Construction of Platinum Thermometers. Ill 



very largely. All the specimens I have tried agree very 

 closely with BenohVs formula. I am inclined to think that 

 the Siemens formula must be wrong. It is at least note- 

 worthy that two of his observations differ by 6 and 3J per 

 cent, respectively from his formula in the direction of agree- 

 ing with Benoit's, and that if allowance is made for the 

 probable errors of the mercury thermometers at 350° the 

 discrepancy may be still further reduced. The resistances appa- 

 rently were only measured to about 1 per cent, in most cases, 

 and the temperatures are given only to the nearest degree. 

 The Siemens formula is undoubtedly superior to that given 

 by Matthiessen, but the evidence in its favour is hardly suffi- 

 cient to justify its continued use in preference to the much 

 simpler and more convenient formula (d). 



The superior capabilities of platinum thermometers as 

 compared with instruments in general use do not appear as 

 yet to be sufficiently appreciated. It will perhaps help to 

 bring out more clearly several of the points mentioned in 

 their construction, if I give a brief summary of some of the 

 advantages which they present in point of range, constancy, 

 and sensitiveness, as compared with the best mercury 

 thermometers. 



The range of a mercury thermometer is obviously limited 

 to temperatures between —40° and +400° C. That of a 

 platinum thermometer may extend from absolute zero to above 

 1500° C. 



The superiority of the platinum thermometer is still greater 

 in point of constancy. It is a fact familiar to those who have 

 studied mercury thermometers, that if a new thermometer be 

 kept at a temperature of 350° 0. for a week or so, its zero 

 will be found to rise by 10 c or 20° owing to the contraction 

 of the glass. After a time this rise reaches a limit ; but if a 

 thermometer which has been thus annealed and allowed to 

 cool slowly be again heated to 350° for a few minutes and 

 allowed to cool rapidly by free exposure to the air, its zero 

 will be found to be depressed by a quantity which varies with 

 different thermometers, but which generally amounts to 

 about 2°. A similar depression, but less in amount, is ob- 

 served if the thermometer is heated to some intermediate 

 point of the scale. The extent of depression also depends on 

 the time during which the thermometer is heated, and on the 

 rate at which it is cooled. Thus even in the best mercury 

 thermometers, which have been carefully annealed by a 

 special process, the position of the zero is constantly shifting 

 in a manner which depends on the past history of the instru- 

 ment. It is consequently very difficult, and in many cases 



