about a grain each, and the platinum 

 stem considerably less than a grain. 



" Experiment 5. — The flask, after having been well air 

 dried and warmed, had boiling oil of vitriol poured into it, 

 and was then closed with an indiarubber stopper carrying 

 the little balance. When a flame was approached, after the 

 air within the flask had reached the ordinary pressure and 

 temperature, the nearest pith came towards it as usual. 

 The pith almost never reached so far as the flame however, 

 but, after several vibrations, took up a position about 20° 

 short of the flame. 



" Experiment 6. — The oil of vitriol was next replaced by 

 water, which was caused to boil in the flask for some time 

 and then allowed to cool down. The inner surface of the 

 glass soon became bedewed with moisture. When the same 

 flame, under the same conditions as in the last experiment 

 was presented, the nearest pith moved towards it as usual. 

 But the motion was in this case much more rapid and the 

 swing farther ; the pith usually went well past the flame, 

 repassed, and then vibrated slowly about 80° from the 

 starting point, or within 10° of the flame. 



" The difference in the motions exhibited in these two 

 cases of the moist and dry air, may be looked upon, I think, 

 as favouring the notion of evaporation as a cause operating 

 in the production of these phenomena. 



" As, in the second case, the air was already saturated 

 with moisture, evaporation could not take place so readily 



