of the Approach and Recession of Bodies. 75 



a position a little above the level of the farther end, the 

 balance swung round, the two ends eventually changing 

 places ; this change of position being repeated at intervals 

 as before. Since in this arrangement, however, the altera- 

 tion in the relative level of the two ends was gradual, and 

 not sudden as it was in the oil of vitriol experiment, the 

 exchange of position between the ends was likewise less 

 sudden, one or more false starts being sometimes made before 

 the change was effected. When time was given this change 

 never failed to take place. 



" In the attempt to eliminate as far as possible the effects 

 of condensation, and so to examine the effects due, as was 

 supposed, to air currents alone, on a balance with ends at 

 different levels, another form of experiment was thus 

 arranged. 



"Experiment 9. — A balance was prepared consisting of a 

 glass stem furnished with a disc of paraffin at the end of 

 one arm, and having the end of the other arm bent in the 

 form of a hook. The hook served to support a number of 

 little glass riders. By adding or removing these little 

 weights the balance could be adjusted so as to swing in any 

 position. It was then suspended within a dry flask. 



"When the paraffin disc was highest it retreated rapidly 

 on the approach of heat, as of the finger held close to the 

 glass. When weights were removed from the other end 

 and the disc was lowest it swung rapidly towards the finger. 

 When the balance was as nearly as possible horizontal the 

 nearest end always approached the heat. A source of 

 cold on being brought near produced just the opposite 

 effects. 



"When balls of paraffin or pith were used the same 

 results were obtained ; and again, when the glass flask was 

 replaced by a tin vessel, only in the latter case the move- 

 ments were rather more rapid. 



" If we consider the course of the air currents within the 

 flask, their effect of difference of level will receive a ready 

 explanation. 



" It has already been shown that the currents rise on the 

 side next the flame, and descend on the other side. As 

 suggested by Mr. F. J. Pirani, during a conversation with 

 him on the subject, these circular currents will not be 

 parallel to one another, but will converge towards the flame.. 

 If we suppose the flask to be divided by a horizontal plane 

 passing through the centre of the flame, then it will be 



