228 MESSRS. R. THRELFALL AND J. A. POLLOCK 



out to us, with the result that he wrote to say he believed they were the three finest 

 thermometers ever made. We mention this to show that we really went to some 

 trouble in the matter, and did not take to platinum thermometry until it was actually 

 forced upon us. We desire to thank M. GUILLAUME and Mr. GRIFFITHS for the 

 trouble they took in the matter. The reason why mercury-in-glass thermometers fail 

 for our purpose is that they are too slow when of the necessary sensitiveness, that 

 they do not give the mean temperature of the thread, and that the zero corrections 

 present difficulties in field work. We are glad to acknowledge that it was M. GUIL- 

 LAUME'S opinion from the first that platinum thermometry would be the most 

 suitable for us. The instrument we at present employ is made simply according to 

 the instructions of Mr. CALLENDER,* but we have made one or two small modifications 

 in the resistance box which was constructed to work it. The platinum wire, of the 

 diameter recommended by Mr. CALLENDER, is wound in a double spiral on a very 

 thin glass tube, 1 centim. in diameter. The tube is mounted on an ebonite plug 

 through which the leads pass, and which serves to screw it in to the inner tube of the 

 balance. The tube is unsupported, except by the plug, and is arranged to lie parallel 

 to the thread, and at a distance of 2 centims. more or less below it, i.e., 2 centims. 

 from thread to centre of tube. 



When we first adopted the platinum thermometer we were of the opinion that the 

 temperature of the thermometer would lag behind that of the thread, but experience 

 has shown that the opposite is the case. When the temperature is rising to a maxi- 

 mum the maximum is always reached and passed some few minutes before the lever 

 reaches its highest point. 



The apparent slowness of the thread in taking up the temperature of the surround- 

 ing space may be due to the following circumstance. Though we are not in possession 

 of the complete mathematical theory of a stretched, weighted, and twisted thread, we 

 have found by experience that an increase of tension of the thread, as by tightening 

 the rosette spring, acts in the same way as an increase in the twist of the thread. 

 Consequently, if the spring takes some time to reach the surrounding temperature, its 

 effect will be the same as if the temperature of the thread itself were lagging, at all 

 events so far as the observer at the microscope is concerned. This appears as follows. 

 The action of a rise of temperature on the tension of the spring is complex. The 

 expansion of the bars increases the tension on the thread ; the decrease in the elastic 

 forces of the spring produces an opposite effect. On the whole the former is probably 

 the krger effect and therefore predominates. Consequently, the rise of temperature 

 will stretch the thread ; this will act like an increase of twist, hence less twist will be 

 required to keep the lever in its sighted position, but this is exactly the effect pro- 

 duced by the temperature increase in the rigidity of the thread. 



* 'Phil. Mag.,' July, 1891. 



