178 



H. V. Gill — Theory of Singing Flames. 



We have performed many experiments to verify the con- 

 clusions of Sondhauss, but we did not make use of a gas- 

 generating flask to produce the gas. We employed ordinary 

 coal gas, which passed into a flask* which was provided with 

 two other tubes (fig. 1\ one being in connection with the sing- 

 ing flame ; the other had one extremity 

 below the level of some water in the 

 flask, the other end being open to the 

 air, thus providing a means of measur- 

 ing the pressure in the gas. With 

 this arrangement we were able to ob- 

 tain all the conditions of a gas-gener- 

 ating flask, with the additional advan- 

 tage that we could regulate the size of 

 the flame, etc., with perfect facility. 

 It is evident that with a flame proceed- 

 ing from a flask in which hydrogen is 

 generated in the ordinary manner, it is 

 almost impossible to regulate the gas 

 supply with any exactness. All our 

 experiments tended to show that the 

 influence of the supply-tube came not 

 from its length, but from the facility 

 with which it allowed the gas to pass 

 to the flame. For with the same sup- 

 ply-tube we were able to obtain any 

 note, either high or low, which we desired, by modifying the 

 size of the flame, its position in 'the tube, and the length of the 

 tube in which it sang. 



In this paper we shall, we think, make it clear that the cause 

 which plays the important part in the production of a musical 

 note by the flame, is one whose effect has not been taken into 

 account by those who have examined this question. 



A brief review of the principal facts hitherto observed will 

 be useful to us in what follows : 



1st. The note produced depends on the length of the tube 

 inside which the flame sings, on the size of the flame, and on 

 its position within the tube. 



2d. The notes are those proper to the tube, account being 

 taken of the temperature of the air inside it. 



3d. The flame must be smaller when it begins than when it 

 is singing well. 



4th. The spontaneous commencement does not seem to be 

 an essential part of the phenomenon. 



* There is no advantage gained from this arrangement in producing the sing- 

 ing flame. The supply-tube may be connected directly with the gas main of the 

 house. 



