MEASUREMENT OF MERCURY VAPOR PRESSURE BY MEANS OF THE KNUDSEN PRESSURE GAUGE BY CHARLES FRANCIS HILL A. B., University of Illinois, 1914. A. M., University of Illinois, 1916. THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN PHYSICS IN THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS 1921 Digitized by the Internet Archive in 2015 https://archive.org/details/measurementofmerOOhill UNIVERSITY OF ILLINOIS THE GRADUATE SCHOOL April 30 j 19&1 I HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MY SUPERVISION BY . CHARLES FRA1ICIR HILL entitled__ jJ£a sii r il: e ;: t , q f^lzrc u ry v a;-cr,.?. r i:s. sury ey izats OF THE KiUDSE:: pre ssu re g auge BE ACCEPTED AS FULFILLING THIS PART OF THE REQUIREMENTS FOR Recommendation concurred in* Committee on Final Examination* *Required for doctor’s degree but not for master’s 4-?'^G£ 9 ■ . - • ' > - 1 TABLE OF CONTENTS I. INTRODUCTION II. HISTORICAL 1. Dalton 2. Avcgadrc 0. Regnault 4. Hertz 5. Van der Plaats a. Table I 8. Hagen 7. Ramsey and Young 8. Morley 9. Knuds en 10. Comparison of Previous Methods .... III. experimental 1. Description of Apparatus and Method 2. Discussion of Accuracy of Method . . 3. Comparison with Previous Methods. . . 4. Data - Tables II to IX 5. Curves IV. CONCLUSIONS 1 2 2 2 3 3 4 5 6 6 7 9 10 14 15 17 24 28 I. INTRODUCTION Within the last thirty years, a large amount of research ha3 been done in partial vacua requiring an accurate knowledge of the gas pressure. Mercury and oil pumps have been used in the production of these vacua, and the gas pressure has been measured by means of the McLeod gauge. Now the total gas pressure must necessarily in- clude the vapor pressures, and as is well known, the McLeod gauge does not measure vapor pressure. Liquid air or other cold agents are usually applied to remove vapors. Often several hours are re- quired to do this, due to the fact that large quantities of vapor are absorbed by the walls of the container. In the case of mercury and vacuum oils a free surface is usually exposed to the vacuum and hence only an equilibrium condition between evaporation and conden- sation can be reached. It has been found in the present experimental work that this equilibrium condition may be as much as one-third or even two-thirds of the saturation pressure, depending upon the amount of surface exposed. Experimenters, generally, have probably not realized the difficulty of removing all vapors. If the saturation pressures of these vapors are accurately known, especially for mercu- ry, it would be of value in vacuum experiments. Most of the pub- lished data on the vapor pressure of mercury at temperatures ranging from 0° to 50° is merely extrapolated from values at high tempera- tures. Only three direct methods have been used over the range 0° to 50°, and these show a lack of agreement, the variations being a large percent. The methods employed are also open to question due to evident sources of error. Because of the meagerness of the data, and the lack of agreement between the different observers, it was decided to attempt a redetermination of the vapor pressure of mercury . ' ■ . . ■ . . . . . ' . . . ■ * r 2 between the temperatures of 0° and 40° C. The use of the Knudsen pressure gauge was considered the most dependable method, since this instrument is independent cf the gas and measures pressures of the order of mercur* vapor. The results obtained with the instrument are given, following a brief review of the methods and results of previous observers. II. HISTORICAL That mercury has a vapor pressure was discovered about 1796 by investigators^ in Holland. Dalton also discovered the fact soon I after and even made a few measurements at high temperatures before 1802. It was shown that mercury could be evaporated at fairly low temperatures but at temperatures of about 20° C, many thought that the vapor pressure was zero. In some experiments about 1824, Fara- day placed some gold foil over mercury. The foil, in the course of a few hours disappeared, and in his explanation of the result, he attributed tne effect as due to