UPON GASES AND LIQUIDS. 535 



thin envelope. The original form is determined by the ascent of 

 the gases evolved from the wick. In the first and second figure, 

 the lateral repulsion, by expanding the flame laterally, produces 

 depression of it. If this action increases, the two outermost 

 apices become inclined upwards towards the equatorial plane, 

 just as the original flame does (PI. IV. figs. 3-5), whence in the 

 first two instances a depression is produced in the centre. But 

 when the poles are situated lower down, the flame, which derives 

 the gas by which it is fed from the wick, cannot leave the wick ; 

 an elevation then occurs in the centre, as seen in the fifth figure. 

 The elevated portion is here less luminous, for the same reason 

 that the lower part is so in the case of the flame of the common 

 wax candle. 



27. In most cases the flame is depressed, and consequently 

 increased, by the magnet. Both these effects are the results of 

 the repulsion produced by the latter. This repulsion however, 

 under altered circumstances, must be capable of producing a 

 diminution, as also an elongation of the flame. The former was 

 accidentally noticed in a small stearine flame, which, when the 

 apices of the poles were depressed considerably below the wick, 

 was in fact extinguished on the excitation of the magnetism ; 

 evidently because the gas which fed the flame, before arriving 

 at it, was repelled laterally. The flame was moreover dimi- 

 nished by the refrigeration arising from the apices of the poles. 



2S. When, on the other hand, the flame is completely above 

 the surface of the poles, and the direction in which it is repelled 

 by the magnet is directed perpendicularly upwards, the flame, 

 if our view be correct, should be elongated instead of being 

 shortened. To confirm this by direct experiment, the parallelo- 

 pipedal halves of the keeper (A) were laid on flat, and retained 

 at a distance of 3*5 millims. A common lamp wick was led 

 close above the middle of the upper angles of the two corre- 

 sponding rectangular surfaces of the poles, and then w ith both 

 ends between them into a vessel situated beneath, and filled with 

 alcohol. When the wick had become soaked, it was set light 

 to above, so that the flame was not diff'used over that portion of 

 the wick which had been conducted to the upper angles. The 

 flame, on closing the circuit, became higher. 



In this experiment the burning portion of the wick must not 

 be advanced from the centre of the upper angles to their extre- 

 mities, because the flame is then simultaneously repelled laterally. 



