DETERMINATION OF ALTITUDES BY BAROMETER 

 Results obtained by computation — Continued. 



125 



No trial was made of Mr. Blodget's empirical table, for, as it was deduced from these very 

 observations, they would not give a fair idea of its value. As has been already mentioned, the 

 method was tested of using successively each place of observation for the lower station, with 

 the next above it for the corresponding upper ; but, as might be expected, errors were found to 

 be propagated, and to be very considerably greater than when the sea level was constantly used 

 for the lower station. 



The result of this investigation, confirming so fully the accuracy of the new method of com- 

 putation, cannot but be satisfactory. When it is remembered that two nice adjustments are 

 required in reading the barometer, it must be considered that a mean error of 9 feet, amounting 

 to only about nine thousandths of an inch of the mercurial column, is a wonderfully close 

 approximation, when the difference of the altitudes computed amounts to nearly 3,000 feet. That 

 this method has slight causes of inaccuracy, even in theory, is not denied ; but this result tends 

 strongly to show that practically their effect is unimportant. It is not supposed that the mean 

 error in the altitudes on our survey is so small as 9 feet ; but that the profile is sufficiently 

 correct for all the purposes of a preliminary reconnaissance, I have not the slightest doubt. 



