an instrument for observing the purity of the atmospheric air, or the quantity of pure dephlogisticated or vital air contained in it, chiefly by means of its diminution on a mixture with nitrous air*. Several kinds of these have been invented, the principal of which are the following:

I. The eudiometer originally used by Dr Priestley is a divided glass tube, into which, after having filled it with common water, and inverted it into the flame, one or more measures of common air, and an equal quantity of the nitrous kind, are introduced by means of a small phial, which is called the measure; and thus the diminution of the volume of the mixture, which is seen at once by means of the graduations of the tube, instantly discovers the purity of the air required.

II. The discovery of this property of nitrous air and the eudiometer by Dr Priestley, soon produced various attempts to improve on the principle, and construct more elegant and accurate machines for discovering the smallest inequality in the constitution of the atmosphere. The first of these was contrived by Mr Landriani; an account of which is published in the fifth volume of M. Rosier's Journal for the year 1775. It consists of a glass tube, fitted by grinding to a cylindrical vessel, to which are joined two glass cocks and a small basin; the whole being fitted to a wooden frame. Quicksilver is used in this instrument instead of water; but the use of that fluid occasions an inconvenience, because the nitrous air acts upon the metal, and renders the experiment ambiguous.

III. In 1777 Mr Magellan published an account of three eudiometers invented by himself. The first of these, represented Plate CLXXXVI. fig. 1, consists of a glass tube MD, about 12 or 15 inches long, and quite cylindrical throughout, having the upper orifice closed with a ground-glass stopple M. A vessel C is joined to the lower part of the tube, and likewise well adapted by grinding. This vessel has three necks, as represented in the figure: one of which serves to join it to the tube M; the other two are ground to those of the phials A and B, whose capacities must be as equal as possible, as well to each other as to the tube MD. Z represents a brass ring which slides up and down the tube MD, and by a finger-screw may be tightened or slackened at pleasure, and set to any place upon it. G is a brass or wooden ruler divided into equal parts, with two semicircular brass pieces, by which it may be easily applied and kept near the glass tube MD, as is shown at F; where it must be kept close to the neck, or upper extremity of the tube, by the notch I. In using this instrument, we must first remove the stopple M, after which the instrument is to be entirely filled with water by dipping it in the tub. The stopple is then to be replaced; taking care that no bubble of air may remain either in the tube, the vessel C, or the two phials A B. The lower part of the instrument, viz. about as far as the middle of the tube, must then be kept under water, and one of the phials A or B, now filled with water, is to be removed from the neck of the vessel C, and filled with the air of which we design to try the purity, in the manner directed under the article Gas; after which it is to be replaced into the neck of the vessel C; and in like manner the other phial must be filled with nitrous air, and replaced in the other neck. Taking the instrument then out of the water, the vessel C must be turned with the bottom upwards, as represented at F; in which case, the two elastic fluids contained in the phials will ascend into the vessel C; where, mixing together, the diminution will be effected. But as soon as the vessel is turned round, the instrument must be plunged in water as far as about the middle of the tube, and the stopple M removed. As the bulk of the two elastic fluids diminishes, the water in the tube MD descends. This instrument is subject to some inaccuracies arising from the greater or lesser height of the column of water in the tube MD, as it is held more or less perpendicular; it may also vary by the very act of putting in the stopple M. Another and still greater fault is, that it cannot admit but one measure of nitrous to one of common air, which is a very uncertain method of estimating the purity of a given kind of respirable air. The divisions on the scale are likewise too large, and it does not seem capable of any great accuracy.

The second kind of eudiometer constructed by M. Magellan is represented fig. 2, and consists of a glass tube TC, two or three feet long, and having a cavity as nearly cylindrical as possible. One of the ends, C, is bent forwards as represented in the figure; the other at T is open, and may terminate in a funnel, to obviate the necessity of using a separate one. The whole tube is flattened by means of two loops to the brass scale CWN. N is a glass phial, having its neck V ground air-tight to the inside of the end of the tube T; the whole phial containing one half of what the tube TC is capable of containing; but the phial ABC, at the other end, must contain three or four times the quantity that N can contain; and the neck of it must also be ground air-tight to the end C of the tube. The scale CWT is divided into 128 parts, the divisions being set from T towards C; and the cavity of the tube between the first and last of them being double the capacity of the phial N. XR is a tin-vessel, which may serve as a case for packing the whole instrument and its appendages; as also for a trough for holding water when experiments are to be made. The glass tube g b, and the glass stopple M, are both ground air-tight to the mouth V of the tube, in order to put it into it occasionally. To use this kind of eudiometer, let the instrument be immersed under water in the tin- vessel; then let the phial \( N \), when filled with water, be put into \( CED \), the inside socket of the tin-vessel. Fill it then with nitrous air; and let this quantity be thrown into the phial \( ABC \), which is to be fixed somewhat tight to the mouth \( C \) of the eudiometer. The same phial \( N \) is afterwards filled with the air of which we wish to try the quality; and raising the end of the instrument \( C \), it is then put into the mouth \( V \). The instrument is then to be placed upright as in the figure, by hanging it on the hook \( W \); and as soon as this last air goes up to the phial \( ABC \), the phial \( N \) is to be taken off, that the diminution of the two mixed airs may be supplied from the water in the tin-vessel: the mouth \( V \) of the eudiometer being all this time held under water. The bent tube \( gh \), having the brass ring \( K \) fitted to it, is then put to the lower end \( V \) of the eudiometer. By observing the surface of the water in the small tube, which thus forms a true syphon with the tube of the instrument, and by means of the brass ring \( K \), the stationary state of diminution in the mixture may be distinguished; which being ascertained, the small tube \( gh \) is taken off from the eudiometer, and the whole instrument laid down for some minutes in the water of the tin-vessel; after which the mouth \( V \) is to be shut up with the glass stopple \( M \); and, reversing the instrument, it is hanged up by the end \( V \) upon the hook \( W \). By this position the whole diminished air of the vessel \( ABC \) goes up to the top, where its real bulk is shown by the scale facing the inside surface of the water. This number being deducted from 128, gives the comparative wholeness of the air already tried without any farther calculation. "But this process (says Mr Magellan) will be still easier, when the last diminution of the two kinds of air is only required in the observation; because no use will then be made of the syphon. In such a case the instrument is left hanging on the hook \( W \) for 48 hours; after which it is laid down under the water of the trough in a horizontal position for 8 or 12 minutes, in order to acquire the same temperature with the water: the mouth \( V \) is then shut up with the stopple \( M \); the instrument is hung by the end \( V \) in a contrary position; and the last real bulk of the good mixed air will then be shown by the number of the brass scale answering to the inside surface of the water.

