VOL. LXVII.] PHILOSOPHICAL TRANSACTIONS. 205 



he concludes that the weight of any column of air may be measured with tliese 

 barometers to .008 inch, though all the errors should lie the same way. The 

 barometrical operations are related at considerable length ; and the computations 

 are made according to Mr. De Luc's method, or rather according to Dr. 

 Horsley's reduction of it to the scales and measures of this country (vide vol. 13, 

 p. 530,) with this difference, that Sir G. reckoned the equation for the expansion 

 of quicksilver = .00323 inch for every degree of Fahrenheit's thermometer in a 

 column of 30 inches, instead of .00312 which Mr. De Luc used ; as Sir G. had 

 collected from some of his own experiments made at Oxford in the beginning of 

 the year 1773 : this difference will not however occasion an alteration in the re- 

 sult of any one of the observations of more than 5 inches, and may therefore 

 be considered as of no account. The barometrical computation was thus made 

 from 3 different sets of observations, the 3 results from which were as below, for 

 the height of the rock at c, above the point a in the plain below : viz. by the 

 1st set, it was 2775.2 feet, or 56.1 feet less than 2831.3 the height by the geo- 

 metrical method ; by the 2d set it came out 2763.2, or 68.1 feet too little ; and 

 by the 3d method 2750.3, or 72 feet too little. 



These observations then seem to prove that the barometrical rules were a little 

 defective as to the true ratio between the gravities of air and quicksilver, viz. in 

 the value of an inch of quicksilver in the Torricellian tube, expressed in inches 

 of the atmosphere with a given temperature. The first comparison gives for 

 this error in defect — I9.8 feet in every 1000 feet; the second, 24.0 feet; and 

 the last, 25.4 feet: the mean of the three is 23.1 feet; and by so much we 

 may conclude that these rules, in greater heights also, will give the difference of 

 elevation too little, viz. by ^ nearly. But it will be fair to make the experiment. 



The Mole is a convenient, insulated mountain, situated about 18 miles east 

 of Geneva, and rising near 5000 feet above the lake, much higher tiian any 

 person had ever made these experiments at, with the required precision. On 

 this summit Sir G. determined to confirm or correct his discovery. After a 

 tedious and difficult ascent up this mountain, its height was taken by the mean 

 of the barometers and thermometers, and found to be 4 1 20 feet ; whereas the 

 geometrical measurement, by means of a base measured at the foot of the 

 mountain, after the manner of the former operation at Mount Saleve, gave full 

 4211.3 feet, for the geometrical height. Hence the mean error of the baro- 

 metrical method, is 21.7 feet on every 1000. This result then justifies the 

 former conclusion, and proves that either the proportional gravity of air and 

 quicksilver is now different from what it was, when M. De Luc made his experi- 

 ments, viz. from 1756 to 176O; or that his or Sir G.'s observations are defec- 

 tive. That my trigonometrical measurements were sufficiently exact, says Sir 

 G. viz. to within 2 or 3 feet, I think I have already shown ; and even that Mr. 



