XXXIV .
Experiments and Observations made in Britain , in order to obtain a Rule for measuring Heights with the Barometer .
By Colonel William Roy .
F. R. S. Read June 12 and 19 , and Nov. 6 and 13 , 1777 .
INTRODUCTIO N. IN philosophical inquiries of every kind , where any point is to be ascertained by experiments , these cannot be repeated too often , nor varied too much , in order to obtain the truth : for even when the utmost precaution hath been used , and the greatest pains have been taken , it rarely happens , that they agree so exactly , as to leave no room for doubt .
Were it possible at all times , to have experiments made in circumstances perfectly similar , a considerable degree of consistency might naturally be expected among the results , whereof the mean would determine the point in question ; but different men , making use of different instruments , have different modes of conducting their operations , each pursuing the tract that seems to him the most likely to insure success .
Hence it is that a variety of results arise , and that things sometimes appear contradictory , or at least present themselves under new forms , difficult at first sight to be accounted for , and therefore apt to mislead , till by a farther investigation of the matter , the true causes are discovered .
Even irregularities of this sort are worthy of being communicated , that others may know what hath happened before , and what , in like cases , they may expect to meet with , in the course of their future inquiries .
Improvements of every kind advance by slow degrees ; and it is not until things have been viewed in every possible light , that the errors , even of our own experiments , are discovered , the points in question ultimately ascertained , and the branch of philosophy depending upon them , gradually brought nearer to perfection .
Ever since the discovery made by TORRICELLI , the barometer hath been applied , by different persons , in different countries , to the measurement of vertical heights , with more or less success , according to the more or less perfect state of the instruments used , and the particular modes of calculation adopted , by the observers .
But of all those who have hitherto employed themselves in this way , none hath bestowed so much time and pains , or succeeded so well , as Mr. DE LUC , of Geneva , F. R. S. In two quarto volumes , published some years since , that gentleman hath given us the history of the barometer and thermometer , with a very curious and elaborate detail of many years experiments , made by him , chiefly on the mountain Saleve .
It would be totally superfluous here to enter into any circumstantial account of the method he makes use of , since that hath already been so fully illustrated by two Fellows of the Royal Society , who have at the same time given formulae and tables , adapted to the measures of this country , ( Phil. Trans. for 1774 , vol. LXV .
No xx .
and xxx .
) that nothing farther can be desired on that head .
It may nevertheless be necessary just to call to remembrance that the rule , deduced from the observations on Saleve , consists of three parts .
1st , The equation for the expansion of the quicksilver in the tube , from the effect of heat , whereby the heights of the columns , in the inferior and superior barometers , are constantly reduced to what they would have been in the fixed temperature of 54~1/ 4 of FAHRENHEIT , independant of the pressure they respectively sustained .
2d , When the mean temperature of the column of air to be measured , is 69~.32 , as indicated by thermometers exposed to the Sun 's rays at its extremities ; then the difference of the common logarithms , of the equated heights of quicksilver in the two barometers , gives the altitude intercepted between them , in toises and thousandth parts , reckoning the three figures to the right hand decimals , and the others integers , the index being neglected .
This temperature of 69~.32 , when the logarithmic differences give the real height without any equation , is reduced to 39~.74 , the new zero of Mr. DE LUC 's scale , when his formula is adapted to English fathoms and thousandth parts , instead of French toises .
And lastly , when the mean temperature of the air is above or below 39~.74 , an equation , amounting to 21/ 10000 parts of the logarithmic height for each degree of difference , is , in the first case to be added to , and in the last substracted from , that result , in order to obtain the real altitude .
In Mr. DE LUC 's book , the experiments for ascertaining the expansion of the quicksilver , are not given in detail ; neither are the particular temperatures of the barometers specified .
The winter season was however chosen forth purpose ; one being left in a cold room , and the other in a closet , heated as high as could conveniently be suffered .
The operation having been repeated several times without any essential difference in the results , this general conclusion is drawn , that between the temperatures of melting ice and boiling water , the expansion of the quicksilver is exactly six French lines , or.532875 decimal parts of an English inch .
But it is to be observed , that the barometer stood then at 28.77525 ; whereas , if it had stood at 30 inches , it would have been .555556 , because the expansion is in proportion to the length of the column .
Farther , the interval between the freezing and boiling points in all thermometers , varies with the height of the barometer , or weight of the atmosphere ; and it is the custom in England to make thermometers when the barometer stands at 30 inches ; that is to say , 1.225 or 13.8 French lines , higher than when Mr. DE LUC 's boiling point was fixed : and since from his experiments it appears , that each line of additional height in the barometer , raises the boiling point 1/ 1134th part of the interval between that and freezing , it follows that 180/ 1134 = 0.158 x 13.8 = 2~.2 , will denote the number of degrees , that Mr. DE LUC 's boiling point is lower than that of English thermometers , which reduces it to 209.8 of FAHRENHEIT , and makes the interval between freezing and boiling only 177.8 degrees .
Hence the expansion.555556 , formerly found , must be increased in the proportion of 177.8 to 180 , which gives for the total .5624297 or.56243 , on a difference of temperature of 180~ .
Thus the expansion for each degree , supposing it to be arithmetical , or uniformly the same in all parts of the scale , will be.00312461 ( a ) .
Having now shown the expansion of quicksilver in the tubes of barometers resulting from the Geneva observations , I shall next proceed to give some account of those I made for that purpose .
They derive their origin from my having very accidentally observed , that a small degree of heat , and of short duration , sensibly affected the length of the column in Mr. RAMSDEN 's portable barometer , whereof a view is given in plate XVI .
The principal parts of this instrument are a simple straight tube , fixed into a wooden cistern , which , for the convenience of carrying , is shut with an ivory screw , and , that being removed , is open when in use .
Fronting this aperture is distinctly seen , the coincidence of the gage-mark , with a line on the rod of an ivory float , swimming on the surface of the quicksilver , which is raised or depressed by a brass screw at the bottom of the cistern .
From this , as a fixed point , the height of the column is readily measured on the scale attached to the frame , always to 1/ 500 ( a ) This paper having lately been communicated to Mr. DE LUC , he hath informed me , that the difference of temperature in his experiments , amounted to about 31~ of REAUMUR , or 72~ of FAHRENHEIT , above freezing : wherefore,.00312461 x 72 =.225 nearly , will denote the rate of expansion from which he deduced that for 180~ ; and within these limits , it will hereafter be oundto differ very little from the result of the present experiments .
part of an inch , by means of a nonius moved with rackwork .
A thermometer is placed near the cistern , whose ball heretofore , was usually enclosed within the wood work , a defect that hath been since remedied .
The threelegged stand , supporting the instrument when in use , serves as a case for it , when inverted and carried from place to place .
Two of these barometers , after the quicksilver in them hath been carefully boiled , being suffered to remain long enough in the same situation , to acquire the same temperature , usually agree in height , or rarely differ from each other more than a few thousandth parts of an inch , which were constantly allowed for in calculating altitudes , as well as in estimating the rate of expansion , in the course of the following experiments .
SECTION I. Experiments on the expansion of quicksilver .
THE experiments made for this purpose were numerous as well as various , and were therefore subdivided into several classes .
To give a minute detail of them all , would be extremely tedious , and now wholly useless , since it was from those of the third class alone , that the rate as well as maximum of expansion was ascertained : wherefore those of the two preceding classes need only be mentioned in a general way .
The first set of the first class comprehended such as were made with one barometer in a cold room , or in the open air , and the other in a room on the same level with the former , where there was constantly a fire , which was occasionally increased , in order to augment the difference of temperature .
When the heated barometer had remained several hours in an angle of the room , the difference of temperature of its quicksilver above that of the coldest , as indicated by their respective attached thermometers , rarely exceeded 10 or 120 , which , from a mean of many observations , gave an expansion of.0333 decimals of an inch , for the 100 comprehended between 32 and 420 of FAHRENHEIT'S thermometer .
So far the result arising in this way , from small differences of temperature , will be found to agree with the third class of experiments .
But when , in the second set of this first class , the difference of temperature was augmented to 20 or 300 , by exposing the barometer within doors to a greater heat , or placing the superior one on the leads , whereby it received the direct and reflected rays of the Sun throughout the greatest part of the day , while the other was kept in the cold area underneath , the rate of expansion for the first 100 exceeded that formerly found nearly in the proportion of three to two , while that for the second and third terms , of 100 each , diminished progressively .
The chief , though not the only cause of this great difference , as will appear hereafter , arose from the position of the ball of the thermometer , originally enclosed within the wood-work of the frame , which prevented it from receiving the heat so readily as the quicksilver in the tube ; at the same time that it retained it longer , and consequently produced results in some degree fallacious .
Finding , from the first class of experiments , that much uncertainty remained with regard to the rate of expansion of quicksilver affected by these smaller degrees of heat , and that it was utterly impossible , from them , to determine its maximum for the 1800 between freezing and boiling ; I resolved to try , how much a column of 30 inches of quicksilver , carefully boiled in a tube , would lengthen , the same being placed with the open end upwards in a tin vessel , occasionally filled with pounded ice and water , and afterwards brought to boil , by means of a charcoal fire placed underneath ?
In this second class , it was easy to see , that the expansion of the tube containing the quicksilver , was necessarily to be taken into the account , and added to that apparently found by experi2 meant .
meant .
This was of course to be done , either by such differences as I could discern and measure , or by those that had resulted from the experience of others .
The nature of the apparatus , employed in this class of experiments , will be easily understood from plate XVII .
where it is represented , as it was used in those of the third class .
In its first state it was not quite so long , and a chasing-dish with a charcoal fire , occupied the place of the cistern holding the quicksilver below .
By means of a circular bit of tin , soldered edgeways in the center of the bottom , and an aperture in the middle of the lid , the tube was kept steadily in the axis of the vessel .
Other openings in the lid , served for the admission of the thermometer , and the application of a deal rod close to the side of the tube , when its height was to be measured .
The longitudinal expansion of the glass was marked by a scratch thereon with a fine edged file at the top of the deal rod , when respectively at the temperatures of freezing and boiling .
The apparent dilatation of the quicksilver was in like manner marked , by the coincidence of its surface with the lower edge of a brass ring embracing the tube .
It having been found impossible to procure tubes whose bores were truly cylindrical , or of any uniform figure , the experiment was repeated , as often as possible , in both ends of the same tube , that the mean might be taken .
But it frequently happened that the tube , which had undergone one or more experiments in one end , broke before any could be made with it in the other .
In this case , the rate of expansion in the last end was taken from that given by such another tube , where it had succeeded in both .
The mean of five results with the best tubes , taken in this way , gave.4901 for the apparent expansion of 30 inches of quicksilver , on 180~ of FAHRENHEIT , between freezing and boiling , which being augmented by the apparent longitudinal dilatation of the glass.0356 x3 =.1068 , the real expansion is.5969 ; exceeding Mr. DE LUC'S by more than 3/ 100ths of an inch .
If , however , Mr. SMEATON'S dilatation of glass , ( .025 x3= .075 ) be substituted , instead of that resulting from these experiments , the real expansion of 30 inches of quicksilver will be.5651 , which does not exceed it quite 3/ 1000 parts of an inch .
In this class of experiments , having attended as diligently as possible to all the circumstances , it seemed to me , that tubes with a small bore , and whose glass was thick , lengthened more than those , which had a larger bore and whose glass was thin : whence I was led to suppose , that solid glass rods would dilate more in proportion , and consequently , show a still more perceptible difference .
With the view of ascertaining this point , I procured four glass rods near three feet long each , and of different diameters , the largest being of the size of the little finger , and the smallest about the thickness of a quill .
One end of each , was somewhat larger than the other , and was made perfectly smooth , as that on which they were to rest when severally measured with the deal rod .
They were then all placed in the tin vessel , in such a manner , as to admit pounded ice rammed very close around them , and the interstices to be filled with water .
Having remained in that state a full half hour , they were severally measured with the deal rod , whose length of 321/ 2 inches was scratched on each with the sharp edge of the file .
This being done , the ice thrown out , and the vessel carefully washed , all the rods were replaced in it , immersed in water , which afterwards was brought to boil .
The fire being kept up , and the ebullition rendered as violent as possible for half an hour , the glass rods were then severally measured , by applying them one after another to the deal rod , standing with them in the boiling water .
The experiment was repeated three times , on as many different days , without its being possible to discern , that any of the glass rods had dilated more than that of deal , from a difference of temperature of 180~ .
In all of them , the freezing mark seemed accurately to coincidewith the top of the deal rod ; whereas the dilatation of the tubes , by the same degree of heat , was always very visible(b ) .
Finding from the second class of experiments , whereof the general result hath now been given , that glass rods seemed not to lengthen more than deal ; and that tubes of different bores , and probably too of different sorts of glass , were susceptible of different degrees of extension , which rendered it impossible , by this means , to ascertain the longitudinal expansion of the quicksilver they contained ; I thought it necessary to recur once more to the barometer , and to try whether it could not be so contrived as to act in water of different temperatures , from freezing to boiling .
This led me to the experiments of the third class : and in order to comprehend them thoroughly , it seems necessary to point out some few alterations which the apparatus underwent .
The center of the bottom being pierced on purpose , a brass socket was prepared for it , wherein a hole was bored conically , to receive the ground-end of a barometer tube , of the ordinary length of 331/ 2 inches ; the tube having been first ground in a separate piece of brass , and ulti(b ) Since these experiments were made , the relative expansion of 18 inches of one of the tubes and one of the rods that had been formerly employed , was found to be , by Mr. CUMMING'S pyrometer , nearly as 4 to 1 , from a heat approaching to that of boiling oil .
mately in the socket itself , fitted it so exactly , as to suffer no water to pass .
The socket being inserted into the aperture at the bottom of the vessel , was firmly soldered to it for the reception of the tube , which was so ground as to reach a full inch and a half below the surface of the brass .
It could not descend farther , the ground parts in both being of the figure of the frustum of an inverted cone .
From the view in the plate it will appear , that underneath the vessel , a separate stand was placed , in order to support the iron cistern containing the quicksilver .
The diameter of the cistern was such , that its stand being occasionally moved , so as to bring one side of it close to the ground part of the tube , the other side projected beyond the bottom of the vessel ; and consequently permitted the rod of a float , resting on the surface of the quicksilver , to rise freely and parallel to the axis of the tube .
The rod was of deal , 1/ 10th of an inch square , carrying on its top a scale , whose zero lay in the lower surface of the float , and whereof the six uppermost inches , from 28 to 34 , were divided into 20ths .
That the whole column of quicksilver might alternately be covered with the freezing mixture and water of different temperatures , and yet permit its surface to be seen , two eyes of plate glass were screwed into sockets , soldered for that purpose opposite to each other , near the top of the vessel , which , in the first set of the third class of experiments , was little more than 29 inches high .
The top of the tube passing through the aperture in the lid , one and a half or two inches of the vacuum generally rose above the vessel .
That the expansion of the column might be measured as nearly as possible in that part of the tube fronting the center of the eyes , more or less quicksilver , according to the state of the atmosphere , was occasionally put into the cistern , to raise or depress the surface of the column to the proper height .
A thin brass ring , whose lower parts were made to spring , embraced with sufficient force the upper part of the tube , permitting it at the same time to be moved freely with the hand .
