ON THE MECHANICAL EQUIVALENT OF HEAT, WITH SUBSIDIARY RESEARCHES ON THE VARIATION OF THE MERCURIAL FROM THE AIR THERMOMETER, AND ON THE VARIATION OF THE SPECIFIC HEAT OF WATER. BY HENRY A. ROWLAND, PROFESSOR OF PHYSIOS IN THE JOHNS HOPKINS UNIVERSITY. Presented June llth, 1879. [REPRINTED FROM THE PROCEEDINGS OF THE AMERICAN ACADEMY OF ARTS AND SCIENCES.] CAMBRIDGE: UNIVERSITY PRESS: JOHN WILSON & SON. 1880. OP ARTS AND SCIENCES. INVESTIGATIONS ON LIGHT AND HEAT, made and published wholly or in part with appropriation from the KUMFOUD FUND. Geol. 75 Lib. QC V. ON THE MECHANICAL EQUIVALENT OF HEAT, WITH SUB- SIDIARY RESEARCHES ON THE VARIATION OF THE MERCURIAL FROM THE AIR THERMOMETER, AND ON THE VARIATION OF THE SPECIFIC HEAT OF WATER. BY HENRY A. ROWLAND,* Professor of Physics in the Johns Hopkins University. Presented June llth, 1879. CONTENTS. I. Introductory Remarks ... . Thermometry .... (a.) General View of Thermome- try (6.) The Mercurial Thermometer 78 (c.) Relation of the Mercurial and Air Thermometers . . 83 1. General and Historical Remarks . . 88 2. Description of Apparatus 90 3. Results of Comparison . 97 (d.) Reduction to the Absolute Scale .... 112 Appendix to Thermometry . . 116 III. Calorimetry ..... 119 (a.) Specific Heat of Water . . 119 (6.) Heat Capacity of the Calo- rimeter .... 131 IV. Determination of Equivalent . . 137 (a.) Historical Remarks . . 137 1. General Review of Meth- ods .... 137 2. Results of Best Deter- minations . . .140 (6.) Description of Apparatus . 165 1. Preliminary Remarks . 155 2. General Description . 157 3. Details . . . .168 (c. ) Theory of the Experiment . 168 1. Estimation of Work done 163 2. Radiation . . . .168 3. Corrections to Thermom- eters, etc. ... in (d.) Results 173 1. Constant Data . . 173 2. Experimental Data and Tables of Results . .174 V. Concluding Remarks, and Criticism of Results and Methods . . 197 I. INTRODUCTORY REMARKS. AMONG the more important constants of nature, the ratio of the heat unit to the unit of mechanical work stands forth prominent, and is used almost daily by the physicist. Yet, when we come to consider * This research was originally to have been performed in connection with Professor Pickering, but the plan was frustrated by the great distance between our residences. An appropriation for this experiment was made by the Ameri- can Academy of Arts and Sciences at Boston, from the fund which was insti- tuted by Count Rumford, and liberal aid was also given by the Trustees of the Johns Hopkins University, who are desirous, as far as they can, to promote original scientific investigation. 76 PROCEEDINGS OF THE AMERICAN ACADEMY the history of the subject carefully, we find that the only experimenter tho has made the determination with anything like the accuracy demanded by modern science, and by a method capable of giving good results, is Joule, whose determination of thirty years ago, con- firmed by some recent results, to-day stands almost, if not quite, alone among accurate results on the subject. But Joule experimented on water of one temperature only, ai did not reduce his results to the air thermometer; so that we are still left in doubt, even to the extent of one per cent, as to the value equivalent on the air thermometer. The reduction of the mercurial to the air thermometer, and thence to the absolute scale, has generally been neglected between and 100 by most physicists, though it is known that they differ several tenths of a degree at the 45 point. In calorimetric researches this may produce an error of over one, and even approaching two per cent, especially when a Geissler thermometer is used, which is the worst in this respect of any that I have experimented on ; and small intervals on the mercurial thermometers differ among themselves more than one per cent from the difference of the glass used in them. Again, as water is necessarily the liquid used in calorimeters, its variation of specific heat with the temperature is a very important factor in the determination of the equivalent. Strange as it may appear, we may be said to know almost nothing about the variation of the specific heat of water with the temperature between and 100 C. Regnault experimented only above 100 C. The experiments of / Him, and of Jamin and Amaury, are absurd, from the amount of variation which they give. Pfaundler and Plattner confined them- selves to points between and 13. Munchausen seems to have made the best experiments, but they must be rejected because he did not reduce to the air thermometer. In the present series of researches, I have sought, firstly, a method of measuring temperatures on the perfect gas thermometer with an accuracy scarcely hitherto attempted, and to this end have made an extended study of the deviation of ordinary thermometers from the air thermometer ; and, secondly, I have sought a method of determin- ing the mechanical equivalent of heat so accurate, and of so extended a range, that the variation of the specific heat of water should follow from the experiments alone. As to whether or not these have been accomplished, the following pages will show. The curious result that the specific heat of water OP ARTS AND SCIENCES. 77 on the air thermometer decreases from to about 30 or 35, after which it increases, seems to be an entirely unique fact in nature, seeing that there is apparently no other substance hitherto experimented upon whose specific heat decreases on rise of temperature without change of state. From a thermodynamic point of view, however, it is of the Bame nature as the decrease of specific heat which takes place after the vaporization of a liquid. The close agreement of my result at 15.7 C. with the old result of Joule, after approximately reducing his to the air thermometer and latitude of Baltimore, and correcting the specific heat of copper, is very satisfactory to us both, as the difference is not greater than 1 in 400, and is probably less. I hope at some future time to make a comparison with Joule's thermometers, when the difference can be accurately stated. II. THERMOMETRY. (a.) General View. The science of thermometry, as ordinarily studied, is based upon the changes produced in bodies by heat. Among these we may men- tion change in volume, pressure, state of aggregation, dissociation, amount and color of light reflected, transmitted, or emitted, hardness, pyro-electric and thermo-electric properties, electric conductivity or specific induction capacity, magnetic properties, thermo-dynamic prop- erties, &c. ; and on each of these may be based a system of ther- mometry, each one of which is perfect in itself, but which differs from all the others widely. Indeed, each method may be applied to nearly all the bodies in nature, and hundreds or thousands of thermometric scales may be produced, which may be made to agree at two fixed points, such as the freezing and boiling points of water, but which will in general differ at nearly, if not all, other points. But from the way in which the science has advanced, it has come to pass that all methods of thermometry in general use to the present time have been reduced to two or three, based respectively on the apparent expansion of mercury in glass and on the absolute expan- sion of some gas, and more lately on the second law of thermo- dynamics. Each of these systems is perfectly correct in itself, and we have no right to designate either of them as incorrect. We must decide a priori on some system, and then express all our results in that system : the accuracy of science demands that there should be no 78 PROCEEDINGS OF THE AMERICAN ACADEMY ambiguity on that subject. In deciding among the three systems, we should be guided by the following rules : 1st. The system should be perfectly definite, so that the same temperature should be indicated, whatever the thermometer. 2d. The system should lead to the most simple laws in nature. ^ Sir William Thomson's absolute system of thermometry, coinciding with that based on the expansion of a perfect gas, satisfies these most nearly. The mercurial thermometer is not definite unless the kind of glass is given, and even then it may vary according to the way the bulb is blown. The gas thermometer, unless the kind of gas is given, is not definite. And, further, if the temperature. as given by either of these thermometers was introduced into the equations of thermo- dynamics, the simplest of them would immediately become compli- cated. Throughout a small range of temperature, these systems agree more or less Completely, and it is the habit even with many eminent physicists to regard them as coincident between the freezing and boil- ing points of water. We shall see, however, that the difference between them is of the highest importance in thermometry, especially where differences of temperature are to be used. For these reasons I have reduced all my measures to the absolute system. The relation between the absolute system and the system based on the expansion of gases has been determined by Joule and Thomson in their experiments on the flow of gases through porous plugs (Philosophical Transactions for 1862, p. 579). Air was one of the most important substances they experimented upon. To measure temperature on the absolute scale, we have thus only to determine the temperature on the air thermometer, and then reduce to the absolute scale. But as the air thermometer is very inconvenient to use, it is generally more convenient to use a mercurial thermometer which has been compared with the air thermometer. Also, for small changes of temperature the air thermometer is not sufficiently sensi- tive, and a mercurial thermometer is necessary for interpolation. 1 shall occupy myself first with a careful study of the mercurial thermometer. (6.) The Mercurial Thermometer. Of the two kinds of mercurial thermometers, the weight ther- mometer is of little importance to our subject. I shall therefore con- fine myself principally to that form having a graduated stem. For OF ARTS AND SCIENCES. 79 convenience in use and in calibration, the principal bulb should be elongated, and another small bulb should be blown at the top. This latter is also of the utmost importance to the accuracy of the instru- ment, and is placed there by nearly all makers of standards.* It is used to place some of the mercury in while calibrating, as well as when a high temperature is to be measured ; also, the mercury in the larger bulb can be made free from air-bubbles by its means. Most standard thermometers are graduated to degrees; but Reg- nault preferred to have his thermometers graduated to parts of equal capacity whose value was arbitrary, and others have used a single millimeter division. 'As thermometers change with age, the last two methods are the best ; and of the two I prefer the latter where the highest accuracy is desired, seeing that it leaves less to the maker and more to the scientist. The cross-section of the tube changes continu- ously from point to point, and therefore the distribution of marks on the tube should be continuous, which would involve a change of the dividing engine for each division. But as the maker divides his tube, he only changes the length of his divisions every now and then, so as to average his errors. This gives a sufficiently exact graduation for large ranges of temperature ; but for small, great errors may be intro- duced. Where there is an arbitrary scale of millimeters, I believe it is possible to calibrate the tube so that the errors shall be less than can be seen with the naked eye, and that the table found shall repre- sent very exactly the gradual variation of the tube. In the calibration of my thermometers with the millimetric scale, I have used several methods, all of which are based upon some graphical method. The first, which gives all the irregularities of the tube with great exactness, is as follows. A portion of the mercury having been put in the upper bulb, so as to leave the tube free, a column about l;) mm - long is separated off. This is moved from point to point of the tube, and its length carefully measured on the dividing engine. It is not generally necessary to move the column its own length every time, but it may be moved 20"""- or 25 nim- , a record of the position of its centre being kept. To eliminate any errors of division or of the dividing engine, readings were then taken on the scale, and the "lengths reduced to their value in scale divisions. The area of the tube at every point is inversely as the length of the column. We shall thus have a series of figures nearly equal to each other, if the tube is good. By subtracting the * Geissler and Casella omit it, which should condemn their thermometers. 80 PROCEEDINGS OP THE AMERICAN ACADEMY smallest from each of the others, and plotting the results as ordinates, with the thermometer scale as abscissas, and drawing a curve through the points so found, we have means of finding the area at any point. The curve should not he drawn exactly through the points, but rather around them, seeing they are the average areas for some distance each side of the point. With good judgment, the curve can be drawn with great accuracy. I then draw ordinates every 10 mra> , and estimate the average area of the tube for that distance, which I set down in a table. As the lengths are uniform, the volume of the tube to any point is found by adding up the areas to that point. But it would be unwise to trust such a method for very long tubes, seeing the mercury column is so short, and the columns are not end to end. Hence I use it only as supplementary to one where the column is about 50 ram - long, and is always moved its own length. This estab- lishes the volumes to a series of points about 50"' m- apart, and the other table is only used to interpolate in this one. There seems to be no practical object in using columns longer than this. Having finally constructed the arbitrary table of volumes, I then test it by reading with the eye the length of a long mercury column. No certain error was thus found at any point of any of the ther- mometers which I have used in these experiments. While measuring the column, great care must be taken to preserve all parts of the tube at a uniform temperature, and only the extreme ends must be touched with the hands, which should be covered with cloth. If V is the volume on this arbitrary scale, the temperature on the mercurial thermometer is found from the formula T =.O V < , where C and f. Q are constants to be determined. If the thermometer contains the and 100 points, we have simply G _ 100 - i o v ~ V Otherwise C is found by comparison with some other thermometer, which must be of the same kind of glass. It is to be carefully noted that the temperature on the mercurial thermometer, as I have defined it, is proportional to the apparent expansion of mercury as measured on the stem. By defining it as proportional to the true volume of mercury in the stem, we have to introduce a correction to ordinary thermometers, as Poggendorf has shown. As I ouly use the mercurial thermometer to compare with the air thermometer, and as either definition is equally correct, I will OF ARTS AND SCIENCES. 