The next operation is to fill the tube, which is commenced by expelling a portion of the air by heat, and immersing the open end in clean mercury, which ought to be recently distilled; when a con siderable quantity has thus entered the bulb, a paper cylinder should be formed by rolling a slip round the end of the tube, into which some mercury being placed, the contents of the half filled bulb are to be brought to a state of ebullition, and the vapour oc cupying the empty space, on condensing, the mer cury will descend from the paper cylinder with rapidity and fill the instrument. Where consider able nicety is required, the mercury should be boiled carefully and repeatedly in the bulb and tube to expel all moisture and particles of air, during which operation it will be prevented from escaping by the upper receptacle already mentioned. When completely filled, enough of mercury may be ex pelled from the top to render its height convenient, and the length of the tube must be proportioned to its entire range.
The tube must next be hermetically sealed, in doing which, if a vacuum above the mercury be de sired, the termination of the tube being brought to a capillary orifice, the bulb must be heated till a few drops of mercury are in the act of escaping, when the flame of a lamp must be suddenly directed so as to close the communication with the atmos phere. With these precautions, and if the thermo meter has throughout been well made, the mercury will run up from the bulb when reversed, even in very fine tubes, unless they have the last degree of capillarity. How far the accuracy of this vacuum is a matter of importance or even of benefit, has lately been questioned, and indeed vague notions seem to have prevailed regarding the resistance which the inclosed air offers to the motion of the mercury in expanding.- It lately occurred to the author of this article, that considering the enor mous effort of fluids to expand almost any degree of pressure which the glass of a thermometer can sus tain, could have no effect in this way; but only per haps a small one by dilating the glass of the bulb, which might have the effect of lowering the sur face of the fluid. To ascertain this point, which seems not before to have been practically examin ed, the following experiment was made. Three thermometers without scales were procured, per fectly similar in size and structure, with flat tubes and small bulbs, two of which were sealed in vacua, and the third was sealed, having the upper part of the tube filled with air at the temperature of 62°. All three were plunged in this state into ice cold water and afterwards into several warmer mixtures, and the most regular of the two vacuum thermome ters was chosen for careful comparison. Several points were accurately marked with waxed thread, and being plunged into water, which by a suitable contrivance was kept very steady in temperature, it was compared with the instrument in which air was sealed, and on the tube of the latter minute file marks were made at the various points. The ex
tremity of the tube of the latter was then broke off, and by a new comparison other marks were made, the results of which are shown in the follow ing table: The first column shows the temperature indicat ed by the correct standard; the second the pressure in atmospheres computed from the reduction of space in which the air was confined, counting the upper limit of the tube = 145°, and hence the bulk of 83° (or corresponding to one atmos phere. The third column marks the constant pressure to which the mercury was exposed after the end of the tube was broke; the fourth, the dif ference of pressure in the two cases; and the fifth, the errors of the thermometer corresponding to these. The last numbers, however, must be consi dered merely approximations, from the smallness of the degrees on the instruments employed; how ever, they sufficiently prove that the effect of any minute quantity of air left in the tubes must be wholly insensible.% The vacuum which has generally been left above mercury is supposed to be the cause of an irregu larity observed in old thermometers, their indica tions• being somewhat too high. This inquiry, a few years ago, produced several series of experi ments by different observers, the discordance of whose results seems to have prevented the final settlement of the question. M. Flaugerguest at tributed the rise of of a degree which he observed in the freezing point of thermometers, to the atmospheric pressure on the exterior of the bulb, as did MM. De la Rive and Marcet of Ge neva, who recommended, that the upper end of thermometers should be left open.t liellani of Mi lan,§ however, denies that this precaution has the desired effect, and imputes the alteration to mole cular action in the parts of the glass, and to the time which that rigid material takes to assume its final form after the operation of blowing. M. Aragoill attributes the alteration to the escape of particles of air from between the mercury and the glass. Professor doubts the existence of the alleged rise at all; we think, however, it has been satisfactorily proved by a great number of comparisons, though we know not if it has been suggested that some of these errors might be ex plained by the use of freezing water instead of melt ing ice in graduation, which was common in the last century, and which usually gives a tempera ture somewhat lower than the melting ice. Per haps the most unequivocal proofs of the fact are those given by Mr. Daniell, in his excellent Meteo rological Essays, who examined two thermometers made in the last century by Mr. Cavendish, who had graduated them only a little above and below the freezing point, and made the degrees very large; in one of these the freezing point had risen 0.°4, and in the other 0.'35." Should the pressure of the air be proven to be the final cause, we should prefer inclosing a little air in the tube to act as a counterpoise, to leaving the extremity open which would admit moisture and injure the delicacy of the instrument.