42 PROCEEDINGS OP THE AMERICAN ACADEMY 



The bulb of the air thermometer is introduced iuto the middle of 

 the copper vessel, and is surrounded by oil, which completely tills the 

 vessel. The stem of the thermometer passes through a groove pro- 

 vided in the upper surface of the vessel. The mercurial thermometers 

 are placed against the back of the vertical channel, with their bulbs 

 projecting into the oil-bath to the level of the air thermometer bulb. 

 The imoccupied space in the channel between the stems of the ther- 

 mometers and the sheet iron front is filled with cotton wool or other 

 similar substance, which may be removed at suitable points to take 

 readings. Auxiliary thermometers are placed with the bulbs at vari- 

 ous suitable heights along the stems of the other thermometers, and by 

 means of them the temperature of the vertical bar, and thus of the 

 thermometer stems, is maintained at a known temperature very nearly 

 the same as that of the batb, so that the stem exposure correction be- 

 comes small and determinate. The massiveness and correspondingly 

 hio'h conductivity and calorific capacity of the whole vessel render it 

 possible to maintain the temperature nearly uniform and constant 

 during the time necessary for the thermometers to acquire the tem- 

 perature of the bath and the readings to be taken. The oil-bath is 

 thoroughly stirred by a sheet-iron stirrer moving vertically. 



The air thermometer used was an imported instrument of the form 

 described by Jolly.* The glass of the bulb was supposed to be of 

 the kind used by him, and the coefficient of expansion used in my 

 computations was based on that which he has given, and was assumed, 

 with sufficient accuracy for the present purposes, to be, — 



t° Mean coeff. from 0^ to P. 



100 0.0000278 



140 286 



180 295 



200 300 



250 31 



300 32 



The mercurial columns were read by means of black hard-rubber 

 clasps sliding upon the tubes, the scale being upon the surface of a 

 glass mirror. Readings were taken to tenths of a millimeter. For 

 the temperature reductions of the mercury columns it was assumed 

 that the coefficient of apparent expansion of the glass scale was 



A = 0.000181 —0.000009 = 0.000172, 

 * Fogg. Ann. Jubelband (1874); Amer. Jour. Sci., vi. 591. 