IV. The third eudiometer constructed by Mr Magellan is represented fig. 3, where \( EN \) represents an uniformly cylindrical glass-tube about two or three feet long, with a large ball \( S \) and a glass stopple \( M \), fitted air-tight to the mouth \( N \), which ought to be wide and funnel-shaped, unless a separate funnel is made use of. \( KL \) is a small syphon with a brass ring \( X \); \( Z \) a small phial, the contents of which do not exceed one third of the ball \( S \), or one half of the glass tube. Lastly, the instrument has a ruler \( T \), divided and stamped like the scale already mentioned, with a glass funnel, which is ground to the mouth \( N \) of the instrument, when this is not funnel-shaped as above directed. When this eudiometer is to be made use of, it must be filled with water, and set in a vertical position, with the mouth \( N \) under the surface of the water in a tub or trough. The phial \( Z \) is to be filled with nitrous air, and thrown into the tube by means of a glass funnel, if the mouth of the eudiometer tube be not sufficiently wide to answer the purpose. The same phial \( Z \) is then to be filled with the air to be tried; after which the syphon \( KL \) is to be immediately added to the mouth \( N \) of the eudiometer under the surface of the water, some of which is to be poured into it. The stationary moment of the greatest diminution of the two airs is watched by means of the ring \( X \); and, when that moment arrives, the syphon \( KL \) is to be taken off; the eudiometer is laid for some minutes under water in an horizontal position or nearly so; but taking care that none of the inclosed air may escape: the mouth \( N \) is then shut up with the glass stopple \( M \), and the instrument is inverted with the mouth \( N \) upwards. Lastly, the space occupied by the residuum of the diminished air is measured by applying to its side the divided ruler or scale, and the result is estimated as has been already explained.

On all these eudiometers it is very obvious to remark, inconvenient that they are complicated and difficult to be used; and it is besides no easy matter to get them made with the requisite accuracy. Mr Cavallo observes also, that the construction of all the three is founded on a supposition that the mixture of nitrous and atmospheric air, after having continued for some time to diminish, increases again; but he informs us that this is a mistake, and that Mr Magellan himself owned it to be so. But the worst of all is, that they are by no means accurate, as appeared from several experiments made by Mr Magellan in Mr Cavallo's presence, with air taken out of the window of the room where the experiments were performed. By the first trial, the diminution was 48 parts out of 132 of the mixture: on a second trial, the same elastic fluids being still used, the diminution was 58 parts out of 132: on a third trial, the diminution was again 48; and by a fourth one, it was 51. Nay, Mr Magellan himself owned that, after many experiments with his eudiometers, he never could obtain any constant result, even when the nitrous and common air which he made use of were precisely of the same quality.

V. A preferable method of discovering the purity of the air by means of an eudiometer is recommended by Mr Fontana; of which Mr Cavallo says, that its accuracy is such as could scarcely be believed by those who have not had an opportunity of observing it. The instrument is originally nothing more than a divided glass tube, though the inventor afterwards added to it a complicated apparatus, which, in Cavallo's opinion, was altogether useless. The first simple eudiometer consisted only of a glass tube, as uniformly cylindrical as possible in its cavity, about 18 inches long, and \( \frac{1}{3} \)th of an inch in diameter in the inside, hermetically sealed at one end (A). The outside of this tube was marked with a diamond, or had circles drawn round it at the distance of three inches from one another, beginning at the closed end of the tube; or at such distances

---

(A) To observe whether the cavity of a glass tube is perfectly cylindrical, pour into it at different times equal quantities of mercury or water, one upon the other; observing each time, by means of a divided ruler, if those equal quantities of liquor fill equal lengths of the tube. flances as are exactly filled by equal measures of elastic fluids. When the parts of these divisions were required, the edge of a ruler, divided into inches and smaller parts, was held against the tube; so that the first division of the ruler might coincide with one of the marks on the tube. The nitrous and atmospheric air are introduced into this tube, in order to be diminished, and the purity of the atmospheric air thus ascertained; but that an equal quantity of elastic fluid may always be certainly introduced, M. Fontana contrived the following instrument as a measure, which cannot be liable to any error. It is represented fig. 4, and consists of a glass tube \(AB\), about two inches long and one inch diameter, closed at the end \(A\), and having a brass piece \(BCDE\) cemented on the other, containing a sliding door \(D\); which when pushed into its proper cavity, shuts the mouth of the tube or measure \(AB\); and when pulled out, as represented in the figure, opens it. To prevent it from being pulled out entirely, a spring \(E\) is screwed upon the flat part of the brass piece, the extremity of which bears upon the head of a brass pin, which passing through a hole, rubs against the door \(D\); and when this is pulled nearly out, the pin, falling into a small cavity, prevents it from coming quite out. The diameter of the brass piece is nearly the same with that of the glass tube \(AB\); and near its mouth \(C\) there are two notches made with a file.

