XXXIV INTRODUCTION. 



station where an observation is made, the lower will be the temperature of 

 the boiling point. 



The difference of elevation between two places therefore can be deduced 

 from the temperature of boiling water observed at each station. It is 

 only necessarj'- to find the barometric pressures which correspond to those 

 temperatures, and, the atmospheric pressures at both places being known, 

 to compute the difference of height by the tables given herein for com- 

 puting heights from barometric observations. 



From the above, it maj^ be seen that the heights determined by 

 means of the temperature of boiling water are less reliable than those 

 deduced from barometric observations. Both derive the difference of alti- 

 tude from the difference of atmospheric pressure. But the temperature of 

 boiling water gives only indirectly the atmospheric pressure, which is 

 given directly by the barometer. This method is thus liable to all the 

 chances of error which may affect the measurements by means of the 

 barometer, besides adding to them new ones peculiar to itself, the prin- 

 cipal of which is the difficulty of ascertaining with the necessary accuracy 

 the true temperature of boiling water. In the present state of ther- 

 mometry it would hardly be safe, indeed, to rely, in the most favorable 

 circumstances, upon quantities so small as hundredths of a degree, even 

 when the thermometer has been constructed with the utmost care ; more- 

 over, the quality of the glass of the instrument, the form and substance 

 of the vessel containing the water, the purity of the water itself, the position 

 at which the bulb of the thermometer is placed, whether in the current of 

 the steam or in the water, — all these circumstances cause no inconsiderable 

 variatioifs to take place in the indications of thermometers observed under 

 the same atmospheric pressure. Owing to these various causes, an obser- 

 vation of the boiling point, differing by one-tenth of a degree from the true 

 temperature, ought to be still admitted as a good one. Now, as the tables 

 show, an error of one-tenth of a degree Centigrade in the temperature of 

 boiling water would cause an error of 2 millimetres in the barometric 

 pressure, or of from 70 to 80 feet in the final result, while with a good 

 barometer the error of pressure will hardly ever exceed one-tenth of a 

 millimetre, making a difference of 3 feet in altitude. 



Notwithstanding these imperfections, the hypsometric thermometer is 

 of the greatest utility to travellers and explorers in rough countries, on 

 account of its being more conveniently transported and much less liable 

 to accidents than the mercurial barometer. A suitable form for it, designed 

 by Regnault (^Aiinales de Chimie et de Physique, Tome xiv, p. 202), consists 

 of an accurate thermometer with long degrees, subdivided into tenths. 

 For observation the bulb is placed, about 2 or 3 centimetres above the 

 surface of the water, in the steam arising from distilled water in a cylin- 

 drical vessel, the water being made to boil by a spirit-lamp. 