It carried along with it a nonius index , projecting as far as the center of the rod , and consequently applying itself to the divisions of the scale , which was kept in its proper position by passing through a slit fitted for it in an arm attached to the lid .
The divisions on the nonius being the same with those of the barometer formerly described , the height of the quicksilver could always be read off to 1/ 500th part of an inch .
The quicksilver having been carefully boiled , as on former occasions , in the tube ; and that being filled completely , and held with its open end upwards ; the tin vessel was inverted over it , and lowered gradually , till the ground end could be inserted into the socket with such a degree of force as to prevent it from being too easily removed .
The finger being then applied closely to the open end of the tube , the whole apparatus was turned up , and placed over the cistern into which the quicksilver had previously been put , great care being taken not to remove the finger till the lower extremity of the tube was fairly immersed into the quicksilver ; when that in the tube was permitted to descend into an equilibrium with the atmosphere .
In the first experiment it was found that the water issued by the eyes , and running down the side of the vessel , fell into the cistern .
In order to remedy this inconveniency , a circular piece of tin was soldered round the upper part of it , immediately below the eyes ; and a flat spout , projecting from it , served as a gutter to throw off the water from the cistern , and from the lamps made use of to bring that in the vessel to boil .
Six lamps , each with a double light , were suspended around the trunk of the vessel , to heat the water as equally as possible ; though any irregularity of this kind was sufficiently guarded against , by constantly mixing it during the operation .
Another lamp of the same kind stood under the cistern , whereby the quicksilver there was kept at the temperature of the water in the vessel , each having its proper thermometer : this last lamp was placed and displaced frequently , during the course of every experiment ; for the heat was very expeditiously communicated to the iron cistern , and thence to the quicksilver it contained ; and both were found to cool very fast , after the lamp was removed .
Such was the state of the apparatus , when the first set of this third class of experiments was made .
In those of the second set , its height was farther augmented by tin soldered to the top , that a tube of the ordinary length might be wholly immersed in boiling water .
The third and last alteration consisted in the occasional application of a detached tin case , equal in diameter to the upper part of the vessel , having a hole in its bottom to admit the top of a long tube to pass .
This case was so contrived , that its bottom stood two inches and a half higher than the lid of the vessel , thereby allowing room for the hand to move the index up or down .
In this state the apparatus is represented in the view ; and its various uses will be best understood from the account of the experiments , which were subdivided into four sets .
Those of the first set were made with tubes of a large bore , upwards of three-tenths of an inch in diameter , of the ordinary length , with a vacuum over the quicksilver of two inches and a half or three inches , part of which reached above the top of the vessel .
The mean of three experiments gave.5258 , for the total dilatation of 30 inches of quicksilver , on 180~ between freezing and boiling ; that , answering to the first 20~ , between 32~ and 52~ , was.0688 ; that , for the 20~ in the middle of the scale , between 112~ and 132~ , was.058 ; and the rate for the last 20~ , between 192~ and 212~ , was only.041 .
From this first set of the third class of experiments , it appeared evident , that the expansion of 30 inches of quicksilver in the barometer , suffering a heat equal to 180~ of FAHRENHEIT , instead of exceeding Mr. DE LUC'S , as appeared to be the case from the results of the open tube , really fell short of it : and instead of being arithmetical or uniformly the same , for equal changes of temperature , was actually progressive ; the expansion answering to the lower part of the scale , being greater than that corresponding to the middle ; which again exceeded that for high temperatures .
In these experiments , when the water had acquired a heat 20 or 30 degrees greater than that of the open air , a certain dustiness was perceived in the vacuum of the tube .
At 100~ of FAHRENHEIT , or thereabout , this appearance had so far increased , as to show clearly , that it could proceed from no other cause than a vapour arising from the surface of the heated quicksilver , quite invisible , till , by its condensation in the cold part of the tube , it was formed into balls , every where adhering to its sides and summit .
These globules were very small near the surface of the water , augmenting gradually as they approached the top of the tube , where they were greatest : their bulk increased with the heat ; and when the water was at or near boiling , they would sometimes unite , and descend by their own gravity , along the sides of the tube , into the general mass .
Hence the progressive diminution of the rate of expansion of the column of quicksilver in the barometer , perceptible even in the first class of experiments , is easily accounted for by the resistance of the elastic vapour(c ) , acting against the top of the tube , which was here colder than the rest .
But in the application of the barometer to the measurement of heights , the whole instrument is of the same temperature ; wherefore , in the second set of this third class of experiments , the tin vessel was heightened , that tubes of the ordinary length , placed in it , might be wholly immersed in boiling water .
The mean of four experiments , which agreed very nearly among themselves , gave.5117 for the total expansion between freezing and boiling ; for the 20~ , between 112~ and 132~.059 ; and for the last 20~ , between 192~ and 212~.046 .
In these ( c ) Having mentioned the circumstances to Mr. RAMSDEN , it first occurred to him , that the resistance of the elastic vapour was the cause of the diminution in the rate of expansion .
experiments ; the tube being wholly covered with boiling water , no condensation of vapour took place in the vacuum ; and therefore no particles of quicksilver were seen adhering to the upper part of the tube .
When the water boiled , the resistance of the vapour was greater than in the preceding set , and the total expansion less .
These two results serve strongly to confirm each other : it is , however , the last that furnishes the data for constructing the table of equation depending upon the heat of the quicksilver in the barometer , of which table we shall give an account hereafter .
Finding , from the comparison of these two sets of experiments with each other , that the maximum and rate of expansion seemed to vary with the length of the vacuum above the quicksilver , I was advised to try(d ) what might be the result , when the vacuum was much longer than in the common barometer .
The third set of experiments of this class was therefore made with a tube somewhat narrower in the bore than the former , and whose vacuum was 14 1/ 2 inches in length , whereof 11 1/ 2 reached above the top of the vessel .
The mean of three observations gave.5443 for the total expansion on 180~ ; that for the first 20~ was.067 ; for ( d ) Dr. BLAGDEN , who afterwards assisted in some of the first experiments with the manometer , proposed that with the long tube .
the 20~ in the middle of the scale.058 ; and for the uppermost 20~.065 : whence the mean rate for every 20~ , is nearly.0605(e ) .
In this set , the condensation in the vacuum of the tube was particularly attended to : it began , as in those of the first set , immediately above the surface of the boiling water , which was always kept an inch or two above the top of the column : the lowermost globules were very small , increasing gradually till they got without the lid of the vessel , where they were the largest ; thence they diminished uniformly upwards , and disappeared entirely three or four inches below the top of the tube .
Though the rate for the middlemost 20~ , in these last experiments , be below the mean , probably from some inaccuracy in observation ; yet , being compared with the former sets , they still serve to corroborate each other : for in these with the long tube , the vacuum seems to have been either completely maintained , or nearly so ; and we accordingly find the maximum of expansion increased , and its rate rendered nearly uniform , as will be farther confirmed from what follows .
( e ) Mr. CAVENDISH , who assisted in the first part of the experiments with the open tube , informed me , that , in those made by his father Lord CHARLES , the difference between the expansion of quicksilver and glass , from 180~ of heat , was.469 .
If to this we add Mr. SMEATON'S dilatation of glass , the total expansion of 30 inches of quicksilver will be.544 , which agrees with the experiments in the long tube , and gives a rate of only.003022 for each degree .
I have already had occasion to mention that a detached tin case was sometimes applied above the vessel , in which state it is represented in the view .
This method was thought of during the operations with the long tube , in order to try whether the vacuum was completely maintained by the temperature of the open air ?
For this purpose the case was placed on the stones of the yard , with a small tube inserted in it , to preserve an open passage in the middle : it was then rammed full of a composition of salt and ice ; and afterwards fixed on the top of the long tube .
The degree of cold thus applied round the greatest part of the vacuum , must have been very great , probably near the zero of FAHRENHEIT ; yet it produced no visible alteration in the height of the column of quicksilver , which still remained in boiling water below , and should have risen , if the vacuum had been formerly incomplete .
As it would have occasioned much trouble to have lengthened the several parts of the apparatus so as to have kept the long tube wholly in boiling water , the counterpart of this last experiment was not made in the accurate manner it ought : nevertheless , the tin case , being emptied of its cold composition , was placed on the tube as before , and filled with boiling water ; which , joining with the intermediate steam arising from that in the vessel below , must have kept the whole nearly in the same temperature .
The consequence of this application was , that the column shortened about 2/ 100ths of an inch ; which seems to prove , that the quicksilver vapour now reached the summit of the tube , and , acting against it , overcame , by so much , the pressure of the atmosphere .
I should now proceed to give some account of the fourth set of this last class of experiments , made on the condensation of the quicksilver , by means of artificial cold , below the temperature of the air .
Previously however to this , it may not be improper to take notice , in a more general way , of some others that were made on expansion ; as tending , with certain circumstances yet to be mentioned , not only to confirm those already described , but likewise to account for many irregularities that occur in operating with barometers .
In the course of the preceding experiments , from accidents of various kinds , it was often necessary to reboil the quicksilver ; and in that operation , many tubes were broken .
The frequent removal of the socket from the bottom of the vessel , in order to get others ground for it , became at last very troublesome ; and made more caution necessary , in boiling such as were ground , especially in frosty weather , which happened to be the case in the last days of March , 177 5 : wherefore it was thought best in the interim to try , what might be the expansion of a column of quicksilver , carefully put into the tube , but not boiled therein ?
With this view , the standard barometer and apparatus were left out during the night of the 29th , that they might acquire the same temperature , which was found next morning to be 34~1/ 2 ; the unboiled quicksilver standing 1/ 100th of an inch higher than that which had been boiled .
The lamps being applied to the vessel , the lengthening of the unboiled column was perceived , on the whole , to be more irregular , and the progressive diminution quicker , than in former experiments ; so as to give , for the maximum of expansion , only.443 for 180~ .
On the morning of the 31st , the unboiled column , which on the preceding day had been the highest , was lower than the other by near 2/ 100ths of an inch , the temperature of both being 31~1/ 2 .
As the water acquired heat from the application of the lamps , the rate of expansion diminished ; and , at boiling , was only.405 for 180~ .
The operation of the 30th seems to point out , in a manner sufficiently conclusive , that the air contained in the unboiled quicksilver , rendered its specific gravity less , than that which had been boiled even a great while before ; since it required a longer columin of the first , to counterbalance the weight of the atmosphere .
And though the vacua might possibly , at the beginning , have been equally complete in both ; yet they could not continue long so : for the air escaping gradually from the unboiled quicksilver , its elasticity increasing with the heat , and uniting with the quicksilver vapour , must have resisted the dilatation of the column , and rendered it less than on former occasions ; which actually appeared from experiment .
This is farther confirmed by the observations of the subsequent day ; for now the unboiled column was become the shortest , owing no doubt to more air having ascended , and rendered the vacuum still more incomplete .
Thus , the causes of resistance increasing , the dilatation is lessened in a superior degree .
The other circumstances to be mentioned , occurred on the 12th of April .
After finishing one of the experiments of the second class , and when the water had cooled to 192~ , the vessel , by accident , received a sudden jolt , whereby the mouth of the tube must have been raised , for a moment , out of the quicksilver in the cistern .
In a few minutes after this , intending to observe how far the column had shortened from the decreasing heat , I was surprised to find , that the quicksilver had wholly disappeared in the tube , and was sunk so low as not to be seen by looking obliquely down at the eye of the vessel .
It was then certain that air , and probably a particle of moisture along with it , had ascended into the upper part of the tube , whereby the vacuum was destroyed in so remarkable a degree .
Since this accident made it necessary to reboil the quicksilver , the water ( then between 180~ and 190~ ) was let out by the cock fixed for that purpose at the bottom of the vessel ; but before it could be entirely drawn off , the tube and its contents , had so sensibly felt the condensing force of the surrounding atmosphere , then about 48~ , that the quicksilver had risen again , and presented itself opposite to the eye of the vessel , something lower indeed than where it formerly stood .
On this discovery , and as soon as water could be boiled for the purpose , the vessel was filled again , when the quicksilver subsided , as before , quite out of sight ; and on drawing off the water a second time , it rose anew , seemingly to its former height .
The appearance , which this accidental circumstance produced , was such , as naturally suggested that farther experiments might have been made , varied as much as possible from each other , by the admission of different quantities of air , or of air and moisture intermixed .
But the nature of the vessel rendering it impossible to see , and consequently to measure , the motion of the quicksilver , occasioned by the alternate expansion and condensation of the elastic vapour contained in the upper part of the tube , and which could not have been accomplished plished without many troublesome alterations in the apparatus , therefore nothing of the kind was attempted .
From the circumstances just now mentioned , it will be readily conceived , how much care is necessary in operating with barometers for the measurement of heights , that the vacua be as nearly as possible complete ; and particularly , that no moisture get up into the tube .
I now proceed to the fourth and last set of experiments .
Having found , from the two first sets of this class , the rate of expansion of a column of quicksilver , in the tube of a barometer of the ordinary length , to be progressive and not arithmetical ; and that its maximum , for the 180~ comprehended between feeezing and boiling , was less than had been supposed ; I thought it proper to try , by means of artificial cold , whether the condensation , for the 32~ below freezing , followed nearly the same law ?
For this purpose the tin vessel , containing the ground tube , was rammed quite full of pounded ice and salt , as well as the tin stand holding the iron cistern below .
In this operation , twelve pounds of ice and four pounds of salt were employed , whereby the mean temperature of the mixture was reduced to + 4~ of FAHRENHEIT .
But before the eyes of the vessel could be sufficiently freed from the composition , so as to permit the surface of the column to be distinctly seen and read off ; it had risen to + 14~ ; the temperature of the air , and also of the standard barometer , being at the same moment 49~1/ 2 .
The observed condensation , arising from this difference of 35~1/ 2 , was 12/ 100ths of an inch ; or .
1189 , when reduced for the height of the barometer , which then stood at 30.296 .
Hence the condensation for 32~ is .
1072 , or.00335 for each degree .
In this day 's experiment , when the temperature of the mixture had risen to 32~ , that of the air and standard barometer was 52~1/ 4 ; whence the reduced difference , for the 20~ between 32~ and 52~ , was found to be.0664 , answerable to former experiments .
The same experiment was repeated two days after , with great care , the vessel being filled no higher than the surface of the quicksilver .
The mean temperature of the mixture was now + 4~ , and that of the standard barometer 49~1/ 4 .
The observed condensation , arising from this difference of 45~1/ 4 , was 162/ 1000 ; or.1594 , when reduced for the height of the barometer , then standing at 30,416 : hence the rate for 32~ is.1127 , or.003,522 for each degree .
When the temperature of the mixture had risen to 32~ , that of the air was 51~ : whence the augmented rate for the 20~ , between 32~ and 52~ , was found to be.0662 .
From the mean of these two experiments it appears , that the condensation of a column of 30 inches of quicksilver in the barometer , affected by the 32~ of cold below freezing , is 1099 : and that the expansion from 20~ of heat , between 32~ and 52~ , is.0663 , a number agreeing perfectly well with former results .
If the condensation.1099 thus found , be added to the expansion .5117 arising from the second class of experiments , we shall have.6216 for the total difference of height of the columns of quicksilver in two barometers , sustaining the same pressure , but differing from each other in their temperatures 212~ of FAHRENHEIT'S thermometer .