81 not further discuss the matter, but will use the first definition, as being the simplest. In the above formula I have implicitly assumed that the apparent expansion is only a function of the temperature ; but in solid bodies like glass there seems to be a progressive change in the volume as time advances, and especially after it has been heated. And hence in mercurial and alcohol thermometers, and probably in general in all thermometers which depend more or less on the expansion of solid bodies, we firyi that the reading of the thermometer depends, not only on its present temperature, but also on that to which it has been sub- jected within a short time ; so that, on heating a thermometer up to a certain temperature, it does not stand at the same point as if it had been cooled from a higher temperature to the given temperature. As these effects are without doubt due to the glass envelope, we might greatly diminish them by using thermometers filled with liquids which expand more than* mercury : there are many of these which expand six or eight times as much, and so the irregularity might be dimin- ished in this ratio. But in this case we should find that the correction for that part of the stem which was outside the vessel whose tem- perature we were determining would be increased in the same propor- tion ; and besides, as all the liquids are quite volatile, or at least wet the glass, there would be an irregularity introduced on that account. A thermometer with liquid in the bulb and mercury in the stem would obviate these inconveniences ; but even in this case the stem would have to be calibrated before the thermometer was made. By a com- parison with the air-thermometer, a proper formula could be obtained for finding the temperature. But I hardly believe that any thermometer superior to the mer- curial can at present be made, that is, any thermometer within the same compass as a mercurial thermometer, and I think that the best result for small ranges of temperature can be obtained with it by studying and avoiding all its sources of error. To judge somewhat of the laws of the change of zero within the limits of temperature which I wished to use, I took thermometer No. G1G3, which had lain in its case during four months at an average temperature of about 20 or 25 C., and observed the zero point, after heating to various temperatures, with the following result. The time of heating was only a few minutes, and the zero point was taken immediately after ; some fifteen minutes, however, being necessary for the thermometer to entirely cool. VOL. xv. (N. s. vn.) 6 82 PROCEEDINGS OF THE AMERICAN ACADEMY TABLE I. SHOWING CHANGE OF ZERO POINT. Temperature of Bulb before finding the Point. Change of Point. Temperature of Bulb before finding the Point. Change of Point. 22.5 700 .115 300 .016 81.0 .170 405 .033 90.0 .281 ' 51.0 .0:59 100.0 .313 CO.O .105 100.0 .347 The second 100 reading was taken after boiling for some time. It is seen that the zero point is always lower after heating, and that in the limits of the table the lowering of the zero is about propor- tional to the square of the increase of temperature above 25 C. This law is not true much above 100, and above a certain tempera- ture the phenomenon is reversed, and the zero point is higher after heating ; but for the given range it seems quite exact. It is not my purpose to make a complete study of this phenomenon with a view to correcting the thermometer, although this has been undertaken by others. But we see from the table that the error can- not exceed certain limits. The range of temperature which I have used in each experiment is from 20 to 30 C., and the temperature rarely rose above 40 C. The change of zero in this range only amounts to 0.03 C. The exact distribution of the error from this cause throughout the scale has never been determined, and it affects my results so little that I have not considered it worth investigating. It seems probable, how- ever, that the error is distributed throughout the scale. IE it were uniformly distributed, the value of each division would be less than before by the ratio of the lowering at zero to the temperature to which the thermometer was heated. The maximum errors produced in my thermometers by this cause would thus amount to 1 in 1300 nearly for the 40 thermometer, and to about 1 in 2000 for the others. Rather than allow for this, it is better to allow time for the thermometer to resume its original state. Only a few observations were made upon the rapidity with which the zero returned to its original position. After heating to 81, the* zero returned from OM70 to 0.148 in two hours and a half. After heating to 100, the zero returned from 0.347 to 110 in nine days, and to -0.022 in one month. Reasoning from this, I OF ARTS AND SCIENCES. 83 should say that in one week thermometers which had not been heated above 40 should be ready for use again, the error being then supposed to be less than 1 in 4000, and this would be partially eliminated by comparing with the air thermometer at the same intervals as the ther- mometer is used, or at least heating to 40 one week before comparing with the air thermometer. As stated before, when a thermometer is heated to a very high point, its zero point is raised instead of lowered, and it seems probable that at some higher point the direction of change is reversed again ; for, after the instrument comes from the maker, the zero point con- stantly rises until it may be 0.6 above the mark on the tube. This gradual change is of no importance in my experiments, as I only use differences of temperature, and also as it Was almost inappreciable in my thermometers. Another source of error in thermometers is that due to the pressure on the bulb. In determining the freezing point, large errors may be made, amounting to several hundredths of a degree, by the pressure of pieces of -ice. In my experiments, the zero point was determined in ice, and then the thermometer was immersed in the water of the com- parator at a depth of about G0 cm . The pressure of this water affected the thermometer to the extent of about 0.01, and a correction was accordingly made. As differences of temperature were only needed, no correction was made for variation in pressure of the air. It does 'not seem to me well to use thermometers with too small a stem, as I have no doubt that they are subject to much greater irreg- ularities than those with a coarse bore. For the capillary action always exerts a pressure on the bulb. Hence, when the mercury rises, the pressure is due to a rising meniscus which causes greater pressure than the falling meniscus. Hence, an apparent friction of the mercu- rial column. Also, the capillary constant of mercury seems to depend on the electric potential of its surface, which may not be constant, and would thus cause an irregularity. My own thermometers did not show any apparent action of this kind, but Pfaundler and Plattner mention such an'action, though they give another reason for it. (c.) Relation of the Mercurial and Air Thermometers. 1. GENERAL AND HISTORICAL REMARKS. Since the time of Dulong and Petit, many experiments have been made on the difference between the mercurial and the air thermometer, 84 PROCEEDINGS OF THE AMERICAN ACADEMY but unfortunately most of them have been at high temperatures. As weight thermometers have been used by some of the best experi- menters, I shall commence by proving that the weight thermometer and stem thermometer give the same temperature ; at the same time, however, obtaining a convenient formula for the comparison of the air thermometer with the mercurial. For the expansion of mercury and of glass the following formulae must hold : For mercury, V = V (1 + a t -f J * 2 + &c.) ; glass, V = V' Q (1 + a t 4- (3 1 2 4- &c.). In both the weight and stem thermometers we must have V '= V. where V and F are the volumes of the glass and of the mercury reduced to zero, and t is the temperature on the air thermometer. The temperature by the weight thermometer is P where P , P t , &c. are the weights of mercury in the bulb at C., t C., &c. Now these weights are directly as the volumes of the mercury at 0. seeing that V is constant. .-. T 100 In the stem thermometers we have F , the volume of mercury at 0, constant, and the volume of the glass that the mercury fills, reduced to 0, variable. As the volume of the glass V' n is the volume reduced to 0, it will be proportional to the volume of bulb plus the volume of the tube as read off on the scale which should be on the tube. OF ARTS AND SCIENCES. 85 At .-. r=ioo which is the same as for the weight thermometer. If the fixed points are and t* instead of and 100, we can write T _ t ,At + Bfl^-Cfl + & c . r=<{i + (<-<<) [f + fl' + fc+o] +&<>} As T and t are nearly equal, and as we shall determine the con- stants experimentally, we may write t = T a t (t 1 t) (b t) + &c., where t is the temperature on the air thermometer, and T that on the mercurial thermometer, and a and b are constants to be determined for each thermometer. The formula might be expanded still further, but I think there are few cases which it will not represent as it is. Considering b as equal to 0, a formula is obtained which has been used by others, and from which some very wrong conclusions have been drawn. In some kinds of glass there are three points which coincide with the air thermome- ter, and it requires at least an equation of the third degree to repre- sent this. The three points in which the two thermometers coincide are given by the roots of the equation t(*-t) (6-0 = 0, and are, therefore, In the following discussion of the historical results, I shall take and 100 as the fixed points. Hence, t' = 100. To obtain a and b, two observations are needed at some points at a distance from and 100. That we may get some idea of the values of the constants in the formula for different kinds of glass, I will discuss some of the experimental results of Regnault and others with this in view. 86 PROCEEDINGS OF THE AMERICAN ACADEMY Renault's results are embodied, for the most part, in tables given on p 239 of the first volume of his Relation des Esper^ces. The Lres given there are obtained from curves drawn to represent he mean of his experiments, and do not contain any theorefcal results. The direct application of my formula to his experiments could hardly be made without immense labor in finding the most probable value the constants. But the following seem to satisfy the experiments quite ' Cristal de Choisy-le-Roi i = 0, = -000 000 82. Verre Ordinaire 6-245, = .000 000 34. VerreVert * = 270, a = .000 000 095. Verrede Suede i = +10, a = .000 000 14. From these values I have calculated the following : TABLE II. REGNAULT'S RESULTS COMPARED WITH THE FORMULA. 1 Choisy-le-Roi. Verre Ordinaire. Verre Vert. Verre de Suede. 1 | 1 1 1 1 1 1 *5 i 1 | 1 I 3 1 1 5 1 1 5 | .2 1 o 5 i 5 100 o o o 120 140 160 180 200 120.12 140 29 160.52 180.80 201 25 120.09 14025 KiO.49 18083 201 .28 +.03 +.04 +.03 -.03 -.03 119.95 139. W 5 159.74 179.63 199.70 119.90 139 80 159.72 179.68 19!l.fi9 +.05 +.05 +.02 .05 + .01 12007 140.21 16(1.40 180.60 20080 120.09 .01 140 22 ! .01 160.391 +.01 180.62 i .02 200.-S9 .09 120.04 120.04 140.11 ! 140.10 160 20 i 160.21 1MI.33 18034 20 ( 50 ! 200 53 +.01 .01 .01 .03 220 221 KJ 221.86 .04 2l9.HOl219.78 + .02 221.20 221.23 .03 220 75 ! 220 78 .03 240 242.55 242.50 239.90! 239 96 -.06 241.60 241.63 -.03 241.16 2*1.08 + .08 200 263 44 203.4H .02 260.20 2C021 -.01 262.15 262.09 +.07 WO 2H4.48 2S452 -.04 280.58 2SO.O>l 02 282.85 262.63 +.22 300 3?0 3(15.72 ::27.25 3115.70 327.20 -.04 -.05 301.08! 301. 12 321.80! 321.80 .04 00 340 349.30 348.88 +.42 434.00 34204 +.36 The formula, as we see from the table, represents all Regnault's curves with great accuracy, and if we turn to his experimental results we shall find that the deviation is far within the limits of the experi- mental errors. The greatest deviation happens at 340, and may be accounted for by an error in drawing the curve, as there are few ex- perimental results so high as this, and the formula seems to agree with them almost as well as Regnault's own curve. The object of comparing the formula with Regnault's results at temperatures so much higher than I need, is simply to test the formula through as great a range of temperatures, and for as many kinds of Corrected from 280.52 in Regnault's table. OF ARTS AND SCIENCES. 87 glass, as possible. If it agrees reasonably well throughout a great range, it will probably be very accurate for a small range, provided we obtain the constants to represent that small range the best. Having obtained a formula to represent any 'series of experiments, we can hardly expect it to hold for points outside our series, or even for interpolating between experiments too far apart, as, very often, a small change in one of the constants may affect the part we have not experimented on in a very marked manner. ' Thus in applying the formula to points between and 100 the value of b will affect the result very much. In the case of the glass Choisy-le-Roi many values of b will satisfy the observations besides 6 = 0. For the ordinary glass, however, b is well determined, and the formula is of more value between and 100. The following table gives the results of the calculation. TABLE III. REGNAOI/T'S RESULTS COMPARED WITH THE FORMULA. Calculated Calculated Calculated . ft = .00000032 a = .00000031 Observed. = .00000044 Thermome- ft = 0. b = 245. A 6 = 260. J ter. Clioisy-le-Roi. Verre Ordinaire. Verre Ordinaire. Verre Ordinaire. o 10 10.00 10.07 .... 10.10 20 19.99 20.12 .... 20 17 . . . 30 29.98 30.15 30.12 +.03 30.21 +.09 40 39.97, 40.17 40.23 -.06 40.23 60 49.96 60.17 50.23 .06 60.23 60 5:>.'.>5 60.15 60.24 .09 60.21 .03 70 09.95 70.12 70.22 .10 70.18 .04 80 79.96 80.09 80.10 .01 80.11 +.01 90 89.97 90.05 90.07 100 100 100 100 ' .... 