Under the same figure the cavity of the brass piece and the parts of the measure are shown separately, viz., \(a\), the glass tube; \(b\), the brass piece; \(c\), the sliding brass door inverted in such a manner as to exhibit the cavity for the pin; \(d\), the pin with the spring and small screw. The inside surface of this measure, as well as of the long tube, should have the polish taken off by rubbing with emery; as this prevents the water, when the experiments are made, from adhering to it in drops, and thus the measurements will be more exact.

To use this apparatus, the long tube must be filled with water; and being inverted in the tub of water described under the article Gas, furnished with a shelf, the measure, being also filled with water, is inverted over a hole in the shelf; and in order to fill it with the elastic fluid required, a phial containing it is brought under the hole; where being inclined a little, part of the gas escapes and passes into the measure. The water then escapes through the notches \(ss\), made with the file in the mouth of the measure, as already mentioned (\(b\)). The door of the measure is then shut by pushing it in as far as it will go; and the measure, being drawn off from the shelf, but still kept under water, is turned with the mouth upwards; by which means the superfluous quantity of elastic fluid, remaining in the cavity of the brass piece by reason of its being separated by the sliding piece, escapes, and has its place occupied by water. The measure being then again inverted with its mouth downwards, is set anywhere on the shelf of the tub; the long tube put over the hole of the shelf, and the air transferred from the measure to this tube, as has already been directed for filling the measure itself.

When M. Fontana made use of this eudiometer, he commonly threw in two measures of respirable air into the tube; then he added one measure of nitrous air; but as soon as the latter was entered, he removed the tube from the shelf, holding it by the upper end, and agitating it for about 20 seconds in the water. The tube was then rested upon the side of the tub, while the measure was again filled with nitrous air; then putting the tube upon the shelf, and holding it as nearly perpendicular as he could, he applied the divided edge of the ruler to it, in order to observe the diminution of the two fluids. After this he threw in a fourth measure of nitrous air; and after shaking and letting it rest for some time, he observed again the diminution of the two elastic fluids.

"That this method (says Mr Cavallo) should be very accurate, may perhaps appear somewhat mysterious; but the mystery will soon vanish, if it be considered that the accurate result depended not so much on the particular construction of the instrument, as on the regular management of it and uniformity of the operation. The exactness of the measure indeed contributed a great deal; but M. Fontana observed, that with exactly the same quantities of nitrous and common air, very different results could be occasioned by their being left a longer or shorter time before the instrument was agitated, or by being agitated much or little, as well as several other circumstances, which to a superficial observer would appear to be of little consequence. He therefore performed the operation always in a similar manner, viz. by agitating the tube always for the same length of time, and always with equal quickness; by which means, when the same elastic fluids were used, the results of the experiments were so nearly the same, that the difference, if any could be observed, might be neglected without any impropriety."

Notwithstanding the accuracy of this instrument, however, M. Fontana found that it was still liable to some small errors arising from the following sources:

1. The elastic fluid within the tube, when the greatest part of it is filled with water, and the tube is kept out of the water excepting its mouth, is not of the same density with the outward or atmospheric air, on account of the pillar of water in the tube; which, according as it is longer or shorter, counterbalances more or less the pressure of the atmosphere upon the quantity of elastic fluid contained in the upper part of the tube; which quantity of elastic fluid of consequence occupies a greater or less space in the tube, according to the greater or less pressure it endures. This error, however, becomes ineffible when the column of water is very short, and the surface of the water on the outside coincides nearly with that on the inside of the tube.

2. The difficulty of keeping the instrument perpendicular in the act of measuring the diminution.

3. The still greater difficulty of observing with what division of the ruler the surface of the water within the tube coincided.

To avoid these errors, M. Fontana made use of the following contrivance. \(AAAA\), fig. 5, represents avoiding strong glass tube about 3 inches diameter, and 18 inches long.

(b) The measure would be filled with elastic fluid though these notches were not made, but not so readily, because the water could not easily get out. inches long, with a foot of glass all made of one piece.

Within about an inch of the mouth of this tube a brass ring is fastened, which contains two brass rings movable upon opposite centres, in the same manner that sea-companies are usually suspended, and which are commonly called gingles. C C C C represents the measuring tube or eudiometer; which is exactly the same with that already described, having lines marked upon its outside surface to show the spaces occupied by equal measures of elastic fluid. The scale B B is adapted to this tube, which is shown separately in fig. 6. It consists of two brass slips A C, A C, connected by two brass rings A A, C C, through which the eudiometer tube passes. To the lowest of these rings a perforated brass piece B B, furnished with cross pins or pivots, is screwed; and, by means of longitudinal cuts, its lower extremity is rendered springy; so that when all the piece A B, A B is put upon the eudiometer tube, the latter cannot slip from within the former, unless the operator forces it. When the eudiometer tube, with the scale, &c., is put together, as represented fig. 5, the cross pins of the piece B B, fig. 6, rest upon the inner ring of the gingle at A A, fig. 5, by which means the tube C C C C is kept perpendicular within the tube A A A A, provided this latter be situated to nearly perpendicular that the former may not touch the side of it, which would prevent it from acquiring the position desired. One of the brass slips A C, fig. 6, is divided into equal parts; 100 of which are equivalent to the space between two of the marks on the eudiometer tube C C C C, fig. 5, and consequently show the parts of a measure. These divisions are numbered from the upper edge of the lower ring connecting the two brass slips, A C, A C.

When this instrument is to be used, one or more measures of respirable air are thrown into the eudiometer tube; a measure of nitrous air is then added; and after shaking the tube for some time, it must be introduced into the large tube A A A A, which for this purpose must be plunged into the water of the tub; for the mouth of the eudiometer tube must not at present be taken out of the water. After it has been introduced into the large tube, the whole is taken out of the water, and set upon the shelf or a table. Now the large tube A A A A is filled with water, and the eudiometer tube suspended perpendicularly in it by means of the cross pins or pivots of the brass piece annexed to the scale, which rests upon the inner ring of the gingle. The operator must then slide the tube C C C C up and down through the scale and brass piece, &c., till the surface of the water within the tube coincides exactly with the upper edge of the lower ring that connects the two brass slips of the scale piece, which may be done very accurately by means of a magnifying glass. The surface of the water within the eudiometer is concave; and when viewed horizontally, it appears like a dark line or limit exceedingly well defined; so that the middle or lowermost point of it may be made to coincide with the edge of the brass ring with great precision, except when some drops of water hang on the outside of the tube, which should therefore be wiped off.

