. I
I I
0.l
0
0.l
rhase Lquilibria in
Hydrocarbon Systems XX. Isobaric Heat Capacity of for the heat capacity of the first seven members of the paraffin series. The absolute values of these data for the higher members of the series are based in part uGon the single value chosen for the specific heat of gaseous n-butane. Since a n accuracy of only 2 per cent was claimed for this value, and since there was some discrepancy between the heat capacity calculated from the normal frequencies as determined by Kassel and the values given by Beeck, it was considered desirable to redetermine these isobaric heat capacities. The experimental work reported in this paper permitted the independent calculation of the isobaric heat capacity a t infinite dilution (i. e., zero pressure) of the paraffin hydrocarbons from propane through n-pentane a t 90" to 320 " F. The normal frequencies for methane are sufficiently well established to make measurements upon this material unnecessary. However, a few values were taken for methane a t the lower temperatures and have been included for comparison.
Gaseous Propane, n-Butane, Isobutane, and n-Pentane' B. H. SAGE, D. C. WEBSTER, AND W. N. LACEY California Institute of Technology, Pasadena, Calif.
Isobaric heat capacities at atmospheric pressure were determined for gaseous propane, n-butane, isobutane, and npentane at temperatures from 90" to 320" F. Several diagrams showing the agreement of these measurements with the data of other investigators are included. t
Method A modification of Eucken's adiabatic method (4) was employed. In a preliminary form this was used in a n investigation by the authors (15). The method is based upon the following general thermodynamic relation:
I
N RECENT years there has been an increasing industrial
interest in thermodynamic data for the paraffin hydrocarbons. The isobaric heat capacity a t atmospheric pressure is important in this regard. Unfortunately there is a scarcity of experimental work aimed toward the direct evaluation of such quantities. Eucken and Parts (6) measured the isobaric heat capacity a t atmospheric pressure of methane and ethane from 65" to 400" F. Lewis and McAdams (11) measured the isobaric specific heat of several gaseous hydrocarbons in a flow calorimeter which was not described. At the present time the science of spectroscopy has progressed sufficiently to permit the accurate calculation of the normal frequencies of methane and possibly of ethane, but such calculations for the hydrocarbons of greater molecular weight appear to be uncertain. Vold (19) calculated the isobaric heat capacity of methane from spectroscopic measurements. Kassel (8) recently reported values of the normal frequencies of several paraffin hydrocarbons above ethane calculated from spectroscopic data. Beeck (1, 2 ) reported measurements upon the "accommodation coefficients" of numerous hydrocarbons. From these measurements and the normal frequencies of ethane, together with a value of the specific heat of n-butane a t approximately 176' F. determined by Sage and Lacey (16),Beeck derived values 1 Previous articles in the series appeared during 1934. 1936. 1936, and in June and October, 1937.
The isobaric heat capacity is dependent upon the absolute temperature, the isobaric thermal expansion, and the isentropic temperature-pressure coefficient. The isobaric thermal expansion was evaluated from a knowledge of the pressure-volume-temperature relations or their equivalent, and the last quantity was measured directly.
Apparatus Figure 1 is a diagram of the apparatus developed for this measurement : The gas was confined in glass vessel A , immersed in oil thermostat B. The pressure existing within the chamber was measured by mercury manometer C. The gas was allowed to expand to atmospheric pressure through quick-opening valve E, which was connected to A by two symmetrically located tubes so arranged as to reduce turbulence within the chamber during the expansion. The change in temperature was measured by means of resistance thermometer D. The gas was admitted to chamber A from storage bulb F , which was located in air thermostat H , maintained at 120' F. All of the connecting tubing and manometers were also located in this thermostat. This increased the accuracy of the pressure measurements by avoiding changes in the density of mercury 1309
1310
INDUSTRIAL AND ENGINEERING CHEMISTRY
in the manometer and also prevented condensation of the less volatile gases, such as n-pentane. This air thermostat was controlled by a mercury-in-glass regulator and showed a maximum deviation in temperature of approximate1 0.1' F. during the course of a series of measurements. VesseTA was shielded from the slight fluctuations of temperature in bath B by metal container G which entirely surrounded the submerged part of this vessel except near the surface of the oil. The temperature of oil thermostat B was controlled by a four-junction copperconstantan thermocouple connected to a high-sensitivity galvanometer through a otentiometer. The li ht beam from this galvanometer actuate$ a photoelectric switcf, which controlled a small portion of the heating current for the thermostat. Exploration with thermocouples indicated that, in some parts of the bath, regular tem erature fluctuations as great as 0.015' F. were encountered. gowever, within the convection shield the temperature fluctuations were less than 0.003' F. at 250' F. The temperature within the convection shield was measured by means of a copper-constantan thermocouple used in conjunction with a s ecial potentiometer with a sensitivity of 0.05 microvolt. It is felieved that the temperahre of the apparatus was known within 0.1' F. throughout the temperature range. The uncertainty in this measurement was primarily due to the lack of suitable primary temperature standards at the higher tem eratures. However, since only the absolute temperature is invoyved, such an uncertainty introduced an error of only about 1 part in 4000 in the final specific heat values.