100 Regnault does not seem to have published any experiments on Choisy-le-Roi glass between and 100, but in the table between pp. 226, 227, there are some results for ordinary glass. The separate observations do not seem to have been very good, but by combining the total number of observations I have found the results given above. The numbers in the fourth column are found by taking the mean of Regnault's results for points as near the given temperature as possible. The agreement is only fair, but we must remember that the same specimens of glass were not used in this experiment as in the others, and that for these specimens the agreement is also poor above 100. The values a = .000 000 44 and b = 260 are much better PROCEEDINGS OP THE AMERICAN ACADEMY above 100 for the given specimens. The table seems to show that between and 100 a thermometer of Choisy-le-Koi almost exactly agrees with the air, thermometer But this is not at all conclusive. Regnault, however, remarks * that be- tween and 100 thermometers of this glass agree more nearly with the air thermometer than those of ordinary glass, though he stat the difference to amount to .1 to .2 of a degree, the mercurial t mometer standing below the air thermometer. With the exception of this remark of Regnault's, no experiments have ever been publi. in which the direction of the deviation was similar to tins. All ( percenters have found the mercurial thermometer to stand above the air thermometer between and 100, and my own expenments agree with this. However, no general rule for all kinds of glass can b< laid down. Boscha has given an excellent study of Regnault's results subject, though I cannot agree with all his conclusions on this subject. In discussing the difference between and 100 he uses a formula of the form and deduces from it the erroneous conclusion that the difference is greatest at 50 C., instead of between 40 and 50. His results for T t at 50 are Choisy-le-Roi ...... .22 Verre Ordinaire ...... -J-.25 VerreVert ....... +.14 Verre de Suede ...... -f- 56 and these are probably somewhat nearly correct, except the negative value for Choisy-le-Roi. With the exception of Regnault, very few observers have taken up this subject. Among these, however, we may mention Recknagel, who has made the determination for common glass between and 100. I have found approximately the constants for my formula in this case, and have calculated the values in the fourth column of the following table. * Comptes Rendus, Ixix. OF ARTS AND SCIENCES. 89 TABLE IV. RECKNAGEL'S RESULTS COMPARED WITH THE FORMULA. Air Thermometer. Mercurial Thermometer. Difference. Observed. Calculated. 10 10.08 10.08 20 20.14 20.14 30 30.18 30.18 40 40.20 40.20 60 50.20 50.20 60 60.18 60.18 70 70.14 70.15 +.01 80 80.10 80.11 +.01 90 90.05 90.06 +.01 100 100.00 b = 290 a = .000 000 33 (100 t)(b t) It will be seen that the values of the constants are not very different from those which satisfy Regnault's experiments. There seems to be no doubt, from all the experiments we have now discussed, that the point of maximum difference is not at 50, but at some less temperature, as 40 to 45, and this agrees with my own experiments, and a recent statement by Ellis in the Philosophical Magazine. And I think the discussion has proved beyond doubt that the formula is sufficiently accurate to express the difference of the mercurial and air thermometers throughout at least a range of 200, and hence is probably very accurate for the range of only 100 between and 100. Hence it is only necessary to find the constants for my thermom- eters. But before doing this it will be well to see how exact the comparison must be. As the thermometers are to be used in a calorimetric research in which differences of temperature enter, the error of the mercurial compared with the air thermometer will be which for the constants used in Recknagel's table becomes Error = ^ 1 = .000 000 33 j 29000. 780 1 -f- 3 1* I . This amounts to nearly one per cent at 0, and thence decreases to 45, after which it increases again. As only 0.2 at the 40 point 90 PROCEEDINGS OP THE AMERICAN ACADEMY produces this large error at 0, it follows that an error of only 0.02 at 40 will produce an error of TT ^ T at 0. At other points the errors will be less. Hence extreme care must be taken in the comparison and the most accurate apparatus must be constructed for the purpose. 2. DESCRIPTION or APPARATUS. The Air Thermometer. In designing the apparatus, I have have had in view the production of a uniform temperature combined with ease of reading the ther- mometers, which must be totally immersed in the water. The uni- formity, however, needed only to apply to the air thermometer and to the bulbs of the mercurial thermometer, as a slight variation in the temperature of the stems is of no consequence. A uniform tempera- ture for the air thermometer is important, because it must take time for a mass of air to heat up to a given temperature within U.01 or less. Fig. 1 gives a section of the apparatus. This consists of a large copper vessel, nickel-plated on the outside, with double walls an inch apart, and made in two parts, so that it could be put together Avater- tight along the line a b. As seen from the dimensions, it required about 28 kilogi'ammes v of water to fill it. Inside of this was the vessel m d efg h k I n, which could be separated along the line c/. In the upper part of this vessel, a piston, q, worked, and could draw the water from the vessel. The top was closed by a loose piece of metal, o p, which fell down and acted as a valve. The bottom of this inner vessel had a false bottom, c /, above which was a row of large holes ; above these was a perforated diaphragm, . The bulb of the air thermometer was at <, with the bulbs of the mercurial thermometers almost touching it. The air thermometer bulb was very much elon- gated, being about 18 cm - long and 3 to 5 cra - in diameter. Although the bulbs of the thermometers were in the inner vessel, the stems were in the outer one, and the reading was accomplished through the thick glass window u v. The change of the temperature was effected by means of a Bunsen burner under the vessel w. The working of the apparatus was as follows. The temperature having been raised to the required point, the piston q was worked to stir up the water ; this it did by drawing the water through the holes at cl and the perforated diaphragm s, and thence up through the OP ARTS AND SCIENCES. 91 apparatus to return on the outside^ When the whole of the water is at a nearly uniform temperature the stirring is stopped, the valve op falls into place, and the connection of the water in the outer and inner vessels is practically closed as far as currents are concerned, and Fig.2. before the water inside can cool a little the outer water must have cooled considerably. So effective was this arrangement that, although some of the ther- mometers read to 0.007 C., yet they would remain perfectly station- ary for several minutes, even when at 40 C. At very high tempera- tures, such as 80 or 90 C., the burner was kept under the vessel w all the time, and supplied the loss of the outer vessel by radiation. The inner vessel would under these circumstances remain at a very 92 PROCEEDINGS OP THE AMERICAN ACADEMY constant temperature. The water in the outer vessel never.differed by more than a small fraction of a degree from that in the inner one. To get the and 100 points the upper parts of the vessel above the line a b were removed, and ice placed around the bulb of the air thermometer, and left for several hours, until no further lowering took place. For the 100 point the copper vessel shown in Fig. 3 was used. The portion y of this vessel fitted directly over the bulb of the air thermometer. On boiling water in x, the steam passed through the tube to the air thermometer. It is with considerable difficulty that the 100 point is accurately reached, and, unless care be taken, the bulb will be at a slightly lower temperature. Not only must the bulb be in the steam, but the walls of the cavity must also be at 100. To accomplish this in this case, a large mass of cloth was heaped over the instrument, and then the water in x vigorously boiled for an hour or so. After fifteen minutes there was generally no perceptible in- crease of temperature, though an hour was allowed so as to make certain. The external appearance of the apparatus is seen in Fig. 2. The method of measuring the pressure was in some respects similar to that used in the air thermometer of Jolly, except that the reading was taken by a cathetometer rather than by a scale on a mirror. .The capillary stem of the air thermometer leaves the water vessel at a, and passes to the tube b, which is joined to the three-way cock c. The lower part of the cock is joined by a rubber tube to another glass tube at d, which can be raised and lowered to any extent, and has also a fine adjustment. These tubes were about 1.5 cm - diameter on the inside, so that there should be little or no error from capillarity. Both tubes were exactly of the same size, and for a similar reason. The three-way cock is used to fill the apparatus with dry air, and also to determine the capacity of the tube above a given mark. In filling the bulb, the air was pumped out about twenty times, and allowed to enter through tubes containing chloride of calcium, sul- phuric acid, and caustic soda, so as to absorb the water and the car- bonic acid. The Cathetometer. The cathetometer was one made by Meyerstein, and was selected because of the form of slide used. The support was round, and the telescope was attached to a sleeve which exactly fitted the support. The greatest error of cathetometers arises from the upright support not being exactly true, so that the telescope will not remain in level OP ARTS AND SCIENCES. 93 at all heights. It is true that the level should be constantly adjusted, but it is also true that an instrument can be made where such an ad- justment is not necessary. And where time is an element in the accuracy, such an instrument should be used. In the present case it was absolutely necessary to read as quickly as possible, so as not to leave time for' the column to change. In the first place the round column, when made, was turned in a lathe to nearly its final dimen- sions. The line joining the centres of the sections must then have been very accurately straight. In the subsequent fitting some slight irregularities must have been introduced, but they could not have been great with good workmanship.* The upright column was fixed, and the telescope moved around it by a sleeve on the other sleeve. Where the objects to be measured are not situated at a very wide angle from each other, this is a good arrangement, and has the advantage that any side of the column can be turned toward the object, and so, even if it were crooked, we could yet turn it into such a position as to nearly eliminate error. It was used at a distance of about HO"" 1 - from the object, and no difficulty was found after practice in setting it on the column to fo mm - at least. The cross hairs made an angle of 45 with the horizontal, as this was found to be the most sensitive arrangement. The scale was carefully calibrated, and the relative errors f for the * The change of level along the portion generally used did not amount to more than .1 of a division, or about .Ol mm - at the mercury column, as this is about the smallest quantity which could be observed on the level. t These amounted to less than .016 mm - at any part. 94 PROCEEDINGS OF THE AMERICAN ACADEMY denoted by the subfix. Then approximat As the height of the barometer varies only very slightly during an experiment, the value of this expression is very nearly which does not depend on the absolute value of the scale divisions But the best manner of testing a cathetometer is to take readings upon an accurate scale placed near the mercury columns to b measured. I tried this with my instrument, and found that it agree< with the scale to within two or three one-hundredths of a millimeter, which was as near as I could read on such an object. In conclusion, every care was taken to eliminate the errors of this instrument, as the possibility of such errors was constantly present in my mind; and it is supposed that the instrumental errors did not amount to more than one or two one-hundredths of a millimeter on the mercury column. The proof of this will be shown in the results obtained. The Barometer. This was of the form designed by Fortin, and was made by James Green of New York. The tube was 2.0 cm diameter nearly on the outside, and about 1.7 cm - on the inside. The correction for capillarity is therefore almost inappreciable, especially as, when it remains constant, it is exactly eliminated from the equation. The depression for this diameter is about .08 mm -, but depends upon the height of the meniscus. The height of the meniscus was generally about l.S 1 ""' ; but according as it was a rising or falling meniscus, it varied from 1.4 to 1.2 mm -. These are the practical values of the variation, and would have been greater if the barometer had not been attached to the wall a little loosely, so as to have a slight motion when handled. Also in use the instrument was slightly tapped before read- OF ARTS AND SCIENCES. 95 ing. The variation of the height of the meniscus from 1.2 to 1.4 mm - would affect the reading only to the extent of .01 to ,02 mm -. The only case where any correction for capillarity is needed is in finding the temperatures of the steam at the 100 point, and will then affect that temperature only to the extent of about 0.005. The scale of the instrument was very nearly standard at C., and was on brass. At the centre of the brass tube which surrounded the barometer, a thermometer was fixed, the bulb being surrounded by brass, and there- fore indicating the temperature of the brass tube. In order that it should also indicate the temperature of the barome- ter, the whole tube and thermometer were wrapped in cloth until a thickness of about 5 or 6 cm ' was laid over the tube, a portion being displaced to read the thermometers. This wrapping of the barometer was very important, and only poor results were obtained before its use; and this'is seen from the fact that 1 on the thermometer indi- cates a correction of ,12 mm- on the barometer, and hence makes a difference of 0.04 on the air thermometer. As this is one of the most important sources of error, I have now devised means of almost entirely eliminating it, and making continual reading of the barometer unnecessary. This I intend doing by an artificial atmosphere, consisting of a large vessel of air in ice, and attached to the open tube of the manometer of the air thermometer. The Thermometers. The standard thermometers used in my experiments are given in the following table. 96 PROCEEDINGS OP THE AMERICAN ACADEMY fl >> ffj 1 -flftrfi Owner or Lender. ( Physical Labora ( Johns Hopkins Uni Prof. Barkei Univ. of Pennsy Chemical Labor Johns Hopkins Un n ^o sS ^-s if I -siSilill ^ >> "^ ! I co 2 : 1_ i CO "3 CC & Q g a J ^ xj JS .S a ^ = - 5 .2 1 5 1 ' 1 : ^ S o 4*. 0> CO CO CN ^-i CD (N O5 OS T uation. | P^ |-i . ri d b b 2 3 a, b "2 "3 ^ C3 '0 ^ 3 CT 1 8 1 1 co o o Tli^ 2 s i H rH i-H i-H I-H C^ rH I-H i i 8> > ^ 2 + O O O CN rH . CO ^0*^ S= 2s o 1 1 1 \ \ ^_ T I ld ~-*~* a-| 49 l|| 58 58 J3 5S 5 CO CO 'o S CO CD O CO CO M* 1 S 3 o W CO CO CSJ CO CO CO 00 g ^ ^ OF ARTS AND SCIENCES. 97 The calibration of the first four thermometers has been described. The calibration of the Kew standard was almost perfect, and no cor- rection was thought necessary. The scale divided on the tube was to half-degrees Fahrenheit; but as the 32 and 212 points were not correct, it was in practice used as a thermometer with arbitrary divisions. The interval between the and 100 points, as Welsh found it, was 180.12, using barometer at 30 inches, or 180.05 as cor- rected to 760 mm - of mercury.