called " the Amphisbcena." 459 



the atmosphere at the other. A thread of quicksilver occupies 

 a portion of the tube, the space between the quicksilver 

 and the closed end being occupied by air. A uniformly 

 graduated scale, of any convenient dimensions, but ivhose zero 

 point is coincident with the closed end, is attached. Since there 

 is no escape for the air contained between the quicksilver and 

 the closed end, this scale serves to determine the volume of 

 this air at any particular moment. The instrument can be 

 suspended vertically with the closed end either upwards or 

 downwards, and in these positions the volumes of the cushioned 

 air, as indicated by the end of the quicksilver thread, can be 

 ascertained. It is clear that when the closed end is upwards 

 this air is subject to a pressure which is the difference between 

 the atmospheric pressure and that due to a column of quick- 

 silver equal in height to the length of the thread. But when 

 the closed end points downwards, the pressure of the cushioned 

 air is the sum of the atmospheric pressure and that due to the 

 length of the thread. 



These two observations, then, enable one to find the relation 

 between the atmospheric pressure and that due to the thread 

 of quicksilver. 



Let H be the unknown height of the barometer ; 

 I be the length of the thread of quicksilver; 

 A be the reading of the volume of air enclosed, when 



the closed end of the tube is upwards; 

 B be the reading of the volume, when the closed end 

 is downwards. 



Then the product of the pressure and volume being the 

 same in the two cases, assuming the temperature constant, 



(H-Z)A = (H + 0B, 

 or 



XT A + B 7 



Thus the following rule applies : — Divide the sum of the 

 readings by their difference, and multiply by the length of 

 the quicksilver thread. 



Since the length of the quicksilver occurs as a factor, it is 

 obviously advantageous that it should be a simple number, as 

 10 inches. Where centimetres are required, 25 centimetres 

 or 50 centimetres would be convenient quantities. 



Here it may be pointed out that the usual correction for 

 temperature may be made once for all, by having the thread 