The series of numbers expressed in the annexed table , agreeing in all essential respects with the expansions found by experiment , will therefore show that which corresponds to any intermediate temperature , for every 10~ of the scale .
Construction and application of the table of equation , for the expansion of the quicksilver in the tubes of barometers .
In the introduction to this paper there was occasion to remark , that in the application of the barometer to the measurement of heights , various modes of calculation had been adopted .
The easiest and best method seems however to be , by means of the tables of common logarithms , which were first thought of by Mr. MARIOTT E , and afterwards applied by Dr. HALLE Y , Mr. BOUGUE R , Mr. DELU C , and others .
They have all proceeded on the supposition , that air is a truly homogeneous and elastic fluid , whose condensations being proportionable to the weights with which it is loaded , its dilatations are in the inverse of the weights ; and in consequence of this law , that the heights of the atmosphere ascended , are in geometrical progression , while the corresponding successive descents of the quicksilver in the tube of the barometer , are in arithmetical progression .
Mr. DELUC makes use of an arithmetical or uniform equation for the heat of the quicksilver in his barometers , whereby their relative heights are reduced to what they would have been in the fixed temperature of 54~1/ 4 of FAHRENHEI T. In the formulae adapting his rule to English measures ( Phil. Trans. vol. LXVII .
No xx .
and xxx .
) hath been shown , that the easiest and simplest method is , to make the difference of temperature of the two barometers the argument for the equation ; and that it is sufficient to reduce either column to what would have been its height in the temperature of the other .
But whatever may heretofore have been the method of using the equation for the heat of the quicksilver , while it was considered as arithmetical ; now that it hath been shown , from the preceding experiments , to be progressive , there seems at least to be propriety in applying to each barometer the equation answering to its particular temperature .
And though , for this purpose , any fixed temperature might have been assumed at pleasure , as that to which both barometers were to be reduced ; yet , the freezing point being fundamental in all thermometers , and that being likewise the zero of the scale for the equation depending on the heat of the air , as will be shown hereafter , it hath been preferred to any other .
From the experiments it appears , that a column of quicksilver of the temperature of 32~ , sustained , by the weight of the atmosphere , to the height of 30 inches in the barometer , when gradually affected by different degrees of heat , suffers a progressive expansion ; and that , having acquired the heat of boiling water , it is lengthened 5117/ 10000 parts of an inch : also , that the same column , suffering a condensation by 32~ of cold , extending to the zero of FAHRENHEI T , is shortened 1099/ 10000 parts , the weight of the atmosphere remaining in both cases unaltered ; but that , in the application of the barometer to the measurement of altitudes , since the pressure and length of the column change with every alteration of vertical height , the equation , depending on the difference of temperature of the quicksilver , will necessarily augment or diminish by a proportionable part of the whole .
Thus , if the weight of the atmosphere should at any time be so great as to sustain 31 inches of quicksilver , the quation for difference of temperature will be just 1/ 30th part more than that for 30 inches ; at 25 inches it will be 5/ 6ths ; at 20 inches 2/ 3ds ; at 15 inches 1/ 2 ; and at 10 inches only 1/ 3d of that deduced from experiment .
It is upon these principles that the annexed table of equation hath been constructed , for differences of temperature extending to 102~ of the thermometer , and for heights of the barometer from 15 to 31 inches ; beyond which limits , it is not probable , that many barometrical observations will be made .
The first or left-hand column , shows the height of the barometer for every half inch , from 31 to 25 ; thence for every inch downwards to 20 ; the 15th inch being half of the observed expansion .
The five next columns towards the right , comprehend the additive equation for condensations , answering to 0~ , 12~ , and 22~ , with their intermediate differences ; those that are progressive , as arising from difference of temperature , being ranged horizontally ; and those that are arithmetical , as depending on the height of the barometer , being placed vertically .
The temperature of 32~ requires no equation , and the thirteen columns from thence towards the right hand , contain the subtractive equations for expansion , corresponding to every 10~ as far as 102~ , with their progressive and arithmetical differences ranged as before .
By means of these differences , the equation for intermediate temperatures may readily be taken out by inspection .
Hence is deduced the first part of the rule for measuring heights by the barometer .
When the temperature of the quicksilver is below 32~ of FAHRENHEI T , add the corresponding equation for condensation to the observed heights of the columns respectively ; when above 32~ , subtract the equation for expansion from the observed heights of the columns respectively ; with which equated heights of quicksilver , expressed in 1000th parts of an inch , the tables of logarithms are to be entered .
SECTION II .
Experiments on the expansion of air in the Manometer .
WITH respect to order of time , the manometrical experiments were made subsequently to the chief part of the barometrical observations , from which alone an approximate rule had previously been deduced for the measurement of heights : nevertheless , in this paper it seemed to me best , that what related to the expansion of air in one instrument , should immediately succeed the expansion of quioksilver in the other .
The thermometer made use of in these experiments is above four feet long .
Its scale extends from 4~ to + 224~ of FAHRENHEIT , each degree being more than 2/ 10ths of an inch : when the barometer stood at 30 inches , its boiling point was fixed in the tin vessel formerly described .
Mr. RAMSDEN'S thermometers generally rise in the same vessel 213~1/ 2 ; and the long thermometer , being placed in the vessel he makes use of to fix his boiling points , rises only to 210~ .
The manometers were of various lengths , from four to upwards of eight feet : they consisted of straight tubes , whose bores were commonly from 1/ 15th to 1/ 25th of an inch in diameter .
The capacity of the tube was carefully measured , by making a column of quicksilver , about three or four inches in length , move along it from one end to the other .
These spaces were severally marked , with a fine edged file , on the tubes ; and transferred from them to long slips of paste-board , for the subsequent construction of the scales respectively belonging to each .
The bulb , attached to one end of the manometer at the glass house , was of the form of a pear , whose point being occasionally opened , dry or moist air could be readily admitted , and the bulb sealed again , without any sensible alteration in its capacity .
The air was confined by means of a column of quicksilver , long or short , and with the bulb downwards or upwards , according to the nature of the proposed experiment .
Here it must be observed that , from the adhesion of the quicksilver to the tube , the instrument will not act truly , except it be in a vertical position ; and even then , it is necessary to give it a small degree of motion , to bring the quicksilver into its true place ; where it will remain in equilibrio , between the exterior pressure of the atmosphere on one side , and the interior elastic force of the confined air on the other .
All the experiments were made when the barometer was at , or near , 30 inches .
When the bulb was downwards , the height of the barometer at the time of observation , augmented , and when upwards , diminished by the number of inches of quicksilver in the tube of the manometer , expressed the density of the confined air .
Pounded ice and water were used to fix a freezing point on the tube ; and by means of salt and ice , the air was farther condensed , generally four , and sometimes five or six degrees below zero .
The thermometer and manometer were then placed in the tin vessel , among water which was brought into violent ebullition ; where having remained a sufficient time , and motion being given to the manometer , a boiling point was marked thereon .
After this the fire was removed , and the gradual descents of the piece of quicksilver , corresponding to every 20 degrees of change of temperature in the thermometer , were successively marked on a deal rod applied to the manometer .
It is to be observed , that both instruments , while in the water , were in circumstances perfectly similar ; that is to say , the ball and bulb were at the bottom of the vessel .
In order to be certain that no air had escaped by the side of the quicksilver during the operation , the manometer was frequently placed a second time in melting ice .
If the barometer had not altered between the beginning and end of the experiment , the quicksilver always became stationary at or near the first mark .
If any sudden change had taken place in the weight of the atmosphere during that interval , the same was noted , and allowance made for it in afterwards proportioning the spaces .
Long tubes , with bores truly cylindrical or of any uniform figure , are scarcely ever met with .
Such however as were used in these experiments , generally tapered in a pretty regular manner from one end to the other .
When the bulb was downwards , and the tube narrowed that way , the column of quicksilver confining the air lengthened in the lower half of the scale , and augmented the pressure above the mean .
In the upper half , the column being shortened , the pressure was diminished below the mean .
In this case , the observed spaces both ways from the center , were diminished in the inverse ratio of the heights of the barometer at each space , compared with its mean height .
If the bore widened towards the bulb when downwards , the observed spaces , each way from the center , were augmented in the same inverse ratio ; but in the experiments on air less dense than the atmosphere , the bulb being upwards , the same equation was applied with contrary signs : and if any extraordinary irregularity took place in the tube , the corresponding spaces were proportioned both ways from that point , whether high or low , that answered to the mean .
The observed and equated manometrical spaces being thus laid down on the paste-board containing the measures of the tube ; the 212~ of the thermometer , in exact proportion to the sections of the bore , were constructed along-side of them : hence the coincidences with each other were easily seen ; and the number of thermometrical degrees answering to each manometrical space , readily transferred into a table prepared for the purpose .
I have already had occasion to remark that , from the operations of the barometer alone , an approximate rule , or mean equation , had been obtained for the measurement of heights ; but as , among the results , irregularities were now and then met with , doubts naturally arose , whether the equation , instead of being considered as uniform , might not follow an increasing or diminishing progression ?
Without an infinite number of observations , in very different temperatures above and below the zero of the scale , this point could not possibly be determined by the barometer : wherefore the first and chief thing proposed to be discovered by the manometrical experiments was , whether common air , occasionally rendered more or less dense , by the addition or subtraction of weight , expanded equally with quicksilver , when affected with the same degrees of heat ?
According to the ratio that took place between the expansion of quicksilver and air , above and below zero , I intended to regulate the barometrical equation already found , without regarding the proportion of the bulb to the bore of the manometer ; or in other words , without paying any attention to the actual expansion of the air confined in that instrument .
But after a great number of these first experiments had been made , it was judged proper to compute the actual expansion of 1000 equal parts of air in the manometer , from a heat of 212~ ; wherefore , in the last , the accurate capacity of the bulb , with respect to the bore , was determined ; at the same time that the original mode of comparing the thermometrical with the manometrical spaces , was still adhered to .
It is easy to conceive in experiments of this very delicate nature , part of which , namely those on air less dense than the atmosphere , were extremely difficult and even laborious , that mathematical exactness was not to be looked for ; and that , notwithstanding every possible precaution was taken , irregularities would occur .
These , however , were not so numerous as might have been expected , nor any way so great as to render the research fruitless : for a few of that kind being thrown out of the total number , the mean of the others , which were very consistent among themselves , served to prove beyond the possibility of doubt , that the expansions of common air did not keep pace with the dilatations of quicksilver .
The manometrical space , answering to the 20~ of the thermometer between 52~ and 72~ , was always found to be greater than any other 20~ of the scale .
Here it is to be understood , that I do not pretend to have ascertained the exact point in that space where the maximum falls : the mean corresponds to the 62d degree , and yet I am inclined to think that it is somewhat lower , perhaps it may be about the 57th : from this point , the condensations of air downwards , and its expansions upwards , follow a diminishing progression , compared with the condensations and dilatations of quicksilver .
The manometrical are equal to the thermometrical spaces , in two points of the scale ; namely , at or near the freezing temperature on one side , and between the 112~ and 132d degrees of the scale on the other .
At the zero and boiling point they are less than the thermometrical spaces .
Whether this maximum of expansion of air , compared with that of quicksilver , be owing to moisture , or any thing else mixed with the former , which is brought into its greatest degree of action , about the temperature of 57o of FAHRENHEIT , must be left to the investigations of future experimenters : I only relate things as I found them after many repetitions , without being able to discover any material difference in the results , even when the air was rendered more or less dense by an addition to , or subtraction from , the weight wherewith it was loaded .
The thermometrical , compared with the manometrical spaces , will therefore appear as in the following table .
Experiments , for determining the actual expansion of common air , in the manometer , affected by the heat of 212o .
For this purpose it became necessary to ascertain , in every manometer , the exact proportion between the capacity of the tube and that of its bulb .
This was done , by weighing the quicksilver that filled them respectively , in a balance that was sensible to a very small fraction of a grain .
The contents of the bulb , and that part of the tube between it and zero , expressed in grains , was called the air in experiment .
The apparent expansion of that air was measured , by the grains that filled the several sections of the tube between zero and the boiling point ; the sum being the total expansion or increase of volume , from a heat of 212o .
The apparent expansion , thus found , was again augmented for the dilatation of the tube , on the following principles .
In the first part of this paper I have shown , that solid glass rods dilate much less than barometer tubes .
The mean between Mr. SMEATON'S and my experiments , gives 14/ 1000 of an inch for the longitudinal extension of every foot of these tubes , by 212o .
From the rate of going of a clock , for near a year , whose pendulum rod is solid glass , its dilatation seems to be one-third part of a steel rod , or 58/ 10000 on a foot , by 212~ .
Now , as the manometers resemble solid rods much more than they do barometer tubes , it is probable their dilatation , even allowing for the greater extension of the bulb , would not exceed 6/ 1000ths of an inch on a foot , or 1/ 1000th part on every two inches .
In this ratio I have therefore augmented the apparent , to obtain the true , capacity of each manometer .
The equation , amounting to about 1/ 220th part of the whole , being less than the common error of such complicated observations , might in fact have been entirely omitted , without producing any material alteration in the results .
Having , in this manner , computed the total increased volume of any number of equal parts of air ( according to the capacity of the bulb and tube in grains ) and very often likewise the partial expansions for intermediate temperatures , expressed by the contents of the corresponding sections of the tube , I then found the ratio answering to 1000 equal parts , which , being divided by the degrees of difference of temperature , gave the mean rate for the whole scale , or the particular rate for any intermediate section of it .
The experiments , considered in this way , are distributed into four classes , whereof the results are comprehended in the four following tables .
The first shows the expansion of air , whose density was much greater than that of the common atmosphere .
The second , which is divided into two sets , contains those on air that sustained a pressure less than the atmosphere .
In the third class , a very short column of quicksilver being employed to confine the air , its density differed little from that we commonly breathe in : this class is likewise subdivided into two sets , and it will hereafter be made use of to regulate the equation depending on the temperature of the air , in the application of the barometer .
The fourth and last class of experiments , were made on air of the common density , artificially moistened by the admission , sometimes of steam , and at others of water , into the bulb ; it is accordingly distinguished into two sets .
TABLE I. Results of experiments on the expansion of air , whose mean density was equal to 21/ 2 atmospheres .
Height of Inches of Density Total expanMean rate the baroquicksilver in sion of 1000 for each equal parts of No meter .
confining inches .
air by 212~ .
degree .
the air .
1 29.7 + 72 .
101.7 451.54 2.12991 2 29.7 + 62.6 92.3 423.23 1.99637 3 29.62 + 50.84 80.46 412.09 1.94382 4 29.66 + 24.88 54.54 439.87 2.07486 5 29.66 + 20.05 49.71 443.24 2.09075 Mean , 75.74 434.00 2.04717 TABLE II .
Results of experiments on the expansion of air of the density of five-six common atmosphere ; and of others on air that was extremely rare , being only pressed about one-fifth of an atmosphere .
Height Inches Density Total expanMean rate Expansion for intermediate temperatures .
of the of in sion of 1000 for each equal parts of No barom .
quickinches .
air by 212~ .
degree .
From 0 32~ 52~ 72~ 92~ ~ silver .
to 32~ .
to 52~ .
to 72~ to 92~ to 132~ to ~ First set .
