1890.] and Freezing Points by the Platinum Thermometer. 221 



this coil to the top of the instrument, and the unimmersed portion of 

 the stem was surrounded by the outer tube of a condenser, and 

 kept at a constant temperature by a flow of tap- water. The resist- 

 ance of this stem was so small that the change in resistance caused 

 by the changes in the temperature of the tap-water might be 

 neglected. 



The diameter of these thermometers was less than -fa of an inch, 

 and their length about 18 inches. They were extremely sensitive, 

 and could therefore be used to trace the rise in temperature due to 

 suffusion, the freezing points of the metals experimented upon being 

 determined by the limit of this rise. 



These thermometers were graduated by the temperature of the 

 boiling points of water, naphthalene, benzophenone, and sulphur, and 

 the freezing point of water. 



The values obtained by Crafts (' Paris, Soc. Chim. Bull.,' vol. 39) 

 were used in the .case of naphthalene and benzophenone, and 

 Regnault's value of the boiling point of sulphur.* The purity of 

 the samples was ascertained by fractional distillation and by the 

 temperature of the melting points. The selection of these tempera- 

 tures was forced upon me by the results of my experiments, and the 

 reasons for their adoption are fully given in the paper. 



The results were plotted in the manner suggested by Callendar 

 (' Phil. Trans.,' A, 1887), and on a scale such that a difference of 

 0'02 could be read with certainty. 



The curves thus obtained differed considerably from each othei 

 and from the curve given by Callendar.t However, intermediate 

 temperatures deduced from these curves showed remarkably close 

 agreement. 



In no case (the total number of experiments exceeds 300) is the 

 divergence of any one experiment from the mean value obtained 

 from all the thermometers as great as 0'2, and if only the results ob- 

 tained from thermometers E, F, and G be taken, the divergence is in 

 no case as great as 0'05. 



The chief difficulties which presented themselves were 



(a.) Variations in the resistance of the connexions between the 

 thermometer coil and the resistance coils. 



(6.) Variations in the temperature of the resistance coils them- 

 selves. 



(c.) The rise in temperature of the thermometer coil due to tho 

 current used when measuring its resistance. 



* Boiling point of naphthalene (760 m.m.) = 218'06. 



benzophenone = 306 0- 08. 



sulphur = 448'34. 



t A discussion of the probable cause of these divergences is given in an appendix 

 to the paper. 



