TRANSACTIOXS OF SECTION A. 



531 



•with more refined apparatus, it was decided to reserve the publicatiou until later. 

 These results, which were obtained from our preliminary experimeuts, stand in 

 close agreement with tlie authors' later results, the error being less than one part in 

 1,000, although carried out with dill'ereut electrical standards and a calorimeter 

 dift'ering materially in design from the one recently employed. 



The present series of experiments was made with a calorimeter in every way 

 similar to the ' water ' calorimeter described by Professor Calleudar and one of 

 the authors,' with the exception that the How tube was smaller in bore. The rest 

 of the apparatus was essentially the same with a few minor changes. 



Preliminary Results. 



These results, obtained in 1897, we reproduce here through the courtesy of 

 Professor Oallendar, who was one of the observers at that time. They were ob- 

 tained with jacket circulation system, without temperature regulator, so that 

 observations talcen at the temperature of the laboratory ouly could be successfully 

 carried out. The calorimeter employed had a flow tube 1 metre long and 

 1 mm. in diameter coiled in a spiral as described by Professor Callendar.- The 

 thermometers were pair A described by one of the authors.^ The Clark cells were 

 S, and S.,, and the resistance standard was one of the manganin coils described iu 

 section 3 b, siijira. 



The only three determinations made at room temperature are as follows : — ■ 



Recent Results. 



The later results were obtained with improved apparatus. The calorimeter 

 had a flow tube about 1 mm. in diameter and 45 cm. long. The thermometers 

 were pair E. 



The mica frame resistance standards and Clark cells X., and X j were used. 

 Redeterminations were made of the value of the resistances iji terms of the 

 standard platinum-silver coils in the possession of the laboratory, which showed 

 that they had remained unchanged to one in 100,000 since the series of tests taken 

 for the measurements for water, and already described. The Clark cells were 

 found to have lowered a little in value since 1900, and a cori-ection was necessary 

 in order to reduce to the mean of a number of new cells. The histoi-y of these 

 original crystal cells, made in 1895, to which X., and X,, belong, shows that from 

 comparisons with new cells made from time to time they have steadilv lowered 

 in value. A new comparison was also made to several Weston cadmium cells. 

 There is no reason to doubt that the values assumed for the electrical standards 

 are the same as those assumed in the measurement of the specific heat of water, to 

 much better than one in 10,000. The thermometers were also recalibrated severa 

 times, as well as further determinations made of the box coil corrections. Hence, 

 even assuming that any error exists for the values assigned to the units used in 

 the calculations for water, the specific heat of mercury may be expressed from 

 these measurements in terms of water, and any possible error eliminated from the 

 relative value. Since also the same apparatus has been used, and as far as possible 

 the same experimental arrangements, systematic sources of error in the method are 

 to a large extent avoided. 



The absolute values of the thermal capacity of mercury are given in the 



' Brit. Assuo. Rep., 1898. 

 Phil. Trans., vol. cxcix. (1902). 



? Jbid.^ vol. 9XCix. p. 182 (1902), sec. 3 c. 



MM 3 