1 29.85 -5.44 24.41 495.455 2.33705 Not observed .
2 29.76 -3.05 26.71 504.340 2.37896 2.27190 2.41666 2.64060 2.55200 2.46040 2.31 3 29.79 -0.53 29.26 470.32 2.21849 1.90437 2.48150 2.63150 2.59650 2.15050 2.12 4 30.09 -5.43 24.66 469.07 2.21259 2.32688 2.53450 2.66250 2.24800 2.25425 ~ 5 29.93 -9.63 20.30 479.20 2.26038 2.14750 2.49500 2.59850 2.24700 2.25950 ~ Mean , 25.17 483.677 2.28140 2.16266 2.48191 2.63327 2.41087 2.28116 ~ Difference of temperature .
~ ~ ooo on 1131/ 2 from 481/ 2 to 162 244 6 Experiment in a heated room in Philpot lane , February 25 , baroo meter 30.03-4.82=25.21 the density of the air , 511/ 2 481/ 2 100 126 62 100 162 118 Second set .
on 1131/ 2 from 481/ 2 to 162 138 In Philpot lane ; tube with a small bore ; barometer 30.03-23.2= 6.77 the density .
The air had been heated red-hot in the bulb 51 481/ 2 1011/ 2 71 7 before it was sealed , 601/ 2 1011/ 2 162 66 The expansion for 212~ , at the mean rate , would be ~ on 212 ~ 32 above zero .
~ In Pulteney street , February 28th ; with the same manometer that 20 from 32 to 52 ~ 8 had been used in the same experiment in Philpot lane , barometer 80 52 132 ~ 30.08-23.2=6.88 the density of the air , 60 132 192 ~ 20 192 212 ~ on 180 from 32 to 212 14 20 32 52 1 20 52 72 2 20 72 92 2 In Pulteney street , April 19th ; tube with a large bore , barometer 20 92 112 1 9 29.8-24.08=5.72 the density of the air , which had been heated red-hot in the bulb before it was sealed , 20 112 132 1 20 132 152 1 20 152 172 1 20 172 192 1 From the experiments of the first class it appears , that 1000 equal parts of common air , loaded with two atmospheres and a half , being affected with a ' heat of 212~ , expands 434 of those parts ; that is to say , in its dilated state , it occupies a space bearing , to that which it originally filled , the proportion of 1434 to 1000 : hence the mean rate of expansion of air of that extraordinary density is 2.04717 for each degree .
From the first set of the second class of experiments it appears , that 1000 equal parts of air , pressed only with 5/ 6ths of an atmosphere , and suffering a heat of 212~ , expands nearly 484 of those parts , whereof the mean rate for each degree is 2.28140 .
The maximum corresponds to that section of the scale between 52~ and 72~ ; and the rate for the extremes is less than the mean .
But in the second set of this class , when the confined air was rendered so extremely rare as to be pressed with only one-fifth of an atmosphere , in which case there was a necessity for heating it red-hot before it was possible to make the quicksilver hang in any tube of a moderate length , the expansion of 1000 equal parts of air is , by the seventh and eighth experiments , diminished to about two-thirds of the usual quantity ; and by the ninth , it is considerably less , amounting only to 141.5 for the 180~ comprehended between freezing and boiling , or 0.78613 for each degree .
The maximum still corresponds to the space between 52~ and 72~ ; and the minimum is constantly at the boiling point .
From these three last experiments it would seem , that the particles of air may be so far removed from each other , by the diminution of pressure , as to lose a very great part of their elastic force ; since , in the ninth experiment , the heat of boiling water applied for an hour together , could only make it occupy a space which , compared with what it filled at freezing , bears the proportion of 1141.5 to 1000 .
From the third class of experiments it appears , that common air , pressed with a single atmosphere , whether taken into the manometer in its natural state , or heated red-hot therein , has the same expansion with air of only five-sixths of that density : for 1000 equal parts of this air expanded 484.21 from 212~ of heat , whereof the mean rate is 2.28401 for each degree .
By comparing this result with that of the first class , and again with that deduced from the second set of the second class , it would seem , that the elastic force of common air is greater than when its density is considerably augmented or diminished by an addition to , or subtraction from , the weight with which it is loaded(f ) ; for , in the first case , it bears the proportion of 484 to 434 ; and in the last , it is ( from the ( f ) This difference between the elastic force of common air , and that which is artificially rendered more or less dense , by the addition or subtraction of weight , particularly the latter , is truly remarkable , and contradicts the experience of BOYLE , MARRIOTTE , &amp;c. It could not arise from the adhesion of the quicksilver to the tube , though in the dense experiments a column of 72 inches was once made use of ; because the constant motion given to the manometer before the spaces were marked , must either have prevented any irregularity whatever , or made the apparent expansion sometimes too great , and at others too little .
But the rare experiments serve to put this matter out of doubt ; for if the adhesion of the quicksilver to the tube had tended to lessen the apparent expansion of the air , beneath the truth in one case , it must have had a direct contrary effect in the other , and augmented it above the truth , which it evidently doth not .
These experiments on the expansion of air less dense than the atmosphere , were extremely difficult and troublesome ; and it was not till after several fruitless attempts that , with the assistance of Dr. LIND , an apparatus was prepared for making them with sufficient accuracy .
The vessel employed for this purpose was made of the brass tube of a large telescope , near four inches in diameter ; it was divided into four pieces , which , when screwed together , made a pot of six feet in height .
This was mounted on a platform laid over the area rails , for the reception of the manometer , which was placed therein , with the bulb uppermost , the lower extremity of the tube passing through a socket at the bottom of the vessel , and then through a collar composed of many thicknesses of flannel .
By means of a brass plate and three long screws , the collar was made to embrace the tube so closely , as to suffer very little water to pass : such as did issue , oozed off along the sides of a paper funnel , bound round the end of the tube , without entering into the bore .
In this position , it required some degree of force to push the manometer up , or draw it down , till the top of the quicksilver confining the air , just appeared without the collar , so as to admit the spaces to be measured , from a fixed point marked on the tube .
The vessel being filled with boiling water , was kept to that temperature by means of lamps suspended the mean of three experiments ) as 484 to 252 , when pressed with only one-fifth of an atmosphere .
The pended around it .
Two thermometers were made use of ; the long one , whose ball stood at the bottom ; and a short one at the top , that descended no lower than just to be immersed in the water .
By some of the first of these experiments , the lamps not being placed directly at the bottom , water was perceived to be a very bad conductor of heat ; for it would boil violently at the top , and the short thermometer there would mark 212~ , while the long one would only mark 185~ or 190~ at the bottom .
By slow degrees the heat would nevertheless descend , and in the space of fifteen or twenty minutes the whole column would become of the same uniform temperature .
But when the apparatus was adapted for experiments on air denser than the atmosphere , in which case a plate of tin was soldered over the hole at the bottom , that it might be placed on a strong fire , the heat was then greatest below , and the long thermometer would mark upwards of 215~ , while the short one stood at 209~ or 210~ .
By desisting from blowing the fire , or removing a part of it , the particles of water suffering the greatest heat would ascend , mix with the rest , and for some little time make the whole column of an uniform temperature .
But the fire being totally removed , the top of the column in cooling was always hottest ; wherefore , in all these experiments , whether on dense or rare air , great care was taken to mix the water thoroughly .
From Mr. DE LUC 's book it appears , that M. AMONTONS found the effect of heat on the air confined in his thermometer , which seems to have been the same sort of instrument with the manometer , proportionable to the weight with which it was loaded .
By this he could not mean that , being of a double density , it had twice the expansion .
I apprehend it must here be understood , that the spaces the air occupied , were inversely as the weights .
That being pressed with a double weight , it only filled half the space ; or with half the weight , a double space .
This is no doubt nearly , though not accurately , the law that it follows .
From these experiments it appears , that there is little difference in the actual expansion or elastic force of air , pressed with an atmosphere + or one third part : yet , when it is rendered extremely rare , its elasticity is wonderfully diminished .
There seems likewise to be a visible diminution in its expansion , when loaded with two atmospheres and a half .
Some of the tubes that I used were near nine The total expansion 484.21 resulting from the third class of experiments , which are very consistent among themselves , being divided into parts proportionable to the manometrical spaces , as measured in degrees of the quicksilver thermometer , we have the expansions for intermediate temperatures , expressed at the bottom of the third table , where , it is to be observed , the rates for every 10~ below 92~ were found by interpolation .
Now as barometrical observations will probably never be made in a temperature higher than 92~ in the shade , nor in one lower than 12~ , if we subtract 26.038 , the expansion answering to 12~ , from 222.006 , that which corresponds to 92~ , we shall have 195.968 for the 80 intermediate degrees ; or 2.45 for the mean rate on each .
This equation , compared with Mr. DE LUC 's , bears the proportion of 245 to 210 , which is a difference of 35/ 100000 on every degree , or one-seventh part of the whole : and though this rate will be found hereafter to feet long .
Had it been possible to have managed them of double or triple that length , so as to have admitted the air to be pressed with a column of 18 or 20 feet of quicksilver , I am persuaded the diminution in the expansion of air of that extraordinary density would have been much more perceptible .
Mr. AMONTONS found , that the condensation of air in his thermometers kept pace with that of spirit of wine , which we are told sollows a decreasing progression with respect to quicksilver : wherefore his experiments agree with these , in making the condensation of air below 57~ follow a decreasing progression , when compared with that of quicksilver .
exceed that deduced from the operations of the barometer in extreme temperatures ; yet they agree exceedingly well with each other for the mean heat of the air , when the barometer will come most frequently into use .
The fourth class of experiments are all that now remain to be mentioned .
The bare inspection of TABLE IV .
will show , how greatly superior the elastic force of moist is to that of dry air .
It is true indeed , that two kinds of irregularities present themselves among the results : first , with regard to the total expansion for 212~ ; and secondly , as to the greatest exertion of the elastic force , which sometimes seems to have taken place before the air has acquired the heat of boiling water .
The first is easily accounted for : it must have arisen from different proportions of moisture being admitted into the same quantity of air , which there was no possibility of ascertaining , the bulbs and their apertures being of very different dimensions .
With regard to the second irregularity , I am rather inclined to think that it may have proceeded from error of observation , it being difficult to determine the accurate temperature near boiling ; especially when any part of the air rose above the top of the vessel , which was sometimes the case , notwithstanding its extraordinary height .
Be that as it may , a very uniform increasing progression will be perceived to take place , from the zero of FAHRENHEIT , as far as 152~ or 172~ ; and even to the boiling point , in those which I esteem the best experiments .
By adhering to the mean result it will appear that air , however moist , having that moisture condensed or separated from it by cold , its expansion differs not sensibly from that of dry air .
Thus the rate for 32~ below freezing 2.22799 , is nearly the same as in dry air ; but no sooner doth the moisture begin to dissolve and mix with the air , by the addition of 20~ of heat , than the difference is perceptible : for instead of 2.46675 , the rate for 20~ above 32~ in dry air , we have 2.588 for that which is moist .
In the next step of 20~ , the rate for dry air is 2.5809 ; whereas that for moist is 2.97 .
In this manner the progression goes on continually increasing , so as to give 7.86854 for the mean rate on each degree of the 212~ , which is near three times and a half the expansion of dry air .
And lastly , the rate for the 20~ between 192~ and 212~ , is twice and one-half the mean rate , and about nine times that which corresponds to the zero of the scale : but if the comparison is drawn from the mean of the fifth , sixth , and ninth experiments , as being probably nearest the truth , the total expansion of moist , will be more than four times that of dry air ; and the rate for the temperature at boiling , will be nearly fifteen times that which corresponds to the zero of FAHRENHEIT .
I am aware it will be alleged , that the proportion of moisture admitted into the manometer in these experiments , is greater than what can ever take place in nature ; and therefore , in order to be able to judge of the degrees of expansion the medium suffers in its more or less dense , and more or less moist states , that not only air near the surface of the earth , but likewise that found at the top of some very high mountain , should have been made use of .
I grant all this : but on the other hand it must be remembered , that these experiments are very recently finished ; that a good hygrometer ( if such can ever be obtained ) a great deal of leisure time , and the vicinity of high mountains , were all necessary for the carrying of such a scheme into execution .
It is for these reasons , and in hopes that other people will , sooner or later , investigate this matter still farther , not only by experiments made on the expansion of air , taken at different heights above the level of the sea in middle latitudes , but likewise on that appertaining to the humid and dry regions of the atmosphere towards the equator and poles , that I have been induced to hasten the communication of this paper .
In the mean time having proved , beyond the possibility of doubt , that a wonderful difference doth exist between the elastic force of dry and moist air , I may be allowed hereafter to reason by analogy , on the probable effects this will produce , in measuring heights with the barometer ; leaving it to others , much better qualified , to consider how far it will affect astronomical refractions .
In the following section I shall therefore give an account of the barometricaI observations made in Britain , and compare them with some others made in distant countries .
SECTION III .
An account of the barometrical observations made in Britain , wherein they are compared with some others of the same kind made in distant countries .
THE revival of the inquiries into that curious and useful branch of philosophy , whereby vertical heights are determined to a great degree of exactness , by the pressure of the atmosphere alone , we owe to Mr. DE LUC ; who hath undoubtedly removed many of the difficulties that formerly occurred in the application of the barometer , and thereby encouraged others to attempt to overcome , some part at least , of such as remain .
If the rule deduced from the observations on Saleve had been absolutely free from exceptions , and if there had not been particular points in the theory concerning which the ingenious author himself seems to have entertained doubts , it would probably have been universally adopted , without undergoing any very scrupulous investigation ; but the observations made at Sun-rising on Saleve , gave results that were defective , or less than the real height .
In certain cases , the equation for high temperatures , remote from the zero of the scale , appeared to follow a diminishing , and in others an increasing progression .
Hence arose some causes of uncertainty , with respect to the specific gravities of quicksilver and air , and the relative expansion of one compared with the other ; especially when the atmosphere happened to have more or less moisture dissolved in it .
It was doubtless from considerations of this sort , that Mr. DE LUC , in his book , so strongly recommends the making of numerous sets of observations , in different countries ; that , by the united labours of all , this interesting part of natural philosophy , might be brought still nearer to perfection .
With this general object in view , I am now to give an account of the principal barometrical observations that have been made in Britain , on heights determined geometrically with great care .
These heights are classed in the following list in six sets , according to the districts of the country wherein they are situated , and nearly in the order of time in which the observations were made .
Height in feet .
St. Paul 's Church-yard ( g ) , North-side , and iron gallery 281 over the dome , Top of Paul's-stairs , and the said gallery , 324 No 1 .
Top of Scotland-yard wharf , and the dining-room of Heights in 422 the Spaniard on Hampstead-heath , and near Great Pulteney-street , and the said dining-room , 352 London .
Pagoda in Kew gardens , 1161/ 2 Gun wharf in Woolwich Warren , and uppermost story 444 of Shooter's-hill inn ( h ) , ( g ) Mr. BANKS , assisted by other gentlemen , measured very accurately with a line the height of the ball of St. Paul 's above the floor of the church , which was found to agree , exceedingly near , with that taken from the engraved section of the building .
The distance of the ball from the dining-room of the Spaniard , was found by a base measured on Hampstead-heath ; and their relative heights by the angle of depression of the hall , taken with the astronomical quadrant from the said dining-room .
