48 



PROCEEDINGS OF THE AMERICAN ACADEMY 



one division of the scale, although through the failure to do so the 

 investigation lucks completeness. 



The correction for the temperature difference between the stem and 

 bulb, i. e. the " stem exposure correction," was made through the use 

 of the customai-y expression, 



n {T—t)k, 

 where 



T =. the temperature of the bulb, in these measurements given by the 



air thermometer. 

 t = the temperature of the exposed stem, 



n = the length expressed in degrees of the exposed alcohol column, 

 k = the coefficient of apparent expansion of the alcohol in the glass 

 of the thermometer. 



The value of k was measured in two ways : first, by an indepen- 

 dent study with the alcohol bulbs in ice and the projecting stems at 

 measured temperatures; secondly, from the results in the comparison 

 with the air thermometer at about 0° C. These gave results in sub- 

 stantial agreement, but the second method was considered more relia- 

 ble, and likely to aid in eliminating constant errors in the comparison 

 since made under the same conditions, and its value of ^ = 0.00091 

 was adopted. 



The readings of the thermometers in ice were taken at intervals, to 

 follow the changes with age and usage, and the instruments were never 

 raised to a temperature above that incidental to the manipulation. 

 The zero point of 8209 rose from — 0°.18 C, in October, 1880, to 

 -^0°.16 in April, 1882. The change in 8208 was probably about 

 the same, but the data do not extend over the same time. 



The following are the mean values obtained by averaging in suitable 

 groups the eighty readings obtained in the comparison with the air 

 thermometer. 



