218 ON DIVERGING STREAMS 



exhibited while using these conical valves. It 

 became necessary to fasten a seat with a hollow 

 cone to the flange, and, in the experiments, the 

 issuing stream of air was made to pass between 

 the cone and its seat. But when this seat was 

 liberated from the flange, and the stream of air 

 suffered to flow, one stream rushed between the 

 cone and the seat, and another between the 

 seat and the flange. And thus the seat of the 

 cone was held in its situation by the two 

 streams of air, without being in contact with 

 any thing else. 



During the experiments, burning paper was 

 placed on the valves, that the flame and smoke 

 might shew whether there was any atmospheric 

 current rushing down upon it. But it was only 

 at the periphery that the flame was drawn down 

 until it came in contact with the stream of air 

 issuing from under the valve, which cut off" the 

 flame as abruptly as it could have been cut 

 through with a knife, apparently from its force 

 and coldness. On the valve the flame blazed 

 in the way in which it ordinarily does, when 

 there is no current of air acting upon it. 



In endeavouring to account for these pheno- 

 mena, it appeared, that the air in the aperture 

 was projected or driven from the aperture, as 