The heights of Paul 's stairs and Scotlandyard wharf , with respect to each other , were found by measuring from them severally to the surface of high water in the Thames .
And the elative heights of the church-yard and floor of the church with respect to the stairs , and of Pulteney-street with regard to the wharf , were obtained by levelling to them respectively .
( h ) The height of Shooter's-hill inn , above Woolwich , was found by a base measured in the meadows from the Warren eastward .
Lord MULGRAVE , Mr. BANKS , and Dr. SOLANDER , assisted in the geometrical operations ; as did Dr. BLAGDEN , Mess .
DE LUC and LLOYD , in the barometrical observations .
Feet .
Station at the East-gate of Castle Menzie 's gardens 7001/ 4 near the village of Weem , and top of Weem Craig , The said station , and top of Bolfracks Cairn , 10761/ 2 The said station , and top of Dull Craig , 12441/ 2 No 2 .
The said station , and top of Knock Farle , 13641/ 2 ~ TayThe said station , and that at the rivulet of Glenmore , bridge in 12791/ 4 below the South observatory on Schihallien , Perthshire .
The said station , and South observatory , 2098 The said station , and Western summit of Schihallien , 3281 Station at the rivulet of Glenmore , and the South ob8183/ 4 servatory , Level of the Clyde at Lanark-bridge , and station in the garden at Lanark , 3621/ 2 No 3 .
Ditto level , and top of Stonebyre hill , 654 near Lanark .
Robinhood 's well , before Carmichael-house , and top 16421/ 2 of Tinto , four feet below the summit of the Cairn , Ditto well , and West end of Carmichael-hill , 4511/ 2 Leith pier-head , and top of the Calton-hill , 344 Leith pier , and summit of Arthur 's Seat , 803 Leith pier , and Kirk-yetton Cairn , on the East-end of No 4 .
the Pentland hills , 1544 near EdinCalton hill , and ditto Cairn , 1200 burgh .
Level of Hawk-hill study , and top of Arthur Seat , 7021/ 2 Hawk-hill observatory , and bottom of the little rock 684 on Arthur Seat , 71/ 4 feet below the summit , Hawk-hill garden-door , and ditto little rock , 7303/ 4 Linhouse , and East Cairn-hill , 5 feet below the summit , 11761/ 2 Ditto , 18 feet below the top , 11651/ 2 No 5 .
Linhouse , and West Cairn-hill , 11 feet below the top , 11781/ 2 near LinDitto , and Corstown-hill , 4 feet below the top , 3861/ 2 house .
Corstown-hill , and West Cairn-hill , 792 Ditto , and East Cairn-hill , 7761/ 2 No 6 .
near CarnarCarnarvon Quay , and Snowdon Peak , 3555 von in North Ditto , and summit of Moel Eilio , 2371 Wales .
To enter into a minute detail of the geometrical operations , whereby the whole of these vertical heights were determined , would be extremely tedious and uninteresting .
That some idea may however be formed of the degree of accuracy with which they were ascertained , it will be sufficient to observe , that the requisite angles were taken with an astronomical quadrant of a foot radius , made by Mr. SISSON , and curiously adapted for the measurement of horizontal or base angles ; which , as well as those of the vertical kind , might always be determined thereby to within ten seconds of the truth .
The bases were measured with care ; and , in order to ascertain the distances , the three angles of each triangle were , as often as possible , actually observed with the quadrant .
That the variation of the line of collimation of the instrument , which was found to alter in carrying , might occasion no error , one or more of the angles of elevation , at each station , were taken on the arc of excess , as well as on the quadrantal arc .
In all cases , the usual ( i ) allowances were made for curvature and refraction : and for the correction of the last , sometimes the angles of de(i ) If the square of the distance be divided by the diameter of the earth , the quotient will give the curvature of the globe on that distance , or the excess of the apparent above the true level : and , by Mr. MASKELYNE 's rule , the square of the distance being divided by the diameter of the earth , augmented by onefourth part , we have the allowance for curvature and refraction ; which last is supposed to raise the object , by an angle equal to that of a great circle subtended by one-tenth part of the distance .
pression as well as of elevation were taken .
When time would permit , the geometrical operations were repeated at the first stations ; or the angles of elevation were observed from some new point connected with the first , and whose relative height , with respect to the others , was known .
Small altitudes were occasionally determined by levelling from one station to the other .
To prove that the vertical heights , assigned to the places in the preceding list , are exceedingly near the truth , I need only mention the following instances , by way of confirmation .
In 1771 , with the assistance of Dr. LIND and his friend Mr. HOY , I measured a base from the observatory of Hawk-hill westward , whereby the height of the summit of Arthur 's Seat , above the telescope of the Hawkhill quadrant , in its horizontal position , was found to be 685.66 feet .
In 1775 , these gentlemen levelled , three several times , from the summit downwards to the said telescope ; and found the vertical distance to be , by the first operation , 686.47 feet ; by the second , 684.43 ; and by the third , 685.25 .
This last , which , from the great care that was used , they considered as the best , differs only three-tenths of a foot from the geometrical result .
They afterwards continued the operation of levelling from Hawk-hill to the pier of Leith , and having repeated it twice , with a difference of only two inches between the results , they found the mean descent to be 117.38 feet : hence Arthur 's Seat is above Leith pier , by the mode of levelling , 802.66 ; and by the mixed method 803 feet .
In 1774 , when the astronomer-royal was carrying on the Society 's experiments for ascertaining the attraction of Schihallien , I found , from my own geometrical operations , depending on a base measured in the plain near Taybridge , the Western summit of the mountain to be 1183 feet above the South observatory .
Of this height , the effect of curvature and refraction amounted to 28.86 feet , on the distance of Bolfracks Cairn from the observatory ; and to 38 feet , on the distance of the said Cairn from the summit of Schihallien .
The result of these operations I communicated to Mr. MASKELYNE , before his trigonometrical operations were begun .
From the data which he hath since been so obliging as to furnish me with , depending on the base in Glenmore at the bottom of Schihallien , and the angles of elevation taken from the Southern extremity of that base , the Western summit of the mountain is 1186.6 feet above the South observatory .
But if the triangle that served to connect the station of the barometer in that valley with the others , and the angles of elevation taken from the same station are made use of , the dif4 Z2 ference ference of height will be 1183.33 feet .
Lastly , Mr. MASKELYNE'S result , from the triangles on the North-side of the mountain , makes it 1180 .
The mean of these three 1183.31 feet , is the height of the Western summit of the mountain above the South observatory , which only exceeds my height by one-third of a foot .
Here it is to be observed , that from the vicinity of these triangles to the mountain , and the shortness of the sides , the greatest curvature amounted only to 16 or 17 inches , which consequently made the effect of refraction next to nothing .
This near agreement between the results seems therefore to prove , that the mode of computation for curvature and refraction , made use of in the Taybridge observations , is just .
By the first angles of elevation , taken from the station of the barometer in Glenmore , the Western summit of Schihallien is 2001.88 feet above it ; from which , if we deduct 1183.33 , there remains 815.55 for the height of the South observatory above the said station : but if the last angles of elevation at the station of the barometer are made use of , the height between it and the observatory comes out 818.97 , whereof the mean is 818.76 feet .
Though these instances are of themselves sufficient to prove , that the geometrical heights may be safely depended upon ; yet , as an example of the method that was always made use of , I shall annex to this paper , a plan of the triangles and detail of the operations for obtaining the height of Snowdon ; because that mountain , at the same time that it is the highest I have measured , is , from its situation , more likely to be visited , and to have experiments repeated upon it , than the remote hills of the North .
I now proceed to give some account of the barometrical observations .
The heights in and near London being so very inconsiderable , it was easily foreseen , that nothing conclusive could be drawn from observations made on them alone .
It was , however , natural enough to try , even on these , whether the rule we had been furnished with would answer ?
A small height of 41 feet 4 inches , which , without inconveniency , could be recurred to at all times of the day , and all seasons of the year , was the first that was made use of .
St. Paul 's , Hampstead , Kew pagoda , and Shooter's-hill , were the next .
The mean results of many observations on the three first , and of several on Shooter's-hill , were found to be defective .
In general the coldest observations , made in the morning and evening , when the temperatures at the two stations differed least from each other , answered best .
In the hottest part of the day , when that difference was the greatest , the results were most defective .
Some months spent in Scotland in the summer of 1774 , afforded opportunities of making barometrical observations on hills of various heights , from three or four hundred , to upwards of three thousand feet , as hath been exhibited in the preceding list .
That season was remarkably cold and wet ; wherefore , in these observations , the mean temperature of the air in the shade was commonly about 55~ .
The hottest never exceeded 63~ in the plain ; and the coldest , namely those on the highest mountains , were generally from 43~ to 48~ .
From the defect found in the results of these observations , which , with respect to temperature , correspond to the mean and hottest of those made at Sun-rising on Saleve , and without any exception whatever , I could easily discover , either that a much greater equation than what the rule directed , must be applied for each degree of heat above the zero of the scale ; or , that the zero itself would fall considerably lower than 39~74 , where Mr. DE LUC'S formula , adapted to English measures , hath fixed it .
This first step towards a correction of the rule , naturally pointed out the second thing to be aimed at , namely , the obtaining of a sufficient number of cold observations , near the zero , and as far as possible below it , that the equation might disappear entirely , and even come to be applied with the contrary sign .
Of this kind the winter seasons of 1774 and 1775 afforded a few on the small heights in and near the metropolis ; but the best I have been furnished with are those which Dr. LIND , assisted by Mr. HOY , was so obliging as to make on Arthur 's Seat near Edinburgh ; and those which Captain CALDERWOOD has since favoured me with on the Cairn-hills , being a part of the Pentland range to the South-west of that city .
By comparing these sets of observations together , it appeared from all of them , that when the air was at or near the freezing temperature , the logarithmic differences gave the real height .
in English fathoms and thousandth parts , without any equation ; and when considerably below that point , the equation was to be substracted , or applied with the sign instead of + .
It was farther perceived , that the same general conclusion might be drawn from the coldest , not only of the Sun-rising , but even of the ordinary observations on Saleve , some reduction of the temperature being in certain cases made , on account of the exposure of the thermometer to the Sun 's rays : hence I was led to suppose , that the morning observations , instead of being made exceptions from the rule , were those , which , it might be presumed , would form the best basis for deducing the equation depending on the heat of the air ; because the mean temperature of the column was then found to differ least from that of its extremities ; whereas in the hottest time of the day , that difference was generally the greatest .
Having been enabled , by means of the cold observations , to form some judgement whereabout the zero of the scale would fall , below which the equation was negative , and above it affirmative ; it followed of course , that the next principal thing to be sought for , was the maximum of equation , or that corresponding to the highest temperatures the climate of our island would afford .
It was partly with the view of obtaining these that I went , in July 1775 , to Snowdon in North Wales .
On this expedition Captain CALDERWOOD was so obliging as to accompany me , and lend me his assistance in the operations for determining the geometrical height of that remarkable mountain .
At that particular period , the weather proved unfavourable for obtaining hot barometrical observations ; but , in other respects , they were very satisfactory , as being in general consistent among themselves , and agreeing sufficiently near with those of the preceding year in Scotland ; at the same time that they were made on a height , as formerly mentioned , greater than any other hitherto measured , with equal care , in Britain .
During the summer of 1776 , Dr. LIND obtained some more hot observations on Arthur 's Seat ; and in the beginning of the following winter , Captain CALDERWOOD made others of the cold kind , on the Cairn-hills in his neighbourhood .
From the combination of the whole of these observations taken together , and a comparison of them with Mr. DE LUC'S , as far as they are similar , I mean to show the agreement or otherwise , between the equation for the heat of the air , as deduced from the barometer and manometer ; but since the British observations , in certain cases , differ considerably in their circumstances from those on Saleve , it is necessary , in the first place , to point out wherein this difference chiefly consists .
In the observations in Britain , the barometers and detached thermometers have been , almost constantly , placed in the open air in the shade , and suffered to remain there generally half an hour , and sometimes a great deal longer , before the corresponding observations were begun , that the quicksilver might have time to take the temperatures of the situations respectively .
They were then observed four times , usually at intervals of ten minutes , the mean of the four being that which is calculated , and called a single observation .
If the time did not admit of so long an interval , the same number of observations were taken at distances of five minutes from each other .
In either case , the extremes never differed above a few thousandth parts of an inch from the mean , so as to render the computations of them separately wholly unnecessary .
Except in very small heights , and chiefly in London , where it was impossible to screen the upper barometer so effectually from the Sun during the time of observation as that below , which generally stood in the shade of some building , the temperature of the quicksilver in the superior(k ) hath been colder than that in the inferior barometer .
The difference was commonly found to be two or three degrees ; sometimes it would amount to six or seven ; rarely , in heights that were considerable , to nine or ten ; and in one instance only to thirteen , where the vertical distance of the instruments was great .
Whether in the plain or on the tops of the highest mountains , the detached thermometers , indicating the temperature of the air , have generally stood something ( k ) I have sometimes found , particularly in frosty weather , that a thermometer placed on the pavement of the North-side of St. Paul 's Church-yard , close to the wall of the building , would stand two degrees lower than that which was exposed on the North-side of the iron gallery over the dome .
The first , no doubt , felt the cold produced by the evaporation from the stones , while that above might be affected by the ascending smoke .
But the most remarkable instance of this kind occurs in one of Dr. LIND'S observations , on the breaking up of the hard frost January 31 , 1776 : at Hawk-hill , at 10h 45~ A.M. the temperature of the open air was 14~ , while that at the summit of Arthur 's Seat was 20~ .
The frost that remained in the ground kept the air extremely cold below , though it had already felt the effects of the thaw at the top of the mountain .
lower than those attached to their respective barometers , until they had remained a considerable time in the same situation , equally shaded from the Sun , when they always agreed : whence it followed , that in these observations , the mean temperature of the air , and equation depending upon it , might always have been determined very near the truth , from the temperature of the quicksilver in the tubes , as shown by the attached thermometers , without ever consulting the detached ones .
Let us now see what were the circumstances attending the observations on Saleve .
Mr. DE LUC'S lowermost barometer stood in the ground-story of a house near Geneva , where it remained unaltered during the whole of his experiments ; while the detached thermometer , indicating the temperature of the air , was exposed on a small eminence , at a little distance , directly to the Sun 's rays : hence we find that , in the observations of high temperatures , the bottom of the column of air is often 12~ or 15~ , and in one case 18~ , hotter than the quicksilver in the tube .
And even in the lowest temperatures , the bottom of the column of air is generally hotter than the quicksilver within doors , contrary to common experience in this country : for in England , in winter , the exterior air in the shade is always colder than the interior air .
This circumstance gives reason to apprehend , that the thermometer suffered not only direct but reflected heat .
The superior barometer was shaded with a parasol from the Sun , while its corresponding detached thermometer was exposed to his rays : wherefore , in the observations of high temperatures , the top of the column of air is usually four or five degrees hotter than the quicksilver in the barometer standing in the same air ; and the mean heat of the column often exceeds very considerably the mean heat of the quicksilver in the tubes .
In many of the coldest of Mr. DE LUC'S observations , as well as in those of mean temperatures of about 50~ or 55~ , the superior barometer is often the hottest of the two , even when the surrounding atmosphere at the top is colder than at the bottom .
