314 Theory of the Pneumatic Paradox. 



diameter. On blowing through the tube, strong jets issued from 

 them, as in similar cases already described. To one end of the 

 tube was then soldered, in such a manner as to offer no obstruc- 

 tion to currents of air, a diverging conical tube, as represented in 

 the accompanying figure, pj„ 7_ 



of which the length was 

 about eight inches; the 

 diameter of the smaller 

 end was of the same diameter with the cylindrical tube, and that 

 of the larger end, one inch and a half On blowing through this 

 compound tube, the flame was drawn into lateral holes made at 

 intervals in the conical part, and in the half of the cylindrical 

 part next to it, although previously to joining the two tubes it had 

 been blown from alt the holes in the cylindrical one. It should 

 seem, that in consequence of the rarefaction produced in the con- 

 ical part, the air in the cylindrical part, meeting with diminished 

 resistance to its progress, rushed forward by virtue of its elasticity 

 and tendency to equilibrium with a velocity much greater than 

 that with which it was expelled from the mouth, and thus became 

 rarer and rarer as it proceeded, and more so than the external air. 

 Analogous to this effect is the well known fact, that the addition 

 of a diverging conical termination to a tube conveying liquids, 

 greatly increases the discharge in the open air, though not in an 

 exhausted receiver. If the cylindrical part of the compound 

 tube be only one eighth of an inch in diameter, flame will not be 

 drawn in at any part of it except very near the place of junction, 

 nor will it be much affected in any way when held near lateral 

 holes in the conical part, more than two or three inches from the 

 same place ; within that distance it will be drawn in. 



If we conceive the conical part of the compound tube to be 

 flattened in such a manner that the opposite sides may be plane 

 and parallel, and to be soldered at the small end perpendicularly 

 to the end of the tube, the interior will represent a portion of the 

 space between the disks, (the whole of which may be considered 

 as composed of a series of flat diverging tubes radiating from a 

 common centre,) with the whole of the tube attached. If we 

 blow through the tube, the results will be similar to those before 

 described, except that lateral jets will escape from all holes in the 

 tube, in consequence of the obstruction the blast meets with at 

 the angle of junction. 