Having ascertained this point, we must next observe which division of the scale coincides with one of the circular divisions marked upon the glass tube C C C C, which will show the parts of a measure. Thus suppose, that when the eudiometer tube is fixed, so that the surface of the water in it coincides perfectly with the edge of the lower brass ring, viz. with the beginning of the divisions; that the 70th division of the scale falls upon the first circular mark, as represented in the figure; then it is plain, that the quantity of elastic fluid contained in the tube is equal to one measure and 70 hundredth parts more. This being observed, and the large tube again immersed in the water, the eudiometer-tube is removed from it, but always taking care that its mouth be not lifted up above the surface of the water. Another measure of nitrous air must now be introduced into the eudiometer-tube; which, after being agitated as already directed, is to be put into the large tube A A A A. The whole is then taken out of the water, and the diminution of the elastic fluid observed as above directed.

Thus the eudiometer tube is kept quite perpendicular, and the pillar of water in it rendered very short, not exceeding half an inch at most. It is easy to perceive, however, that if the operator, when furnished with the eudiometer-tube only, keeps it so far immersed in the water of the tube when he observes the divisions, that the water within the tube may be nearly equal with the edge of the tube; the large tube A A A A may be spared, and the operation will thus become much more easy and expeditious. Little difference can happen from the position of the tube; because the brass ring acquires the position of the water so well, and the difference occasioned by a few degrees deviation from the true perpendicular is so small, that it can scarce be perceived.

VI. M. Sauffre of Geneva has invented an eudiometer, which he supposes to be more exact than any future's eudiometer of those hitherto described. His apparatus consists of the following parts. 1. A cylindrical glass bottle with a ground stopple, capable of containing about five ounces and a half, and which serves as a receiver for mixing the two airs. 2. A small glass phial, whose capacity is nearly equal to one third of that of the recipient, and serves for a measure. 3. A small pair of scales which may weigh very exactly. 4. Several glass bottles for containing the nitrous or other air to be used, and which may supply the place of the recipient when broken. The whole of this apparatus may be easily packed into a box, and thus transported from place to place, and even to the summits of very high mountains. The method of using it is as follows.

1. The receiver is to be filled with water, closed exactly with its glass stopper, wiped on the outside, and weighed very exactly. Being then immersed in a vessel full of water, and held with the mouth downwards, the stopple is removed, and, by means of a funnel, two measures of common and one of nitrous air are introduced into it one after another; these diminish as soon as they come into contact; in consequence of which the water enters the recipient in proportionable quantity. After being stopped and well shaken, to promote the diminution the receiver is to be opened under water; then stopped and shaken, and so on for three times successively. At last the bottle is stopped under water, taken out, wiped very clean and dry, and weighed exactly as before. It is plain, that now when the bottle is filled partly with elastic fluid and partly with water; it must be lighter than when quite full of water; the weight of it then being subtracted from the former, the remainder shows that quantity of water which would fill the space occupied by the diminished elastic fluid. Now, in making experiments with airs of different degrees of purity, the above mentioned remainder will be greater when the diminution is less, or when the air is more impure, and vice versa; and thus the comparative purity between two different kinds of airs may be determined.

On this method it is obvious to remark, that notwithstanding the encomiums bestowed on it by the inventor, it is subject to many inconveniences and errors, principally arising from the inaccuracy of the measure, and the difficulty of stopping the bottle without occasioning a pressure upon the contained elastic fluid, which being variable, must occasion some error in the weight of the bottle.

VII. To avoid the inconveniences to which all these instruments are subject, Mr Cavallo employs a glass tube with its scale and measure, such as is represented fig. 5, the length of the tube being about 16 or 17 inches, and between \( \frac{1}{2} \) and \( \frac{3}{4} \) of an inch in diameter, and of as equal a bore as possible throughout; having one end sealed hermetically, and the other shaped like a funnel, though not very wide. The whole of this apparatus is represented fig. 7, where \( AB \) is the glass tube, to the upper end of which a loop \( AEC \) should be fastened, made of waxed silk-lace, with several cords threads \( CC, DD, EE, \) &c., in order to suspend the instrument to a hook \( AB \), fig. 8, which should either be fastened to that side of the tube opposite to the shelf, or so constructed that it may be easily fixed and removed again as occasion requires; or it may be made of thick brass wire, the lower extremity of which fits a hole made in the side of the tub. The brass piece with the scale, which slides upon the eudiometer, is formed of two brass slips \( FG, HI \) (fig. 7.), joined by two brass rings, to which they are soldered. One hundred divisions are marked upon one of those brass slips, beginning from the upper edge of the lower ring \( GI \), and all together equal to the space contained between two of the marks or measures made upon the glass tube; so that they show the parts of a measure. An hundred divisions are likewise marked upon the other brass slip \( HI \), beginning from the lower edge of the upper ring \( FH \).—The following directions are given by Mr Cavallo for marking these divisions. "When the tube \( AB \) is filled with water, a measure of air should be thrown into it in the manner already directed; the tube must then be suspended to the hook by the loop, as represented fig. 8, so high, that the surface of the water within the tube may be very near the surface of the water in the tub, two inches, for instance, above it; then looking horizontally through the tube, a mark should be made by sticking a bit of soft wax upon the tube, just coinciding with the lower part of the surface of the water within it; in which place afterwards a circular mark should be made with the edge of a flint, or with a piece of agate or diamond, but not so deep as to endanger the breaking of the tube. Thus the first measure is marked; and in like manner may any other one be marked. The attentive practitioner, however, should never venture to mark the tube with an indelible stroke after one trial, lest he should be mistaken. The proper method is to mark them first with wax, and then repeat the operation once or twice, in order to correct some errors that might escape the first time; after which the mark may be made with a diamond, flint, or perhaps more conveniently with a file. The polish of the inside of both tube and measure should be taken off with emery; which is a very laborious operation, though it is particularly necessary that the measure should be done in this manner."

