72 SCIENCE PROGRESS. 



temperature than the central portions, must be a severe strain 

 on the wire. It is also evident that any crack or flaw in 

 the surface will tend to be intensified by the local develop- 

 ment of greater heat, and if the wire is heated to a tempera- 

 ture near its melting point where it begins to be appreci- 

 ably volatile, this action must inevitably produce serious re- 

 sults. If a wire which has been thus treated be examined 

 under the microscope its surface will generally be found to 

 be cracked and scored in a manner which is in itself amply 

 sufficient to account for the increased resistance and 

 brittleness." 



With the exception of objections based on the apparent 

 complexity of the instrument (which I deal with later) the 

 above are the only serious adverse criticisms which it is 

 necessary to notice. 



Before adducing the evidence on the other side, it would 

 be as well to briefly explain the terminology which has been 

 found to be convenient. Callendar in 1887 ((3), p. 163) 

 suggested the use of the term "platinum temperature " to 

 denote the reading on a scale so constructed that a rise of 

 i° of that scale would at any temperature increase the re- 

 sistance of a platinum wire by ihu of the difference between 

 its resistance at ioo° and at o°C. Hence if R is the resist- 

 ance of a platinum wire at a temperature / (measured in 

 degrees Centigrade by the air thermometer), and if R x and 

 R are the resistances of the same wire at 100 C. and o°C. 

 respectively, and if pt stand for the platinum temperature, 

 then 



R- R 



*' = rt^r; x io °- 



In the same paper Callendar communicated to the Royal 

 Society the results of an exhaustive series of comparisons 

 of the resistance of a platinum wire and the temperature as 

 indicated by the air thermometer in whose bulb it was 

 wound. No one can, I think, rise from the study of this 

 important communication without admitting that the experi- 

 mental evidence produced was sufficient to establish the 

 following conclusions (p. 161, ibid.): — 



(a) The self-consistency of the platinum thermometer 



