18 



The manufacturer, in adjusting a good barometer, endeavors to 

 eliminate as completely as possible, or at least to reduce to a very 

 small quantity, the several corrections mentioned above, viz, cor- 

 rection for capillarity, for imperfect vacuum, and for instrumental 

 error. This he can do by sliding the scale up or down a small frac- 

 tion of an inch until he finds by repeated trials and comparative 

 readings with a standard instrument that the new barometer, when 

 corrected for temperature, as described below, gives the same or 

 nearly the same readings as the standard. Any slight outstanding 

 difference that may finally remain then becomes the " correction for 

 instrumental error, including capillarity," or briefly, " correction for 

 scale errors and capillarity." 



By comparing a barometer in a partial vacuum, so as to ascertain 

 the " correction for scale errors and capillarity " at several pressures, 

 such, for example, as at each inch between 20 and 30 inches of 

 pressure, it has been learned in a few interesting cases that very 

 great differences in the correction may be found at different points 

 of the scale. These differences amounted in one case to eighty-three 

 thousands of an inch between 25 and 30 inches, and could not be 

 explained by any error of the scale or by any influence other than that 

 of the irregular capillarity action at different points of the tube. 

 These investigations demonstrated the necessity of ascertaining the 

 correction for scale errors and capillarity of each instrument for the 

 ]:>articular pressure at which that instrument is to be used. 



(4) Correction for temperature. — The temperature of a barometer 

 affects the accuracy of its readings in two ways. First, the metal 

 scale expands and contracts with changing temperatures, and is, 

 therefore, continually changing its length. Second, the mercury 

 itself expands and contracts much more than the scale. The 30 

 cubic inches of warm mercury in a barometer tube at, say, a tempera- 

 ture of 80° F., will be more than 1 ounce lighter than the same volume 

 of mercury at the freezing temperature. 



The true pressure of the air, therefore, is not shown by the observed 

 height of the mercurial column until we take into account both the 

 temperature of the scale and the density of the mercury. 



For this reason barometric readings require to be reduced to a read- 

 ing which would have been obtained had the mercury and scale been 

 at certain standard temperatures. 



The standard temperature adopted for the mercury is always that 

 of melting ice— that is, 0° C, or 32° F. 



When the readings of the scale are taken in inches, the standard 

 temperature for the scale reduction is then 62° F. If, however, the 

 metric unit of length is used, the standard temperature is then 0° C. 

 In the latter case the same temperature serves for both the scale 

 and the mercury. 



There is thus a disparity between the temperatures at which Eng- 

 lish and metric scales are of standard length ; moreover, tables of 

 barometric corrections for temperature usually give the reduction for 

 both the scale and the mercury in one correction, whence it follows 

 from these two circumstances that the corrections in English and in 

 metric tables are not mutually convertible. An error is therefore 

 introduced if the uncorrected reading of a mercurial barometer 

 expressed in metric units is converted into English units, or vice 