To use this eudiometer, fill the tube with water, taking care that no bubbles of air remain in it; and inverting it with the mouth downwards, leave it in the water leaning against the side of the tub. Fill the measure then with the elastic fluid whose purity is to be tried. Put the eudiometer tube upon the shelf of the tub, keeping it perpendicular, and with the mouth exactly upon the hole of the shelf, and throw the measure of air into it; fill it again with the same air, and throw this likewise into the tube. Then fill it with nitrous air, and throw this also into the tube, which must be shaken immediately after the operation by moving it alternately up and down in the water of the tub for about a quarter of a minute. It is then left a short time at rest and suspended by the hook formerly mentioned, so that the surface of the water in the inside may be about two inches above that in the tub; when the brass scale is slid upon it till the upper edge of the lower ring coincide with the middle part of the surface of the water within the tube, and then we may observe which division of the scale coincides with any of those on the tube; by which means the quantity of elastic fluid remaining in the tube may be clearly seen, even to the hundredth-part of a measure. The following directions are given by our author for noting down the results in a clear and accurate manner.

"1. The two measures first introduced into the tube are expressed by a Roman number; after which the noting single measure of nitrous air is expressed by another Roman number; and the measures, with the parts of a measure remaining in the tube after diminution, are expressed by common numbers with decimals.—Thus, suppose, that after introducing two measures of common and one of nitrous air, and after shaking in the manner above directed, the quantity of fluid remaining in the eudiometer is such, that when the upper edge of the lower ring of the scale coincides with the lower point of the surface of the water in the tube, the 56th division of the scale falls against the second circular division on the tube, then this diminution is marked thus II, I, 2,56; signifying that two measures of common and one of nitrous air, after diminution by being mixed together, occupy the space of two measures and 56 hundredth-parts of a measure.—Lastly, after marking the first diminution, throw a second measure of nitrous air into the tube; shake the instrument; and after a little rest, observe this second diminution: which, supposing it to have reduced the whole bulk to three measures and seven hundredth-parts, is thus marked down, II, II, 3,07. Sometimes one, two, or three measures of nitrous air must still be added, in order to observe the diminution of some very pure species of respirable air. The divisions which begin from the upper ring of the scale-piece of the eudiometer are useful when the quantity of elastic fluid..." contained in it is so small, that the edge of the lower brass ring cannot be raised so high as to coincide with the edge of the water within the tube on account of the silk loop; in which case the under edge of the upper ring is brought to that point; and we must then observe which of those divisions coincides with the first circular division upon the tube. If it be asked, Why the two or more measures of nitrous air are not thrown into the tube all at once, and the last diminution noted? the answer is, That in this method, the effects of similar experiments have not been found equally uniform with those tried in the above mentioned manner.

2. "In this operation care should be taken to shake the tube immediately after the nitrous air has been thrown into it, and to leave it at rest afterwards for some time; otherwise the results of similar experiments are far from being alike. It is also necessary to observe, that by holding the measure or the eudiometer tube with the hand, which is warmer than the water of the tub, the elastic fluid undergoes some degree of rarefaction, so that the event of the experiments may often be rendered precarious. For this reason the instruments should be held only with the extremities of the fingers and thumb; and before the door of the measure be shut, or the point of the scale on the eudiometer tube be fixed, those instruments should be left a short time by themselves, keeping the hands and breath at a sufficient distance from them."

The following are some particulars necessary to be observed in making experiments of this kind.

1. When respirable air is mixed with nitrous air, their joint bulk is diminished, and the diminution is greater when the air is purer, ceteris paribus, and vice versa.

2. On mixing the two airs together all at once, the ensuing diminution is greater than if the same quantity of nitrous air be added to an equal quantity of respirable air at different times: and hence it follows, that the quicker the two sorts of elastic fluids are mixed together, the greater is the diminution, and contrary-wise.

3. Nitrous air of different quality occasions a different degree of diminution with respirable air; and therefore care should be taken to use such materials as afford air always of the same quality. The most proper substance for this purpose is very pure quicksilver; a quarter of an ounce, or even less, with a proper quantity of diluted nitrous acid, will produce a great deal of nitrous air, which is always of the same quality, provided the metal be always of equal purity; but with other metals, as brass, copper, &c., the nitrous air made at one time is often different from that made at another, and therefore occasions a greater or less diminution when mixed with common air though precisely of the same sort.

4. The quality of nitrous air is impaired by keeping, especially when in contact with water; and for this reason it ought to be prepared fresh every two or three days.

5. In performing these experiments, it should be carefully remarked, that no mistake arise from heat or cold; as the elastic fluids are easily contracted or expanded by any variation of temperature.

6. Though the greatest diminution takes place immediately after mixing the respirable and nitrous airs together, especially when they are agitated, yet they continue to diminish a little for some time after; for which reason the diminution should be observed always at a certain time after the mixture is made. The whole process indeed ought always to be performed in an uniform manner, otherwise the results will be frequently very dissimilar.

7. It must be remarked, that the surface of the water which lies contiguous to the elastic fluid contained in a small vessel, is very far from being a plane, or even from being always of a similar figure in the same vessel, on account of the attraction or repulsion between the substance of the glass and water. This is altered by many circumstances, particularly by the adhesion of extraneous bodies; whence it is very improper to use common open phials for this purpose. We must also take into consideration the drops of water adhering to the sides of the vessel, and the quality of the water in which the operation is performed.