some sort of an affinity between the gold and the mercury and not to the evaporation cf the mercury. This view that the vapor pressure was zero at lew temperatures, seems to have been rather general until about 1850. I _ 3 In 1833, using temperatures between 2b0° and 300°, Avagadro made measurements on the vapor pressure cf mercury in a barometer tube. Using Dalton’s equation, P = ab^, he extrapolated back to 0°; however, over such a long range the extrapolation would have but little value, and it is now known that Dalton's equation does not 1. J.R. Dieman, Van trost Wijk, Bondt. Chemische and Physische Oefanngen, 18S7. 2. Phil. Trans. , 1826. 3. Ann. de Pogg. , 1833, T27, p 60. . 1 . 3 hold except for the region in which its constants are determined. Regnault^, using practically the same method, made measure- ments in 1862 at temperatures above 100 c ; he had, however, consider- able trouble in getting readings at the lower temperatures. Regnault used the empirical equation cf Biot, Leg P = a + bC^ + cK* 1 , and extrapolated to 0°. His results over the range 0° to 40° are now known to be 10 to 100 times too large since values which he gives can be measured accurately by later methods. About 1880, due to the invention of the Toepler and Sprengel air pumps and also the McLeod gauge, the question cf the vapor press- ure of mercury became of greater importance. Rood claimed .000002mm. for the Sprengel pump, ana Hagen .00001 mm. for the Toepler pump. If mercury had an appreciable vapor pressure, this fact would have a de- cided influence upon the resultant pressures reached by the pumps. Several investigations were car- ried out immediately and published between the years cf 1880 and 188S. 5 Among the first was that cf Hertz , his measurements being considered as among the mere accurate ones made at higher temperatures. The apparatus used, consisted of a differential manometer as shown in Fig.l. The system was pumped out and closed. Then A was placed at the desired temperature, which caused a difference in level of the two arms of 4. Memoirs de I'nstitute, 1862, T 26, p 506. 5. Wied. Ann., 15, Vol.17, p 193. . : ■ . . . . . . . . . . . . 4 the manometer C. By letting in air until the two arms of C balanced, the vapor pressure of mercury could be read as the difference in the heights of the two arms of B. The temperature range was frcrr 89.4" to 206 p C. hertz used the theoretical equation, i — s— c -W T P = KT" irp e w by means of which an extrapolation was made to zero. While tne equation has some theo- retical foundation, the constants are all calculated by applying the equation to the data, and all of the quantities become experimentally determined. Hertz took readings below 89.4° but did net consider them since his error was of the order of magnitude of the quantities measured. Even at 117° his error was ¥jo from the mean, and errors at the lew points affect extrapolated values to a great extent, especially when the values sought are as small as those of mercury vapor, and tin extrapolation is over a long range. L In 1886, van der Plaats published results by taking readings in the temperature range of 0 ° to 20° C, a region in which no other data had been taken except that of Hagen. While the percent error cf nis measurements is comparatively large, the data must be given considerable weight since the method was direct. Values from his mean curve have been given preference in tables, probably due to the fact that the method was direct. Van der Plaats’ obtained his read- ings by passing dry gas through water, through sulphuric acid, and then through mercury until saturated, after which the mercury was collected by means of geld and pumice stone. The amount of water and mercury collected was weighed. Knowing the volume, temperature, and pressure of the gas, and the vapor pressure cf water he could calculate 6. Rec. Trans. Chim. 5, p 49, 1886. . ' . 5 vapor pressure of the mercury. Van der Plaats' readings show consid- erable variation, nevertheless they are consistent enough to indi- cate that the order must be correct. The readings are given below. Temperature 0° . 