In order to construct a good barometer, it is indispensably necessary that the mercury be perfectly free from impuri ties, and carefully purged of air ; this is obtained by boiling it. The particles of air and moisture which cling obstinately to the sides of the tube must also be ex pelled by heat ; the mercury must then be introduced slowly and continuously in a hot state, and while the tube con tinues hot. Since the time of Deluo it has been usual to boil the mercury in the tube before inverting it and forming the vacuum; but doubts now begin to be en tertained among the most skilful makers of the expediency of this very trouble some process. The mercury is partially oxydated by boiling, and a thin crust formed, which keeps the column sus pended at a greater height, and obstructs the freedom of the motion. It is impor tant that the diameter of the tube be not very small; for it is found that the mer cury moves with more freedom in a tube of considerable width, the oscillations following the atmospheric changes with more promptitude than in one of smaller dimensions ; besides which, there is less disturbance from capillary attraction. The interior diameter should in every case exceed one-fourth of an inch.
The value of the barometer as a scien tific instrument depends on the purity of the mercury, and the total exclusion of atmospheric air. By proper care in the construction, it is, perhaps, possible to expel every partiCle of air from the mer cury and the interior of the tube when the barometer is made ; but it seems doubtful if, by any means whatever, it can be preserved for a considerable length of time in this state. The most carefully constructed barometers are liable to a slow and gradual deterioration by the in trusion of air, which has been supposed to insinuate itself between the metal and the tube, and not through the mercury. To obviate this inconvenience Professor Daniell conceived the ingenious idea of fixing to the open end of the tube of the cistern barometer a substance having a greater affinity than glass for mercury. " I caused," says he, " a small thin piece of platinum tube to be made, about the third of an inch in length, and of the diameter of the glass tube ; this was care fully welded to its open end,. so that the barometer tube terminated in a ring of platinum. The tube was filled and boiled as usual, and the infiltration of air was completely prevented by the adhesion of the mercury both to the interior and ex terior surface of the platinum guard. I have no doubt that a mere ring of wire welded, or even cemented, upon the ex terior surface of the glass, which would be a much easier and iess expensive operation, would be a sufficient protec tion, as the slightest line of perfect con tact must effectually arrest the passage of the air" In all barometic observations there are, in general, two essential corrections to be made ; one for the capillarity or depres sion of the mercury in the tube, and the other for temperature. Pure mercury in
a glass tube always assumes a convex surface. The following are the correc tions for tubes of different diameters, ac cording to the theory of Mr. Ivory.
Diem. of Tuba. Depreesion. Diem. of Tube. Dammam. Inedee. Inedee. Inches.
1403. 40 - - 0153.
15 - - 0863. .45 - - 0112.
20 - - 0581. 50 - - -0083.
25 - - -0407. .60 - - 0044.
30 - - -0292. 70 - - -0023.
.35 - - -0211. 80 - - 0012.
These corrections, which must always be applied to cistern barometers, show that wide tubes ought to bepreferred ; in fact, when the diameter of the tube exceeds half an inch, they may be safely omitted. In siphon barometers having both branches of the same diameter, the depression is equal at both ends ; conse quently the effect is destroyed, and no correction is required. This is a consi derable advantage ; for notwithstanding the most elaborate calculations, some un certainty must always remain with re gard to the exact amount of the capillary repulsion.
The correction for the temperature, which is the most important, depends on the expansion of the mercury, and the expansion of the scale on which the divi sions are marked. If we make a= the height of the thermometer in degrees above the freezing point, x = the *no tional part of its bulk which mercury ex pands for one degree of heat on Fahren heit's scale, y = the fractional part of its length by which the scale increases, h= the observed height of the barometer ; then the height which would have been observed had the thermometer stood at the freezing point is 7/,-2 a (xy).
The expansion of mercury in part of its bulk is 0001001. The scale is generally of seine mixed metal of which the expan sion is not very well ascertained: sup posing it to be equal to that of copper, the expansion would be 0000096; therefore it will be sufficiently accurate to neglect the temperature of the scale, and assume that of the mercury to be 0001. Hence the following practical rule for reducing an observed height to the corresponding height at the temperature of the freezing point : " Subtract the ten-thousandth part of the observed altitude for every degree of Fahrenheit above 32." Sup pose the thermometer 54° and the baro meter 30 inches, the correction will be (54-32) X 30 X -0001 = 066, to be sub tracted from 30 inches. In order to find the value of this correction a thermome ter must be attached to the barometer, and observed at the same time. A table, showing the correction for temperature for every degree of Fahrenheit from 30° to 90°, and for every difference of half an inch in the height of the mercury from 28 to 30.5 inches, was constructed by Professor Schumacher, and is given by Mr. Bally in the Phil. Trans. for 1837.