OF ARTS AND SCIENCES. 305 



Di(T. in Length 



1880, April 12, 



April 13. 



If the changes of temperature are gradual, the two steel bars will 

 reach a 6tate of rest under a constant temperature in about twelve 

 hours ; but if they are subjected to an abrupt and violent change of 

 temperature, it is not safe to make the comparisons till after the lapse 

 of from forty-eight to sixty hours. 



Baily's comparisons of the yard of the Royal Society certainly show 

 traces of an error of the kind I have here described. On the other 

 hand, Clarke's comparisons do not show any marked evidence of their 

 existence. But in the former case, the com paring-room was not well 

 adapted to the maintenance of a steady temperature, while in the lat- 

 ter it appears to have been admirably constructed. As observations 

 are still being made with the apparatus at Southampton, it would be 

 interesting to see if results corresponding in a general way with those 

 I have found could be obtained from a special series of comparisons 

 arranged for this specific investigation. 



An investigation of the character which I have indicated is espe- 

 cially necessary in the measurement of base lines in which the standard 

 unit is necessarily suhject to greater and more rapid variations of tem- 

 perature than take place in a well arranged and protected comparing- 

 room. Still further observations are necessary in order to determine 

 vol. xv. (n. s. vii.) 20 



