328 MR. CLIVE CUTHBERTSOX OX THE 



difficulty in two hoxirs. The tubes, which were about 37 centims. long, were placed 

 in the axis of the furnace, which projected about 10 centims. beyond each end. 



The next difficulty which threatened to be serious was that due to the shimmering 

 produced by currents of unequally heated air. The distance between the two mirrors 

 of the refractometer cannot be indefinitely increased ; and the gradient of temperature 

 between the furnace and the air outside must necessarily be steep. In fact, the 

 whole fall of temperature occurred in a distance of about six inches. It appeared 

 probable that the mirage effects would make it impossible to read the bands. This 

 danger was, however, completely met by a device previously used by M. LE Roux. 

 In each end of the furnace three diaphragms of ordinary plate glass were inserted, 

 separated by distances of about an inch, and the internal cylinder of the furnace was 

 prolonged by a paper tube four inches long, which deflected the air currents. By this 

 means all flicker was avoided, and, during an experiment, the bands appeared to pass 

 perfectly steadily, even when the temperature was at its highest point. So far as 

 this cause of error was concerned, there was no difficulty in reading the bands 

 to yjy of their breadth . 



Having determined the form of the apparatus, the next point for consideration was 

 the means to be adopted for estimating the amount of vapour present in the tube. 



Mercury was the first substance upon which experiments were made, and, since its 

 vapour pressure at various temperatures has been determined with considerable 

 accuracy by RAMSAY and YOUNG* and others, the plan originally tried was to 

 measure, by means of thermometers, the temperature of the coldest part of the tube, 

 and to assume that the density of vapour throughout the tube was that which 

 corresponded to this temperature. Numerous experiments were made with this 

 method, but the results were not satisfactory. The errors due to the calculation of 

 the appropriate stem correction, and to the difference of temperature between the 

 outside and the inside of the tube, were sufficient to cause a variation of 25 per 

 cent, in the results, and the method was consequently abandoned. 



The alternative plan, next adopted, was to weigh a quantity of the substance 

 sufficient to cause a convenient number of bands to cross the field, to introduce this 

 into the tube, and then to heat it in a vacuum until the whole amount had passed 

 into the form of vapour. This method proved far superior to the other, and was 

 perfectly successful. All measurements of temperature and pressure are avoided, 

 and the density of the vapour is calculated from the weight of the substance 

 vapourised and the volume of the tube, both of which can be measured very 

 precisely. 



Before stating the results obtained for different substances, it will be convenient 

 to describe the dimensions of the apparatus and the details of the procedure adopted 

 in an experiment. 



The tubes employed in the refractometer were of moderately hard glass, 36 to 

 * RAMSAY and YOUNG, 'Trans, of Chem. Soc.,' vol. 49, p. 37, 1886. 