This circumstance is easily accounted for : wood is known to be a bad conductor of heat , to receive it slowly , and retain it long : that barometer , which was moved about from place to place upon the mountain , with a very short interval between the observations ( as is sufficiently evident from the great number of stations it passed through in a limited time ) must have acquired and retained heat superior to that of the atmosphere , and communicated it to the tube with which it was in contact .
Some difference would no doubt arise from this cause , if the temperatures of the quicksilver in the tube and attached thermometer did not keep exactly pace with each other .
The last point to be mentioned is still more remarkable than the rest ; it is briefly this : in the observations at Sun-rising on Saleve , though the superior quicksilver is the coldest ; yet the top of the column of air is commonly five or six , and sometimes eight or nine degrees , warmer than the bottom .
Having thus shown the steps that were taken , for obtaining the coldest and hottest barometrical observations that the climate of this island would afford , the mode of observing , and wherein the circumstances attending them differed from those on Saleve , I shall now point out the general result .
In order to avoid repetitions as much as possible , it is necessary , once for all , to remark , that the computations of the British observations , by the rule hereafter to be given , are subdivided into their respective classes .
Each table contains 15 columns , which their titles sufficiently explain , that the principles from which the rule was deduced , the result and error , might all appear in one view .
The last column towards the righthand shows the ratio of the weight of quicksilver to air , the columns of the first in the barometers being severally reduced to the mean temperature of the last .
By comparing the tables it will be found , that the observations for extreme temperatures belong to the Edinburgh class of observations ( No 4 .
) it being thought best , in this case , to omit the few hot ones obtained on the inconsiderable heights near London : the mean of the coldest , answering to the temperature of 21~.75 , make the logarithmic excess 29/ 1000 ; and the mean of the hottest , corresponding to the temperature of 69~.6 , give a defect of 81/ 1000 .
Now the sum of the two equations 110/ 1000 , being divided by the difference of temperature 47~.85 , we have nearly 2.3 for the mean rate of the equation on each degree , which is less than that resulting from the operations of the manometer .
Again , from the mean of the very best observations , as being made on the greatest heights , when the temperature of the air is 52~ , it appears , that the defect is from 49 to 50/ 1000 , or 2.5 for each degree nearly , which agrees perfectly well with the manometrical expansion .
In this case , the ratio of the weight of quicksilver to air is as 11377 to 1 ; greater very considerably than 11232 to 1 , assigned to them by Mr. DE LUC , when the temperature is 69~.32 , answering to the zero of his scale , without any allowance for the diminution of pressure on his columns , which should have rendered air still comparatively lighter .
From the British observations , made on the most considerable heights , it appears , that when the temperature of the air is 28~.2 , the ratio of its weight , with respect to that of quicksilver , is as 1 to 10552 : hence the increase of the weight of air , on every degree of difference of temperature between 28~.2 and 52~.5 , amounts to 34.4 ; and hence we have 52~.5-4~.2=48~.3 for the temperature of the air in Britain , when its weight would be 1/ 11232 of that of quicksilver ; and consequently agree with Mr. DE LUC'S , though the heat would differ from his 21~ .
It will no doubt be remarked , that the equation for the air , resulting from the operations of the barometer , falls short of that given by the manometer .
Part of the difference , I apprehend , may arise from the small number of barometrical observations obtained in extreme temperatures .
I shall , nevertheless , adduce reasons hereafter for supposing , that it really should diminish , because of the drier and less elastic state of the superior air , compared with that taken into the manometer at the earth 's surface .
In the mean time , since both instruments agree in the equation for 52~ , which is a heat that the barometer will very frequently be used in , it seems best to adhere to the mean manometrical result 2.45~ in fixing the zero of the scale , which is obtained in the following manner .
Divide the excess or defect , expressed in 1000th parts of the logarithmic result , by 2.45 , the mean expansion of air for each degree of the thermometer ; the quotient will give the number of degrees , in the first case , to be added to , and in the last subtracted from , the temperature of the air in the observation ; the sum or difference answers to the zero of the scale , or that temperature when the logarithmic result gives the real height in English fathoms and 1000th parts .
According to this mode of computation , we have , from the aggregate of the several classes of British observations , the place of the zero as follows : Zero .
By the 1st class of observations in and near London , ooo between the temperatures of 25.5 and 71.2 at 32.2 2d , near Taybridge , 46.1 62.9 31.1 3d , near Lanark , 44 .
62 .
32.8 4th , near Edinburgh , 17 .
70.7 31.3 5th , near Linhouse , 26.1 46.5 29.9 6th , near Carnarvon , 49.1 62.3 32 9 Mean place of the zero at 31.7 The number 31~.7 differing so very little from 32~ , we may hereafter consider that remarkable point of FAHRENHEIT'S thermometer , as the zero of the scale depending on the temperature of the air ; and hence is deduced the second part , of the rule for measuring heights with the barometer .
When the mean temperature of the column of air to be measured is at 32~ of FAHRENHEIT , the difference of the common logarithms of the equated heights of quicksilver in the inferior and superior barometers , expressed in 1000th parts of an inch , gives the real height in fathoms and 1000th parts , the three figures towards the right-hand being decimals , and the rest integers ; which , being multiplied by fix , gives the result in feet .
Let us next consider , in a general way , how far this will correspond , or otherwise , with Mr. DE LUC'S observations in extreme temperatures .
I have already had occasion to remark , that when the temperature of the air was at 69~.32 , as indicated by thermometers exposed to the Sun 's rays , Mr. DE LUC found that the differences of the common logarithms of the heights of the barometers at the two stations , gave the altitude between them , in French toises and 1000th parts : in which case the specific gravity of quicksilver to air was as 11232 to 1 .
When his formula is adapted to English measures , the zero of the scale necessarily descends to 39.74 , where the English fathom bears the same proportion to the modulus of the common logarithms , as , in the former case , the French toise did to that modulus , the equation for the intermediate temperature being now applied with the contrary sign .
As it hath been shown , that the British observations differ in their circumstances from those on Saleve , and require a greater equation , it is unnecessary to enter into any minute comparison of the two sets : nevertheless , that some idea may be formed of the cause , of part at least , of the difference that takes place between them , I have collected into one view , the computations of such as were made in extreme temperatures ; namely , the coldest of those at Sun-rising ( though the whole of that class were considered as exceptions from the rule ) ; the coldest and hottest of the ordinary observations ; also those on the Dole(l ) , at Genoa , and at Turin , whereby the heights of the lake of Geneva and of Turin , above the sea at Genoa , were obtained .
In the table it will be observed , that there is a column for the reduced temperature of the air , on account of the exposition of the thermometer to the Sun 's ( l ) Having recomputed the whole of Mr. DE LUC'S observations on Saleve , and classed them according to the months in which they were made , I intended , at one time , to have given a general table , comprehending the mean results of all of them : however , this is now become unnecessary , since a very respectable and ingenious member of this Society hath had opportunities of making many curious and interesting observations on those very heights , which cannot fail of being perfectly satisfactory ; and who , at my request , was so obliging as to determine the height of the Dole geometrically .
On this mountain Mr. DE LUC had made barometrical observations , whose results differing considerably from the altitude , 4182 feet above the lake of Geneva , as taken by Mr. FATIO DE DUILLIER , made me suspect there was an error .
In fact it appears , from the last measurement , that the summit of the Dole is 4293 feet above the surface of the lake , which gives for the vertical distance of Mr. DE LUC 's barometer 4210 feet .
rays : I apprehend that I have not exceeded , but rather fallen short , in the reduction , to what would have been indicated by thermometers in the shade , perfectly free from direct and reflected heat , and with sufficient time allowed between the observations .
Be this as it may , it is of no importance , as no other conclusion is drawn from these observations , than that of showing what , in my apprehension , might probably have been the case , if another mode had been adopted .
From the table it appears , that when the temperature of the air is at 29~.5 , the logarithmic excess is 9/ 1000 ; and at 75~.5 reduced temperature , the defect is 96/ 1000 .
The sum of the two equations 105/ 1000 being divided by the difference of temperature 46~ , we have , as in the British observations , nearly 2.3 for each degree , which is greater than that applied by Mr. DE LUC'S rule , in the proportion of 23 to 21 .
That too small an equation hath been made use of in these hottest observations , supposing the original zero and temperature to remain , is sufficiently evident : for 96/ 1000 being divided by 42~ the difference of temperature , we have , as before , 2.3 very nearly for the equation answering to each degree .
Farther , if we consider the ratio of the weight of quicksilver to air , actually resulting from the observations themselves , the same kind of error ( for I cannot see it in any other light ) still exists .
Thus , in the coldest of the morning , as well as in the ordinary observations , when the temperature is at or near freezing , the mean ratio of the weight of quicksilver to air , is about 10850 to 1 .
When the observed and reduced temperatrues are respectively 41~ and 35~ , the ratio between them is that of 11295 to 1 , answering nearly to what hath been assigned to them when the heat is 69~.32 .
Again , in the hottest observations of the 14th and 15th of July 1759 , and 20th of July 1760 , on the highest , and consequently the best stations , when the observed and reduced temperatures are respectively 81~.7 and 75~.6 , quicksilver is to air as 12650 to 1 .
Now if we reduce this number 12650 by a proportionable part , for the degrees of difference between Mr. DE LUC'S zero 69~.32 , and the observed and reduced temperatures respectively , we shall have , in the first case , 12200 ; and in the last , 12390 to 1 , for the ratio of the weight of quicksilver to air ; either of which exceeds very considerably 11232 , which hath been assigned to them .
With regard to the observations on the Dole , the defect is 81.4/ 1000 , answering to the observed temperature of 66~.6 , and which is only reduced to 65~.2 .
On this great height , the ratio of the weight of quicksilver to air(* ) is that of 12595 to 1 .
Mr. DE LA CAILLE'S observations at the Cape of Good Hope , annexed to the table containing Mr. DE LUC'S , give a defect of 58.7/ 1000 , when the temperature seems to have been about 58~ , in which case quicksilver was 11687 times heavier than air .
Now if , from the aggregate of these observations , the same method be adopted , as was made use of in the British , for finding the zero of the scale , we shall have it as follows : By Mr. DE LUC'S equation for the air and By the manometrical equation observed temperature .
and reduced temperature .
Coldest of the morno ooooo ing observations , 25.2 to 30.5 at 33.7 from 25.2 to 30.5 at 33.12 from Coldest of the ordinary observations , 27.1 41.9 38.7 26 .
35 .
32.97 Hottest of the ordinary observations , 76 .
84.5 36.2 73.5 77 .
36.32 On the Dole , 59.2 71.5 27.6 58 .
70 .
32 .
Light house of Genoa , 75 .
79 .
26 .
75 .
79 .
33.40 DE LA CAILLE'S , 58 .
30 .
58 .
33.35 Zero at 32.03 Zero at 33.52 ( m ) It will even be found , though the calculations are not inserted in the table , that the hottest of Mr. DE LUC'S morning observations , June 8th , 1758 , at the 15th station , answering to the mean temperature of 57~.5 , and which I consider as the best , because no reduction is necessary for the exposition of the thermometer , agree with the manometrical experiments , in requiring a greater equation than is wanted in extreme temperatures : for in this case , the mean of two observations gives a defect of 65.5/ 1000 for 25~.5 above freezing , which is 2.57 for each degree ; the ratio of the weight of quicksilver to air being that of 12196 to 1 .
From the mean of these observations , though the results are irregular among themselves , it appears sufficiently evident , that if the morning observations on Saleve had been retained , instead of being made exceptions from the rule , the zero of the scale would have descended about 8~ ; viz. from 69~.3 to 61~.4 of FAHRENHEIT , supposing always the equation 2.1 for each degree of temperature , and the French toise , as the standard measure , to have been adhered to : for the French toise bears to the English fathom , the proportion of 106575 to 100000 ; wherefore 6575/ 106575 = 61.69/ 2.1 = 29~.4 + 32~ = 69~.4 , denotes the relative positions of the two zeros , the intermediate equation 61.69/ 1000 being to be subtracted , when the toise is made use of .
But it hath been shown , that the mean expansion of air is really greater , for such temperatures at least as the barometer can be applied in , than what Mr. DE LUC supposed it , in the proportion of 245 to 210 ; whence it follows , that 61.69/ 2.45 = 25~.18 + 32~ = 57~.18 , will denote the relative positions of the two zeros : which , instead of almost 30~ , are only distant from each other a little more than 25~ .
From what hath been said it is easy to see , that in calculating heights according to Mr. DE LUC'S rule , when the temperature of the air is below his zero , which we may take at 40~ , the English measure being used , the common error in the result will be equal to the sum of the two equations , 2.1+2.45=4.55 for each degree ; which amounts to 36.4/ 1000 parts for the 8~ that the zero is too high .
Above 40~ , the former error 36.4/ 1000 will be augmented by the difference of the equations for each degree that the temperature is above his zero , viz. 2.45-2.1=0.35/ 1000 .
In either case it is to be observed , that the progressive rate of equation for the heat of the quicksilver is not here taken into the account ; because it will not produce any material difference , unless one barometer is much hotter than the other , at the same time that their vertical distance is very great .
Thus the 32d degree of FAHRENHEIT , or freezing temperature , which is fundamental in all thermometers , happens , somewhat remarkably , to be the zero of the scale , when the English fathom bears such proportion to the modulus of the common logarithms , as that their difference , in computing heights by the barometer , brings out the result in fathoms .
No other proportion of a measure will do it : for if we suppose twenty-four of different lengths , between ours and the French toise , each surpassing the other by 263/ 100000 of that toise , the zero of the scale , in computing heights by these measures respectively , will ascend a single de3 gree gree on each ; and the French toise being the 25th , will have its zero nearly at the 57th degree : about which temperature the expansion of air appears , from the experiments , to be at its maximum .
From that point , therefore , the equation will diminish both ways , though by a quicker progression for condensation , than it doth for dilatation .
Having thus compared , in a general way , the results of the British observations with those of Mr. DE LUC , pointed out what seem to be the chief causes of the constant defect found in his rule , and thereby obtained , it is hoped , some corrections tending to improve the theory of the barometer , when applied to the measurement of heights in middle latitudes ; it remains to show the principles , whereon the table for the equation of the air hath been constructed .
Previously however to this , it may be proper to compare , with as much brevity as possible , these observations , with others that have been made towards the Pole and at the Equator : from which it will appear probable , that the rule which answers in middle latitudes , will not in the frigid and torrid zones .
In 1773 , Captain PHIPPS , now Lord MULGRAVE , commanding two of his Majesty 's ships then sent on discoveries towards the North Pole , measured geometrically , with great care , the height of a mountain in Hakluyt 's Island near Spitzbergen , and found it to be 1503 feet above the level of the sea .
On the morning of the 18th of August , the following observations , at the sea-shore and top of the mountain , were made with a single barometer , wherein the quicksilver had not been boiled .
At 6 h. A.M. Barometer at the shore , 30.040 therm. 50~ 7 h. 45~ A.M. Ditto at the top of the mountain , 28.266 42 8 h. 45~A.M .
Ditto at the top of ditto , 28.258 42 11 h. 45~ A.M. Ditto at the sea-shore , 30.032 44 Whence we have the following computations , equated for the times corresponding to the two observations at the top .