VOL. 29, NO. 11
the open-circuit type, and current was flowing through the wire for only about 0.2 second for any one measurement. A galvanometer with a period of approximately 1.5 seconds was employed; a current of 2 X 10-9 ampere flowing for 0.2 second was sufficient t o give a visible deflection. Considerable effort was expanded to determine the optimum size of wire for the resistance thermometer. Small wire (0.2mil or less) gave too great a tem erature rise above that of the gas due to the passage of the sma5 bridge current through it for the time of the resistance measurement. On the other hand, wire as large as 3 mils did not follow the temperature of the gas close enough in the time intervals involved.
4 0.147
DEFLECTION
MILLIMETERS
FIGURE 2. REPRESENTATIVE EXPERIMENTAL READINGS
For the measurements reported, a current of a proximately ampere was used in the bridge circuit. f n the course of the preliminary investigation, the current through the resistance thermometer was increased to approximately 5 x 10-8 ampere without seriously affecting the results, although there was a considerable rise in the temperature of the wire over that of the gas. Because of thermal conductivity, etc., the elapse of tihe after the expansion before heat diffused into the gas in the vicinity of the resistance thermometer was short, and some care was required to obtain the resistance of the thermometer before this diffusion occurred. To avoid the use of photographic recording equjpment for this rapid measurement of resistance, the galvanometer used in the bridge circuit was employed only as a null instrument. For a given gas at one temperature, the same pressure change was em lo ed for a series of expansions. By this means the bridge coupd {e adjusted so that only small deflections of the galvanometer occurred immediately after the expansion of the gas. 1X
FIGURE 1. DIAGRAM OF APPARATUS
Manometer C was constructed of glass, approximately 0.75 inch in diameter, and was filled with carefully purified mercury. The initial pressure within the apparatus was determined by measuring the difference in height of the mercury surfaces with a cathetometer, with which the reading of difference could be reproduced to 5 X 1 0 - 4 inch. The differential pressures used varied between 1.5 and 3.3 inches. The uncertainty of measuring the change in pressure was considered to be less than 0.1 per cent. The barometric pressure was determined by a freshly calibrated mercury-in-glassbarometer located near the apparatus. In order to prevent undue heat conduction from the gas by the resistance thermometer leads, a small auxiliary mercury thermostat, J, was installed around the leads in the upper part of vessel A. The temperature of this thermostat was controlled by a differential thermocouple, operating between the oil bath and the mercury in the thermostat. It was found necessary t o maintain the mercury approximately 0.5' F. above the temperature of the bath in order to avoid small convection currents in the gas within chamber A. Resistance thermometer D, which was used to determine the change in temperature of the gas due t o isentropic expansion, was constructed of approximately 4 inches of 1.2,5-mil annealed platinum wire. This resistance thermometer was connected to a bridge in which all contact resistances were so shunted as to avoid any significant errors in measuring the small changes in resistance involved in these determinations. The bridge was of
Figure 2 gives the readings for a series of expansions with isobutane for identical changes in pressure. The bridge settings are plotted as ordinates and the deflections of the galvanometer as abscissas. Since the time involved for each measurement was less than the period of the galvanometer, these points do not represent the true equilibrium positions of that instrument. However, the point of zero deflection does correspond to zero current in the galvanometer circuit and is the true null position. Curves of this type were obtained for each of the experimental points reported. The uncertainty in determining the change in resistance for a given change in pressure was not much greater than 1 part in 1000. Because of the widely different rates of diffusion of the gases investigated, the time available for measurement varied from approximately 0.2 second for hydrogen to 1.25 seconds for pentane a t the lower temperatures.