00042 4° . 00058 7 C . 00071 9° . 00073 10° . 00077 11° .00083 12° . 001C1 15° .001 14° .00101 16° .00115 18° .00133 20° . 00133 Pressure .000515 .00048 .00044 .00084 .000735 .00058 . 00093 It is readily seen that if two or three values are omitted the others fall within +15/o of the mean curve. His error could be accounted for by the small quantities of collected mercury that he had to measure. Otherwise, the method seems difficult to criticise; however, it should net give values too high. 7 In 1882, Hagen published results which he got by taking readings between 0° and 200°, in 50° steps. The apparatus is shown in Fig. 2. The system was pumped while heating in t order to rid the walls of vapors and gas- es, after which the mercury was admitted through the capillary at the bottom. After the pressure had come into equi- librium, the tubes at the points E and F were sealed off. The mercury then stood 7. Wied. Ann., 1882, Vol.lS, p 610, 6 at the same level at A and. B. By raising the whole apparatus to a temperature, T, the space above the mercury became saturated with vapor. If C was now placed in carbon dioxide snow and ether, Hagen thought that all of the vapcrs would be removed and the difference in pressure read between the heights of the two branches, A and B, would thus be tne vapor pressure of mercury. This difference was read by means of a cathetometer. What Hagen measured, of course, was an equilibrium condition between evaporation and condensation. His values should therefore be too low. However, his results, like those of Regnault, are 10 to 100 times too high. His data also in- creases about 20% from 0° to 20° and 40% from 90 c to 100°, while vapor pressures, as is well known, should have a decreasing percent increase as the temperature increases. Q Another important set of data, that of Ramsey and Young , was published in 1886. The relation of the absolute temperatures of wa- ter and mercury for which the two have the same vapor pressure, is well enough known sc tnat if a few values are determined the ones between may be calculated. They applied this principle, getting readings from 135° to 520°C. , and then extrapolated to zero, using Regnault 1 s equation. At the high temperatures, the results of Ramsey and Young and those of Hertz agree very closely. Ramsey and Young, however, used a much greater tenperature range, and thus theii data should give a more accurate extrapolation. o In 1904, E.W. Morley* used a method similar to that of Van der Plaats, except that the evaporated mercury was determined by weighing the whole mercury sample before and after ex T aporat ion, and 8. Journal Chem. See. , 1886, 49, p 37. 9. Phil. Mag. , 1904, p 662. 7 the difference taken. He made determinations at temperatures 18 r , 30° , 40°, 50°, 60°, and 70°, and calculated a mean curve by means of Dalton's formula, P = ab ^ . Morley's readings at 16°, 30°, and 40° are from 8 Jo to 20 fo below his mean curve, hence it appears that his extrapolated values can not be depended upon to be so very accurate. Again, the uniform slope of the curve is net in agreement with what one would expect for vapor pressure curves; however, the small amount of data and the variation of the lower points could easily account for this. The variation at the low temperatures is not surprising if one considers that he musl detect a less of 4.5 mg. of mercury from a large sample of mercury, even after a run of 13 days. The last data of importance published was that of Knudsen J ‘ J in 1909. Knudsen first developed an equation*' 1 ' for the flow of a 12 gas through a tube in a time t. Later, this equation was extended to the case of a tube with a plate over one end, the plate contain- ing a small opening so that the resistance to flow was mostly at the opening. His equation is as follows, _ - P'-P” G “ w x +w 2 where G is the mass of gas that passes through, is the density at the temperature considered, P' and P" are the pressures at the ends of the tube, W 1 and W are the resistances of the tube and opening, and t is the time. W* and W" are calculated from dimensions and are gotten from theoretical considerations. Apparatus, based upon the 10. Ann. der Physik, lb 09 , 29, p 179. • H rH Ann. der Physik, 1908-9, 28, p 75. 