32 Messrs. C. W. Waidner and F. Mallory on 



termined. But a scale of temperature defined in this way 

 would vary with each different specimen of glass used, and 

 would not be exactly the same even for the same kind of 

 glass. The mean scale of a number of French hard-glass 

 thermometers kept at the Bureau International, which have 

 been compared directly with the gas-thermometers (H, N, 

 and C0 2 ), defines temperature on the French hard-glass 

 ("verre-dur") scale. The slight deviations of the hard-glass 

 scale of any other Tonnelot thermometer which is compared 

 with these standai'ds is observed, and these corrections serve 

 to reduce the indications of this thermometer to the mean 

 French hard-glass scale. 



Method of Comparison. 



The Tonnelot thermometer was adjusted with its stem in a 

 horizontal position in the comparison-tank, and with its bulb 

 near the centre of the inner brass box. The micrometer- 

 telescope was then adjusted by means of a level, so that its 

 axis of collimation was perpendicular to the stem of the 

 thermometer. When these adjustments had been made, no 

 certain difference could be detected between observations 

 made with the divisions in front of the stem and those made 

 with divisions back of the stem. The glass stem of the 

 platinum thermometer passed from the outside through the 

 sides of the wooden box and comparison -tank into the brass 

 box, where its coil was almost in contact with the bulb of the 

 mercury thermometer. Before taking an observation the 

 top and bottom of this box were opened, and the entire 

 contents of the tank thoroughly stirred ; the lids were then 

 tightly closed, and a series of observations taken. A single 

 series included a setting on the division below the meniscus, 

 on the meniscus, and on the division above the meniscus, 

 repeated three times, but in inverse order. No certain 

 variation of temperature was ever detected during the time 

 required to take a series of observations. The box containing 

 the bulbs was then opened, the water again thoroughly 

 stirred, the box closed, and a similar series of observations 

 taken. If the temperature of the water in the tank was 

 below that of the room, after each stirrring there was a 

 slight rise in temperature : if the temperature of the water 

 in the tank was above that of the room, there would be a 

 slight decrease after each stirring : steam was therefore 

 passed through the heating-coil to make sure of a rising- 

 meniscus. Usually four such series of obseiwatiens were 

 taken at each temperature. Immediately after the com- 



