469 



When the apparatus was arranged, the door of the room closed, 

 and the atmosphere in a quiescent state, it was found that a con- 

 stant regular rotation of the disc was established, and kept up by 

 the upward current of air through the apparatus, and continued so 

 long as the atmosphere of the room was quiet ; but that agitations 

 of the surrounding air either rendered the rotation uncertain, or 

 reversed it. 



Having thus ascertained that the current of air within the vertical 

 tube possessed sufficient force to cause the rotation of a lightly 

 suspended fly-disc, the question arose, what influence elongation or 

 shortening of the tube would exert on the velocity of the current. 

 For this purpose three tubes of precisely similar construction, but 

 with long limbs of 12, 24, and 48 inches respectively, were fitted as 

 before with fly- discs, and placed near each other in the centre of the 

 room. 



In nineteen observations, the number of revolutions in the tube, 

 with a long limb of 12 inches, varied from 0*75 to 4'5 per minute; 

 in that with a long limb of 24 inches, from 1'5 to 9'0, and in that 

 with the long limb of 48 inches, from 3' 75 to 14*0 per minute. 

 The gross number of revolutions in the three tubes, in the nineteen 

 observations, were respectively 51'25, 111'25, and 199'75; and the 

 mean revolutions per minute, 2' 697, 5'855, 10'513, which, allow- 

 ing for errors of observation, yield the ratios ] , 2, 4 nearly ; so that 

 it may be said that the velocity of the revolutions is in a direct ratio 

 to the lengths of the vertical tubes. 



The influence of the conoidal form of the tube being suggested 

 by Dr. Roget as worthy of investigation, a tube, 96 inches long by 

 3 inches diameter below and 6 inches above, was fitted to a rec- 

 tangular tube containing the rotating disc. Another tube of the 

 same length, 3 inches in diameter throughout, was placed near the 

 conical tube as a term of comparison. The revolutions of the disc 

 in the conical tube were more rapid than in that of uniform diameter, 

 in the proportion of 8'8 to 3'0. When the position of the cone was 

 reversed, and the entrance and exit orifices were equal, the revolu- 

 tions still continued more rapid than in the tube of uniform 

 diameter. 



To determine the influence of the area of the tube on the velocity 

 of the current, four tubes, 96 inches in length in the long, and 