* At the present time it is 179.68,f showing a change of 1 part in 486 in twenty-five years. This fact shows that the ordinary method of correcting for change of zero is not correct, and that the coefficient of expansion of glass changes with time4 I have not been able to find any reference to the kind of glass used in this thermometer. But in a report by Mr. Welsh we find a com- parison, made on March 19, 1852, of some of his thermometers with two other thermometers, one by Pastre", examined and approved by Regnault, and the other by Troughton and Simms. The thermometer which I used was made a little more than a year after this ; and it is reasonable to suppose that the glass was from the same source as the standards Nos. 4 and 14 there used. We also know that Regnault was consulted as to the methods, and that the apparatus for calibration was obtained under his direction. I reproduce the table here with some alterations, the principal one of which is the correction of the Troughton and Simms thermometers, so as to read correctly at 32 and 212, the calibration being assumed correct, but the divisions arbitrary. * Boiling point, Welsh, Aug. 17, 1853, 212.17 ; barometer 30* . Freezing point, " " " 82.05. Boiling point, Rowland, June 22, 1878, 2 12.46; barometer 760 mm .. Freezing point, " " " 32.78. The freezing point was taken before the boiling point in either case, t 179.70, as determined again in January, 1879. t The increase shown here is 1 in 80 nearly ! It is evidently connected with the change of zero; for when glass has been heated to 100, the mean coefficient of expansion between and 100 often changes as much, as 1 in 50. Hence it is not strange that it should change 1 in 80 in twenty-five years. I believe this fact has been noticed in the case of standards of length. VOL. xv. (N. s. vn.) 98 PROCEEDINGS OP THE AMERICAN ACADEMY TABLE VI. COMPARISON BY WELSH, 1852. Mean of Kew Standards Nos. 4 and 14. Fastre231, Kegnault. A Kew. Troughton and Simms (Royal Society). A Kew. 32.00 32.00 32.00 38!? 1 45.04 38.72 45.03 +.01 -.01 38.70 45.03 .01 .01 49.96 49.96 .00 49.96 .00 55.34 55.37 +.03 55.34 .00 60.07 60.05 .02 60.06 .01 65.39 65.41 +.02 65.36 .03 69.93 69.95 +.02 69.93 .00 74.69 74.69 .00 74.72 +.03 80.05 80.06 +.01 80.14 +.09 85.30 85.33 +.03 85.44 +.14 90.60 90.51 +.01 90.56 +.06 95.26 101.77 95.24 101.77 .02 .00 95.40 101.94 4-.14 +.15 . 109.16 109.15 .01 109.25 +.08 212.00 212.00 .00 212.00 .00 It is seeruhat the Kew standards and the Fastre agree perfectly, but that the Troughton and Simms standard stands above the Kew ther- mometers at 100 F. The Geissler standard was made by Geissler of Bonn, and its scale was on a piece of milk glass, enclosed in a tube with the stem. The calibration was fair, the greatest error being about 0.015 C., at 50 C. ; but no correction for calibration was made, as the instrument was only used as a check for the other thermometers. 3. RESULTS OF COMPARISON. Calculation of Air Thermometer. This has already been described, and it only remains to discuss the formula and constants, and the accuracy with which the different quantities must be known. The well-known formula for the air thermometer is H-h + T= l + at> _ Solving with reference to T, and placing in a more convenient form, we have H-h' nearly, OF ARTS AND SCIENCES. 99 where and For the first bulb, For the second bulb, = .0058. To discuss the error of T due to errors in the constants, we must replace a by its experimental value, seeing that it was determined with the same apparatus as that by which T was found. As it does not change very much, we may write approximately T= 100 H h V l + yt From this formula we can obtain by differentiation the error in each of the quantities, which would make an error of one tenth of one per cent in T. The values are for T = 40 nearly ; t = 20 ; ff lw h = 270""" ; and h = 750 mm -. If x is the variable, A x = dx dT ' *JL _L_ - ru d JL d T looo - u * d T TABLE VII. ERRORS PRODUCING AN ERROR IN T OF 1 IN 1000 AT 40 C. H H 100 orA. V T ha a - constant, a b a ^5 constant. 1,90 k^/con*. b ino -b a -^ constant. Absolute ralue, Ax n mm. 27 mm. .005 .00074 .00087 .0047 .00087 Relative value, Az X ... ... 0.9 .10 .12 .62 ... From this table it would seem that there should be no difficulty in determining the 40 point on the air thermometer to at least 1 in 2000 ; and experience has justified this result. The principal difficulty is in the determination of H, seeing that this includes errors in reading the barometer as well as the cathetometer. For this reason, as men- tioned before, I have designed another instrument for future use, in which the barometer is nearly dispensed with by use of an artificial atmosphere of constant pressure. The value of ^ does not seem to affect the result to any great extent; and if it was omitted altogether, the error would be only 100 PROCEEDINGS OF THE AMERICAN ACADEMY about 1 in 1,000, assuming that the temperature t was the same at the determination of the zero point, the 40 point, and the 100 point. It seldom varied much. The coefficient of expansion of the glass influences the result very slightly, especially if we know the difference of the mean coefficients between and 100, and say 10 and -f 10. This difference I at first determined from Regnault's tables, but afterwards made a deter- mination of it, and have applied the correction.* The table given by Regnault is for one specimen of glass only ; and I sought to better it by taking the expansion at 100 from the mean of the five specimens given by Regnault on p. 231 of the first volume of his Relation des Experiences, and reducing ' the numbers on page 237 in the same proportion. I thus found the values given in the second column of the following table. TABLE VIII. COEFFICIENT OF EXPANSION OF THE GLASS OF THE AIR THERMOMETER, ACCORDING TO THE AIR THERMOMETER. Tempera- ture ac- cording to Air Ther- mometer. Values of 6 used for a first Calculation b from Regnault's Table, Glass No. 5. Experimental Results. Apparent Coefficient of Expansion of Mercury. b, using Regnault's Value for Mercury .t 6, using Recknagel'g Value for Mercury 4 6, using Wullnel's Value for Mercury . 20 40 60 80 100 .0000252 .0000253 .0000256 .0000259 .0000262 .0000264 .0000263 .0000264 .0000267 .0000270 .0000273 .0000276 .00015410 .00015395 .00015391 .00015381 .0000254 .0000258 .0000261 .0000277 .0000264 .0000266 .0000267 .0000277 .0000273 .0000276 .0000278 .0000287 The second column contains the values which I have used, and one of the last three columns contains my experimental results, the last being probably the best. The errors by the use of the second column compared with the last are as follows : TTT ^ from using b m 6 40 = .0000008 instead of .0000011 ; TuW from using b m = .0000264 instead of .0000287 ; or, ffa-Q for both together. * This was determined by means of a large weight thermometer in which the mercury had been carefully boiled. The glass was from the same tube as that of the air thermometer, and they were cut from it within a few inches of each other. t Relation des Experiences, i. 328. t Pogg. Ann., cxxiii. 135. Experimental Physik, Wullner, i. 67. OP ARTS AND SCIENCES. 101 As the error is so small, I have not thought it worth while to entirely recalculate the tables, but have calculated a table of corrections as follows, and have so corrected them : TABLE IX. TABLE OP CORRECTIONS. Tf T Calculated Corrected Correction. Temperature. Temperature. 8 10 9.9971 .0029 20 19.9946 .0054 30 29.9924 0076 40 39.9907 .0093 50 49.9894 .0106 60 69.9865 .0135 80 79.9880 .0120 100 100. T = T' {1 + 373 (b' m - b m ) - (273 + T) (V - &)}. T T 1 {1 .000858 -f (273 -f T'} (b b')}. T = .99975 T' approximately between and 40. This last is true within less than T ^ 7 of a degree. The two bulbs of the air thermometer used were from the same piece of glass tubing, and consequently had nearly, if not quite, the same coefficient of expansion. In the reduction of the barometer and other mercurial columns to zero, the coefficient .000162 was used, seeing that all the scales were of brass. In the tables the readings of the thermometers are reduced to volumes of the tube from the tables of calibration, and they are cor- rected for the pressure of water, which increased their reading, except at 0, by about 0.01 C. The order of the readings was as follows in each observation : 1st, barometer; 2d, cathetometer; 3d, thermometers forward and backward ; 4th, cathetometer ; 5th, barometer, &c., repeating the same once or twice at each temperature. In the later observations, two series like the above were taken, and the water stirred between them. The following results were obtained at various times for the value of a with the first bulb : 102 PROCEEDINGS OP THE AMERICAN ACADEMY Mean .0036664 .0036670 .0036658 .0036664 .0036676 .00366664 obtained by using the coefficient of expansion of glass .0000264 at 100 or a = 0036698, using the coefficient .0000287. The thermometers Nos. 6163, 6165, 6166, were always taken out of the bath when the temperature of 40 was reached, except on November 14, when they remained in throughout the whole experi- The thermometer readings are reduced to volumes by the tables of calibration. TABLE X. IST SERIES, Nov. 14, 1877. Relative Weight. Air Thermometer. V 6163. V 6166. 7 6167. Temperature by 6167. A 4 o 115.33 21.25 6.147 4 17.1425 422.84 255.80 15.685 17.661 .236 4 23.793 634.71 341.05 19.157 24.089 .296 5 30.582 653.49 431.71 22.833 80.896 .814 2 38.569 793.18 27.175 38.935 .366 2 61.040 33.864 61.320 .280 4 59.137 38.256 69.452 .816 The first four series, Tables X. to XIII., were made with one bulb to the air thermometer. A new bulb was now made, whose capacity was 192.0 ccm -, that of the old being 201.98 c - cm -. The value of -p for the new bulb was .0058. The values of h' and a were obtained as follows : a hf June 8th .00366790 753.876 June 22d .00366977 753.805 June 25th .00366779 753.837 Mean .0036685 753.84 This value of a is calculated with the old coefficient for glass, new would have given .0036717. The OF ARTS AND SCIENCES. 103 W 53 5 gg 8 Jf fc-|7 '"*'*'*^-* 1 <1 ;i *'*'*T>50Ou3o Jih ooioo s^ i as ^ s s ^ s 104 ' PROCEEDINGS OF THE AMERICAN ACADEMY fi- J ^ iii s s s s T* t-- t 1 * CO OS 00 -H rH g 35 3 f S S5 a? s CO CO ^tl 00 g O W N c$ t- as i-i S 8 oo ec oj to to to S S OF ARTS AND SCIENCES. 105 *> it- Difference Column reduced ~ 8 S i s s s 10 cd ca t^ S S & & If 5 S 2 -N S S 3 g 8 S 5! c *Q CO CC ^ CO l~ t^ O 00 O r-! 2 3 $ co rh co S S 58 !o lo S 11 ft !J S at 11 P CO cc cc -^ 10 106 PROCEEDINGS OF THE AMERICAN ACADEMY 5 q q i + I o o 5 So o co 3" e S 3 TP eo to & S S 5 ll S S r- O CO t^- Nil- Ili ll 2 8 S I .8 I -f S ^ S S? x 8 9 S 2 5! S S S 8 8 o> oo 10 -j( O CD 5 *-l (N (N (M OP ARTS AND SCIENCES. 107 + + + + + + I" I t-< (N t-. ^ S So ? S S co Oi I-* I-H o i>- < 100 T*t-COiOiOOOcO S S ^ S S g S I S 88 I S 3 I S g s ji fe o' i |U I J_I_ f: S S 5 S
(1 +34.8 m).
Hence m = ~^ .
We thus obtain the following results :
Date. Weight. Value of m.
Nov. 14 1 .000236
Nov. 20 2 .000218
Jan. 25 3 .000217
Feb. 11 4 .000197
June 8 3 .000215
June 22 2 .000216
Mean .000213
OF ARTS AND SCIENCES. Ill
The results for m are then as follows :
From direct comparison of No. 6167 with the air thermometer .000187
From " " " No. 6163 " " .000154
From comparison of No. 6163 with No. 6167 .000213
The first and last are undoubtedly the most exact numerically, but
they apply to No. 6167, and are also, especially the first, derived
from somewhat higher temperatures than the 20 point, where the
correction is the most important. The value of m, as determined in
either of these ways, depends upon the determination of a difference of
temperature amounting to 0.30, and hence should be quite exact.
The value of m, as obtained from the direct comparison of No. 6163
with the air thermometer, depends upon the determination of a differ-
ence of about 0.05 between the mercurial and the air thermometer.
At the same time, the comparison is direct, the temperatures are the
same as we wish to use, and the glass is the same. I have combined
the results as follows :
m from No. 6167 .000200
m " 6163 .000154
Mean .00018*
It now remains to deduce from the tables the ratios of the constants
for the different thermometers.
The proper method of forming the equations of condition are as
follows, applying the method to the first series:
Weight.
4 21.25 C tll = 115.33 C, v
4 255.80 C UI = 422.84 C, v
4 341.05 C in = 534.71 G t v
5 431.71 C UI = 653.49 C, v
where G UI is the constant for No. 6166, C, is that for No. 6163,
and v is a constant to be eliminated. Dividing by (7/, the equations
S~i
can be solved for --. The following table gives the results.
* See Appendix to Thermometry, where it is finally thought best to reject
the value from No. 6167 altogether.
112
PROCEEDINGS OP THE AMERICAN ACADEMY
TABLE XVI. RATIOS OF CONSTANTS.
6163
6166
6166
6165
6165
Date.
Weight.
6167
6167
6163
6163
6166
Nov. 14
1
.031009
.040058
1.3111
Nov. 20
2
.030985
.040670
1.3128
Jan. 25
3
....
....
1.3122
Feb. 11
4
1.3115
8.6588
6.1449
June 8
3
....
1.3108
8.0605
6.1469
June 22
2
1.3122
8.0588
6.1428
Mean j
.030993
'.00005
.040666
.000003
1.31175
.0004
8.0594
.0002
6.1451
.0004
From these we have the following, as the final most probable
results :
' . C n =8.0601 G t ,
. C UI = 1.31175 (7,,
C, = .031003 (7 iT ,
C,, .24991 C iv ,
C UI = .040661 <7 iv ,
of which the last three are only used to calculate the temperatures on
the mercurial thermometer, and hence are of little importance in the
remainder of this paper.
The value of C" which we have found for the old value of the
coefficient of expansion of glass was
6^ = . 056976;
and hence, corrected to the new coefficient, it is, as I have shown,
C, =.056962.
Hence,
<7 ;/ = .45912,
, = .074720.
And we have finally the three following equations to reduce the ther-
mometers to temperatures on the air thermometer:
Thermometer No. 6163:
T=. 056962 V-t '-.miST(4Q-T) (l - .003 (7*+ 40)).
Thermometer No. 6165 :
^=.45912 V"- C __. 00018 T(T- 40) (l _ .003 (7+40)).
OF ARTS AND SCIENCES. 113
Thermometer No. 6166 :
= . 074720 r'"*,/" .00018 T(T 40) (l .
where V, V", and V" are the volumes of the tube obtained by
calibration ; t ', t ", and t '" are constants depending on the zero point,
and of little importance where a difference of temperature is to be
measured ; and T is the temperature on the air thermometer.
On the mercurial thermometer, using the and 100 points as fixed,
we have the following by comparison with No. 6167 :
Thermometer No. 6163 ; t = .057400 V t ;
Thermometer No. 6165 ; t = .46265 V t Q ;
Thermometer No. 6166 ; t = .4)75281 V t v
The Kew Standard.
The Kew standard must be treated separately from the above, as
the glass is not the same. This thermometer has been treated as if
its scale was arbitrary.
In order to have variety, I have merely plotted all the results
with this thermometer, including those given in the Appendix, and
drawn a curve through them. Owing to the thermometer being- only
divided to F., the readings could not be taken with great accuracy,
and so the results are not very accordant ; but I have done the best I
could, and the result probably represents the correction to at least
0.02 or 0.03 at every point.
('/) Reduction to the Absolute Scale.
The correction to the air thermometer to reduce to the absolute
scale has been given by Joule and Thomson, in the Philosophical
Transactions for 1854 ; but as the formula there used is not correct,
I have recalculated a table from the new formula used by them in
their paper of 1862.
