BAROMETRICAL MEASUREMENTS. 209 



In view of these measurements, Humboldt makes the following remarks : 

 These measures, taken at different periods, vary from 1,700 to 2,600 toises; and, what is remarkable 

 enough, the results obtained by geometrical operations differ more from each other, than those which 

 were found by the barometer. It has, nevertheless, been extremely wrong to cite this want of harmony 

 as a proof of the uncertainty of all measurements of mountains. Angles, the value of which is deter- 

 mined by imperfect graphometers ; bases that have not been levelled, or the length of which has been 

 determined by the log; triangles that give an excessively acute angle at tlie sumimt of the moimtain; 

 heights of the barometer, without any notice taken of the temperature of the air and of the mercury ; un- 

 questionably are not means calculated to lead to accurate results. Of fourteen trigonometrical and 

 barometrical operations above indicated, the four following only can be considered as true measurements : 



Borda, by trigonometry, 1905 toises. 



do by means of the barometer, 1976 " 



Lamanon, the same, 1902 " 



Cordier, do 1920 " 



HumbokU's Pcnmwl Narrative, Vol. 1 and 2, in one. 



It is worthy of remark, tliat the mean of the five geometrical measurements made on land 

 differs from the extremes by 422 and 493 toises, while the mean of the barometrical measure- 

 ments differs from the extremes by less than 53 and 59 toises. It is remarkable, also, that 

 Humboldt, after elaborate discussion with the details of the operations before him, should 

 select but one geometrical result out of nine, and three barometrical ones out of four, as the 

 most suitable to derive his mean from. It appears, also, that the range of the differences of 

 the four measurements adopted as true ones, is 74 toises, or 3'9 per cent of the whole eleva- 

 tion. This justifies a remark made in an early part of this article, that the two barometrical 

 results should be regarded as accordant ; since the difference is only 2*4 per cent of the ele- 

 vation of the summit above the lower station, which is more than one-third less than Humboldt 

 considered consistent with reasonable agreement. 



The measurements which I have made, although by no means so numerous as could be 

 desired, are sufficient to fix with considerable accuracjr the position of that extensive tract of 

 table land interposed between Lakes Ontario and Champlain. Racket lake, a beautiful sheet 

 of water in Hamilton county, embracing a surface of probably twenty or twenty-five square 

 miles, is situated near the geographical centre of it, and may be regarded as its summit, par- 

 ticularly of that portion which lies west of the Adirondack mountains. The area of this tract 

 is httle, if any, less than 10,000 square miles ; embracing nearly the whole of Essex, Hamil- 

 ton and Warren counties, the southern and western parts of Clinton, the southern half of 

 Franklin, the southeastern angle of St. Lawrence and northern half of Herkimer counties. 

 These estimates are founded partly upon my own observations, and partly upon information 

 derived from other sources, and should therefore be regarded as vague approximations only to 

 topographical accuracy. 



The natural features of this tract are prominent, and in some respects singularly interesting. 

 The eastern division, commencing within a few miles of Lake Champlain, forms the base 



Geol. 2d Dist. 27 