8. In case the experiment is to take up some hours, in order to observe the last diminution, it will be proper to notice, by a good barometer, if the gravity of the atmosphere has suffered any alteration during that time; for a difference in its pressure may occasion some difference in the result of the experiments.

9. A simple apparatus is always to be preferred to a more complicated one, even though the latter should appear to have some advantage over it in point of accuracy. Complex machines are not only expensive, and subject to be easily put out of order, but occasion frequent mistakes, on account of the operator having generally many things to do and keep in proper order; whence it is easy to overlook some of them.

It has already been remarked, that one source of error in the experiments made with eudiometers is the sources of inequality of the column of water in the tube by which error in the mixture of elastic fluids is confined. For example, if a cubic inch of air, taken near the apparatus where the experiment is to be performed, be introduced into a long tube previously filled and inverted in water, so that the surface of the water in the tube may be 20 inches higher than that in the basin, the air in the upper part will then be found to occupy considerably larger space than if the column of water was shorter; because in the former case the pressure of the water in the tube partly counterbalances the pressure of the atmosphere, so that the latter is less able to resist the elasticity of the confined air. The difference will be much greater if quicksilver be made use of instead of water, as the weight of that fluid is much greater than that of water. To avoid this, it has been directed to manage matters so that the surface of the fluid on the outside may nearly correspond with that in the inside of the tube; but this is sometimes impracticable, especially where quicksilver is used, with which the error is more considerable than with water: in such cases, therefore, we must have recourse to calculation, and deduce the real quantity of elastic fluid from the apparent space it occupies in a receiver, which is partly filled with it and partly with water or some other gross fluid. For this purpose it must be remembered, that the spaces into which air or any other elastic fluid is contracted, are to one another in the inverse ratio of the pressures which confine these elastic fluids; hence the space occupied by a quantity of elastic fluid $AB$, (fig. 9.) confining in the tube AC inverted in quicksilver, and filled with it as far as B, is to the space which the same quantity of fluid occupies out of the tube, as the pressure which acts upon it when out of the tube is to the pressure which acts upon it in the tube; that is, as the height of the barometer, to the same height of the barometer deducting the height BC of the quicksilver in the tube. Thus, suppose that the length AB of the tube occupied by an elastic fluid is three inches, and that the length BC, filled with quicksilver, is 20 inches; it is required to determine the length of the same tube, which the same quantity of elastic fluid would occupy if the surface of the quicksilver in the basin was brought even with B, viz. if the said elastic fluid was only acted upon by the pressure of the atmosphere. First observe the height of the barometer at that time, which suppose to be 30 inches; then say, As the height of the barometer is to the same height deducting the height of the quicksilver CB in the inverted tube AB; so is the space AB to the real space required; that is, \( \frac{3}{30} : \frac{30 - 20}{30} = 1 : 1 \); so that one inch is the length of the tube AC which the quantity of elastic fluid AB would occupy, if the surface B of the quicksilver in it was brought even with that of the quicksilver in the basin. Here, however, we must suppose the tube AC to be perfectly cylindrical; otherwise the calculation would become very intricate by being adapted to the form of the vessel.

VIII. In the 73rd volume of the Philosophical Transactions, we have an account of a new endiometer by Mr Cavendish. He prefers the Abbe Fontana's to all the rest: the great improvement in which (he says) is, that as the tube is long and narrow, and the orifice of the funnel not much less than the bore of the tube, and the measure made to deliver its contents very quick, the air rises slowly up the tube in one continued column; so that there is time to take the tube off the funnel, and to shake it before the airs come quite into contact; by which means the diminution is much greater and more certain than it would otherwise be. Thus, if equal measures of nitrous and common air are mixed together in this manner, the bulk of the mixture will, in general, be about one measure; but if the airs are suffered to remain in contact about a quarter of a minute before they are shaken, the bulk will hardly be less than one measure and one fifth; and it will be very different according to the length of time they are suffered to remain before they are shaken. In like manner, if, through any fault in the apparatus, the air rises in bubbles, as in that case it is impossible to shake the tube soon enough, the diminution is always less than it ought to be. Another very considerable advantage arising from the method of mixing the airs just mentioned is, that the diminution takes place in its full extent almost instantly; but if they are allowed to remain for some time in contact before they are shaken, the mixture will continue diminishing for many hours afterwards.

The reason of these differences, according to our author, is, that in the Abbe Fontana's method, the water is shaken briskly up and down in the tube while the airs are mixing; by which means every small portion of nitrous air must be in contact with water either at the instant it mixes with the common air, or at least immediately after; and it seems that the water, by absorbing the nitrous acid the moment it is formed, greatly contributes to the quickness of the diminution, as well as to the quantity of it. Hence Mr Cavendish was induced to try whether the diminution of adding would not be more certain and regular, if one of the airs were added to the other slowly and in small airs slowly bubbles, the vessel being kept shaking all the while; so that the mixture was made; and on trial he found that this method fully answered his expectations.

The apparatus used by our author is, i. A cylindrical glass vessel A (fig. 10), with brass caps at top and bottom. To the upper cap a brass cock B is fitted; the bottom cap is open, but made to fit close into the brass socket D, and is fixed into it in the same manner as a bayonet is on a musket. This socket has a small hole E in its bottom, and is fastened to the board of the tub by the bent brass F G, in such a manner that b, the top of the cock, may be about half an inch under water; consequently, if the vessel A is placed in its socket with any quantity of air in it, and the cock is then opened, the air will run out by the cock; but will do so very slowly, as it can escape no farther than the water can enter by the small hole E to supply its place.

2. Besides this vessel, there are three glass bottles like M, fig. 11, having each a flat brass cap at bottom to make it stand steady, and a ring at top to suspend it; also some glass measures of different sizes, as B fig. 12, having a flat brass cap at bottom with a wooden handle. These are filled with the air to be measured, then set upon the brass knob C fitted to the board of the tub below the surface of the water, which drives out some of the air, leaving only the proper quantity.

