142 Dr. E. J. Mills on Thermometry. 



to be supposed, the upper plate was negatively charged on its 

 lower surface, with which it lay on the inductric layer, and 

 had therefore to be inverted in order to give similar discharges 

 to those of the first plate. 



If we fix our attention specially on the upper ebonite plate, 

 we find: — 



(1) That the effect of induction through it began almost 

 instantaneously (for immediately after having laid this upper 

 plate on the excited lower one I could put on, lift, and dis- 

 charge the cover) ; and that this effect of induction had disap- 

 peared also almost instantaneously when I tried to use the 

 plate alone for the electrophorus. 



But, further, we have seen (2) that slowly another inherent 

 change took place : a penetration of electricity occurred which 

 made the plate more duringly effective as an electrophorus- 

 disk. 



XIV. Remarks on Thermometry. 

 By Edmund J. Mills, D.Sc, F.R.S.* 



THE July number of this Magazine contains an expression 

 of the desire of Professors Thorpe and Rucker for further 

 information on certain points contained in a memoir by myself, 

 and recently published by the Royal Society of Edinburgh. 

 In view of the increasing interest now taken in thermometry, 

 I am willing to comply with their request ; but it is only just 

 to state that their object would have been gained at a much 

 earlier date by a private communication, instead of the much 

 longer and indirect method of addressing me in the pages of 

 this Journal. 



1. The Exposure Correction. — The results of very many ex- 

 periments with four thermometers have led me to propose a 

 new formula for this correction, viz. 



y=(«+^N)(T-<)N, 



instead of the ordinary one 



2/=«(T-0N, 

 where u = '0001545 ; for if we regard a as an unknown 

 quantity, and determine it by actual measurements, we shall 

 find that it is a linear function of the exposure. Both a and 

 (8 are found to depend in part on the individual thermometer. 

 If we put #=a + /3N, it is easy to calculate when x (i. e. the 

 total correction factor) agrees first for two or more ther- 



* Communicated Iby the Author 