Materials The methane used in this investigation was obtained from the Buttonwillow field in California. An analysis of this gas showed it to contain 99.7 volume per cent methane and 0.3 per cent carbon dioxide. This gas was passed over barium perchlorate and ascarite, a t 500 pounds per square inch pressure, to remove water and carbon dioxide. The samples of propane, isobutane, n-butane, and n-pentane were obtained from the Phillips Petroleum Company. Their special analysis showed that the propane and isobutane contained not more than 0.03 per cent impurities. The analysis of the n-butane sample was: n-butane, 99.7mole per cent; isobutane, 0.3 mole per cent. Their analysis indicated that the sample of n-pentane contained 99.3 mole per cent n-pentane and 0.7 mole per cent iso-
.
NOVEMBER, 1937
INDUSTRIAL AND ENGINEERING CHEMISTRY
pentane. The purity of the samples was further substantiated by the degree of constancy of their vapor pressure during isothermal condensation. Since nearly all of the impurities in the samples were isomers of the substance in question, their presence had a negligible effect upon the reported values of heat capacity. The nitrogen used in calibrating the instrument was obtained from the Linde Air Products Company and probably contained less than 0.1 per cent impurities. The hydrogen, which was also used in calibrating the instrument, was prepared by electrolysis of dilute sulfuric acid. The gas was passed over sodium hydroxide and barium perchlorate to remove acid and water vapor, respectively.
1311
0.68
0.5
c
0
0.5
Calibration
0.4
An instrument of this type had several limitations when used for the absolute evaluation of the change in temperature due to an isentropic change in 0.4 pressure. Since the wire had a n appreciable heat capacity, the gas in its vicinity did not change in temperature as much as an isolated sample. The time involved before thermal diffusion set in was short, and the wire probably did not quite follow 50 IO0 150 200 250 the change in gas temperature. At the ends of the TEMPERATURE wire, heat was conducted to the wire from the leads. FIGURE 3. ISOBARIC HEATCAPACITY AT INFINITE DILUTION FOR In any actual expansions minor frictional processes METHANE AND ETHANE were inevitable, and the actual expansion therefore digressed slightly from an isentropic path. Because ofthese limitations in the use of thk instrument, it appeared deform for the direct evaluation of the isobaric heat capacity sirable to avoid the direct calibration of the resistance therfrom finite changes in state. For this reason a modification mometer. Instead, the calibration was accomplished by measof this equation was used in the calculation. uring the change in resistance of the thermometer for known The following thermodynamic relation is valid for a system changes in pressure for gases of known heat capacity. Nitroa t equilibrium : gen and hydrogen were chosen for the purpose because of their ease of preparation in the pure state and because of the reliable spectroscopic heat capacity information available. The values of the isobaric heat capacity of these gases at infinite dilution The Joule-Thomson coefficient ,u and the isobaric heat capacity reported by Davis and Johnston (S), which were calculated are related to the isothermal enthalpy-pressure coefficient in from spectroscopic data, were chosen as reference values. the following way: The calibrations of the resistance thermometer by means of the expansion of hydrogen and of nitrogen agreed with one an= -Pep (3) other within 0.25 per cent. This agreement was considered satisfactory for gases of such different behavior, since the abThe specific volume is related to the existing pressure, temsolute specific value of the heat capacity and the pressureperature, specific gas constant b, and the compressibility facvolume-temperature relations of these gases were involved in tor of gas Z by t,he equation: the results. Since the molecular weight of nitrogen is someZbT what closer to the molecular weight of the gases investigated v=-(4) P and its thermal conductivity is lower, it was used as the reference gas in this investigation. The accuracy of the heat If the values of the isothermal enthalpy-pressure coefficient capacities reported are therefore directly dependent upon the from Equation 3 and the specific volume from Equation 4 values chosen for the isobaric heat capacity of nitrogen. are substituted in Equation 2, we obtain: These values a t infinite dilution were calculated from the spectroscopic values of Davis and Johnston (3) and are given (5) in B. t. u. per pound per O F.: OF.
(%)*
40'F. 0.2483
lOO'F. 0.2484
160°F. 0.2486
220'F. 0.2490
280'F. 0 2496
340'F. 0.2505
The reported heat capacities of the hydrocarbons may be changed if there is need to modify the spectroscopic values for the heat capacity of nitrogen calculated by Davis and Johnston.
