[ 128 ] 



XV. 



ON THE MEASUREMENT OF SMALL PEESSUEES. By PEO- 

 FESSOE GEOEGE FEANCIS FITZGEEALD, M.A., F.E.S., 

 F.T.O.D. ; and J. JOLY, M.A. 



[Read April 18, 1888.] 



The difficulty in measuring small pressures is not on account of 

 the smallness of the forces concerned : a pressure of the millionth 

 of an atmosphere is about the weight of a milligram per sq. cm., 

 and this on any considerable area is quite a large force compared 

 with the weights measured every day in balances. The difficulty 

 arises from its being so hard to make the surface pressed on freely 

 movable. We may use fluids in tubes — in horizontal or nearly 

 horizontal tubes, for instance — but the sticking of the surface of 

 the liquid and the tube, due to capillarity, which becomes serious 

 on account of the necessity for using small tubes, if the liquid is to 

 stand with free surfaces nearly at right angles to the tube, when it 

 is nearly horizontal, has seriously interfered with this method of 

 observation. Herr Toppler has recently used a liquid in the bend 

 of a tube bent at a very obtuse angle, and has observed the motion 

 of the surface by means of a microscope. This method seems 

 capable of considerable accuracy. The capillary sticking might be 

 entirely obviated by making the tube large. To whatever extent 

 it exists it would generally diminish the effect of pressure, and its 

 irregularity of action would introduce an unnecessary uncertainty 

 into the measurement. In the case of water, the difference of 

 level due to a pressure of the millionth of an atmosphere is the 

 hundredth of a mm., an amount that can be easily measured 

 by means of a microscope. A float on the surface with a fine 

 mark attached would be a convenient object to observe with 

 the microscope. The difficulty in this case is to keep the float 

 steady, but it could probably be anchored at the junction of two 

 crossed spider lines, or by suspending part of its weight by a 

 stretched spider line. With a pretty large float — and there is 

 no reason in many cases why it should not be large — there would 



