the Thermal Unit. 453 



If it be thought that the advantage of adopting what I 

 believe is termed a "round number''' as a multiple of the 

 absolute unit is more than counterbalanced by the selection 

 of so low a temperature as one near 10° for the standard 

 therm, I would suggest the following alternative proposals: — 



(I.) That 41*89 x 10 6 ergs be termed a " Rowland." 

 (II.) That (pending the results of further investigations) 



we assume that a therm at 15° is a standard 



therm. 



(III.) In order to enable observers to express their results 

 in terms of a standard therm, we (pending the 

 results of further investigations) assume the validity 

 of the following expressions over the temperature 

 ranges indicated: — 



Range 10° to 15°, 



Qi5=4{i+ -000414(15-0}; 



Range 15° to 20°, 



Q 15 =a{l-'000284(*-15)}. 



I think it improbable that the error introduced by propo- 

 sition II. exceeds 1 in 800 in either case ; at all events 1 

 believe it to be as close an approximation as the present state 

 of our knowledge permits. 



An inspection of Plate VII. will show that the actual mean 

 of all the selected observers at 10° and at 15° is as follows: — 



At 10°. At 15°. 



Assuming Joule (R)* (41-971 + -023) xlO 6 ; (41*891 + "023) x 10 6 

 „ Joule (S)* (41-958 + -029) x 10 6 ; (41-875 + -029) x 10 6 



but I repeat that greater importance should be attached to 

 the results of Rowland's experiments than to those of other 

 observers. Hence I selected 42 at 10° and 41'89 at 15° as 

 the most probable values. 



With regard to proposition (III.) our position is more un- 

 certain, but when we remember the entirely dissimilar method 

 of experiment adopted by the observers upon whose results 

 these formulas are based, the comparatively close agreement 

 over these ranges carries weight. 



Another subject to which the attention of the Committee 

 should be directed is the determination of the relation between 

 the mean calorie and the standard therm, suggested in pro- 

 position (II.) . Until this is accomplished we shall be unable 



* See note, p. 448, ante. 



