of Water at High Temperatures. 119 



in the last series, so that they may be repeated without obstacle 

 to success. 



§ 4. The difficulties that beset the inquiry come from three 

 sources. 1. The glass tube that contains the water must be 

 vertical ; and, to prevent the liquid column from breaking up by 

 the formation of vapour-bubbles below its summit, the tempera- 

 ture should be an increasing one upwards, whereas the natural 

 gradient of a simple heater is a decreasing one upwards. 2. 

 The hard German glass, though more impervious to the solvent 

 power of water than soft glass, begins to whiten and cloud below 

 300°, and generally becomes mottled with opake patches just at 

 the very part of the tube where clear vision is indispensable. 

 There is also a certain amount of the liquid absorbed, apparently, 

 but which may be partially recovered by heating. 3. To obtain 

 certainty as to the mean temperature of the liquid column, it 

 was thought proper to use as thermometer a tube of the same 

 length as the water-tube, filled with a liquid of great expansi- 

 bility and high boiling-point. To graduate such thermometers 

 by comparison with ordinary bulb-thermometers, it was neces- 

 sary to have an open bath of a similar liquid and transparent in a 

 glass vessel sufficiently deep to contain the tube held vertical. 

 The fumes of such liquids heated towards 300° become over- 

 powering, and reliable comparisons can hardly be obtained 

 above 250°. 



§ 5. The ordinary chemical thermometers, having a vacuum at 

 the top of the stem, are not to be depended upon at temperatures 

 near the boiling-point of mercury, as vapour is sometimes gene- 

 rated below. Besides they are graduated so as to require the 

 stem to be of the same temperature as the bulb. I therefore 

 had one made with short stem and bulb strong enough to with- 

 stand high pressures, containing air over the mercury. It was 

 about the same length as the water-tubes, and 100° measured 

 about |ths of an inch. After being heated to 400°, it was gradu- 

 ated and marked off in divisions of 10°, and did not alter in its 

 reading afterwards. It is represented by PI. III. fig. 7. Another 

 two of a similar kind were also provided ; and, each being gradu- 

 ated by itself and afterwards compared with the others, the highest 

 temperatures would have been well determined if it depended only 

 upon thermometers. 



§ 6. Graduating tubes. — Fig. 3 (PI. II.) is a full-size represent- 

 ation of the last of the six water-tubes employed. It was pre- 

 pared by first cleaning out, before sealing, with sulphuric acid, 

 then by distilled water, then dried by heating. The bottom 

 was then closed at the flame, and allowed to remain in it until 

 the glass thickened. The top was then sealed before filling, 

 employing a very small Bunsen-flame. In operating with 