In mixing the airs together, our author commonly adds the respirable slowly to the nitrous; to do which, a proper quantity of nitrous gas is put into the bottle M, by means of one of the measures already described, and another quantity of respirable air is put into the vessel A, by first filling it with this air, and then putting it on the knob C, as was done by the measure; after which the vessel A is fixed in the socket, and the bottle M placed with its mouth over the cock. The quantities of air made use of, and the diminution of the mixture, are determined by weighing the vessels under water in the following manner. From one end of a balance, placed in such a manner as to hang over the tub of water, a forked wire is suspended, to each end of which fork is fixed a fine copper wire; and in trying the experiment, the vessel A, with the respirable air in it, is first weighed by suspending it from one of those copper wires, so that it may remain entirely under water. The bottle M, with the proper quantity of nitrous air in it, is then hung in the same manner on the other wire, and the weight of both together determined. The air is then let out of the vessel A into the bottle M, and the weight of both vessels together found a second time; by which we know the diminution of bulk the airs suffer on being mixed. Lastly, the bottle M is taken off, and the vessel A weighed again by itself, which gives the quantity of respirable air made use of. It is needless to determine the quantity of nitrous air by weight; because, as the quantity... quantity used is always sufficient to produce a full diminution, a small difference therein makes no sensible one in the diminution. No sensible error can arise from any difference in the specific gravity of the air; for the thing found by weighing the vessel is the difference of weight of the included air and an equal bulk of water; which, as air is no less than 800 times lighter than water, is very nearly equal to the weight of a quantity of water equal in bulk to the included air. A common balance is not convenient for weighing the bottles under water, without some addition to it: for the lower the vessel of air sinks under water, the more the air is compressed; which makes the vessel heavier, and thereby causes that end of the beam to preponderate. Hence we must either have the index placed below the beam, as in many effay-balances; or by some other means remove the centre of gravity of the beam so much below the centre of suspension, as to make the balance vibrate, notwithstanding the tendency which the compressibility of the air in the vessels has to prevent it.

In this manner of determining the quantities of the air by weight, care must be taken to proportion the lengths of the copper wires in such a manner that the surface of the water in A and M shall be on the same level, when both have the usual quantity of air in them; as otherwise some errors will arise from the air being more compressed in one than the other. This precaution, indeed, does not entirely take away the error, as the level of the water in M is not the same after the airs are mixed that it was before; but in vessels of the size used by our author, this error could never be equal to the fiftieth part of the whole; which therefore is quite inconsiderable: but even if it was much greater, it could be of no consequence, as it would always be the same in trying the same kind of air.

The vessel A (fig. 10.), used in these experiments, holds 282 grains of water, and is the quantity denominated one measure by our author. There are three bottles for making the mixture, with a measure B (fig. 12.) for the nitrous air adapted to each. The first of these holds three measures, and the corresponding measure one and one-fourth of the former measure; the second bottle holds six, and the corresponding measure two and a half; the third holds 12, and the corresponding measure five. The first bottle and measure are made use of in trying common air, and the others for the dephlogisticated or purer kinds. As the same quantity of respirable air is always made use of, 1½ measure of nitrous air is added to one of the common atmospheric kind; and in trying very pure dephlogisticated air, five measures of the nitrous kind are made use of; and our author is of opinion, that there is no kind of air so pure as to require a greater quantity of nitrous air. The way by which it is known whether a sufficient quantity of nitrous air has been added, is to observe the bulk of the mixture; for if that is not less than one measure, that is, than the respirable air alone, it is a sign that the quantity of nitrous air is sufficient, or that it will produce the proper diminution, unless it be very impure. It must be observed, however, that though the quantity of respirable air will always be nearly the same, as being put in by measure, yet the observed diminution will commonly require some correction. For example, suppose that the observed diminution was 2.353 measures, and that the quantity of respirable air was found to be .985 of a measure; then the observed diminution must be increased by .035, in order to have the true diminution, or that which would have been produced if the respirable air made use of had been exactly one measure; whence the true diminution is 2.388.

In weighing common air, our author somewhat abridges the process above described. He does not weigh the vessel A, but only the bottle M with the nitrous air in it; then mixes the airs, and again weighs the same bottle with the mixture in it, and finds the increase of weight; which added to one measure, is very nearly the true diminution whether the quantity of common air made use of was a little more or a little less than one measure. The reason of this is, that as the diminution produced by the mixture of common and nitrous air is only a little greater than the bulk of the common air, the bulk of the mixture will be very nearly the same whether the bulk of the common air be a little greater or a little less than one measure. Let us suppose, for example, that the quantity of common air made use of is exactly one measure, and that the diminution of bulk on mixing is 1.08 of a measure; then must the increase of the weight of the bottle M, on adding the common air, be .08 of a measure. Let us next suppose that the quantity of common air made use of is 1.02 of a measure; then will the diminution, on adding the nitrous air, be 1.08 + \(\frac{1.02}{1.00}\) or 1.1016 of a measure; and consequently the increase of the weight of the bottle M will be 1.1016 - 1.02, or .0816 of a measure, almost exactly the same as if precisely one measure of common air had been made use of.