Calculations The experimental measurements yielded information concerning the change in temperature resulting from a finite isentropic change in pressure. Equation 1 is not in a suitable
If Equation 5 is combined with Equation 1, the following expression is obtained for the isobaric heat capacity: c p =
ZbT
V
(6)
The change in temperature resulting from a finite isentropic expansion may be expressed by the following equation: (7)
1312
VOL. 29, NO. 11
INDUSTRIAL AND ENGINEERING CHEMISTRY
For such a process Equation 6 may be rewritten in the following integral form:
This equation can be solved rigorously for the isobaric heat capacity without simplifying assumptions, but the solution is tedious and there is no need for the rigorous evaluation of this integral for the accuracy required in this case. The specific volume may be related to the prevailing pressure, temperature, and value of ( b Z / b P ) , by the following equation :
If it is assumed that, for the changes of state involved in the isentropic expansion over a small pressure range, the derivatives (bp/bP),, (bCp/bP),, and (bzZ/bTbP),are constant, average values of p , Cp,and 2 may be used. Equations 8 and 9 may then be combined and integrated. Using the asterisk to indicate an average value, the following equation is obtained :
for hydrogen were based upon the work of Holborn ( B ) , and the values for the Joule-Thomson coefficient were taken from the International Critical Tables (7). The values of ( b Z / dT)?, which were needed in calculating the specific heat of methane, were based upon the work of Kvalnes and Caddy (IO). The necessary values for propane were obtained from published pressure-volume-temperature data (16) and JouleThomson measurements (14) by the authors. The information for n-butane was from the same source (9, 17). The necessary pressure-volume-temperature information for n-pentane was based upon the work of Young (do), and the JouleThomson coefficients were taken from published values by the authors (9). The data for isobutane were obtained from work by the authors which is to be published shortly. It is believed that the above auxiliary information is of sufficient accuracy so that no uncertainty greater than 0.2 per cent was introduced by its use. The heat capacities calculated from Equations 10 and 11 are for the average pressure and temperature existing during the isentropic expansion. Since this type of information is difficult to correlate, all of the experimentally measured values were converted to those for infinite dilution (i. e., zero pressure) by the graphical solution of the following integral expression ( I d ) : CP, = CP*
Equation 10 was used for calculating the heat capacities of all the gases with the exception of methane, for which no suitable Joule-Thomson information was available. In this case a similar type of derivation gave the following expression for the isobaric heat capacity in terms of the average values of the compressibility factor, Z*, the compressibility-temperature coefficient, (bZ/dT),*, the absolute temperature, and the change of temperature resulting from an isentropic change of pressure:
+ Jp*
+ CP
[IJ($)p
dP
(12)
This equation was also employed in converting the data of Davis and Johnston (3) to the average pressure of the expansion in connection with the calibration of the apparatus. Since Joule-Thomson data were not available for methane, the small correction to the heat capacity was determined by graphical evaluation of the following equation: Cp, =
Cp*
+ bT
P
P
Results Figure 3 presents the results of the experimental work for methane which have been converted to infinite dilution. The values of heat capacity are given in B. t. u. per pound per O F. The necessary values of (bZ/bP),for nitrogen were taken They are numerically equal (within small limits) to the dimenfrom the work of Smith and Taylor (18), and the values of sionless quantity, specific heat. The experimental measurethe Joule-Thomson coefficient were obtained from the work ments of Eucken and Parts (6) for methane and ethane have of Roebuck and Osterberg (IS). The values of (bZ/bP), been included, alone: with the mectroscopic values' reported by Beeck (1) a n d Vold (19). The experimental measurements of Eucken and Parts were converted to infinite dilution by Equation 13. The present measurements for methane, indicated by circled points, agree within the experimental uncertainty with the spectroscopic values. Eucken's values for methane agree a t the lower temperatures but are approximately 4 per cent higher than the spectroscopic values a t 250° F. His experimental measurements for ethane agree with the values reported by Beeck a t the lower temperatures but are approximately 2.5 per cent below themat 210"F. V a l u e s calculated from the equation proposed by Lewis and McAdams (11) TEMPERATURE OF. are Dresented. These results are considerably higher than the spectroscopic FIGURE 4. ISOBARIC HEATCAPACITY AT I N F I N I T E DILUTION FOR P R O P A N E AND:,nvalues for both methane and ethane. P ~ NANE T
NOVEMBER, 1937
INDUSTRIAL AND ENGINEERING CHEMISTRY
1313
