30 Trnn'<arfio)!s. 



The temperature of tube and contents was controlled by a weak ice- 

 and-salt solution in a felt-covered glass jar. The Beckmann thermometer 

 was fitted by a stretched rubber coverino-tube into a tapered hole, 13-15 mm. 

 diameter, in a circular brass cap C, 62 mm. diameter, having a square top 

 and screwing into a female screw on the head B, provided also with a recess 

 containing a lead washer engaging with the flange projecting from the top 

 of B. 



The zero-point of the thermometer was found by inserting the bulb 

 in distilled water contained in an ice sheath prepared by freezing distilled 

 water round a Jena-glass tube and releasing the latter by warm water. A 

 normal thermometer reading to (>02°C., and known to be without appre- 

 ciable error throughout the range employed, was treated similarly, and the 

 two carefully compared up to about 5° C. in a continuously stirred water- 

 bath of rising temperature. 



The pressure-effect on the thermometer was considerable, and for the 

 first thermometer used was not satisfactorily found, owing to its accidental 

 destruction. Hence the first series of twenty experiments, otherwise per- 

 haps the best, are not given here. For the second thermometer the pressure- 

 effect was determined by filling the experimental tube with mercury, raising 

 the pressure to the maximum, and taking simultaneous readings on thermo- 

 meter and manometer as the pressure was progressively reduced. The curve 

 plotted from the results is nearly a straight line. It is assumed that this 

 line can be continued backwards to a point of no pressure. 



On setting up the apparatus all air was driven out of the connecting 

 tubes by a stream of carbon-dioxide, and out of the freezing-tube by satu- 

 rating the contained water at 80 lb. pressure and removing the pressure. 

 The gas evolved from the water in four or five such operations completely 

 expelled the last traces of air. In each experiment, to insure saturation,, 

 the freezing-tube A was surrounded by ice-water, and the contained water 

 stirred continuously some 15 to 30 minutes at a pressure about 10 lb. above 

 that at which readings were to be taken. The gas was then allowed to 

 escape until bubbles were evolved from the water. When, under continuous 

 stirring, the manometer no longer rose, owing to gas evolved from solution, 

 the temperature was reduced by a freezing-bath some 5° to 7° below final 

 freezing-point. The formation of ice (in flakes throughout the solution) 

 was followed by a sudden change in the click of the stirrer and by a rapid 

 rise in the Beckmann thermometer. With the first thermometer used the 

 temperature usually came to a final steady value, while the pressure slowly 

 rose after freezing had progressed a little, owing to the gas frozen out of 

 solution. 



In the experiments taken as trustworthy the rise of pressure during 

 freezing never amounts to more than 1 cm. of mercury. The pressure used 

 in calculation is that shown during the steady period of thermometer, or an 

 average of the pressures recorded if they varied for that period. 



Skrh<:s a, wnii Oiucixal Thermometer (20 Experiments). 



Results are of (jualitative value onlv, as, owing to an explosion, tlie 

 pressure-eft'ect on thermometer was not accurately determined. 



Details of one complete experiment are given to show the method en\- 

 ployed throughout the work. In Series B and Series C tables of corrected 

 results only are given. 



