VOL. LXVII.] PHILOSOPHICAL TRANSACTION'S. 211 



It has been suspected, in consequence of some experiments made by a very 

 ingenious member of this society, that air does not expand uniformly with quick- 

 silver; or that the degrees of heat shown by a quicksilver-thermometer would 

 be expressed on a manometer, or air-thermometer, by unequal spaces in a cer- 

 tain geometrical ratio. I do not deny this proposition ; but I have also very 

 little reason to assent to it, if I may trust my own experiments, which certainly 

 evince that this ratio, if not truly arithmetical, is so nearly so as to occasion no 

 sensible error in the measuring of heights with the barometer; and that is all 

 I contend for. The small differences that are seen in the above table of this 

 expansion, deduced from a mean of 14° or of 40°, I would attribute rather to 

 the errors of observation than to any actual irregularity in nature. If however 

 this progression be insisted on, it should seem that the degree of the air's 

 expansion increases with an increase of heat; and that the difference of volume 

 or density from 1° of heat, any where within the limits abovementioned, would 

 be about one part in 5000 from what I take it at a mean. I should not have 

 insisted so long on this circumstance, but in respect to the known accuracy of 

 the author of this hypothesis. Neither do I find any reason to believe, that the 

 expansion of air varies with its density. I have tried air whose density or 

 pressure was equal to 23i inches, and also to 40 inches; but the dilatation, 

 with equal volumes and equal degrees of heat, was very nearly the same in both 

 cases. I might add a great deal more on these manometrical experiments, but 

 am afraid it would be more tedious than useful. I proceed therefore to the 

 expansion of quicksilver. 



This experiment was made with a tube, something like a thermometer, but 

 considerably larger than the ordinary size, and open at one end; it was filled 

 with quicksilver to a certain height, and then exposed to the temperatures of 

 freezing and boiling repeatedly, the barometer being at 30 inches: the difference 

 of the volume in each instance was determined afterwards by accurately weigh- 

 ing the contents. I thus found, that if the quicksilver at freezing be supposed 

 to be divided into 1 3 1 1 Q parts, the increase of volume by a heat of boiling 

 water became equal to 208 of these parts = ■^^, and -^^ X t-to = -, , \ g-^- ; 

 and such would be the expansion for each degree of the thermometer, com- 

 mencing from the freezing point, = O.OO262 inch on a column of 30 inches 

 of the barometer, if the glass had suffered no expansion during the experiment. 

 This however has been found to be with 180° of heat = ^^ in solidity (viz. 

 the cube of its longitudinal expansion) and T-hi- X t4-o = -rrlo-a- = 0.00042 



posed to deserve equal confidence, the error of the tables in common use, in ih above circum- 

 stances, would amount to half that quantity, and therefore probably will be thought scarcely worth 

 correcting. I have mentioned this in order to obviate the conclusions that have been drawn by some 

 persons from Mr. De Luc's theory. — Orig. 



E E 2 