a t the lower temperatures. The previous measurements for n-butane by the authors (16)are approximately 3 per cent lower than the present values. This discrepancy is primarily due to uncertainties in the pressure-volume-temperature relations of n-butane a t the time these earlier no 0 data were calculated. However, a discrepancy of approximately 1.5 per cent still remained between the old and new values even after recalculation of the old data. This is probably partly due to the larger wire used in the previous work. Consideration of Figures 4 and 5 shows that the maximum deviation from the probable value is approximately 0.5 per cent and that the average deviation is less than 0.2 per cent. These results were TEMPERATURE obtained while employing changes in HEATCAPACITY AT INFINITE DILUTION FOR ISOBUTANE AND pressure which varied from 1.5 to 3.3 FIGURE 5. ISOBARIC TL-BUTANE inches of mercury for a given material a t a single temperature. Such variations did not cause any appreciable change in the resulting values of The experimental results for propane and n-pentane are heat capacity. The individual measurements of the change given in Figure 4. The vaIues reported by Beeck (1) are of temperature and pressure were made with a n uncertainty of included as well as values calculated from the Lewis and approximately 0.1 per cent; but because of changes in the conMcAdams equation (11), which were corrected to infinite dition of the surface of the wire, variation in the frictional dilution. This latter equation gives values for propane apeffects involved, andother factors beyond control, it is believed proximately 1.25 per cent higher than the present measurethat the results are trustworthy only to approximately 0.7 per ments. On the other hand, this equation gives values apcent. Again, it must be emphasized that the absolute values proximately 3.5 per cent lower than the measured values for of the results are entirely dependent upon the values chosen n-pentane. The values of (bC,/bT), for propane and nfor the isobaric heat capacity of nitrogen reported above. pentane, as determined by the Lewis-McAdams equation, The isobaric heat capacity a t atmospheric pressure for proare nearly the same as the values from the present experipane, n-butane, isobutane, and n-pentane are recorded in ments. In general, the values of heat capacity reported by Table I for a series of temperatures. These values were obBeeck for propane and n-pentane are somewhat lower than those from the present work in this temperature range. Howtained from the curves of heat capacity a t infinite dilution by use of Equation 12. ever, his value for (bC,/bT), is sufficiently greater for both The isothermal variation with molecular weight of the isopropane and n-pentane so that curves for the two sets of baric molal heat capacity a t infinite dilution in B. t. u. per data intersect at temperatures above those shown in Figure 4. pound mole per O F. is presented in Figure 6. The dotted The earlier values for aroaane reDorted by the authors (16) are approximately 2.5 pei c e h lower'than the present information. Recalculation of these earlier data indicated that this diserepancy was due in part to uncert a i n t i es in the pressure-volume-temperature measurements a t low pressures, which were used a t that time. Another source of error in the older data was the relatively large (40 B & S gage) thermocouple employed, which caused R rather large change in the calibration of the instrument with molecular weight of the gas involved and thus introduced an additional uncertainty. The results for n-butane and isobutane are presented in Figure 5. This work indicated that the isobaric heat capacity a t infinite dilution for n-butane was approximately 1per cent higher than the corresponding values for isobutane. The Lewis and McAdams equation (11) shows satisfactory agreement with the information here presented, although it gives R somewhat smaller value (bC,/dT)=than that indicated by the experimental work. Reeck's values, which are MOLECULAR WEIGHT the same for both n-butane and isobutane, again agree with the present FIGURE 6. ISOTHERMAL VARIATION IN MOLAL HEATCAPACITY AT INFINITE DILUTION measurements a t 200" F. but are smaller WITH MOLECULAR WEIGHT FOR PARAFFIN HYDROCARBONS OF.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1314
portion of the curves has been extended to include the molal heat capacity of hydrogen (3). All of the data fall on smooth curves, with the exception of the spectroscopic values for ethane. This discrepancy may be due to errors in the heat capacity of the hydrocarbons of higher molecular weight, but rather large discrepancies would be required to bring these data into line with the ethane values. On the other hand, Eucken's experimental values as indicated in Figure 3 are somewhat lower than the results reported by Beeck.
gases of considerably greater molecular weight than n-pentane. However, it is not applicable to gases of lower molecular weight than propane, or possibly ethane, since the linear change in molal heat capacity with molecular weight breaks down in this region as is indicated in Figure 6.