That equation, which originated with Rankine, can be placed in
the form &i/Vl~ '
where p, v, and p. are the pressure, volume, and absolute temperature
of a given weight of the air ; D is its density referred to air at C.
and 760 mm - pressure ; fj, is the absolute temperature of the freezing
point; and m is a constant which for air is 0.33 C.
VOL. xv. (N. s. vn.) 8
14 PROCEEDINGS OP THE AMERICAN ACADEMY
For the air thermometer with constant volume
T = 100 pt ~ Po -,
PlGO PO
or, since D = 1,
,,- ^ = T- . 00088 T =,
from which I have calculated the following table of corrections.
TABLE XVII. REDUCTION OF AIR THERMOMETER TO ABSOLUTE SCALE.
T
Air Thermometer.
A* Mo
Absolute Temperature.
A
or Correction to
Air Thermometer.
10
9.9972
.0028
20
19.9952
.0048
30
29.9939
.0061
40
39.9933
.0067
50
49.9932
.0068
60
. 59.9937
.0063
70
69.9946
.0054
80
79.9956
.0044
90
89.9978
.0022
100
100.000
200
200.037
+.037
300
300.092
+.092
400
400.157
+.157
500
500.228
+.228
It is a curious circumstance, that the point of maximum difference
occurs at about the same point as in the comparison of the mercurial
and air thermometers.
From the previous formula, and from this table of corrections, the
following tables were constructed.
^ 7 '
OF ARTS AND SCIENCES.
115
TABLE XVIII. THERMOMETKB No. 6163.
= Q
11
i
li^
111
H
i!
II
If!
ill!
ll
iS
1 Reading in
meters on
Temperature
curiiil Them
and 100
Temperature
curial Them
and 40 fi
Air Thermc
1
Temperate
Absolute Set
0C
1 Reading in
meters on
Temperature
curial Them
and 100
Temperature
curial Them
and 40 fi
Air Thermc
II
Temperatu
Absolute Sea
0C.
50
.923
.917
-.911
.911
240
20.557
20.409
20.350
20.345
58.1
250
21.670
21.515
21.457
21.452
60
+.217
+.215
+.214
+.214
260
22.776
22.616
22.559
22.564
70
1.356
1.336
1.328
1.328
270
23.884
23.713
23.657
23.652
80
2.494
2.475
2.461
2.460
280
24.989
24.810
24.755
24.750
90
3.631
3.604
3.584
3.583
290
26.093
25.907
25.854
25.848
100
4.767
4.733
4.707
4.706
300
27.200
27.006
26.956
26.950
110
5.903
5.860
5.829
5.827
310
28.311
28.108
28.060
28.056
120
7.036
6.986
6.950
6.948
320
29.425
29.214
29.169
29.163
130
8.170
8.111
8.071
8.069
330
30.541
30.324
30.282
30.276
140
9.304
9.237
9.193
9.190
340
31.662
31.436
31.398
31.392
150
10.436
10.361
10.314
10.311
350
32.782
32.548
32.514
32.508
160
11.568
11.485
11.435
11.432
360
33.903
33.660
33.630
33.624
170
12.700
12.608
12.556
12.553
370
35.023
34.773
34.748
34.742
180
13.829
13.730
13.676
13.672
380
36.143
35.884
35.864
35.857
190
14.957
14.850
14.794
14.790
390
37.261
36.994
36.979
36.972
200
16.081
15.906
15.909
15.905
400
38.377
38.103
38.094
38.087
210
17.203
17.080
17.022
17.018
410
39.492
39.210
39.206
39.199
220
18322
18.191
18.132
18.127
420
40.604
40.314
40.316
40.309
230
19.440
19.301
19.242
19.237
TABLE XIX. THERMOMETER No. 6165.
s
1
!fi
lilt
|
si
B-3
21
ifj
11!
|
si
EJ
.S o
sPa,
5-S2
2
SSd
a -
So
g Q tea 2
3
3 d
ii
Temperatu
curial Thei
and 101
Temperatu
curial Thei
and 40
Air Then
1
Tempera
Absolute S
bo _
-
Temperatu
curial Thei
and 1(K
Temperatu
curial Thei
and 40
Air Then
II
-
po
'!
30
-.464
-.460
-.457
-457
230
17.198
17.067
17.009
17.005
35
240
18.056
17.920
17.861
17.857
40
+.463
+.460
+.457
+.457
250
18.917
18.773
18.714
18.709
50
1.387
1.376
1.368
1.368
260
19.771
19.621
19.562
19.557
60
2.307
2.290
2.276
2.275
270
20.621
20.465
20.406
20.401
70
3.216
3.192
3.174
3.173
280
21.469
21.306
21.247
21.242
80
4.122
4.092
4.069
4.068
290
22.308
22.139
22.081
22.076
90
5.022
4.984
4.957
4.955
300
23.144
22.969
22.912
22.907
100
5.916
5.872
5.841
5.839
310
23.974
23.792
23.736
23.731
110
6.804
6.753
6.714
6.712
320
24.796
24.607
24.552
24.547
120
7.685
7.628
7.590
7.588
330
25.618
25.424 1 25.370
25.365
130
8.564
8.500
8.459
8.456
340
26.433
26.232
46.180
26.174
140
9.439
9.368
9.324
9.321
350
27.245
27.038
16.987
26.981
150
10.309
10.232
10.186
10.183
360
28.049
27.837
27.788
27.782
160
11.174
11.091
11.042
11.039
370
28.856
28.637
28.590
28.584
170
12.038
11.947
11.896
11.893
380
29.651
29.426
29.382
29.376
180
12.900
12.802
12.749
12.746
390
30.449
30.218
30.176
30.170
190
13.760
13.655
13.601
13.598
400
31.249
31.011
30.971
30.965
200
14.619
14.508
14.453
14.450
410
32.073
31.829
31.782
31.786
210
15.479
15.362
15.305
15.302
420
32.861
32.611
32.577
32.581
220
16.340
16.215
16.157
16.153
116
PROCEEDINGS OF THE AMERICAN ACADEMY
TABLE XX. THERMOMETER No. 6166.
1*
111
ifi
1
?. a
2
il
ill
If*
13
1
11
*tfl
H
Ji
Temperature <
curial Therm(
and 100 i
Temperature <
curial Therm <
and 40 f
Temperatui
Air Thermoi
Temperatu
Absolute Sea
0C.
2 'ia
J?B
Temperature
curial Therm
and 100
Temperature
curial Therm
and 40
1
Temperatu
Absolute Sot
0C.
20
.036
-.036
.034
-.034
230
16.478
16.356
16.298
16.294
30
40
+.770
1.574
+.764
1.562
+.759
1.553
+.759
1.553
240
250
17.259
18.042
17.132
17.908
17.074
17.849
17.070
17.845
50
2.368
2.350
2.336
2.335
260
18.825
18.686
18.627
18.622
60
3.156
3.133
3.115
3.114
270
19.609
19.464
19.405
19.400
70
3.941
3.911
3.889
3.888
280
20.392
20.241
20.182
20.177
80
4.726
4.691
4.665
4.664
290
21.170
21019
20.960
20.955
90
5.509
5.468
5.438
5.436
300
21.735
21.793
21.735
21.730
100
6.293
6.246
6.212
6.210
310
22.511
22.569
22.511
22.506
110
7.076
7024
6.988
6.986
320
23.292
23.349
23.292
23.287
120
7.862
7.804
7.765
7.763
330
24.075
24.131
24.075
24.070
130
8.649
8.585
8.544
8.542
340
24.855
24.910
24.855
24.850
140
9.437
9.367
9.323
9.321
350
25.634
25.687
25634
25.628
150
10.228
10.151
10.105
10.102
360
26.415
26.466
26.412
26.406
160
11017
10.935
10.887
10.884
370
27.441
27.245
27.195
27.189
170
11.805
11.717
11.667
11.664
380
28.240
28.030
27.982
27.976
180
12.589
12.496
12.444
12.441
390
29.030
28.814
28.768
28.762
190
13.370
13.271
13.217
13.214
400
29.819
29.597
29.550
29.544
200
14.148
14.043
13.988
13.984
410
30.608
30.381
30.339
30.333
210
14.923
14.812
14.756
14.752
420
31.396
31.162
31.123
31.117
220
15.699
15.583
15.526
15.522
430
32.189
31.950
31.914
31.908
In using these tables a correction is of course to be made should
the zero point change.
TABLE XXI. CORRECTION OF KEW STANDARD TO THE ABSOLUTE SCALE.
Temperature C.
Correction in
Degrees C.
10
.03
20
.05
30
.06
40 .07
50
.07
60
.06
70
-.04
80
.02
90
.01
100
OF ARTS AND SCIENCES.
117
Appendix to Thermometry.
The last of January, 1879, Mr. S. W. Holraan, of the Massachusetts
Institute of Technology, came to Baltimore to compare some ther-
mometers with the air thermometer ; and by his kindness I will give
here the results of the comparison which we then made together.
As in this comparison some thermometers made by Fastre in 1851
were used, the results are of the greatest interest.
'The tables are calculated with the newest value for the coefficient
of expansion of glass. The calibration of all the thermometers, except
the two by Casella, has been examined, and found good. The Casella
thermometers had no reservoir at the top, and could not thus be readily
calibrated after being made. The Geissler also had none, but I suc-
ceeded in separating a column.
The absence of a reservoir at the top should immediately condemn
a standard, for there is no certainty in the work done with it.
TABLE XXII. SEVENTH SERIES.
Original Readings.
Reduced Readings.
Air
6163
Ther-
Kew
Reduced
Kew
mome-
ter.
6163.
7334
Baudin.
Stand-
ard
32374
Casella.
Geissler.
to Air
Ther-
7334
Baudin.
Stand-
ard
32374
Casella.
Geissler.
No. 104.
mome-
No.104.
ter.
*58.83
.11
32.68
+.20
+.69
6
a
t.43
63.5
3360
.71
.52
.52
.51
6.08
113.0
43.65
6.33
6.08
6.11
6.13
12.68
171.55
12.59
55.47
12.91
13.42
12.65
12.73
12.68
1270
12.82
20.49
242.0
20.48
69.56
20.77
21.29
20.49
20.63
20.57
20.56
20.74
24.55
278.8
24.50
76.90
24.80
25.33
24.54
24.66
24.61
24.59
24.81
29.51
323.9
29.49
85.88
29.80
30.32
29.52
29.66
29.61
29.58
29.83
39.45
4131
39.43
103.72
39.76
40.22
39.47
39.62
39.53
39.54
39.80
39.15
4107
39.15
103.23
39.48
39.08
39.20
39.34
39.26
39.26
39.56
51.17
51.10
124.84
51.49
51.83
....
51:32
51.29
51.26
51.49
61.12
61.05
142.73
61.47
61.69
61.29
61.24
61.23
61.41
70.74
' |
70.57
159.87
71.00
71.14
.....
70.83
70.78
70.76
70.92
80.09
79.74
176.50
80.31
80.25
80.02
80.04
80.06
80.10
80.39
80.15
177.23
80.74
.80.66
.
80.43
80.44
80.49
80.51
89.95
89.63
194.35
90.22
90.11
89.93
89.97
89.97
90.03
89.92
89.59
194.22
90.18
90.06
.
89.89
89.90
89.93
89.98
100.00
99.69
212.37
100.06
99.32
. . .
100.00
100.00
100.00
100.00
* The original readings in ice were 58.68 and 58 45, to which .15 was added
to allow for the pressure of water in the comparator. This, of course, gives
the same final result as if .15 were subtracted from eacli of the other tempera-
tures. No correction was made to the others.
t Probably some error of reading.
118
PROCEEDINGS OP THE AMERICAN- ACADEMY
TABLE XXIII EIGHTH SERIES.
Original Readings.
Reduced Readings.
Air
Ther-
mome-
6163
376
7316
Baudin.
368
Fastr<5.
3236
Casella.
6163
Reduced
to Air
Ther-
376
Fastrd.
7316
Baudin
868.
Fastr<5.
3235
Casella.
ter.
Fastre.
mome-
ter.
*5860
111.3
.23
87.6
32.80
5
ft
3.67
90.7
130.0
106.25
39.35
3.61
3.64
3.64
3.65
11.55
161.6
170.9
11.40
147.2
53.70
11.56
11.60
il.64
11.62
11. Q3
20.72
243.7
217.9
20.59
194.2
70.15
20.70
20.75
20.84
20.80
20.79
32.19
347.4
276.9
32.09
253.2
90.80
32.17
32.24
32.34
32.28
32.29
39.36
411.85
313.85
39.26
290.1
103.68
39.36
39.43
39.52
39.48
39.45
50.71
372.0
50.57
248.2
123.65
i
50.75
50.84
50.80
50.57
60.10
. '.
420.0
59.92
39(5.45
140.80
.
60.10
60.19
60.21
60.12
73.82
490.6
73.59
466.85
165.68
73.84
73.87
73.93
73.97
86.50
655.25
86.16
531.22
188.20
.
86.48
86.51
86.56
86.56
550.2
85.21
525.95
186.42
86.45
85.60
85.45
85.51
100.66
624.93
99.70
600.58
212.45
. . .
100.00
100.00
100.00
100.00
From these tables we would draw the inference that No. 6163
represents the air thermometer with considerable accuracy. At the
same time, both tables would give a smaller value of m than I have
used, and not very far from the value found before by direct compari-
son, namely, .00015.
The difference from using m = .00018 would be a little over
0.01 C. at the 20 point.
All the other thermometers stand above the air thermometer,
between and 100, by amounts ranging between about 0.05 and
0.35 C., none standing below. Indeed, no table has ever been
published showing any thermometer standing below the air ther-
mometer between and 100. By inference from experiments above
100 on crystal glass by Regnault, thermometers of this glass should
stand below, but it never seems to have been proved by direct experi-
ment. The Fastre thermometers are probably made of this glass,
and my Baudins certainly contain lead ; and yet these stand above,
though only to a small amount, in the case of the Fastre's.
