Temperature of Mercury in a Siphon Barometer. 257 



The imperfections of a barometer may chiefly be classed under 

 the following heads. 



1. An erroneous scale. 



2. An imperfect vacuum. 



3. Variableness of the friction of the mercury on the interior 

 surface of the tube. 



4. The want of equality and uniformity of those parts of the 

 two branches within the range of the mercurial surfaces. 



The first of these must evidently occasion uncertain results in 

 the calculated temperature, as well as in every other object of re- 

 search, and the above formula furnishes no means of detecting or 

 appreciating the error of the scale. An imperfect vacuum, al- 

 though it influences the length of the column, and on this account 

 injures the instrument, has no effect upon the calculated tempera- 

 ture. For, in this respect, it is evidently immaterial whether the 

 inferior surface falls, and the superior one rises by atmospheric 

 pressure alone, or whether it is modified by the elasticity of the 

 enclosed air. 



The third imperfection, although little attention has been be- 

 stowed upon it hitherto, is one to which barometers of all con- 

 structions are exposed, and is probably among the most difRcnlt to 

 rectify. The eflfects of this are to sustain the column at a height 

 diflferent from that due to the atmospheric pressure, and also to 

 change ihe forms of its convex terminations. Both these circum- 

 stances occasion error as it respects the height of the column, while 

 the latter, or change of form only, can affect the correctness of the 

 calculated temperature. 



The altitudes of the segments which terminate the column are 

 found to vary more or less at different times in the same barome- 

 ter. In some, the variation is quite inconsiderable, if necessary 

 care is taken in observing ; while in others, one or both the ter- 

 minal surfaces may assume all forms from a plane to an apparent 

 hemisphere. When the atmospheric pressure is increasing, the 

 inferior meniscus tends to become less convex, and the superior 

 one more so. For, as the column lengthens, and the mercury 

 consequently descends in the shorter branch of the siphon, the 

 cylindric surface of mercury which is in immediate contact with 

 the glass tube is retarded by its friction against the glass, and re- 

 tards in its turn, though in a less degree, the next concentric sur- 

 face of mercury ; and this the succeeding one, and so on ; the 

 central filament being least retarded of all. This would cause 



