On the Limit of the Liquid State. 



297 



ings than those with wide bores. They were dried by passing dried 

 hydrogen through them for over two hours and keeping them warm 

 all the time. 



In order to obtain readings which would always be near the truth, 

 and be more independent of accidental errors, an apparatus was con- 

 structed with two manometer tubes, and the manometers were made 

 of tubes as wide as was consistent with the strain they were destined 

 to bear. The apparatus as used is shown in fig. 1, where the two 

 manometers are shown fixed in the two upright branches, while the 

 pressure screw is at the right hand, and the working tube at the left. 

 The air-bath has been drawn as though it were transparent, to show 

 the internal arrangement, The working tube is recurved, so that the 

 liquid to be experimented upon is contained between the mercury and 

 the sealed top of the tube. The air-bath consists of two cylindrical 

 baths with holes in the lids for passing the working tube through, 

 and an outside cover which keeps the heat from the lamp from being 

 too quickly radiated. The internal baths are supported by one of 

 Fletcher's solid flame burners, and the bottoms are covered by a layer 

 of non-conducting cement. The outside cover has openings at the 

 top for the escape of the burnt gases, and its top is covered with a 

 thick layer of asbestos wool, to prevent cooling. The whole of the 

 baths and cover were made of iron, as the high temperature used 

 caused copper to scale heavily. 



Two thermometers were used in the bath, one on each side of the 

 working tube ; and at first each thermometer had a little one fixed to 

 it for temperature corrections, but it was subsequently found that one 

 hung between the two gave quite as much accuracy. 



The two thermometers used were of soda-glass with cylindrical 

 bore, and registered the same temperatures to within o, 5, being- 

 chosen from a number. They were heated and cooled from 270° to 0° 

 over seventy times, and one sent to Kew, where it was compared with 

 the standard up to 100, and the stem calibrated and the corrections up 

 to 350 calculated. The zero points of both thermometers rose from 

 0°*2 to 2 0, 2 during the preliminary heating and cooling. The changes 

 were determined daily. The temperatures given in this paper may, 

 therefore, be considered practically correct. 



Two small thermometers were fixed to the manometers for tempera- 

 ture corrections. 



After trying several stands I found that shown in fig. 1 to be the 

 most convenient and steady ; it is simply a large block of wood with 

 a groove cut in it, in which the tube lies, the two upright arms 

 preventing any movement of the apparatus. The packing of the 

 joints has been described before, and I would only add that I find it 

 better to face the India-rubber packing with leather by fixing a piece 

 of fine soft leather to the face of the plug with india-rubber solution. 



