238 



Prof. J. Dewar. Coefficients of the Cubical [Apr. 16, 



air can now be obtained in many laboratories suggests that its 

 application to work of this kind would in some cases be a convenience, 

 and the present investigation was undertaken with the desire of 

 ascertaining what accuracy could be attained, and how the method 

 could be applied to inorganic or organic substances which occur in the 

 form of fine crystals. The use of a mixture of varying composition 

 and density like liquid air necessitates a determination of its density 

 with accuracy and rapidity before and during the course of the 

 experiments. For this purpose, in the experiments about to be 

 detailed, the liquid air that had been allowed to evaporate for twenty- 

 four hours in advance was used in large silver-coated vacuum vessels 

 of some 3 litres capacity. In order to ascertain the density of the 

 liquid, a polished silver ball, which had been weighed once for all in 

 liquid oxygen, was weighed in the sample of liquid air, and from the 

 relative weights thus found the density of the liquid air could be 

 approximately determined, assuming that of liquid oxygen to be 

 1'137.* To prevent any disturbing ebullition in the liquid-air flask in 

 which the weighings took place, and to reduce the rate of its evapora- 

 tion to a minimum during the course of an experiment, the substance 

 to be used was previously cooled in a supplementary vessel containing 

 liquid air and then transferred to the large flask. To avoid as far as 

 possible the formation of cracks in the bodies during the process of 

 immersion in the liquid air, it was found advisable to cool them slowly 

 in the air of the vacuum flask first, and then to lower them into the 

 liquid. In this way, with proper care and attention, results were 

 obtained comparable in accuracy with the density taken in liquid 

 oxygen. Substances like solid carbonic acid and ice were weighed in 

 the cool, gaseous air of the vacuum vessel, and their weights subse- 

 quently corrected for buoyancy. The temperature of the densest 

 and lightest samples of liquid air were ascertained by the hydrogen 

 thermometer, and that of the others deduced by graphic interpola- 

 tion. As the entire range of temperature through which the bodies 

 were cooled amounted to about 200°, a degree or two up or down 

 has no real influence on the results ; the extreme range of temperature 

 in the air samples was from 83° *8 to 86°*1 Abs. 



When the body to be examined was a salt, it was employed in the 

 form of a compressed block. One experiment was, however, made in 

 a section of a large crystal of chrome alum. The salt, previously 

 reduced to a fine powder, was moistened with water and compressed in 

 a cylindrical steel mould under great hydraulic pressure. During 

 compression the saturated salt solution drained away, and finally a 

 cylindrical block of some 50 grammes of the salt was obtained free 

 from porosity and hard enough to allow its surface to be polished. In 



* As the correction due to the contraction of the silver ball between the tem- 

 perature of boiling oxygen and that of the air sample is small, it may be neglected. 