7 h. 45~ 30.038 46~-046 = 29.992 +70 .
43~ = 1573 42~ A.M. 28.266 42-031 = 28.235 =44.5 41 8 h. 45~ 30.036 45~-043 = 29.993 +78.3 43 = 1581.3 42 A.M. 28.258 42-031 = 28.227 =49.5 41 Mean 1577.1 = +47/ 1000 Quicksilver to air as 10224 to 1 .
From these observations it appears that , instead of the usual equation 24.5/ 1000 , to be added to the logarithmic result , in order to obtain the true height in Britain , when the temperature is 42~ , there is an excess of 47/ 1000 : and , instead of the usual ratio of the weight of quicksilver to columns of air , of equal altitude and temperature in Britain , namely about 11200 , we have that of 10224 to 1 .
Thus air at Spitzbergen seems to be specifically heavier , than that affected with the same heat and pressure in the middle latitudes : whence it follows that , instead of 32~ which is found to be the zero of the scale about the middle of the temperate zone , we shall have 47/ 245 = 19~.2+42~ = 61~.2 for the zero at Spitzbergen , within 10~ of the North Pole .
It is much to be regretted , that the French academicians , when employed in measuring the degrees of the meridian in Peru , were not supplied with better barometers , and that they made not observations at corresponding times ; since the scene of their operations was undoubtedly preferable to any other on the surface of the globe , for determining many curious points with respect to the modifications of the atmosphere in the torrid zone : nevertheless , by attending diligently to what Mr. BOUGUER ( n ) hath told us , of the steadiness of the barometer ( n ) He says , that at the South Sea , REAUMUR'S thermometer , in the morning before Sun-rising , stood at 19~ , 20~ , or 21~ ; and in the afternoon at 26~ , 27~ , or 28~ .
The respective means correspond to 76~1/ 2 and 92~1/ 2 of FAHRENHEIT , and make the mean heat of the day 84~1/ 2 .
At Quito the temperature continued at 14~ or 15~ , answering to 65~1/ 2 of FAHRENHEIT .
At the summits of Coracon and Pichincha , the thermometer stood in the morning several degrees below freezing , and varied 17~ in the heat of the afternoon ; whence the mean temperature at these highest stations , would probably be about 43~1/ 2 of FAHRENHEIT .
He farther says , that in the torrid zone , whatever may be the meter throughout the year ; the uniformity of the mean temperature in every assigned station ; and his mode of computing , by means of the tables of common logarithms , the altitudes of the Cordillero mountains above the valley that extends itself between them ; it will be no difficult matter to discover , nearly at least , what sort of equation became necessary ; and what were the relative weights of quicksilver and air of the mean temperature , not only in that high region of the atmosphere , but also at the level of the sea .
Thus , by inspecting the table of computations , it will appear , that columns of air , whose bases were removed six or eight thousand feet from the level of the sea , and whose heights equalled that distance , when the temperature was 55~ of FAHRENHEIT , as determined from the mean between the coldest of the morning and hottest of the afternoon , the mean logarithmic defect was only 36.3/ 1000 : whereas , in measuring heights near the level of the sea , in middle latitudes , the common equation for that temperature is about 57/ 1000 .
The mean ratio of the weight of quicksilver to air , on these long columns comprehended respectively between Carabourou and Quito , and the mean heat in any assigned station , it continues uniformly the same throughout the year .
In this rough estimation of the temperature in Peru , it seemed unnecessary to examine , whether the true thermometer of REAUMUR was used or not ; as it could produce no material difference , except at the very hottest stations .
summits of Pichincha and Coracon , is that of 16793 to 1 .
On the altitude of 1534 feet , intercepted between Carabourou and Quito , which short section of the column is about half-way between the level of the sea , and the summits of the Cordilleros , the mean temperature being 66~ , the ratio is that of 15089 to 1 : hence it seems probable , that quicksilver would have to the different sections of the general column of air , comprehended between the sea and the top of Coracon , nearly the following ratios : Temp. At the level of the South Sea , 841/ 2 13100 to 1 Half-way from thence to Carabourou , 751/ 2 14100 At Carabourou , 661/ 2 15100 Half-way from thence to Coracon , 55 16100 At the summit of Coracon , 431/ 2 17100 Whereof the mean is , 65 15100 Mr. BOUGUER tells us , that the barometer in the torrid zone varies not at the sea-shore above two and a half , or at most three lines throughout the whole year .
At Popayan , its variation is only a line and a half ; and at Quito a single line .
Now let us suppose , that an altitude had been measured with the barometer at the level of the South-sea , where the descent of quicksilver at the upper station , was exactly an inch in the mean heat of the day , answering to 84~1/ 2 .
On the former supposition of the weight of quicksilver to air , the height would be 13100 inches or 1091.7 feet .
29.930 841/ 2-169=29.761 Hence =890.6 feet , 28.930 841/ 2-169=28.761 the logarithmic result , which is defective 201.1 , or nearly 226/ 1000 parts .
Now this equation being divided by 2.45 the mean expansion of air , we have nearly 92~ for the difference between 84~1/ 2 , the temperature of the observation , and the zero of the scale , which reduces it to -7~1/ 2 of FAHRENHEIT .
If it should be thought that I have supposed the air to be too light at the level of the sea under the equator , let it be taken to quicksilver only , at a mean between 13100 and 12672 , which seems to have been the ratio of their weights at Genoa , when Mr. DE LUC'S temperature was 79~ , and we shall have 12881 inches , or 1073.4 feet of air , for the counterpoise to the inch of quicksilver in the barometer : hence 1073.4-890.6 = 1828/ 890.6 = 205/ 2.45 = 83~.7 , will denote the number of degrees that the zero of the scale would , in that case , be below the temperature of the air , which brings it to within less than a degree of the cipher of FAHRENHEIT .
But in middle latitudes the zero of the scale is at 32~ , and the equation , applicable at the level of the sea for the heat of ~ , is at most only 132/ 1000 instead of 205/ 1000 .
Mr. BOUGUER found , that the rule which his experience had furnished him with , for computing heights with the barometer between the ranges of the Cordilleros , namely , that of deducting 1/ 30th part from the number of toises expressed by the logarithmic differences , which agrees nearly with the equation 36.3/ 1000 which I have made use of in the table of computations , would not answer when he came to apply it at the level of the sea .
He tells us , indeed , that the elasticities of the air , above and below , are there , as well as in Europe , exactly proportionable to its condensations ; and even , that the intensity of the elastic force , or spring of the air , is every where equal in all places of the torrid zone that are considerably elevated .
The real condensations in each place are proportionable to the weights of the superior columns causing the compression ; these condensations being in geometrical , the heights are in arithmetical progression .
But below the same law doth not take place ; because the intensity of the elastic force is really considerably less at the level of the sea , than it is at one or two hundred toises above it , notwithstanding the effect of the heat , which should render it greater .
It is to be observed , that Mr. BOUGUER hath not given us the observations whereon he founded this last deduction ; and his note on the text , which I apprehend , nevertheless , conveys his true meaning , is contradictory to it : for there he says , that the dilatation occasioned by the heat throughout the day , changed the distribution of the weight with regard to all the places situated within the Cordilleros , as well as on other mountains , and made the lower sections of the columns contain less and the upper sections more air , than they should have done , had it been a perfectly elastic fluid .
Having now mentioned all the barometrical observations that have come to my knowledge , tending any way to throw light on this very intricate subject , it remains to sum up , from the whole , the general principles whereon I have proceeded in constructing the table of equation for the heat of the air .
It will be remembered , that I have more than once remarked , that in the British observations , when the temperature was 52~ , the defect was 49 or 50/ 1000 , the lowermost barometer standing at or near the level of the sea ; but in the observations on Tinto , a considerable hill appertaining to the third class , whose base is elevated 700 feet above the level of the Clyde at Glasgow , when the temperature was 52~ , I found the equation to be little more than 45/ 1000 .
Again , these two facts being compared with the aggregate result of Mr. DE LUC'S observations , where the lowermost barometer stood about 1300 feet above the sea , the equation for the same temperature seemed not to exceed 42/ 1000 .
Lastly , these circumstances being confronted with the results of Mr. BOUGUER'S observations , where the lowermost barometer stood from 6000 to 8000 feet above the sea , the mean equation for 55~ was only 36.3/ 1000 , which gives 34/ 1000 for the heat of 52~ .
Now these Peruvian observations , which I believe to be exceedingly good from the steadiness of the barometer in that part of the world , being substituted in lieu of those not yet obtained in our own quarter of the globe , there seemed to me to be a necessity for concluding , that the equation for middle latitudes , with any assigned temperature above or below the zero of the scale , diminished as the height of the place above the sea increased ; which consequently implied , that the magnitude of the logarithmic terms increased faster than the dilatations of the air .
But when the comparison was carried yet farther , and the observations in Peru and at Spitzbergen were fairly brought into one view , there appeared to be sufficient grounds for suspecting , if not absolutely for concluding , that there could be no fixed zero for the scale depending on the temperature of the air ; but that it would change with the density of the atmosphere appertaining to the latitudes , climates , or zones of the earth , where the observations were made .
On this supposition it was natural for the mind to form to itself some general hypothesis , which might serve to account for the appearances ; and the first that presented itself was the following : that the atmosphere surrounding our globe migth possibly be composed of particles , whose specific gravities were really different ; that the lightest were placed at the equator ; and that the density of the others gradually increased from thence towards the poles , where the heaviest of all had their position ( o ) It is a well known and established fact , that in the middle latitudes , a North or North-east wind constantly raises the barometer , and generally higher as its continuance is longer .
The contrary happens when a South or South-west wind blows ; for I believe it is commonly lowest when the duration and strength of the wind from ( o ) .
It was suggested by Dr. GEORGE FORDYCE , that equatorial and Greenland air might be brought bottled up , and weighed in this country in air of the respective temperatures , by means of a curious balance whereof he is possessed , in order to see whether any difference could be discovered in their specific gravities .
A thought of the same kind , but more easily put to experiment , occurred to Lieutenant GLENIE , of the Royal Artillery , namely , that of weighing equatorial and polar sea-water .
To this gentleman I am obliged for his assistance in part of the manometrical experiments , as well as in several of the computations .
that quarter have been the greatest .
Thus the North-east wind , by blowing for any length of time , brings into the middle latitudes a mass of air heavier than that which naturally appertains to the region , and raises the barometer above its mean height .
The continuance of a South-wester carries off the heavy air , deposits a much lighter body in its stead , and never fails to sink the barometer below its mean height : hence , in the middle parts of Europe , there is a difference of about two inches and a quarter between the highest and lowest states of the barometer .
But supposing it to be only two inches , the difference of pressure still amounts to 1/ 15th part of the whole weight of the atmosphere .
Now it is evident from the Peruvian observations , that the greatest fluctuation of the barometer , which is at the level of the sea , doth not exceed 0.226 of an inch , or 1/ 133d part of the whole pressure ; and if the change should be no greater at the poles , which I think not improbable , it follows , that the measurement of heights by means of the barometer , in middle latitudes , will be more precarious and uncertain than in the torrid and frigid zones .
Such in general were the first ideas which the comparison of the operations of the barometer with the effects of the North-east and South-west wind ( p ) on that instrument , suggested with regard to the different densities of the atmosphere in the different zones of the earth .
But since the experiments on the expansion of moist air have shown its elasticity to be so much greater than that which is dry , I apprehend , that the simple principle of heat and moisture may suffice to account for all the phenomena .
Thus it is universally admitted , that there is a greater degree of humidity and heat in the air , near the earth 's surface , than there is in the higher regions of the atmosphere .
The elasticity or expansion of the lowermost section ( q ) of every column of air , whether long ( p ) I have been well informed , that in China the North-west wind raises the barometer most , and is highly electrical ; it is at the same time the driest and the coldest ; and at Canton , under the Northern tropic , there is frequently ice .
On the East-coasts of North America the severity of the North-west wind is universally remarked ; and there can scarcely be a doubt , that the inhabitants of California , and other parts on the West-side of that great Continent , will , like those on the West of Europe , feel the strong effects of a Northeast wind .
The extraordinary dryness and density of the wind from the North pole , seems therefore to be occasioned by its passing over the Continent of Europe and Asia on one side , and that of North America on the other .
Those who live on the East and West-coasts of South America , will find the driest and coldest winds come to them respectively from the South-west and Southeast .
As the winds seem to be colder , drier , and denser , in proportion to the extent of land they pass over from the poles towards the equator , so they appear to be more moist , warm , and light , in proportion to the extent of Ocean they pass over from the equator towards the poles .
Hence the humidity , warmth , and lightness , of the Atlantic wind to the inhabitants of Europe .
( q ) Mr. DE LUC seems to have suspected something of this kind towards the long or short , will consequently be greater than the uppermost section of it ; for the heat , by dissolving the moisture , produces a vapour lighter than air , which mixing with its particles , removes them farther from each other , increases the elasticity of the general mass , and diminishes its specific gravity comparatively more than it doth that of the section immediately above it , where there is less heat and less moisture .
Hence I infer , that the equation for the air , in any assigned vertical , will gradually diminish as the elevation of the place above the sea increases , and that it will vanish at the top of the atmosphere .
This is in some respect confirmed by the experiments on the expansion of rare air ; for from them it appears , when the particles are very far removed from each other , by a great diminution of pressure , as is undoubtedly the case in the higher regions of the atmosphere , they lose a great part of their elastic force .
Thus the equation , answering to any particular temperature , above or below the zero of the scale , at different heights above the surface , will , I apprehend , be expressed by the ordinates to a curve of the hyperbolic order , whose curend of his 8th chap .
sir les difficultes a vaincre : and in that which follows , he gives proofs of the lightness of vapours with regard to air , saying , that they point out fire to be their vehicle .
He afterwards quotes NEWTON with respect to the lightness of a humid atmosphere compared with one that is dry .
vature may be supposed to change fast near the surface of the earth , and differ insensibly from a straight line at great heights above it .
With regard to the latitudinal equation , the same principle of heat and moisture seems to make it probable , that such will become necessary in operating with the barometer ; for it is well known , that there is a great degree of humidity in the air between the tropics ; and , on the contrary , that the polar atmospheres are very dry .
The heat and moisture being greatest at the equator , there the elasticity or equation will likewise be the greatest at the level of the sea ; and the zero of the scale will necessarily descend to a lower point of the thermometer , than that to which it corresponds in middle latitudes .
As the elasticity of the air at the level of the sea , or equal heights above it , with the same degree of heat , will always be proportionable to the quantity of moisture dissolved in it , therefore it will gradually diminish from the equator towards the poles ; that is to say , the zero of the scale will ascend in the thermometer , coincide with the 32d degree in the middle latitudes , and , in its motion upwards , will give the equation to be applied with the contrary sign in high latitudes .
Hence I infer , that every latitude , climate , or zone , will not only have its particular zero , but also its particular curve , whose ordinates will always measure the equations applicable in the respective situations .
The equatorial curve will probably change the fastest , and the others become gradually flatter , as they approach towards the poles , where the greater , but more uniform , density of the atmosphere may occasion it to differ little from a straight line .
I apprehend , however , that even at the pole some small diminution might be found to take place in the equation , was it possible , in that region , to prove it by experiments at a sufficient height above the level of the sea .