12. Ann. der Physik, 1S08-9, 28, p 999. . I I 8 above theory was then developed by which the vapor pressure of mercu- ry could be determined. This apparatus is shown in Fig. 3. A glass tube with mercury placed as shown is evacuated and then sealed off. The small opening is placed between the two compartments, A and B. A is placed in a temperature bath while B is kept at a low temperature by means of carbon dioxide snow and benzol. P% the pressure in B, is considered zero since the vapor pressure of mercury at this low temperature would be negligible. The mass of vapor that passes through in a time t is then measured. In the equation, the pressure P' is considered that at the opening and to be the vapor pressure of mercury, and can be calculated if , W* and W" are known. Readings from 0° to 154° C were taken. A criticism of this method should include a criticism of both the theory and the experimental method. With the weight of evidence of other methods indicating that his results are too low, the equa- tion should certainly be tested further before great weight can be given to uhe results. The method tends to give values too low* An appreciable time is required for mercury to evaporate, and since P* is continually being relieved, this would tend to give a value below that of the saturation pressure. Any residual gas in the tube would also hinder diffusion and thus lower the calculated value. All otner values, except the extrapolated values of Hertz, are far above those of Knudsen over the range from 0 C to 30 r C. . Xnudsen's curve has the greatest slope of any ether investigator. 9 and, although lowest at 0° is highest at 154°, his highest tempera- ture measured. in order to compare results, the following table is given. Pressure is expressed in miliirr.et ers of mercury. TABLE I T Regnault Hagen Hertz Rams ey Young Van der Plaats Morley Knudsen Hill 0° .03 .015 .00019 . 00047 .0004 . 000184 . 00035 10° .0268 .018 , 0005 . 0008 . 0008 .0005 . 000775 20° .0372 .021 . 0013 . 0013 .0015 .00118 .00182 30° .053 .026 .0029 .003 .00273 . 00407 40° .0767 . 0 33 .0063 .008 . 006 .006 .00787 50° . 112 .042 .013 .015 .011 .0126 . 0080C 60° .021 70° . 04 Regnault’s and Hagen’s data agree fairly well, but both are now known to be too high since values as large as they give could be readily measured by later methods. At higher temperatures of 100° and above, the results of Ramsey and Young, and those of Hertz differ by only a small percent, and either is probably within a small per- cent of the correct values at those high temperatures. Hertz dis- carded his readings below about 90° because his error was large com- pared to the magnitude measured. Errors in his lowest temperature readings would affect his extrapolated values to a large extent. Ramsey and Young used a wider temperature range but their lowest point was 135° C. , which gives them a longer extrapolation. The weight of the two sets of data would probably be about the same. Knudsen used a direct method over a range from 0° to 154° and obtained consistent results. His data also agree with the extra- polated values of Herts. However, his method would tend to give values too low. The methods of Morley and van der Plaats, would also be expected to give results too lew if in error, and yet they are about two or three times as high as Knudsen' s, and agree fairly I . 