THE MEASUREMENT OF TEMPERATURE. 65 



SECTION I. 



Errors arising from faulty determination of temperature 

 are, I believe, much more common than is usually supposed 

 to be the case, and in an earlier number of this review 1 I 

 have called attention to some of the lamentable consequences 

 arising from such errors. True, the labour involved in the 

 complete standardisation of a mercury thermometer is ap- 

 palling. In addition to the necessary observations of the 

 accuracy of its " fixed points " it includes : — 



(1) A calibration of the bore. 



(2) The determination of the temporary changes of 

 zero and repeated observation of its permanent rise. 



(3) The estimation of differences caused by movements 

 from the horizontal to the vertical position. 



(4) Observation of the effect of changes in the external 

 pressure. 



(5) Determination of the differences resulting from the 

 rate of rise of temperature. 2 



(6) The application of the correction (at best but ap- 

 proximate) for any unimmersed portion of the stem. 



(7) The reduction from the mercury in glass scale to the 

 air scale. 



No man whose time is valuable can, as a preliminary to 

 some investigation, afford to spend months in the deter- 

 mination of such quantities, or, having done so, contemplate 

 unmoved the fact that a careless motion of the hand may 

 render all his labour of no avail. Let us suppose, however, 

 that all the above quantities have been successfully ascer- 

 tained, the labours of the observer are by no means ended, 

 for one determination of temperature is of itself a lengthy 

 operation. I can best illustrate this by an extract from my 

 note-book, of the observations and corrections in a single 

 case (Tonnelot's hard glass thermometer, No. 11,048). 



1 April, 1894. 



~ No accurate determination can be made by a mercury thermometer 

 whose temperature is falling (see Phil. Trans., 184 A, p. 442). 



