1919] on Clock Escapements 453 



— by Harrisoii's gridiron construction ; or that of Graham by the 

 expansion of mercury in the bob ; or, again, by the zinc and iron 

 combination used in many turret-clocks ; or, best of all, by availing 

 ourselves of the low expansion nickel-steel recently introduced by 

 Guillaume. xilso, for added security, the whole clock must be 

 enclosed in a thermostatic chamber, as is done by Prof. Sampson at 

 the Royal Observatory at Edinburgh. The other condition is much 

 more difficult. There is, besides the almost inevitable friction of the 

 escapement, the effect of the buoyancy of the air. This last can be 

 avoided by enclosing the whole clock in a glass case, tightly fitted, 

 in which the air can be slightly rarefied and maintained at a constant 

 pressure below that of the atmosphere. This would seem to offer a 

 very satisfactory solution of the difficulty. Temperature error and 

 buoyancy erroi* having thus been to a great extent mastered, we come 

 back to the forces connected with the maintenance and recording of 

 the motion as the principal sources of uncertainty. And let no one 

 suppose that little has been effected. Perfection in this, as in other 

 human pursuits, is doubtless unattainable, but we approach it 

 asymptotically, and we are farther along the asymptote than might 

 be imagined. Prof. Sampson tells us that in his thermostatic 

 chamber and barostatic cases, and with the Riefler, Cottingham, and 

 Synchronome escapements which he is studying, the errors average 

 no more than one-hundredth of a second per day — that is, at the 

 rate of one minute in sixteen years, if the clock could run so long 

 without stopping — truly an almost miraculous accuracy, unrivalled, I 

 imagine, in any physical measurement. Anyone, therefore, who 

 hopes to improve upon this has a difficult task before him. If it is 

 true that le mieux est Vennemi du bien^ it must be acknowledged that 

 Je mieux has against him a most formidable antagonist. 



[The lecture was illustrated by a number of working models.] 



[A. T. H.] 