10 well with each other, and with the results of Ramsey and Young and with the data taken in the present experimental work. While van der Plaate and Mcrley can not claim a high percent of accuracy due to the error in observations, yet their method would not be expected to be in error by as much as Knudsen's data indicates. The conclusion from the above discussion is, that the weight of experimental evidence seems to indicate that the correct values for the vapor pressure of mercury are probably of the order of the data taken by Morley and van der Plaats. III. EXPERIMENTAL After noting the disagreement of previous observations, it is evident that a set of measurements by a direct and dependable method at ordinary working temperatures would be worth while. The Knudsen"^ pressure gauge, if accurately calibrated, will give consistent and accurate readings on pressures of the magnitude of mercury vapor and has been used as low as 10“® mm. The instrument may be used as an absolute manometer; however, the one used in this case was not ar- ranged to be used in this manner, -lienee it was necessary to cali- brate it by means of seme other gauge. The principle upon which it depends is that if a strip of platinum, say, is heated in a partial vacuum, the molecules of the residual has will become heated by con- tact with the strip and fly off. If a vane free to turn is placed before the heated strip, these molecules will strike and turn the vane, if the mean free path of the molecules is greater than the distance between the vane and foil. The pressure is proportional to the number of molecules present and thus the deflection gives a measurement of the pressure. Now the deflection at zero pressure 13. Ann. der Physik, 1910, 32, 4, pp 809-43; Phys.Rev. 12, pp 70- 30, 1918. 11 is zero, so if the pressures are read on another gauge and the deflections on the Knudsen gauge, curves may he drawn with the ori- gin as an accurately determined point. This fact made it possible to use a McLeod gauge for calibration, since it may be read fairly accurately to about .0005 mm. A special Pyrex McLeod gauge was made for the purpose. This was desirable since the rest of the apparatus was made of pyrex. The volume tube was made rather large so as tc do away with friction and surface tension of the mercury as much as possible. The gauge would still read consistently to .0005 mm. if the mercury and glass were kept clean. The apparatus was connected as shown in Fig. 4. The Knudsen gauge. A, and the sample container, B. were fixed rigidly in a tight box at 1,2,3, so that the calibration would remain constant and the temperature could be con- trolled. A tube led off to the liquid air trap, tc the McLeod gauge and to the pump as shown. The pump- ing system consisted of a Langmuir condensation pump in series with Gaede Rotary oil and mercury pumps. With all of the vapors removed and kept from the Knudsen gauge by means of liquid air on the trap, C, the pressure was made about .005 mm. and then by reducing the pressure in steps, reading the pressures on the McLeod gauge and the deflections on the Knudsen gauge, a set of calibration curves was drarra. for the currents .3, .4, .b and .6 amperes flowing through the platinum strip. In order tc gain 12 accuracy, the curves were drawn on 40 cm. co-ordinate paper. Prac- tically all of the points fell accurately on the curves, showing that the McLeod gauge was consistent at least. (See Fig. 5 and 6.) The sample of mercury was then introduced into the container and with liquid air on the trap and pumps running, the mercury was distilled out of the container to the walls of the tube and hack again by heating slowly. This process of distillation was carried out a number of times. Then with warm water on the sample, the whole tube was heated to a temperature of 350° to 300°C. for several hours, to drive the vapors and absorbed gases out of the walls of the tube. The apparatus was then sealed off at E. The vapor press- ure of mercury at any temperature may be gotten by allowing the sys- tem to reach a constant temperature for a time, taking the total pressure, and then, after driving in all of the vapor by heating, with liquid air on the sample container, measure the residual press- ure. The difference between the total and residual pressures is the vapor pressure of mercury. In this way readings were taken at a number of temperatures between 0° and 35° C. , and the tube was again opened and the same process of distillation and heating carried out and tiie tube sealed off again. This process was continued until minimum values for the vapor pressure