Nomenclature = isobaric heat ppacity, B. t. u./lb./" F. T = temperature, R. (" F. abs.)
C p
P = pressure, lb./sq. in. abs. V = specific volume, cu. ft./lb.
HEATCAPACITY AT ATMOSPHERIC PRESSURE H TABLEI. ISOBARIC Temp,
Pro$iane
F. 70 100 130 160 190 220 260 280 310 340 a
0.4044 0.4162 0.4261 0.4373 0.4489 0.4608 0.4726 0.4843 0.4963 0.508
n-Butane Isobutane B. t. u. pm Zb. per ' F. 0.3888 0.3946 0,3988 0.4062 0.4166 0.4096 0.4212 0.4286 0.4332 0.4409 0.4458 0,4636 0.4663 0.4694 0.4734 0.4787 0.4881 0.503
.... ....
n-Pentane
....
0.4040 0.4138 0.4246 0.4360 0.4479 0.4604 0.4734 0.486
....
Extrapolated value.
The experimental information indicated that there was a linear isothermal variation in the molal heat capacity with molecular weight for the paraffin hydrocarbons from propane through n-pentane. Lewis and McAdarns proposed this type of relation but extended it to include methane and ethane. The results of the present experimental work may be described by the fouowing equation in which M is the molecular weight and t is the temperature in " F.: M C p = 4.9
+ [0.314+ 5.20 X 10-4t - 7.38X 10-9t2
- 4.49 x
10-1ot31(~ - 8)
(14)
Values calculated from Equation 14 agree with the experimental values with an average deviation of less than 0.2 per cent. This equation is known to apply only a t infinite dilution for the paraffin hydrocarbons including propane, the butanes, and n-pentane, for temperatures between 70" and 360" F. Spectroscopic considerations (8) indicate that there may be a linear relation between molal heat capacity and molecular weight for the hydrocarbons of greater molecular weight than those here investigated. If this is the case, Equation 14 may be applicable with reasonable accuracy to hydrocarbon
EXPERIMENTAL COTTON MILLIN
THE
VOL. 29, NO. 11
enthalpy Joule-Thomson coefficient, O F./lb./sq. in. b = specific gas constant 2 = compressibility factor (PV/bT) S = entropy p
= =
Acknowledgment This work was done as part of a general program of investigation being conducted by Research Project 37 of the American Petroleum Institute. The authors acknowledge the financial assistance and cooperation of the institute in making this work possible. The Los Angeles Gas and Electric Corporation kindly furnished the sample of Buttonwillow gas used.
Literature Cited Beeok, J. Chem. Phys., 4, 680 (1936). Ibid., 5, 268 (1937). Davis and Johnston, J . Am. Chem. Soc., 56, 1045 (1934). Euoken and Lude, von, 2.physik. Chem., B5, 413 (1929). Euoken and Parts, Ibid., B20, 184 (1933). Holborn, Ann. Physilc, 63, 674 (1920). International Critical Tables, New York, McGraw-Hill Book Go., 1929. Kassel, J . Chem. Phys., 4, 435 (1936). Kennedy, Sage, and Lacey, IND. ENQ.CHEM.,28,718 (1936). Kvalnes and Gaddy, J. Am. Chem. SOC.,53, 394 (1931). Lewis and McAdams, Chem. & M e t . Eng., 36, 336 (1929). Lewis and Randall, "Thermodynamics," p. 69, New York, MoGraw-Hill Book Co., 1923. Roebuck and Osterberg, P h y s . Rev., 48, 480 (1935). Sage, Kennedy, and Laoey, IND.ENQ.CHEM.,28, 601 (1936). Sage and Lacey, Ibid., 27, 1484 (1935). Sage, Schaafsma, and Laoey, Ibid., 26, 1218 (1934). Sage, Webster, and Laoey, Ibid., 29, 1188 (1937). Smith and Taylor, J . Am. Chem. Sac., 45, 2107 (1923). Vold, Ibid., 57, 1192 (1935). Young, Sci. Proc. Roy. D u b l i n SOC.,13, 310 (1912) RECDITED June 22, 1937.
TEXTILESECTION, NATIONALBUREAU OF STANDARDS