The Geissler still seems to retain its pre-eminence as having the
greatest error of the lot.
The Baudin thermometers agree well together, but are evidently
made from another lot of glass from the No. 6 1 67 used before. These
last two depart less from the air thermometer. The explanation is
plain, as Baudin had manufactured more than one thousand ther-
* See note on preceding page.
OF ARTS AND SCIENCES.
119
mometers between the two, and so had probably used up the first
stock of glass. And even glass of the same lot differs, especially as
Regnault has shown that the method of working it before the blow-
pipe affects it very greatly.
It is very easy to test whether the calorimeter thermometers are of
the same glass as any of the others, by testing whether they agree
with No. 6163 throughout the whole range of 40. The difference
in the values of m for the two kinds of glass will then be about
.003 of the difference between them at 20, the and 40 points
agreeing. The only difficulty is in calibrating or reading the 100
thermometers accurately enough.
The Baudin thermometers were very well calibrated, and were
graduated to -fa C., and so were best adapted to this kind of work.
Hence I have constructed the following tables, making the and 40
points agree.
TABLE XXIV. COMPARISON OF 6163 AND THE BAUDIN STANDARDS.
6163
Mercurial
and 40 fixed.
7334*
Difference.
6163
Mercurial
and 40 fixed.
7316.*
Difference.
12.699
20.547
24.604
29.564
39.337
12.673
20.553
24.567
29.550
39.337
+.026
.006
+.037
+ 014
11.609
20.762
32.203
39.358
11.584
20.746
32.211
39.358
+.025
+.016
.008
Taking the average of the two, it would seem that No. 6163 stood
about .015 higher than the mean of 7334 and 7316 at the 20 point,
or 6163 has a higher value of m by .000045 than the others.
These differ about .17 from the air thermometer at 40, which gives
the value of m about .000104. Whence m for 6163 is .00015, as we
have found before by direct comparison with the air thermometer.
I am inclined to think that the former value, .00018, is too large,
and to take .00015, which is the value found by direct comparison, as
the true value. As the change, however, only makes at most a differ-
ence of 0.01 at any one point, and as I have already used the previous
value in all calculations, I have not thought it worth while to go over
all my work again, but will refer to the matter again in the final
results, and then reduce the final results to this value.
* A correction of 0.01 was made to the zero points of these thermometers
on account of the pressure of the water.
120 PROCEEDINGS OP THE AMERICAN ACADEMY
III. CALORIMETRY.
(a) Specific Heat of Water.
The first observers on the specific heat of water, such as De Luc,
completed the experiment with a view of testing the thermometer ;
and it is curious to note that both De Luc and Flaugergues found the
temperature of the mixture less than the mean of the two equal por-
tions of which it was composed, and hence the specific heat of cold
water higher than that of warm.
The experiments of Flaugergues were apparently the best, and he
found as follows : *
3 parts of water at and 1 part at 80 R. gave 19.86 R.
2 parts of " " 2 parts " " 39.81 R.
1 part of " " 3 parts " " 59.87 R.
But it is not at all certain that any correction was made for the
specific heat of the vessel, or whether the loss by evaporation or
radiation was guarded against.
The first experiments of any accuracy on .this subject seem to have
been made by F. E. Newmann in 1831.f He finds that the specific
heat of water at the boiling point is 1.0127 times that at about 28 C.
(22 R.).
The next observer seems to have been Regnault,t who, in 1840,
found the mean specific heat between 100 C. and 16 C. to be
1.00709 and 1.00890 times that at about 14.
But the principal experiments on the subject were published by
Regnault in 1850, and these have been accepted to the present time.
It is unfortunate that these experiments were all made by mixing
water above 100 with water at ordinary temperatures, it being
assumed that water at ordinary temperatures changed little, if any.
An interpolation formula was then found to represent the results;
and it was assumed that the same formula held at ordinary tempera-
ture, or even as low as C. It is true that Regnault experimented
on the subject at points around 4 C. by determining the specific heat
of lead in water at various temperatures ; but the results were not of
sufficient accuracy to warrant any conclusions except that the variation
was not great.
* Gehler, Phys. Worterbuch, i. 641.
t Pogg. Ann., xxiii. 40. j Ibid., li. 72.
Pogg. Ann., Ixxix. 241 ; also, Rel. d. Exp., i. 729.
OF ARTS AND SCIENCES. 121
Boscha has attempted to correct Regnault's results so as to reduce
them to the air thermometer ; but Regnault, in Comptes Rendus, has
not accepted the correction, as the results were already reduced to the
air thermometer.
Him (Comptes Rendus, Ixx. 592, 831) has given the results of
some experiments on the specific heat of water at low temperatures,
which give the absurd result that the specific heat of water increases
about six or seven per cent between zero and 13! The method of
experiment was to immerse the bulb of a water thermometer in the
water of the calorimeter, until the water had contracted just so much,
when it was withdrawn. The idea of thus giviug equal quanti-
ties of heat to the water was excellent, but could not be carried
into execution without a great amount of error. Indeed, experi-
ments so full of error only confuse the physicist, and are worse than
useless.
The experiments of Jamin and Amaury, by the heating of water by
electricity, were better in principle, and, if carried out with care,
would doubtless give good results. But no particular care seems to
have been taken to determine the variatiorfof the resistance of the
wire with accuracy, and the measurement of the temperature is
passed over as if it were a very simple, instead of an immensely diffi-
cult matter. Their results are thus to be rejected ; and, indeed,
Regnault does not accept them, but believes there is very little change
between 5 and 25.
In PoggendorfFs Annalen for 1870 a paper by Pfaundler and Plat-
ter appeared, giving the results of experiments around 4 C., and
deducing the remarkable result that water from to 10 C. varied as
much as twenty per cent, in specific heat, and in a very irregular man-
ner, first decreasing, then increasing, and again decreasing. But
soon after another paper appeared, showing that the results of the
previous experiments were entirely erroneous.
The new experiments, which extended up to 13 C., seemed to give
an increase of specific heat up to about 6, after which there was
apparently a decrease. It is to be noted that Geissler's thermometers
were used, which I have found to depart more than any other from
the air thermometer.
But as the range of temperature is very small, the reduction to the
air thermometer will not affect the results very much, though it will
somewhat decrease the apparent change of specific heat.
In the Journal de Physique for November, 1878, there is a notice of
some experiments of M. von Munchausen on the specific heat of
122 PROCEEDINGS OF THE AMERICAN ACADEMY
water. The method was that of mixture in an open vessel, where
evaporation might interfere very much with the experiment. No
reference is made to the thermometer, but it seems not improbable
that it was one from Geissler ; in which case the error would be very
great, as the range was large, and reached even up to 70 C. The
error of the Geissler would be in the direction of making the specific
heat increase more rapidly than it should. The formula he gives for
the specific heat of water at the temperature t is
1 _|_ .000302 t.
Assuming that the thermometer was from Geissler, the formula,
reduced to the air thermometer, would become approximately
1 _ .00009 t + .0000015 t 2 .
Had the thermometer been similar to that of Recknagel, it would
have been 1 -f .000045 t + .000001 t\
It is to be noted that the first formula would actually give a decrease
of specific heat at first, and then an increase.
As all these results vary so very much from each other, we can
hardly say that we know anything about the specific heat of water
between and 100, though Regnault's results above that temperature
are probably very nearly correct.
It seems to me probable that my results with the mechanical
equivalent apparatus give the variation of the specific heat of water
with considerable accuracy ; indeed, far surpassing any results which
we can obtain by the method of mixture. It is a curious result of
those experiments, that at low temperatures, or up to about 30 C.,
the specific heat of water is about constant on the mercurial ther-
mometer made by Baudin, but decreases to a minimum at about 30
when the reduction is mvde to the air thermometer or the absolute scale,
or, indeed, the Kew standard.
' As this curious and interesting result depends upon the accurate
comparison of the mercurial with the air thermometer, I have spent
the greater part of a year in the study of the comparison, but have
not been able to find any error, and am now thoroughly convinced of
the truth"of this decrease of the specific heat.' But to make certain,
I have instituted the following independent series of investigations
on the specific heat of water, using, however, the same thermome-
ters.
The apparatus is shown in Fig. 4. A copper vessel, A, about 20 cm -
OP ARTS AND SCIENCES.
123
in diameter and 23 cm high, rests upon a tripod. In its interior is a
three-way stopcock, communicating with the small interior vessel ,
the vessel A, and the vulcanite spout G. By turning it, the vessel B
could be filled with water, and its temperature measured by the ther-
mometer D, after which it could be delivered through the spout into
the calorimeter. As the.vessel B, the stopcock, and most of the spout,
were within the vessel A, and thus surrounded by water, and as the
vulcanite tube was very thin, the water could be delivered into the
calorimeter without appreciable change of temperature. The proof
of this will follow later.
The calorimeter, JE, was of very thin copper, nickel-plated very
thinly. A hole in the back at F allowed the delivery spout to enter,
and two openings on top admitted the thermometers. A wire attached
to a stirrer also passed through the top. The calorimeter had a
capacity of about three litres, and weighed complete about 388.3
grammes. Its calorific capacity was estimated at 35.4 grammes. It
rested on three vulcanite pieces, to prevent conduction to the jacket.
124 PROCEEDINGS OP THE AMERICAN ACADEMY
Around the calorimeter on all sides was a water-jacket, nickel-
plated on its interior, to make the radiation perfectly definite.
The calorific capacity of the thermometers, including the immersed
stem and the mercury of the bulb, was estimated as follows: 14 cm - of
stem weighed about 3.8*% and had a capacity of .8^; 10"- of
mercury had a capacity of .3**- ; total, I.!* 1 -.
Often the vessel B was removed, and the water allowed to flow
directly into the calorimeter.
The following is the process followed during one experiment at low
temperatures. The vessel A was filled with clean broken ice, the open-
ing into the stopcock being covered with fine gauze to prevent any
small particles of ice from flowing out. The whole was then covered
with cloth, to prevent melting. The vessel was then filled with water,
and the two thermometers immersed to get the zero points. The
calorimeter being about two thirds filled with water, and having been
weighed, was then put in position, the holes corked up, and one ther-
mometer placed in it, the other being in the melting ice. An obser-
vation of its temperature was then taken every minute, it being
frequently stirred.
When enough observations had been obtained in this way, the cork
was taken out of the aperture .Fand the spout inserted, and the water
allowed to run for a given time, or until the calorimeter was full. It
was then removed, the cork replaced, and the second thermometer
removed from the ice to the calorimeter. Observations were then
taken as before, and the vessel again weighed.
Two thermometers were used in the way specified, so that one
might approach the final temperature from above and the other from
below. But no regular difference was ever observed, and so some
experiments were made with both thermometers in the calorimeter
during the whole experiment.
The principal sources of error are as follows :
1st. Thermometers lag behind their true reading. This was not
noticed, and would probably be greater in thermometers with very
fine stems like Geissler's. At any rate, it was almost eliminated in
the experiment by using two thermometers.
2d. The water may be changed in temperature in passing through
the spout. This was eliminated by allowing the water to run some
time before it went into the calorimeter. The spout being very thin,
and made of vulcanite, covered on the outside with cloth, it is not
thought that there was any appreciable error. It will be discussed
more at length below, and an experiment given to prove this.
OP ARTS AND SCIENCES.
125
3d. The top of the calorimeter not being in contact with the water
rt. temperature may be uncertain. To eliminate this, the
was often at the temperature of the air to commence with. Also the
water was sometimes violently agitated just before taking the final
leading previous to letting in the cold water. Even if the tempera
ture of this part was take* as that of the air, the error would scarcely
ever be of sufficient importance to vitiate the conclusions
5th Some water might remain in the spout whose temperature
might be different from the rest. This was guarded against.
6 h. Evaporation. Impossible, as the calorimeter was closed.
7th. The introduction of cold water may cause dew to be depos-
octrred rimeter ' The ex P eri <*^ were rejected where this
The corrections for the protruding thermometer stem, for radiation
&c, were made as usual, the radiation being estimated by a series of
observations before and after the experiment, as is usual in determin-
ing the specific heat of solids.
June 14, 1878. Wr* Experiment.
Time.
41
42
43
44
Ther. 6168.
296.75
296.7
296.7
296.65
Points.
6163, 57.9 Air, 21 C.
6165, 34.8 Jacket about 25 C.
6166, 20.5
44-44f Water running.
*6i 218.7 251.7
4 H 218.8 251.8
*8 218.9 252.0
Calorimeter before
after
Water at added
Thermometer
2043.0
2853.3
810.3
1.1
Total' at
811.4
Temperature before 296.6
Correction for -}- .2
Calorimeter before
Weight of vessel
Water
Capacity of calorim.
thermom
2043.0
388.3
TesI?
35.4
1.1
296.8 = 26.597
Correction for stem -{- .019
Initial temp, of calorimeter 26.616
Total capacity
1691.2
126 PROCEEDINGS OP THE AMERICAN ACADEMY
218 . 6 + .2 = 218.8 = 17.994 251.6 - 1 = 251.5 = 17.962
Correction for stem 006 Correction for stem 006
1^988 17 ' 956
Mean temperature of mixture, 17.972.
Mean specific heat 18 1691.2 X 8.644 __ l m5
Me^n specific heat 18 27 811.4 X 17-972
June 14. Second Experiment.
Calorimeter before 2016.3; temperature 361.4 by No. 6163.
after 3047.0; " 244.5 and 288.7.
Air, 21 C.; jacket about 27.
361.4 + .2 = 361.6 = 33.803, or 33.863 when' corrected for stem.
