METHODS EMPLOYED IN CALIBEATION OF MEECUEIAL THERMOMETERS. 151 



the apparatus used for measuring the threads were sufificiently sensitive 

 to make the accumulated errors of thread-measurement insensible when 

 compared with the error of reading introduced when the thermometer 

 is used. 



The Committee have been able to study the results produced by 

 different instruments, as well as by different methods, as will appear from 

 the following restmie of the observations considered in the preparation of 

 this Report. 



Some time ago Professors Thorpe and Riicker obtained a number of 

 mercurial thermometers for the purpose of comparison with the air ther- 

 mometer. A full description of these instruments is given in Table I. :— 



Table I. 



Those numbered 561-2-3 were constructed and calibrated in the Kew 

 Observatory, and those indicated by the symbols O,, Oo, O3, in the 

 Physical Laboratory of the Owens College. In the case of the Kew 

 instruments, the method of calculation adopted was that introduced into 

 the Observatory by the late Mr. "Welsh, and explained by him in the 

 ' Report of the British Association ' for 1853, p. 35, which, as it is prac- 

 tically the same as that of Gay-Lussac, need not be further described. 



The Owens College instruments were furnished with a uniform scale. 



In addition to the construction of the thermometers, the authorities 

 of the Observatory and the Owens College undertook to perform, in 

 accordance with the directions of Professors Thorpe and Riicker, the 

 measurements necessary for their correction by Bessel's method. The 

 appai'atus used in each case was an excellent dividing engine. The 

 accuracy of the measurements and of the correction curves obtained is 

 evident from the fact that the difference between any one measurement 

 of the length of a mercurial thread expressed in terms of the corrected 

 scale, and the mean length of that thread, equals or exceeds 0°'01 C. (i.e. 

 about 0*1 m.m.), in eleven only out of a total of 880 observations made 

 in all on the six instruments. 



(8) The application of Bessel's method, therefore, to the Kew instru- 

 ments, afforded an excellent test of Welsh's method by which they had 

 been calibrated. The result was reported to Section A of the British 

 Association at York (Report 1881, p. 541), in a paper from which the 

 following is an extract : — 



