85 



portable as any other, I directed my attention to this object. 

 Finding that it was not practicable to enable so small a double 

 bellows to sustain incessant working without being soon worn 

 out, I discarded that plan ; and, recollecting the centrifugal 

 bellows as improved by Papin, in which fans revolving eccen- 

 trically in a circular tympanum, with a large opening in its 

 centre, were sufficient to create a current of air, I had several 

 such made in succession, and at length ascertained the smallest 

 that would answer the purpose. Papin's construction failed, 

 but by some alterations and additions it succeeded admirably 

 in giving an equal blast. With these fans I connected a very 

 small train of wheel-work, which, after many modifications, 

 imparted to them the exact velocity necessary for supplying 

 the proper ratio of air to the naphtha vapour. This machine, 

 actuated by a mainspring and maintaining power, afforded a 

 constant blast for eight hours ; it might then be wound up 

 without stopping the fans or extinguishing the light. 



" Many difficulties still arose, such as the necessity of ap- 

 portioning the diameter of the holes in the burner ; the capa- 

 city and number of the air passages in the burner; the length 

 of the burner, and its distance from the naphtha ; the number 

 and situation of the holes in the tube which distributed the 

 atmospheric air to the naphtha vapour. Various trials sur- 

 mounted all of them. 



" But a chief difficulty was to find a remedy for the conse- 

 quences of change of temperature in the apartment where the 

 lamp burns. If the temperature be much raised, the lamp will 

 smoke ; if it be much lowered, the light will become feeble ; 

 but if maintained without much change of temperature, the 

 light will not alter. Several methods were tried ; such as 

 cooling or heating the naphtha by increased or diminished 

 speed of the fans ; removing the supply of naphtha farther 

 from, or bringing it nearer to the source of heat, and other 

 minor expedients ; but none of them acted satisfactorily. 



«' A method was then adopted which proved successful. In 