of mercury were obtained on three successive sets of readings. It was thought that the mercury could be considered pure at this point. Another calibration of the gauge was now carried out and a second sample of mercury purified with nitric acid was introduced and the process of distillation and heating repeated. The readings on this sample checked within a small percent cf the three readings on the first sample. Of the four sets ' * ' ! t > ■ - 13 of data, one or two points differ from a mean curve by about 6^>, the rest are all within +3$, which is about the accuracy one would ex- pect from the method, as will be shown later. Using the above method, the vapor pressure of mercury was measured at nineteen points between the temperatures of -.7° and 34. 9° C. The temperature was measured by means of two tenth-degree thermometers, placed at different points in the box. To insure a uniform temperature, the air was circulated by a fan within the box but driven by a motor outside, the whole fan system being entirely disconnected from the box and the pier supporting it. This was necessary to prevent jarring of the Knuds en gauge. Bafflers, H, were introduced to direct the circulation of the air. The regulation of temperature was easily accomplished. For those above room temperature, two heating coils, D, were used, through which the current could be controlled. For those below room temperature, the window was opened and an electric fan made to blow air on the box, -.7° being reached in this way, the unusually warm winter preventing lower temperatures. An attempt was made to control the low temperatures by turning the nozzle of a carbon dioxide tank into the box but this lacked constancy. In general the temperatures were held very constant, probably to within +. 1° , or at most to ±. 2° . If the thermometer changed as much as .2 the change was detected on the Knuds en gauge; and it is doubtful if the apparatus would change as readily as the thermometers. The Knuds en gauge was read by means cf a lamp and scale at about one meter distance. In order to hold the zero point, or rather to hold tne vane and foil at constant distance, the scale was rigidly fixed to the floor. Whenever the vane became displaced, it could be • ■ • . . 14 brought back to its zero point by merely restoring the zero point of the scale. One set of calibration curves was used for three sets cf data, and between each set the calibration was checked, so that it was known to remain constant. Before comparing the present data with that of former observ- ers, it would probably be best to consider some of the sources of error in this research and their magnitudes. In the first place the method would be expected to give values too high, since all impuri- ties would tend to increase the vapor pressure in the tube. Hence, the first sample of mercury was distilled more than twenty times in a vacuum, probably 95 $ of the sample being lost. The second sample was first treated with nitric acid and then distilled 8 times with the pump running. The data on it checked within a small percent of the mean curve of the first sample. This indicates that the error due to other vapors was negligible. It was thought that the sample container, cooled to the temperature of liquid air, might absorb some cf the residual gas, but tnis point was tested before the sam- ple was introduced and no effect detected. Special care was taken to keep the mercury and the glass cf the McLeod gauge clean to prevent friction. It was found that upon taking several readings at the same pressure, the gauge could be read within +2^ except a t the low pressures and these seem to check well with the higher readings as shown by the curves. The accurate point at the origin helped to take care of such errors. The read- ings of the Knudsen gauge differ over a range cf two or three percent however, much of this is due to the fact that the readings were taken from curves and a range of about one percent could be obtained by drawing a new curve for the same points. Hence, the error of the ■ . • . . i r.ufi I :|'V . i . ■ ■ 15 Knudsen gauge was apparently within +2fo. The temperature readings apparently did net introduce an error larger than 2 jo, since the Knudsen gauge would detect small changes. The total error, therefor*, is within +6 $>. The fact that this is about the variation of the indi- vidual readings from the mean curve, indicates that the principle sources of error have been accounted for, and that the accuracy of the data is what one should expect from the method employed. The actual readings are given on the following pages, begin- ning with Table II. The two sets of calibration curves are shown in the form of curves in order to show hew the readings fit the curves. The values determined for the vapor pressure of mercury are shown in Fig. 7, a smooth curve is drawn through the points. The values for each 2°, beginning with 0°C and taken from the mean curve are given in Table IX. . In order to compare results with those of former methods. Table I and the mean curves of Fig. 8, will be referred to. The data for all methods is given in 10° intervals. In the first place, if the slope of the curves is considered, the percent increase should decrease as the temperature increases. The slope of the present data as measured in this way, is about a mean of the slope of other methods. Van der Plaats" curve has such a slight rise from 0° to 20° that if it is extrapolated above 20 ° it will not agree with other methods. This may be accounted for by the variations of his obser- vations. Morley's curve rises faster but the low values at 16°, 30° , and 40°, would make his curve too low in that region. The slope of the present data is greater than that of Morley’s and agrees with Ramsey and Young. The methods of Morley and van der Plaats should tend to give values too low, while the present method should give . . ■ . . . . . . , . ' «0 16 values too high if in error, and. yet the three agree fairly well and with that of Ramsey and Young. On the other hand, Knudsen's method would tend to give values too low, and it does net 3eerr. probable that the other three methods could be in error as much as indicated by Knudsen t s data. His results are approximately 5C$> lower over the interval 0° to 20° C. , . CTiCn^OO CD CJ1 tfc* Cn CD 01 iP* 03 CD Cl tt'' Ol CD CD ^ CD 17 TABLE II First set of calibration data Pressures read on the McLeod gauge, and deflections on Knudsen gauge. Shown graphically in Fig. 5 10.0 10.0 10.0 S. 8 9.7 R Deflect ion McLeod Pressure Mean 14.71 4.71 .005 18.04 8.04 .005 22.02 12.02 . 00495 26.4 16.4 . 004S8 .005 14.1 4.1 .004 16.98 6.98 .004 20.41 10.41 .004 24.15 14. 15 . C0395 .004 12.98 2.98 . 0026 15.2 5.2 . 00255 17.81 7. 81 . 00255 20.7 10.7 .0025 , 00255 11.96 3. 16 .0016 13.6 3. 8 .0016 15.4 5,6 .0016 17.48 7.68 .00155 . 0016 10.8 1.1 . 00075 11.6 1.9 . 00072 12.58 13.62 2.88 3.92 . 00067 .00071 ? ' TABLE III 18 Data for the determination of vapor pressure of mercury at vari- OUt temperatures as read by the Knuds en Gauge First Run • Total Pressures I R. R Deflect ion Pressure Mean Tempera- 0 Curve ture . 3 10.1 13.0 2.9 .0024 .4 15.1 5.0 .00337 23° ,5 17.45 7.35 .00235 . 6 20. 9.9 .0023 . 002355 Residual G as Pressure .3 9.82 10.05 .23 .00012 .4 10. 15 . 33 .00012 .5 10.3 .48 . 00011 .6 10.5 .68 .00011 .000115 Total Pressur es .3 10.25 14.5 4.25 .00422 .4 17.5 7.25 .00422 29.9° .5 20.9 10.65 .00416 . 6 24.5 14. 35 .00407 .00416 .3 10.45 14.05 3, 6 .00336 .4 16.65 6.2 .00328 26.8° .5 19.55 9.1 .00324 .6 22.65 12.2 .00314 .00323 . 3 10.4 13. 32 2.92 . 00243 .4 15.46 5.06 .0034 .5 17.85 7.45 .00237 22.8° . 6 20.4 10.0 . 00235 .00239 . 3 10.15 15.52 5.37 . 00694 34.9° ,4 19.22 9.07 . 00603 .5 23.4 13.35 .00575 . 6 28.0 17.85 .0056 . 00584 . 3 10.32 11. 7 1.38 .0009 .4 12. 65 2. 33 .00088 11.8° .5 13. 86 3.54 . 0009 .6 15.15 4.83 .0009 .000895 .3 10.68 11.7 1.02 .00062 .4 12.4 1.72 .00064 5.8° .5 13.25 3.57 .00062 ,6 14. 2 3.52 .00063 . 00063 .3 10.1 12.6 3.5 .00196 .4 14.5 4.4 .00196 20. 3° .5 16. 6 6. 5 .00194 . 6 18.9 8.8 . 00194 .00195 12 TABLE III (Continued) The apparatus was heated to 100° for a short time with liquid air on sample container to drive in mercury and the following residual pressures at 24. 