The table of the equation , depending on the heat of the air , annexed to this paper , is constructed for middle latitudes .
It extends to temperatures from 12~ to 92~ of FAHRENHEIT ; and for situations so greatly elevated above the sea , as to make the mean barometer between the two stations stand no higher than 19 inches .
As the equation corresponding to the lower parts of the atmosphere , contained in the right-hand columns , will come more frequently into use than that appertaining to the higher regions , comprehended in those on the left ; therefore , in the first , it is given for every half ; and in the last only , for every whole inch of descent of quicksilver in the tube .
The equation found in the column of 29 inches , corresponds exactly with the expansion of air resulting from the manometrical experiments ; and the ratio of diminution , in the temperature of 52~ , hath been taken from the Peruvian observations , supposing it to decrease uniformly 2/ 1000 on each inch , or 16/ 1000 on the eight inches between 29 and 21 .
For the sake of simplicity , as well as from the want of sufficient data for ascertaining the lengths of the ordinates of the curve , the arithmetical hath been preferred to any progressive diminution that might have been adopted , though by this mode the results would have agreed better with some of my own , as well as Mr. DE LUC'S observations .
In each of the columns the equations for particular temperatures , compared with that for 12~ or 92~ , are reciprocally proportionable , so that the maximum of the rate always corresponds to the space between 52~ and 72~ , as indicated by the manometer .
It will be observed , that though the equation in the table is only given for every 10~ of difference of temperature , yet , by the intermediate rates for single degrees in the columns respectively , and the ratio of diminution for the height of the mean barometer above the sea , expressed in that towards the right-hand , the equation for any particular temperature may be readily obtained .
The application of this table makes the third part of the rule , for measuring heights with the barometer .
When the mean temperature of the column of air is above 32~ of FAHRENHEIT'S thermometer in the shade , add the equation corresponding to the temperature and height of quicksilver in the mean barometer to the logarithmic altitude ; when below 32~ , subtract the equation from the logarithmic altitude ; the sum in the first case , and difference in the last , gives the real height .
Besides the table of equation for the air , adapted to the measurement of the greatest accessible heights the barometer can possibly be applied to in middle latitudes , I have annexed , for the use of those who may prefer simplicity , and still doubt of the vertical diminution , a thermometrical scale of the equation , suited to English and French measures , with their respective thermometers .
It will readily be conceived , that the divisions , expressing the 1000th parts in this scale ( r ) , are unequal , since they follow the inverse ratio of the thermometrical compared with the manometrical degrees .
Where these last are the greatest , as between 52~ and 72~ , the divisions expressing the equation are the smallest , because a greater ( r ) Any scale of this kind , unless it had been mechanically divided by a mathematical instrument-maker , could not be rendered very exact ; and it may be expected , that the imperfections in the original will be augmented in copying by the engraver , notwithstanding the utmost care on his part : wherefore , on the left-hand side of the plate , I have annexed the number of degrees and decimal parts of FAHRENHEIT , below the temperature of 91~.88 , corresponding to every 10/ 1000th parts of the equation , by which means the unequal scale may , at any time , be divided with sufficient accuracy .
number of them correspond to the same thermometrical space .
When the height is required in fathoms , the zero of FAHRENHEIT corresponds to -71.72 , and the boiling point to +412.49 : the sum of the two equations 484.21 is the actual expansion of common air from the heat of 212~ .
When the French toise is made use of as the measure , the zero of the scale hath been shown to coincide with 57~.18 of FAHRENHEIT , or +11~1/ 4 of REAUMUR .
The negative equation 134~.72 answering to -14~1/ 4 of REAUMUR , and the positive 349~.49 corresponding to +80~ , or the boiling point , being added together , make again 484.21 .
In order to convey a more distinct idea of the effect which heat produces in the dilatation of different kinds of air , compared with quicksilver , along with the scale for the equation I have placed another , expressing the actual and relative expansions , resulting from the mean of the experiments , for every 20~ of difference of temperature .
This scale is intended to give a comparative view of the manometrical with the thermometrical spaces , mentioned in the second section .
I shall now close this paper , which hath already greatly exceeded the limits I wished to have been able to prescribe to it , with a few remarks on the error of the rule , perceivable in the tables of computation , and the meaV O L. LXVII .
5E sures sures that should , in my opinion , be taken to bring the theory of the barometer to a still greater degree of persection , such as I believe it to be really capable of .
By inspection of the tables containing the computations of the British observations , it will be seen , that the error of the rule is in general very small .
In the London class it is greatest on Shooter's-hill , making the height five feet too little .
In those at Taybridge , one of the observations on Schihallien gives a defective result of 291/ 2 feet ; but this is easily accounted for , as it certainly arose from the short time given to the barometer to lose the heat it had acquired in carrying upwards , those destined to observe at the summit arriving there too late , that is to say , towards the expiration of the whole hour which the inferior barometer had been observed in .
One of the observations on Carmichael-hill , though a small height , is defective eight feet , which I ascribe to the South-west wind and humidity of the air .
From the same cause I would account for the uniform defect in the first part of the Edinburgh observations : in the last part of these , the circumstances having changed , the error hath the contrary sign .
In the Linhouse class of observations , the same cause of defect appears on the 1st of December , 1775 , and on the 20th of November , and 17th of December , 1776 .
The only result which I consider as very irregular , and do not pretend to account for , is that for the height of Moel Eilio , a hill situated between Carnarvon and Snowdon : the real altitude 2371 feet , is exceeded by the barometrical result 21 feet , though the circumstances were such as , in other cases , generally make it fall short .
At the bottom of the tables of computations I have occasionally substituted Mr. DE LUC'S equation for the air , in calculating one or more of the greatest heights , that the difference between the two methods might become more obvious .
Thus the first observation on Schihallien is defective 671/ 2 feet ; the mean of those on Tinto 29 feet ; Moel Eilio 41 feet ; and Snowdon 81 feet .
With respect to the results that the rule produces on Mr. DE LUC'S heights it will be observed , that it answers very well in the cold observations , which , with his rule , were often defective 60 or 70 feet ; but gives too much in those that are hot .
If , however , the whole of these hot observations had been included , the apparent error would have been less ; for the mean defect was taken at 96/ 1000 for the hottest temperature , whereas it sometimes amounts to 110/ 1000 .
On the height of the Dole the rule errs in defect ; and on the mean of Mr. DE LA CAILLE'S observations , at the Cape of Good Hope , it exceeds the truth .
By substituting Mr. DE LUC'S equation for the air , in the computations of the Dole and Table-hill , the respective results are defective 96 and 62.6 feet .
To the British observations a table is annexed , containing the barometrical computations of altitudes not yet determined geometrically .
In the chief part of these the inferior barometer stood at Belmont-castle , the seat of the lord privy-seal for Scotland , by whose directions the corresponding observations were made .
This table likewise comprehends Mr. BANKS'S observations in 1772 , for the height of the South-pap of Jura , above Freeport in the island of Isla , and those he made the same year , to obtain the height of Mount Hecla , above Hafniford in Iceland .
Lastly , it is to be observed , that in the application of the table , the equation found in the columns 291/ 2 , 30 , and 301/ 2 , will never come into use , except in the measurement of short columns of air , whose bases stand at , or not much above , the level of the sea ( s ) .
In an island , whose ( s ) Having been accustomed , from the beginning , to call the station of the inferior barometer the place of observation , and to suppose the mean height of its quicksilver to denote the elevation of the place above the sea , for the sake of simplicity I adapted the formula to the height of quicksilver in that barometer , and made all the computations in the tables accordingly .
But it having been suggested to me , first by Sir GEORGE SHUCKBURGH , and afterwards by Mr. DE LUC , that this mode , though the easiest , was not strictly accurate , nor consistent with the principles whereon a vertical diminution of the equation for whose climate is so very variable as that of Britain , settled weather should be chosen as the best time for observations .
With any sudden fall of the barometer , in any assigned station below its mean height , it is apprehended that the rule will have a tendency to give defective results ; and the contrary should happen when , from the increased weight of the atmosphere , it rises much above the mean height .
From what hath been said in the course of this paper , it will be perceived , that though the error of the rule is in general very small , yet now and then such irregularities do occur as plainly show , that something still remains to be done , in order to perfect the theory of the barometer .
The existence , or otherwise , of a latitudinal equation being a point of the greatest consequence , should be determined with so much care as to leave no doubt remaining on that head .
And as this can only be effected by differences that are extremely obvious , the observations for that purpose should be made at the equator , and as near as possible to the poles ( t ) .
Peru is no doubt the best situation the heat of the air was founded , I have since changed it to the mean barometer , or middle of the column of air intercepted between the two stations .
In this way all the great heights have been re-computed : the smaller altitudes , not being sensibly affected by the alteration , continue as at first .
( t ) Some idea may be formed what altitudes on the surface of the globe are situation on the globe for conclusive equatorial observations ; but as it would be found very difficult to carry any scheme of that kind into execution , such as may be more easily obtained in our West India islands , which have the highest mountains , would be very satisfactory with respect to the expansion and weight of moist air , at different heights above the surface .
At the tops of the mountains in the torrid zone , the observations would always be sufficiently cold ; but it would be of use likewise , to have the coldest possible at the level of the sea , under or near one of the tropics , when the Sun was in the other .
With regard to observations in the frigid zones , Spitzbergen seems to be as proper a situation as any ; though others may no doubt be found in the Northern parts of the Russian empire : and it is presumed , that the Petersburg academy would direct the experiments to be made .
accessible to man , by considering the height above the sea of the inferior line of perpetual snow .
In the middle of the torrid zone it appears , from Mr. BOUGUER'S observations , to be elevated 5201 yards , and 4476 about the tropics .
In middle latitudes there is everlasting snow on the mountains at the height of 3300 yards .
In the latitude of 80~ North , Lord MULGRAVE found the inferior line of snow to be only about 400 yards above the sea : whence we may conclude , that the surface of the earth , at the pole itself , is for ever covered with snow .
The Peak of Teneriffe , AEtna , the mountains of Auvergne and Roussillon , as well as Hecla in Iceland , are all very proper for observations in intermediate latitudes .
Within the island of Great Britain , Ben Nevis seems to be the best mountain for barometrical observations , because of its great height , its vicinity to the sea , and that there is very good ground close to its foot ( which is rarely the case in the Highlands ) for the measurement of the base that would be made use of in the geometrical operations .
One of the chief causes of error in barometrical computations , I apprehend , proceeds from the mode ( though simplicity is in its favour ) of estimating the temperature of the column of air from that of its extremities , which must be faulty ( u ) in proportion as the height and difference of temperature are great .
Where very accurate conclusions are expected , simultaneous observations , at different times of the day , and different seasons of the year , should be made with several barometers , placed at different heights , each accompanied with a thermometer and manometer .
By this method , the progression of temperature , as well as the law of diminution of the equation , from the position of the inferior barometer above the sea ( if such diminution doth really take place ) ( u ) This is taken notice of by Mr. DE LUC .
would be obtained with certainty .
Supposing , for instance , Ben Nevis was divided into four sections , five barometers , with as many observers , would be necessary .
This number may seem great , but the expense of people employed in that way would be inconsiderable .
And if it should be judged proper , there could not surely be any great difficulty in providing reasonable accommodation for an observer , who should live a whole year at the top of the mountain , while another made corresponding observations below .
But the perfecting of the theory of the barometer is not the only advantage that would accrue from a combination of these observations ; for , while they were carrying on in different climates , or zones of the earth , good opportunities would offer of determining the refractions , as well as the force of gravity and figure of the globe , from the vibrations of the pendulum .
The mean expansion of common air is already found to be greater than what was formerly supposed ; wherefore the mean refraction will be altered proportionably .
And since the expansion of moist air is found to be so much greater than that of common air , a larger field for inquiry and investigation is now laid open .
With respect to the experiments with the pendulum , Mr. BOUGUER seems to have been the only person , so far as I know , who hath taken the density of the medium in which it performed its vibrations into the account , and given us its length at the equator in vacuo .
But if we are to judge of the density of the air in the frigid zone from the barometrical observations at Spitzbergen , the pendulum there must have lost so much of its weight , as to have lessened considerably the number of vibrations below what they would have been in vacuo , in the same temperature .
Having considered the effect that this would produce , I collected the best experiments that have hitherto been made with the pendulum into one view , and having applied the equation that the density of the air , in which they severally vibrated , seemed to require ; I found from computation , that the ratio of the diameters of the earth is ( as Mr. BOUGUER supposed it ) nearly that of 178 to 179 , instead of 229 to 230 , as estimated by Sir ISAAC NEWTON , and which agrees very nearly with the mean result from the measurement of the degrees of the meridian .
The experiments with the pendulum are so simple and easy , may be repeated so often in all situations , and are so much more consistent with each other , than the measured lengths of degrees of latitude , that it appears to be incomparably the best method for determining the figure of the earth .
And if it should really be found so flat a spheroid as the penduV O L. LXVII .
5F lum lum seems to make it , both parallaxes and refractions , will require correction .
Upon the whole , though I wished to be concise in the recital of the experiments and observations contained in this paper , yet I found it necessary at the same time to be explicit .
Some of them were either entirely new , or managed in a different manner from what they had formerly been .
This forced me into a comparison of many minute circumstances attending the operations , and to a tedious , though necessary , combination of the various results .
Without taking a comprehensive view of the whole matter , and stating every thing with fairness and candour , I could not convey to others the ideas I entertained of it myself ; nor enable them to judge , how far I had been just in the conclusions already drawn , or consistent in my suppositions concerning such points as are yet left doubtful .
If I have been obliged to differ from Mr. DE LUC , it is because the British observations , as well as his own ( considered by their extremes ) seem to authorize it : he is himself too candid to suppose , that I have had criticism in view , or indeed any other object , than that of contributing my mite towards the discovery of the truth , from the very good foundation which he hath already laid for it .
I am aware it may be alleged , that I have rendered the theory of measuring heights by the barometer so much more complicate and difficult , as perhaps to deter others from applying it to useful purposes .
To this I answer , that though it seem utterly impossible to render what is really intricate in its nature , extremely simple ; yet that the best and surest method of arriving at simplicity at last will , in the first place , be to ascertain the limits of deviation of the rule , by a proper number of good observations , made in circumstances and situations as different as possible from each other .
In the present state of the matter , I doubt not but the barometer will be found to give results sufficiently near the truth for all ordinary purposes , the nicer business of levelling alone excepted .
It is the only instrument by which the relative heights of places , in very great and distant tracts of country , can easily and speedily be obtained , by the pressure of the atmosphere alone .
The method of using it is attainable by all , requiring only a little habit , and some degree of attention to prevent the admission of air into the tube .
Few people are qualified for the tedious and very laborious operations of accurate geometrical measurements .
Mountainous countries rarely afford bases of sufficient length , which , to avoid error , must be measured again and again with the utmost care .
Instruments of the most expensive kinds must be employed to take the angles ; at the same time that a thorough knowledge of their use , and a scrupulous attention to their various adjustments , become indispensably necessary .
In short , the facility of one method , compared with the other , is so exceedingly obvious as to need nothing else to recommend it as a subject very curious and useful , and therefore well worthy of the researches of philosophers , till , by their united labours , it hath been brought to perfection .

