'GO 



3^ 



60 Mr. A. R. McLeod orc the 



Curves are plotted in fig. 2. 



Fig. 2. 







LA 



GSF 



3H A 



LC0HOL i 



sylu 



4DEF 



t. 



Fig-2. 







































G=+ o 



2.8 







&'=»185. 































\ ' 



) 



3 







Time ia 







Seconds 



































g' 









G=o 



~* ; 













































90 120 



Time m Seconds. 



Hence from these values and the curves in figs. 1 and 2, 

 we see that cylinders of the dimensions given are much 

 superior to spheres, and the alcohol cylinder is the best of 

 the four *. The times taken for the temperature to drop from 

 5° C. (when £ = 0) to within a given fraction of the steady 

 lag are approximately proportional to the values of the 

 steady lags, so that the steady lags alone will suffice to 

 compare the rates of response of two thermometers to out- 

 side temperature changes. For example, the times taken 

 to drop to half the steady lags are respectively 100 s , 112 s , 

 67 s , and 44 s nearly. 



The effect of the glass, will be to increase the value of c. 

 If the thickness of the glass wall is a small fraction of the 

 inner radius, which we have taken as c, the correction should 

 be a fraction of the same order, and the lags will be similarly 

 affected. 



At the point of inversion, the lag changes continuously 

 from its value at time t 1 to the steady value on the new 

 gradient. Hence the greatest error in the thermometer 

 reading cannot exceed the greater of the two steady lags. 



* See however para. 5. 