244.5 -j- .2 = 244.7 = 20.865 ; no correction for stem.
288.7 .1 = 288.6 = 20.846 ;
Mean, 20.855.
Mean specific heat between and 21 ^ ^^gg
Mean specific heat between 21 and 34
June 14. Third Experiment.
Calorimeter before 1961.8; temperature 293.6 by No. 6166.
" ' after 3044.6; " 243.7 and 213.0.
Air and jacket, about 18 C.
393.6 .1 = 393.5 = 29.036, or 29.077 when corrected for stem.
243.7 .1 = 243.6 = 17.349 ; no correction for stem.
213.0 4- .2 = 213.2 = 17.374 ;
Mean, 17.361.
Mean specific heat between and 17 _ QQOI
Mean specific heat between 17 and 29 ~
It is to be observed that thermometer No. 6166 in all cases gave
temperatures about 0.02 or 0.03 below No. 6163. This difference
is undoubtedly in the determination of the zero points, as on June 15
the zero points were found to be 20.4 and 58.0. As one has gone up
and the other down, the mean of the temperatures needs no cor-
rection.
OP ARTS AND SCIENCES. 127
i
June 15.
Calorimeter before 2068.2 ; temperature 364.6 by No. 6166.
after 2929.2 ; " 249.7 and 217.7.
Air and jacket at about 22 C.
264.6 == 26.766, or 26.782 when Corrected for stem.
249.7 = 17.822, or 17.812 "
217.7 + .1 =217.8 = 17.884, or 17.874 when corrected for stem.
Rejected on account of great difference in final temperatures by the
two thermometers, which was probably due to some error in reading.
June 21.
Calorimeter before 2002.7 ; temperature 330.3 by No. 6163.
" after 3075.2; " 221.9 and 256.6.
.-
Air and jacket, 21 C.
330.3 -f- .1 = 330.4 = 30.321, or 30.359 when corrected for stem.
221.9 + - 1 == 222 - = 18.349, or 18.343
256.6 -j- .0 = 256.6 = 18.358, or 18.352 " "
Mean, 18.347.
Specific heat between and 18
Specific heat between 18 and 30
June 21.
Calorimeter before 2073.8 ; temperature 347.8 by No. 6166.
" after 2986.8 ; " 234.5 and 206.6.
Air and jacket, about 21 C.
347.8 -|-.0 = 347.8 = 25.457, or 25.471 when corrected for stem.
234.5 -f -0 = 234.5 = 16.643, or 16.636 " "
206.6 -j-.l = 206.7 = 16.651, or 16.644
Mean, 16.640.
Specific heat between and 17 99971
Specific heat between 17. and 25
Rejected because dew was formed on the calorimeter.
A series was now tried with both thermometers in the calorimeter
from the beginning.
128 ' PROCEEDINGS OF THE AMERICAN ACADEMY
June 25.
Calor. before 2220.3 ; temperat. 325.6 by No. 6166 ; 309.9 by No. 6165.
after 3031.4; 233.4 224.6
Air, 24.2 C. ; jacket, 23.5.
325 6 4- .0 = 325.6'= 23.725, or 23.72'6 when corrected for stem.
309'.9 + '.2 = 310.1 = 23.739, or 23.740
233.4 -f -0 = 233.4 = 16.558, or 16.545
224.6 + .2 = 224.8 = 16.562, or 16.549
Means, 28.733 and 16.547.
Specific heat between and 16 _ 1 QQ ^
Specific heat between 16 and 24
June 25.
Calor. before 2278.6; temperat. 340.35 by No. 6166; 324.1 by No. 6165.
after 3130.2; " 242.5 " 232.8 " "
Air, 23.5 C. ; jacket, 22.5.
340.35 + .0 = 340.35 = 24 .877, or 24.881 when corrected for stem.
324.1 +.2=324.3 = 24.899, or 24.903 " "
242.5 -f .0 = 242.5 = 17.264, or 17.253 " "
232.8 +.2 = 233.0 = 17.261, or 17.250 " "
Specific heat between and 17 ^7
Specific heat between 17 and 25
, June 25.
Calor. before 2316.35 ; temperat. 386.1 by No. 6166 ; 368.4 by No. 61 65.
" after 2966.90; " 295.4 " " 281.7 " "
Air, 23.5 C.; -jacket, 22.5.
386.1 -f .0 = 386.1 = 28.455, or 28.465 when corrected for stem.
368.4 -f- .2 = 368.6 = 28.472, or 28.482 "
295.4 -f -0 = 295.4 = 21.374, or 21.368
281.7 -j- '-2 = 281.9 = 21.400, or 21.394 " "
Means, 28.473 and 21.381.
Specific heat between and 21
Specific heat between 21 and 28 *
Two experiments were made on June 23 with warm water in
vessel A, readings being taken of the temperature of the water, as it
OF ARTS AND SCIENCES. 129
flowed out, by one thermojneter, which was then transferred to the
calorimeter as before.
June 23.
Water in A while running, 314.15 by No. 6163'.
Calor. before 1530.9 ; temperat. 281.1 by No. 6166.
" after 2996.3; % " 328.4byNo.6166; 272.7byNo.6163.
314.15 + .1 = 314.25 = 28.526, or 28.552 when corrected for stem.
281.1 -{-.0 = 281.1 = 20.262, or 20.258 " " "
328.4 4- -0 = 328.4 = 23.945, or 23.950
272.7 -f- 1 = 272.8 = 23.960, or 23.966
Specific heat between 20 and 24 _
Specific heat between 24 and 29 ~
June 23.
Water in A while running, 383.9 by No. 6163.
Calor. before 1624.9 ; temperat. 286.75 by 6166.
after 3048.2 ; 392.45 by 6166, and 318.1 by 6163.
383.9 -f- .1 = 384.0 = 36.303, or 36.357 when corrected for stem.
286.75 -j- .0 = 286.75 = 20.702, or 20.700
392.45 -f -0 = 392.45 = 28.954, or 28.980
318.1 -f .1 = 318.2 = 28.9.64, or 28.992
Specific heat between 21 and 29
Specific heat between 29 and 36
To test the apparatus, and also to check the estimated specific heat
of the calorimeter, the water was almost entirely poured out of the
calorimeter, and warm water placed in the vessel A, which was then
allowed to flow into the calorimeter.
Water in A while running, 309.0 by No. 6163.
Calor. before 391.3; temperat. 314.5 by 6166.
" after 3129.0; " 308.3 by 6166, and 378.5 by 6163.
Air about 21 C.
Therefore, water lost 0.078, and calorimeter gained 5. Hence
the capacity of the calorimeter is 39.
Another experiment, more carefully made, in which the range was
greater, gave 35.
VOL. xv. (N. s. vn.) 9
130 PROCEEDINGS OP THE AMERICAN ACADEMY
The close agreement of these with the estimated amount is, of
course, only accidental, for they depend upon an estimation of only
0.08 and 0.12 respectively. But they at least show that the
water is delivered into the calorimeter without much change of
temperature.
A few experiments were made as follows between ordinary tempera-
tures and 100, seeing that this has already been determined by Reg-
nault.
Two thermometers were placed in the calorimeter, the temperature
of which was about 5 below that of the atmosphere. The vessel B
was then filled, and the water let into the calorimeter, by which the
temperature was nearly brought to that of the atmosphere ; the opera-
tion was then immediately repeated, by which the temperature rose
about 5 above the atmosphere. The temperature of the boiling
water was given by a thermometer whose 100 was taken several
times.
As only the rise of temperature is needed, the zero points of the
thermometers in the calorimeter are unnecessary, except to know that
they are within 0.02 of correct.
Jujie 18.
Temperature of boiling water, 99.9.
Calor. before 2684.7; temperat. 259.2 by 6166, and 248.3 by 6165.
" after 2993.2; " 381.0 363.4 "
259.3 = 18.568, or 18.555 when corrected for stem.
248.3 = 18.564, or 18.551
381.0 = 28.054, or 28.065
363.4 = 28.055, or 28.066
Specific heat 28 100
"Specific heal 18 ~^28" =
Other experiments gave 1.0015 and 1.0060, the mean of all of
which is 1.0033. Regnault's formula gives 1.005 ; but going directly
to his experiments, we get about 1.004, the other quantity being for
The agreement is very satisfactory, though one would expect my
small apparatus to lose more of the heat of the boiling water than
Regnault's. Indeed, for high temperatures my apparatus is much
r to Regnault's, and so I have not attempted any further
experiments at high temperatures.
OP ARTS AND SCIENCES. 131
My only object was to confirm by this method the results deduced
from the experiments on the mechanical equivalent ; and this I have
done, for the experiments nearly all show that the specific heat of
water decreases to about 30, after which it increases. But the
mechanical equivalent experiments give by far the most accurate
solution of the problem ; and, indeed, give it with an accuracy hitherto
unattempted in experiments of this nature.
But whether water increases or decreases in specific heat from to
30 depends upon the determination of the reduction to the air ther-
mometer. According to the mercurial thermometers Nos. 6163, 6165,
and 6166, treating them only as mercurial thermometers, the specific heat
of water up to 30 is nearly constant, but by the air thermometer, or by
the Kew standard or Fastre, it decreases.
Full and complete tables of comparison are published, and from
them any one can satisfy himself of the facts in the case.
I am myself satisfied that I have obtained a very near approxi-
mation to absolute temperatures, and accept them as the standard.
And by this standard the specific heat of water undoubtedly decreases
from to about 30.
To show that I have not arrived at this result rashly, I may men-
tion that I fought against a conclusion so much at variance with my
preconceived notions, but was forced at last to accept it, after studying
it for more than a year, and making frequent comparisons of ther-
mometers, and examinations of all other sources of error.
However remarkable this fact may be, being the first instance of
the decrease of the specific heat with rise of temperature, it is no
more remarkable than the contraction of water to 4. Indeed, in
both cases the water hardly seems to have recovered from freezing.
The specific heat of melting ice is infinite. Why is it necessary that
the specific heat should instantly fall, and then recover as the tempera-
ture rises ? Is it not more natural to suppose that it continues to fall
even after the ice is melted, and then to rise again as the specific heat
approaches infinity at the boiling point? And of all the bodies which
we should select as probably exhibiting this property, water is cer-
tainly the first.
(J.) Heat Capacity of Calorimeter.
During the construction of the calorimeter, pieces of all the material
were saved in order to obtain the specific heat. The calorimeter which
Joule used was put together with screws, and with little or no solder.
But in my calorimeter it was necessary to use solder, as it was of a
132 PROCEEDINGS OF THE AMERICAN ACADEMY
much more complicated pattern. The total capacity of the solder
used was only about ^ of the total capacity including the water;
and if we should neglect the whole, and call it copper, the error would
be only about T sW Hence H was considered 8ufficient to wei S h the
solder before and after use, being careful to weigh the scraps. The
error in the weight of solder could not possibly have been as great as
ten per cent, which would affect the capacity only 1 part in 12,000.
To determine the nickel used in plating, the calorimeter was weighed
before and after plating; but it weighed less after than before, owing
to the polishing of the copper. But I estimated the amount from the
thickness of a loose portion of the plating. I thus found the approxi-
mate weight of nickel, but as it was so small, I counted it as copper.
The following are the constituents of the calorimeter :
Thick sheet copper .
25.1 per cent.
Thin .
45.7 "
Cast brass
17.9 "
Rolled or drawn brass
5.7 "
Solder . . ' .
4.0 "
Steel . .. ' .
1.6 "
100.0
Nickel . . . . .3
To determine the mean specific heat, the basket of a Regnault's
apparatus was filled with the scraps in the above proportion, allowing
the basket of brass gauze, which was very light, to count toward the
drawn brass. The specific heat was then determined between 20
and 100, and between about 10 and 40. Between 20 and 100
the ordinary steam apparatus was used, but between 10 and 40 a
special apparatus filled with water was used, the water being around
the tube containing the basket, in the same manner as the steam is in
the original apparatus. In the calorimeter a stirrer was used, so that
the basket and water should rapidly attain the same temperature.
The water was weighed before and after the experiment, to allow for
evaporation. A correction of about 1 part in 1,000 was made, on
account of the heat lost by the basket in passing from the apparatus
to the calorimeter, in the 100 series, but no correction was made in
the other series. The thermometers in the calorimeter were Nos. 6163
and 6166 in the different experiments.
The principal difficulty in the determination is in the correction for
radiation, and for the heat which still remains in the basket after some
OF ARTS AND SCIENCES. 133
time. After the basket has descended into the water, it commences
to give out heat to the water ; this, in turn, radiates heat ; and the
temperature we measure is dependent upon both these quantities.
Let T = temperature of the basket at the time t
" T' = " " " "
" T" = " " " " oo
" 6 " water " t
u Ql ,
0" = " C
0" = T"
We may then put approximately
T T" = (T' T") e~7,
where c is a constant. But
T' T" TIT
hence t
0' = (d 6') (1 fi-^T).
To find c we have
1 0" tf
where 0" can be estimated sufficiently accurately to find C' approxi-
mately.
These formulae apply when there is no radiation. "When radiation
takes place, we may write, therefore, when t is not too small,
where C is a coefficient of radiation, and t is a quantity which must be
subtracted from t, as the temperature of the calorimeter does not
rise instantaneously. To estimate < , T a being the temperature of the
air, we have, according to Newton's law of cooling,
C(t < ) = jr=jr f( d ~ Ta) dt nearly '
6" 6'
- D y>
where it is to be noted that _ , is nearly a constant for all values
of 0" T a according to Newton's law of cooling.
134 PROCEEDINGS OF THE AMERICAN ACADEMY
The temperature reaches a maximum nearly at the time
and if O m is the maximum temperature, we have the value of 6" as
follows :
6" = T" = 6 m + (t m + c * );
and this is the final temperature provided there was no loss of heat.
