262 



dividual results from the formula is only Yi^th. of a degree (without 

 regard to sign). 



Difference from 



The agreement with M. Regnault's table is also extremely close ; 

 and considering the ordinary limits of error of such observations, the 

 writer considers it nearly indifferent for elevations under 13,000 feet 

 which method of calculation be used. 



The consistency of the results shows that the method of observa- 

 tion (which differs in some respects from that commonly used) and 

 the graduation of the thermometers were satisfactory. 



On carefully examining Dr Joseph Hooker's detailed results 

 (obligingly communicated by him), which that naturalist considered 

 to be incompatible with Professor Forbes's formula, it is shown that 

 the inconsistencies of observation are so considerable, that it is diffi- 

 cult to give a decided preference to one formula rather than another, 

 for the purpose of representing them ; but that up to heights of at 

 least 13,000 feet, a linear formula, or one which assumes the lower- 

 ing of the boiling point to be exactly proportional to the height, 

 seems to express the observations as well as any other ; and the rate 

 of diminution is almost the same as that deduced from Professor 

 Forbes's observation, or a lowering of 1° for 538 feet of ascent. 



The author has little doubt that M. Regnault's table (which was 

 not published when he last wrote) does really represent the law ac- 

 cording to which water boils more accurately than the simpler linear 

 formula, though the difference is in most cases insensible. For all 

 ordinary heights (or up to 12,000 feet) Regnault's table may be 

 more accurately represented by the formula 



h = 535 T. 



Where h is the height in English feet, T the lowering of the boil- 

 ing point in Fahrenheit's degrees, reckoning from 212°. But he 



