the mercury suffices to fill up the cistern just to the throat of the con- 

 tracted portion at G. By screwing up the milled head H, the plate 

 G closes against the bottom of the cistern and completely imprisons 

 the mercury with only a little free space for expansion. If the 

 barometer is now turned erect the mercurial column can not descend 

 unless the screw H is loosened, whereupon the mercury flows into 

 the previously unoccupied space below the plate G and permits the 

 column to resume its normal level. 



23. The capacity correction required for a barometer with fixed 

 cistern and true scale graduated in standard units, and not con- 

 tracted, as explained in paragraph 19, can be determined from care- 

 ful measurements of the internal diameters of the tube and cistern 

 made before the barometer is filled, but the accuracy of this correc- 

 tion should always be checked by subsequent comparisons, as indi- 

 cated below. 



24. Cap<icity correction^ hov>. fownd. — It is evident that by sliding 

 the scale of a fixed-cistern barometer up or down it can be so adjusted 

 that a reading at some one point is just right ; for example, we may 

 place the 30-inch mark so that when the top of the column is at this 

 mark the surface of the mercury in the cistern is just 30 inches 

 below. If the sectional area of the tube is a, and that of the cistern 

 A^ then, if the mercurial column in the tube rises one scale division, 



the fall in the cistern will be only the -j part of one division. That 

 is, the correction for a scale reading just one division about the 30- 

 inch mark is : + -j divisions ; for a reading two divisions above the 



correction is : +2-^» etc. This, expressed in a mathematical formula, 

 becomes — 



Correction : G={h—R^-7\ 



in which R^ is the reading at which the correction is zero and h is 

 the observed reading, uncorrected for temperature. This may be 

 reduced to the following simpler form — 



C^nh — m; 



in which m and n are two quantities whose values are best determined 

 from a complete series of readings of the actual height of the mer- 

 curial column, as compared with the reading of the top of the column 

 of the fixed-cistern barometer. As the level of the mercury in the 

 cistern is generally not visible in barometers of this type, the direct 

 measurements of the heights of the column can not be made, and the 

 necessary actual heights must, therefore, be obtained from readings 

 of some standard barometer. 



To determine the values of m and n accurately by comparisons, 

 observations should be made over a greater range of pressures than 

 ordinarily occur from day to day, and the best results will require 

 observations under pressure artifically changed to suit. 



25. Changes of temperature may cause the sectional areas of the 

 tubes and cistern to have a different relation than that assumed in 