When the final temperature of the water is nearly equal to that
of the air, G will be small, but the time t m of reaching the maximum
will be great. If a is a constant, we can put G = a (0" T a ), and
O(t m _j- c t ) will be a minimum, when
That is, the temperature of the air must be lower than the tempera-
ture of the water, so that T a = 0" as nearly as possible ; but the for-
mula shows that this method makes the corrections greater than if we
make T a = 6', the reason being that the maximum temperature is
not reached until after an infinite time. It will in practice, however,
be found best to make the temperature of the water at the beginning
about that of the air. It is by far the best and easiest method to
make all the corrections graphically, and I have constructed the fol-
lowing graphical method from the formulae.
First make a series of measurements of the temperature of the
water of the calorimeter, before and after the basket is dipped, together
with the times. Then plot them on a piece of paper as in Fig. 5,
making the scale sufficiently large to insure accuracy. Five or ten
centimeters to a degree are sufficient.
n abed is the plot of the. temperature of the water of the calo-
rimeter, the time being indicated by the horizontal line. Continue
the line dc until it meets the line la. Draw a horizontal line
through the point /. At any point, b, of the curve, draw a tangent
and also a vertical line b g ; the distance e g will be nearly the value of
the constant c in the formulas. Lay off //equal to c, and draw the
line fh k through the point h, which indicates the temperature of the
atmosphere or of the vessel surrounding the calorimeter. Draw a
vertical line, j k, through the point k. From the point of maximum,
OF ARTS AND SCIENCES.
135
c, draw a line, j c, parallel to dm, and where it meets kj will be the
required point, and will give the value of d". Hence, the rise of tem-
perature, corrected for all errors, will be kj.
This method, of course, only applies to cases where the final tem-
perature of the calorimeter is^ greater than that of the air ; otherwise
there will be no maximum.
In practice, the line dm is not straight, but becomes more and more
nearly parallel to the base line. This is partly due to the constant
decrease of the difference of temperature between the calorimeter and
the air, but is too great for that to account for it. I have traced it to
the thin metal jacket surrounding the calorimeter, and I must con-
demn, in the strongest possible manner, all such arrangements of calo-
rimeters as have such a thin metal jacket around them. The jacket is
of an uncertain temperature, between that of the calorimeter and the
air. When the calorimeter changes in temperature, the jacket follows
it, but only after some time ; hence, the heat lost in radiation is uncer-
tain. The true method is to have a water jacket of constant tempera-
ture, and then the rate of decrease of temperature will be nearly
constant for a long time.
The following results have been obtained by Mr. Jacques, Fellow
of the University, though the first was obtained by myself. Correc-
tions' were, of course, made for the amount of thermometer stem in
the air.
Temperature. Mean Specific Heat.
24 to 100 .0915
26 " 100 .0915
25 " 100 .0896 ,-'
13 " 39 .0895
14 " 38 .0885
9 " 41 .0910
136 PROCEEDINGS OP THE AMERICAN ACADEMY
To reduce these to the mean temperature of to 40, I have used
the rate of increase foundry B^de for copper. They then become., .
for the mean from to 40,
.0897
.0897 .
.0878
.0893
.0883
.0906
Mean .0892 dc .00027
As the capacity of the calorimeter is about four per cent of that of
the total capacity, including the water, this probable error is about
J fo T of the total capacity, and may thus be considered as satisfactory.
I have also computed the mean specific heat as follows, from other
observers :
Copper between 20 and 100 nearly.
.0949 Dulong.
.0935 ) ^
>0952 }Regnault.
.0933 Be-de.
.0930 Kopp.
' .0940
This reduced to between and 40 by Bede's formula gives .0922.
Hence we have the following for the calorimeter: *
Per cent. Specific Heat between and 40 0.
Copper 91.4 .0922
Zinc .7 .0896
Tin 3.6 .0550
Lead 2.7 .0310
Steel 1.6 .1110
Mean .0895
The close agreement of this number with the experimental result
can only be accidental, as the reduction to the air thermometer would
decrease it somewhat, and so make it even lower than mine. How-
* The cast brass was composed of 28 parts of copper, 2 of tin, 1 of zinc, and
1 of lead. The rolled brass was assumed to have the same composition. The
solder was assumed to be made of equal parts of tin and lead.
OF ARTS AND SCIENCES. 137
ever, the difference could not at most amount to more than 0.5 per
cent, which is very satisfactory.
The total capacity of the calorimeter is reckoned as follows :
Weight of calorimeter 3.8712 kilogrammes.
" screws ' .0016 "
" part of suspending wires .0052 "
Total weight 3.8780 "
Capacity = 3.878 X -0892 = .3459 kilogrammes.
To this must be added the capacity of the thermometer bulb and sev-
eral inches of the stem, and of a tube used as a safety valve, and we
must subtract the capacity of a part of the shaft which was joined to
the shaft turning the paddles. Hence,
.3459
+ .0011 "
-f .0010
- .0010 Jtyr
Capacity = .3470
As this is only about four per cent of the total capacity, it is not
necessary to consider the variation of this quantity with the tempera-
ture through the range from to 40 which I have used.
IV. DETERMINATION OF EQUIVALENT.
(a.) Historical Remarks.
The history of the determination of the mechanical equivalent of
heat is that of thermodynamics, and as such it is impossible to give
it at length here.
I shall simply refer to the few experiments which a priori seem to
possess the greatest value, and which have been made rather for the
determination of the quantity than for the illustration of a method,
and shall criticise them to the best of my ability, to find, if possible,
the cause of the great discrepancies.
1. GEXERAL REVIEW OF METHODS.
Whenever heat and mechanical energy are converted the one into
the other, we are able by measuring the amounts of each to obtain
the ratio. Every equation of thermodynamics proper is an equation
138 PROCEEDINGS OF THE AMERICAN ACADEMY
between mechanical energy and heat, and so should be able to give us
the mechanical equivalent. Besides this, we are able to measure a
certain amount of electrical energy in both mechanical and heat units,
and thus to also get the ratio. Chemical energy can be measured in
heat units, and can also be made to produce an electric current of
known mechanical energy. Indeed, we may sum up as follows the
different kinds of energy whose conversion into one another may fur-
nish us with the mechanical equivalent of heat. And the problem in
general would be the ratio by which each kind of energy may be con-
verted into each of the others, or into mechanical or absolute units.
a. Mechanical energy.
b. Heat.
c. Electrical energy.
d. Magnetic "
e. Gravitation u
/. Radiant "
g. Chemical "
h. Capillary "
Of these different kinds of energy, only the first five can be meas-
ured other than by their conversion into other forms of energy,
although Sir William Thomson, by the introduction of such terms as
" cubic mile of sunlight," has made some progress in the case of radia-
tion. Hence for these five only can the ratio be known.
Mechanical energy is measured by the force multiplied by the dis-
tance through which the force acts, and also by the mass of a body
multiplied by half the square of its velocity. Heat is usually referred
to the quantity required to raise a certain amount of water so many
degrees, though hitherto the temperature of the water and the reduc-
tion to the air thermometer have been almost neglected.
The energy of electricity at rest is the quantity multiplied by half
the potential ; or of a current, it is the strength of current multiplied
by the electro-motive force, and by the time; or for all attractive
forces varying inversely as the square of the distance, Sir William
Thomson has given the expression
where It is the resultant force at any point in space, and the integral
is taken throughout space.
These last three kinds of energy are already measured in absolute
OP ARTS AND SCIENCES. 139
measure, and hence their ratios are accurately known. The only
ratio, then, that remains is that of heat to one of the others, and this
must be determined by experiment alone.
But although we cannot measure /, g, h in general, yet we can
often measure off equal amounts of energy of these kinds. Thus,
although we cannot predict what quantities of heat are produced when
two atoms of different substances unite, yet, when the same quantities
of the same substances unite to produce the same compound, we are
safe in assuming that the same quantity of chemical energy comes
into play.
According to these principles, I have divided the methods into direct
and indirect.
Direct methods are those where b is converted directly or indirectly
into a, c, d, or e, or vice versa.
Indirect methods are those where some kind of energy, as g, is con-
verted into 6, and also into a, c, d, or e.
In this classification I have made the arrangement with respect to
the kinds of energy which are measured, and not to the intermediate
steps. Thus Joule's method with the magneto-electric machine would
be classed as mechanical energy into heat, although it is first con-
verted into electrical energy. The table does not pretend to be com-
plete, but gives, as it were, a bird's-eye view of the subject. It could
be extended by including more complicated transformations ; and, in-
deed, the symmetrical form in which it is placed suggests many other
transformations. As it stands, however, it includes all methods so far
used, besides many more.
In the table of indirect methods, the kind of energy mentioned first
is to be eliminated from the result by measuring it both in terms of
heat and one of the other kinds of energy, whose value is known in
absolute or mechanical units.
It is to be noted that, although it is theoretically possible to measure
magnetic energy in absolute units, yet it cannot be done practically
with any great accuracy, and is thus useless in the determination of
the equivalent. It could be thus left out from the direct methods
without harm, as also out of the next to last term in the indirect
methods.
140
PROCEEDINGS OF THE AMERICAN ACADEMY
TABLE XXV. SYNOPSIS OP METHODS FOB OBTAINING THE MECHANICAL
' EQUIVALENT OF HEAT.
re. Expansion or compression ac-
cording to adiabatic curve.
b. Expansion or compression ac
cording to isothermal curve.
1. Reversible Process
c. Expansion or compression ac-
cording to any curve with re-
generator.
f-Heat| M -So a n E " rgy -
d. Electro-magnetic engine driven
by thermo-electric pile in a cir-
cuit of no resistance.
( n. Friction, percussion, etc.
*
2. Irreversible Process \ b. Heat from magneto-electric cur-
( rents, or electric machine.
f
a Thermo-electric currents
{1. Reversible Process
b, Pyro-electric phenomena (prob-
2
ably).
a. Heating of wire by current, or
<4
2. Irreversible Process
heat produced by discharge of
electric battery.
I y. Heat, Magnetic Energy
1. Reversible Process
o. Thermo-electric current magnet-
izing a magnet in a circuit of
{a. Mechanical Energy.
b. Electrical "
M *
c. Magnetic
d. Gravitation
Crooke's radiometer.
Thermo-electric pile.
Thermo-electric pile with electro-
magnet in circuit.
a Mechanical Energy
( 1. Cannon.
< 2. Electro-magnetic machine run by
( galv. battery..
. Chemical Energy, Heat . .
(Combustion, etc.)
b. Electrical "
c. Magnetic
Current from battery.
( Electro-magnet, magnetized by a
I battery current.
d. Gravitation "
a. Mechanical Energy.
Movement of liquid by capillarity.
y. Capillary energy, Heat . . .
6. Electrical
{Electrical currents from capillary
action at surface of mercury.
(Heat produced when a liq-
c Magnetic "
uid is absorbed by a po-
it. Gravitation "
Raising of liquid by capillarity.
rous solid.)
(a. Mechanical Energy
{Magneto-electric or electro-magnetic
machine. Electric attraction.
S. Electrical energy, Heat . . (
(Heat generated in a wire by
6. Magnetic "
c. Gravitation "
Electro-magnet.
an electrical current.)
ia. Mechanical Energy
{Armature attracted by a permanent
magnet.
t
Magnetic Energy, Heat . .
(Heat generated on demag-
6. Electrical "
{Induced current on demagnetizing
a magnet.
netizing a magnet.)
c. Gravitation "
(. Gravitation Energy, Heat .
(Heat generated by a fall-
a. Mechanical Energy.
6. Electrical
c. Magnetic "
Velocity imparted to a falling body.
ing.body.)
2. RESULTS OF BEST DETERMINATIONS.
On the basis of this table of methods I have arranged the following
table, showing the principal results so far obtained.
In giving the indirect results, many persons have only measured
one of the transformations required ; and as it would lengthen out the
OP ARTS AND SCIENCES.
141
TABLE XXVL-H,.*,.,
OF Exr ERIMENTAL RE8ULm
Me
0<
thod in
neral. Method in Particular.
.
Observe
Date
Result.
Compression of air
Expansion " . ! .
Joule 2
184
443.8
Theory of gases (see below) .
Joule 2
184
437.8
' vapors (see below) .
Experiments on steam-engine
Him 7
185
413.0
Him 7
186
420-432
Expansion and contraction of meta
Edlund 8
1865
443.6
4301
428.3
Boring of cannon ....
Friction of water in tubea .
in calorimeter
in calorimeter
. . in calorimeter
Junction of mercury in calonmete
plates of iron . . .
Rumforc
Joule 3
Joule*
Joule 5
Joule 6
Joule 6
Joule 6
1798
1843
1845
1847
1850
1850
1850
940ft.lbs.
424.6
488.3
428.9
423.9
424.7
4252
metals
metals in mercury calo
Him 7
Favre 9
1857
1858
371.6
413.2
metals ....
Joring of metals
Vater in balance a frottement . '.
Flow of liquids under strong pressur
Crushing of lead .
Him 7
Him 7
Him?
Him 7
TT- _T
1858
1858
1860
1860-
400-450
425.0
4320
432.0
Friction of metals .
-TUttl'
1860-
425.0
Water in calorimeter ....
Joule
1876
1878
426.6
423.9
a
Heating by magneto electric cur- )
rents . .' t
Joule 8
1843
460.0
Heat generated in a disc between )
the poles of a magnet . . . j
vww.
1870
435~.2
434.9
435.8
437.4
A ft
Heat developed in wire of known (
absolute resistance . . . . "
Quintus
1857
399.7
(
so Weber
D . do. do. J
.