The same bottle is made use of, viz. that which holds three measures, when the nitrous is added to the respirable air. In this experiment the bottle M is first to be reweighed without any air in it, and then weighed again spirable, when full of respirable air, which gives the quantity of the latter made use of. The nitrous air is then put into the vessel A, and weighed together with the bottle M; after which, having mixed them together, the diminution takes place, and they are weighed again, in order to discover its quantity. In this method a smaller quantity of nitrous air is necessary than in the former. In the first method, it was found that the diminution was scarce sensibly less when one measure of nitrous air was used than with a much larger quantity; so that one measure may be accounted fully sufficient. Our author, however, chose to employ 1\(\frac{1}{4}\) measure, lest the nitrous air should be impure. There was no sensible diminution whether the orifice of the vessel A opening into the bottle M was \(\frac{1}{10}\)-th or \(\frac{1}{5}\)-th of an inch; that is, whether the air escaped in small or large bubbles: the diminution was rather greater when the bottle was shaken briskly than otherwise; but all the difference that could be perceived between these two methods of shaking did not exceed .01 of a measure. The diminution, however, was remarkably less when the bottle was not shaken at all; being at first only .09; in about three minutes it increased to .092; and after being shaken for about a minute, it increased to .099; but when gently shaken at first, the diminution was 1.08 on mixing, and did not sensibly increase after that time. Some difference was found to arise from the length of time the air took up in passing from one vessel to another. When it took up 80 seconds, for instance, in passing from the one bottle into the other, there was a difference of 5 hundredth-parts more than when it took up only 22 seconds, and about 2 hundredth-parts more than when it took up 45 seconds; but at other times the difference was less. As the hole in the plate Dd, however, was always the same in our author's experiments, the time taken up by the air in passing from one vessel into the other varied so little that no perceptible difference could arise from that cause. A greater difference arose from the size of the bottles and quality of the water made use of. When the small bottle, holding three measures, was used, and filled with distilled water, the diminution of common air was usually 1.08; but when the bottle was filled with water from the tub, it was .05 less. Using the bottle which held 12 measures, and filled with distilled water, the diminution was about 1.15; and with the same bottle filled with water from the tub it was usually 1.08. "The reason of this (says Mr Cavendish) is, that water has the power of absorbing a small quantity of nitrous air; and the more deploglificated the water is, the more of this air it can absorb. If the water is of such a nature as to froth or form bubbles on letting in the common air, the diminution is remarkably less than in other water. In general the diminution was nearly as great with rain as with distilled water; but sometimes the former would froth a good deal; in which case it was no better than water fouled with oak-shavings. This difference of diminution, according to the nature of the water, is a very great inconvenience, and seems to be the chief cause of uncertainty in trying the purity of the air; but it is by no means peculiar to this method, being equally great in that of Fontana's. In his method indeed it makes little difference whether the water be disposed to froth or not; but this is no great advantage, as it is easy to find water which will not froth; though it shows plainly how little any of the experiments hitherto made on the purity of air can be depended upon." The best method of obviating this inconvenience is to be always careful to use the same kind of water: our author always made use of distilled water; but found that even this was sometimes endowed with a greater power of absorbing nitrous air than at others: and with a view to remedy this, he made the following experiment. Some distilled water being purged of its air by boiling, one part was kept for a week in a bottle with deploglificated air, and frequently shaken; the other part being treated in the same manner with phloglificated air. By a mean of three different trials the test of common air tried with the first of these waters was 1.139; the diminution suffered by shaking nitrous air in it for two seconds being about 0.285. The test of the same air tried with the other water was 1.054, and the diminution by nitrous air only 0.09; the heat of the water in the tub and of the distilled waters being 45°. The heat of the water in the tub and the distilled waters was then raised to 67°; when the test of the same air tried by the first water was 1.100, and by the latter 1.044; the diminution of nitrous air with the first water being 0.235; by the latter 0.089. Hence it might seem that the observed test ought to be corrected by the subtraction of \(\frac{1}{3}\)ths of the diminution which nitrous air suffers by being shaken in the water, and adding 0.02 for every three degrees of heat above 0; but though this correction will undoubtedly diminish the error, he is of opinion that it will not by any means take it away entirely; and from some circumstances it appears that distilled water possesses a property of absorbing different quantities of nitrous air independent of its heat.

In the second method, viz. when the nitrous acid is added to the common air, the diminution is considerably less than in the other; the reason of which is, that when nitrous and common air are mixed together, the former is deprived of part of its phlogiston, and thereby converted into phloglificated nitrous acid, and in that state is absorbed by the water; besides that the common air is phloglificated, and thereby diminished: so that the whole diminution on mixing is equal to the bulk of nitrous air which is turned into acid, added to the diminution which the common air suffers by being phloglificated. Now it appears, that when a small quantity of nitrous air comes in contact with a large one of common air, the former is more completely deprived of its phlogiston, and absorbed by the water in a more deploglificated state than when a small quantity of common air comes into contact with a large quantity of nitrous: in the second method, therefore, where small portions of nitrous air come in contact with a large quantity of common air, the former, as has been just observed, is more deprived of its phlogiston; and therefore a smaller quantity of it is required to phloglificate the common air than in the former method, where small portions of common air come in contact with a large quantity of nitrous air; so that a less quantity of the nitrous air is absorbed in the second method than in the first. The common air most probably suffers an equal diminution in both cases.

Another proof that a smaller quantity of nitrous air is required in this method than the former is, that if common air be mixed with a quantity of nitrous air not sufficient to phloglificate it, the mixture will be more phloglificated if the nitrous be added slowly to the common air without being in contact with water; the mixture will be found to be still more phloglificated than in the second method where the two airs are in contact with water at the time of mixing. The final result of Mr Cavendish's experiments on this subject is, Conclusions that nitrous air used in the first method does not phloglificate common air more than three-fourths of the extreme quantity used in the second way; and not too much, as one half of the quantity used in the third way, viz. by adding the nitrous air slowly to the other, without being in contact with water.

With respect to the quality of nitrous air used in these experiments, our author observes that it may vary in two respects. 1. In purity; that is, in being more or less mixed with phloglificated or other air. 2. In two parcels of equally pure air, it is possible that one parcel may contain more phlogiston than the other. A difference in the second respect will cause an error in the test, in whatever proportion it be mixed with the respirable air; but if it differs in the first respect, it will scarcely cause any error unless it be uncommonly impure; provided care is taken to use a quantity sufficient to make a full diminution. It must be observed, however, that if the nitrous air be mixed with fixed air, an error will be occasioned, because part of the latter is absorbed while the test is trying; but this will hardly be