OF ARTS AND SCIENCES. 95 



ing. Tlie variation of the height of the menisetis from 1.2 to 1.4 mm - 

 would affect the reading only to the extent of .01 to .02 mm -. 



The only case where any correction for capillarity is needed is in 

 finding the temperatures of the steam at the 100° point, and will then 

 affect that temperature only to the extent of about 0°.00u. 



The scale of the instrument was very nearly standard at 0° C., and 

 was on brass. 



At the centre of the brass tube which surrounded the barometer, a 

 thermometer was fixed, the bulb being surrounded by brass, and there- 

 fore indicating the temperature of the brass tube. 



In order that it should also indicate the temperature of the barome- 

 ter, the whole tube and thermometer were wrapped in cloth until a 

 thickness of about 5 or G cra - was laid over the tube, a portion being 

 displaced to read the thermometers. This wrapping of the barometer 

 was very important, and only poor results were obtained before its 

 use; and this is seen from the fact that 1° on the thermometer indi- 

 cates a correction of ,12 mm - on the barometer, and hence makes a 

 difference of 0°.04 on the air thermometer. 



As this is one of the most important sources of error, I have now 

 devised means of almost entirely eliminating it, and making continual 

 reading of the barometer unnecessary. This I intend doing by an 

 artificial atmosphere, consisting of a large vessel of air in ice, and 

 attached to the open tube of the manometer of the air thermometer. 



The Thermometers. 



The standard thermometers used in my experiments are given in 

 the following table. 



