Deluc, in his work, does not describe, so far as we know, the precautions he took to ensure the accu racy of the upper term of adjustment in the spirit thermometer, for which he employs boiling water as in the case of mercury. He plunges the instru ment (the fluid of which boils at Fahrenheit,) into freezing and then into boiling water, divides the interval into 80 degrees, which he compares with the mercurial thermometer at every 5°, and gives the defect as the error of expansion. But we are disposed to ask Deluc, who blames Reau mur for not suffering the spirit in his thermometer to boil, what does he gain by this procedure; how does he pretend to mark the dilatation of a fluid above a temperature when a considerable portion of it must confessedly pass into vapour P Deluc gives no explanation of the point, and we find his table of the inequality of the expansion of alcohol quoted in almost every work on the subject, without the slightest comment on its derivation; the discrepan cies arc so enormous, (amounting to above 12° 1'ahr. at a maximum,) as to convince us, if correct, that such a fluid is absolutely unfit for thermometric measurement. :NI. Riot, however, in his excellent work already quoted, says expressly, that the or dinary method of graduating alcohol from mercu rial thermometers, is extremely incorrect, for noth ing is easier than to raise the spirit thermometer to the temperature of the boiling point by carefully freeing the tube of air and permitting an elastic atmosphere of vapour to be formed above it, which shall prevent, by means of pressure, the vaporiza tion of the fluid, and enable it to attain an unnatu ral temperature.* This we readily admit; but in the first place, it presupposes the conversion of a portion of the fluid into an invisible and elastic form, which obviously lowers the height of the re mainder in the tube, since it is abstracted from it; and what notice, the quantity of vapour required to repress by its elasticity the tendency of the alcohol to vapourize, must vary with the length which the tube may happen to be, so as to occupy it with vapour, having the great elasticity of 67 inches of mercuryt to enable it to reach the tem perature of boiling water. Not to speak of the natural contraction of the fluid itself under this pressure, or the expansion of the glass bulb, we would ask by what rule we can expect a substance in so unnatural a condition to be a true measure of equal increments of heat by expansion, when that very substance would be naturally expanded at the same temperature, not by the previous laws of di latation, but after rendering latent an immense quantity of heat, into a subtile vapour many hun dred times rarer than in its fluid state. We are aware that in a real vacuum it may be said that alcohol boils far below 175°, and that it is repressed below as above that temperature, by the force of its own vapour, but this very force is so small, differ ing from that at the natural boiling point by a frac tion of an atmosphere, and therefore the conversion of so small a quantity of fluid into vapour being necessary, that it may well be overlooked; besides, that in theory we do not think there ought to be, and in practice there never is an exclusion of air from above the thermometric column, for we shall soon show how much the value of such a vacuum has been overrated in mercurial thermometers.
But the most satisfactory proof of the inaccuracy of Deluc's method is obtained from the following consideration, which is the only attempt made, though an indirect one, which we have been able to discover, to correct the important misconceptions which have prevailed on the subject. Dalton, whose accuracy as an experimentalist is undoubted, has confirmed the assertion of Deluc, that while spirit of wine expands 35 parts between the freezing point and 122° Fahr., it expands 45 between
and 212°, but the same chemist has stated, that taking 170° as the upper point, which may be de termined with great accuracy by a careful compari son with the mercurial thermometer, he found the spirit which occupied 1000 parts at 50° to fill 1079 at 170° and at 110°, the mean distance where the error should have been at a maximum, he found it to occupy 1039 parts, or half a division below the true mean; from which it appears that almost the entire error exists, as was to be expected, in the 40° P. of Deluc's thermometer nearest the boiling point of water, which, therefore, it is manifestly unfair to carry down through the scale, and charge the whole instrument with an enormous error of ex pansion, amounting at a maximum, to one sixteenth of the whole interval between boiling and freezing water. We shall now quote the latest table of the expansions of alcohol, from experiments performed on the method of Deluc by Dr. De Wildt,f for every 5° of the octogesimal scale.
Here we have another question to ask regarding this table, which shows a violence done to the tural properties of bodies not less than that of ing alcohol to bear the heat of boiling water. How, we should be glad to know, does Dr. De Wildt tend to measure the dilatation of mercury at — 45 — Fahrenheit P He must have forgot that the congelation of mercury has not hitherto been measured by this fanciful scale of the alcohol thermometer, but by the mercurial one itself, and found to coincide with 39° of Fahrenheit's scale,§ a point determined beyond a doubt, and which is somewhat greater, though not nearly so much as De Wildt would have us to believe, than the perature expressed by most spirit thermometers constructed in the ordinary way, as we hope it ever will be cor.structed, by a careful comparison with a good mercurial one. Hence it appears that De Wildt actually compared alcohol with mercury 37 degrees below its freezing point, if we are by any means to judge of temperature measured by a solid just reduced from fluidity. This, however, is im possible, since mercury contracts enormously after solidification,* we may therefore approximate to the real temperature of the first point in the table in this way; supposing the contractions of absolute alcohol equable with the temperature, as we have endeavoured to prove they are nearly, when at a moderate distance from the boiling point of the fluid, we may say, comparing it with the mercurial thermometer at the first term given above congela tion (— 30 R) as 20.32 : 30:: 28.50: 42.07. Whence we may estimate the thermometer to have sunk about 3 R = 7 F. too low, a circumstance quite accidental, since below the point of congelation, mercury sometimes contracts suddenly, so as to in dicate a cold of several hundred negative degrees, sometimes contracts but little, indicating a tem perature of — 50 or — 60 without moving. On the whole, therefore, we conclude that these expe riments have done gross injustice to the spirit ther mometer, and though the irregularities of that fluid are in all probability much greater than those of mercury, they have been exaggerated by the inju dicious choice of the highest fixed point. Below zero of Fahrenheit, alcohol appears to be defective in its expansion, though by no means so much so as stated in De Wildt's table, (which makes the error about degrees of Fahrenheit at the freezing point of mercury,) for in the numerous experiments of Hutchins, the spirit thermometers, at extremely low temperatures, sometimes did not differ so much from the mercurial ones, as these from one another, and they occasionally indicated temperatures not only as low, but even lower than the mercurial ones at a point higher on the scale than some of those at which Dr. De Wildt has at tempted to compare the instruments.