112 



being both accurately ground together, no air can possibly remain 

 between them, so that each time the piston is drawn up, a va- 

 cuum is left behind it until it passes the receiver-hole, when a fresh 

 portion of air must expand from the receiver into the barrel, this 

 is expelled as before, and so the exhaustion must proceed without 

 limit, at least as far as the expansive power of the air will per- 

 mit. 



I have not had time since the instrument was finished, to deter- 

 mine accurately its powers, but I know them to be considerable : 

 When the valve is disengaged from its frame (and of course no 

 longer opened mechanically) it still performs pretty well, bringing 

 the mercury in the barometer gage to within f of an inch of the 

 barometric column, although the valve requires some force to open 

 it against the pressure of the atmosphere ; but in its ordinary con- 

 dition, (attached to the frame) the difference is only tV of an inch, 

 and on a dry day I have seen it less. Very few air-pumps exceed 

 this, and even these do not without precations of manipulation, 

 which, if used in the present instance, would encrease its power. 



In one respect, only, I am not perfectly satisfied with it. In the 

 ascending part of the stroke, the mercury sinks in the gage, and 

 rises again as soon as the piston passes the hole, and these oscil- 

 lations in some experiments may be inconvenient. 



They are however only troublesome at the commencement of 

 the exhaustion, diminishing with every stroke of the pump, and 

 therefore do not produce any serious annoyance. But I ha\e de- 

 vised a means of preventing them, which is also useful on another 

 account, as it makes the i/p as well as the downstroke, efficient in 

 exhausting from the receiver, and the instrument becomes equiva- 

 lent to a double barrelled pump. This improvement consists in 



