278 PHILOSOPHICAL TRANSACTIONS. [aNNO 1792. 



larged. But tlie chief reason why there may sometimes seem to be a loss, at 

 other times no loss at all, 1 apprehend to be, that more of the fluid will adhere 

 to the upper part of the tube, on filling it, at one time than at another. In the 

 use of this instrument also, a small error may arise from the stopper not being 

 always put in exactly alike; in which case the capacity of the instrument would 

 be altered ; and, of course, the divisions on its stem would not give the expan- 

 sion of the fluid accurately. Care was always taken in these experiments to put 

 the stopper in as nearly alike as possible; but it might not perhaps always be 

 done exactly so. 



It is also obvious, that experiments with this instrument will be affected by 

 another source of error, if made in the manner which is recommended, namely, 

 by heating the fluid up from 6o° to lOO", and cooling it down again to 30°: for 

 it must be evident that the whole length of the tube will then be left wet by the 

 fluid, in s'nkingfrom 100° to 30°, and consequently the expansion will be made 

 to appear too great. The effect of this circumstance will be very considerable; 

 but in the use of this instrument we have no certain means of ascertaining with 

 accuracy the quantity of error occasioned by it, because that quantity falls in 

 with other errors. 



The former experiments were all made with that kind of instrument which 

 has a tube rising from the side of the ball, witii a ground-glass stopper inserted 

 into it; an instrument we have seen by no means to be considered as sufficiently 

 accurate for ascertaining the expansion of fluids; I therefore constructed one, 

 similar to the other of the 2 recommended by Mr. Ramsden, which has 2 tubes 

 rising from the ball, one on each side. Having charged this instrument with 

 some of the same spirit employed in the former experiments, and brought it to 

 the temperature of 60", the spirit in the 1 tubes was found to stand at 4 above 

 O. It was then cooled down to 30°, when the spirit in the 2 tubes was found to 

 stand at l6l below O, tiie instrument being always so held as to bring it to the 

 same point in both tubes. I then heated it up to 100°, and it stood in the 2 

 tubes at 236 above O. I cooled it again to 30°, when it was found to stand in 

 the 2 tubes at 162 below O. It was again heated up to 100°, and it stood in the 

 2 tubes at 236 above 0. I then brought it again to the temperature of 6o°, and 

 found it to stand in the 2 tubes at no more than 3 above O. 



It appears from the above experiments, that the contraction of the spirit from 

 6o° to 30° is l6l + 4 = 165, and the expansion in heating up again to 60°, the 

 same. On heating up from 6o° to 100°, 236 — 4 = 232, and therefore the 

 total expansion from 30° to 100° is 307; but in cooling down from 100° to 30°, 

 the total expansion will be 236 + 162 = 398; the former quantity differs 0.55, 

 in defect, and the latter 0.43 of a division, in excess, from the experiments by 

 weight. Now it is evident that the method of heating up from 30° to 1 00° can 