3° C. taken I R o R Deflection Pr essur Curve . 3 10.35 10.4 .15 .00008 .4 10.5 .25 . 00002 .5 10.63 .37 . 000088 . 6 10. 75 .5 . 00008 M ean . 000085 Tempera- ture Applying Charles Law, the residual pressure at various temperatures was calculated 5° 00008 10° 0000813 15° 0000835 20 c ...... . . 0000858 25° ...... , .000085 50° 0000865 35° . . 000088 Oi cn go cn cn w 01 oi if* co 20 TABLE IV Second Run of Data. o R Deflect ion Pressure Mean Tempera Curve ture 5 13. 44 3.94 . 00245 14.55 5. 05 .0024 23° 16.9 7.4 .00234 19.6 10.1 . 00236 .00239 7 11.25 1.55 .00103 13.35 2.65 . 00103 13° 13.55 3.85 .00100 14.9 5.2 .00098 .00101 Heated to : 130° with liquid ai r cn container and the following residual pressures taken. 75 9.33 .08 .00005 33° 9.86 .11 .000043 S. 8 .15 . 000043 10.0 .35 .00004 . 000044 ... .. O- O! 03 CD CJi 03 CD Ul 03 03 CJ3 ^ 03 21 I R 0 9.8 10.0 10.0 9.7 10.2 10 . 15 TABLE V Third Run of Bata R Deflection 13.0 3.2 15 . 3 5.5 18.0 8.3 30.3 11.1 14 . 33 4.32 17.3 7.3 20.7 10.7 34.5 14 . 5 15.0 5.0 18.55 8.55 22,5 12.5 26.75 16.75 Pressure Mean Temp Curve tu .00277 .00274 25 ° .00276 .00273 .00275 . 00432 .00438 30 . 8 ° .00423 .00416 .00427 .0055 .00556 34 . 3 ° .0054 .00514 .00540 11.05 1 . 35 12.03 2.32 13.19 3 . 43 14.5 4.8 10.8 .6 11.17 .37 11.7 1.5 12.35 2.15 .00087 .00088 . 00089 ,00088 .00033 .00036 .00036 .00036 y precept ible .00002 .00003 .00003 11 . 6 ° .00088 -. 7 ° .00035 .00002 Residual Pressure deflection hardl 10.19 .04 10.33 .08 10.38 .13 H CD TABLE VI 22 Second Set of Calibration Data I R 0 R Deflection Pressure (McLeod) Mean Tempera- ture . 3 10. 12 13.97 3.85 .0037 .4 16. 8 6. 68 .00365 . 003675 .5 20.1 9.98 .6 23.65 13.53 . 3 10. 12 12.67 2.55 .002 .4 14.5 4.38 .002 .002 .5 16. 78 6 . 66 . 6 19.16 9.04 . 3 10.12 11.8 1.68 .00105 .4 13.08 2.S6 .0011 .5 14.5 4.38 ,0011 . 00108 .3 10. OS 11. 7 1.08 .00068 .4 11.97 1.88 .0006 . 00064 .5 12. 89 2. 8 . 6 13.9 3.81 TABLE VII Data on Second Sample I *0 R Deflection Pressure Mean Tempera- Curve tur e .3 10.5 14.0 3.5 .00318 .4 16.5 6.0 .00316 25. 6 C .5 19.6 9.1 .00320 . 6 22.6 12.1 .00316 .00316 .3 10.6 14. 7 4.1 . 00402 .4 17. 75 7.15 .00406 28.6° .5 20.9 10,3 .00388 . 6 24.4 13. 8 .00380 . 00394 .3 10. 85 12.56 1.71 .00112 .4 14.0 3. 15 .00118 11.4° .5 15.55 4.7 .0012 . 6 17. 2 6, 3b .00116 .00116 Residual Pressure . 3 10.4 1C. 8 .4 .00023 .4 11.1 .7 . 00024 .5 11.45 1.05 . 00024 . 6 11.8 1.4 . 00024 .00024 ,3 10.65 13.95 3. 3 .0029 .4 16.35 5.7 .00292 24° .5 IS. 0 8.35 .00281 . 5 22.1 11.45 .00284 .00286 J i TABLE VIII 23 Values for the Vapor Pressure of Mercury at Various Temperatures T emperature Pressure Temperature Pressure mm. mercury mm. mercury -.7° . . . . . 00033 24.0° . . . . . 00262 5.8° . . . . . 00055 25.0° . . . . .00273 11.4° ... . . 000S2 25.6° . . . . .00292 11.6° . . . . .00083 26.8° . . . . .003145 11.8° ... . .000815 28.6° . . . . .0037 13.0° . . . . .00097 29.9° . . . . .004075 20.3° ... . .00187 30.8° . . . . .00425 22.8° ... . . 00230 34.3° . . . . . 00538 23.0° ... . . 00224 34.9° . . . . .00575 23.0° . . . . . 00235 TABLE IX Values for Each 2 C as Taken From Mean Curve 0.0° ... . .000350 2.0° ... . .000412 4.0° . . . . .000487 6.0° ... . .000572 8.0° ... . .000662 10.0° ... . .000775 12.0° . . . . .000895 14.0° ... . .00105 16.0° . . . . .00126 18.0° ... . .00150 20.0° . . . . .00182 22.0° . . . . . 00214 24.0° . . . . . 00334 36.0° . . . . . 003 28.0° . . . . . 0035 30.0° ... . . 00407 32.0° . . . . . 00467 34.0° . . . . . 00535 36.0° . . . . . 00607 38.0° .... . 00693 40.0° ... . . 008 Fig. 5 Fig. S Fig.? 27 Fig. 8 38 IV CONCLUSIONS The following conclusions m ay be drawn: 1. A direct method for the measurement of mercury vapor pressure has been employed in which advantage was taken of the pres- ent day methods of obtaining a high vacuum, and its measurement by means of the Knuds en gauge. 3. Nineteen readings on the vapor pressure of mercury have been taken between the temperatures of -.7° and 34.9° Centigrade. These values lie within ±