Equilibrium Studies upon the Bucher Process. - Industrial

Publication Date: October 1919. ACS Legacy Archive. Cite this:Ind. Eng. Chem. 1919, 11, 10, 946-950. Note: In lieu of an abstract, this is the article...
0 downloads 0 Views 689KB Size
946

T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y

the hydrochloric acid absorbed was titrated with silver nitrate. One volume' of methyl chloride should give one volume of hydrochloric acid.

Vol.

II,

KO.I O

ACKNOWLEDGMENT

The authors wish t o thank Dr. G. B. Taylor, Mr. G. W. Jones, and Mr. W. L. Parker for their valuable suggestions and assistance in this work. BUREAUOP MINES D. c.

WASHINGTON,

EQUILIBRIUM STUDIES UPON THE BUCHER PROCESS B y J. B. FERGUSON AND P. D. V. MANNING Received April 19, 1919

Fro 3

On nine samples of the methyl chloride, the combustion data gave a value of from 86 per cent t o 93.3 per cent. The hotter t h e platinum spiral in the combustion pipette, t h e greater the percentage of methyl chloride obtained. The corresponding values obtained b y titrating t h e hydrochloric acid formed varied from 62.7 per cent t o 79.8 per cent. This discrepancy may be due t o part of t h e hydrochloric acid not being absorbed by the 5 per cent sodium hydroxide solution and later being absorbed by the potassium hydroxide pipette on the Orsat. This amount of hydrochloric acid would, therefore, not appear in t h e titration. This method, moreover, is not suitable where other combustible gases are mixed with the methyl chloride. CONCLUSIONS

The glacial acetic acid method is by far the most rapid method of determining methyl chloride in mixtures, and in mixtures with a methyl chloride content of 40 per cent or more, the results lie within the limits of experimental error. Below 40 per cent, the probable error of the determination is about 4 or 5 per cent. Water vapor and air decrease the absorption of methyl chloride by glacial acetic acid. The partial pressures method gives results on methyl chloride-natural gas mixtures comparing with the glacial acetic acid results on the same sample t o a probable difference of 0.6 per cent. The combustion method is not suitable on mixtures of methyl chloride with other combustible gases, and on pure methyl chloride the temperature required for complete combustion is too high for the platinum spiral t o last long. For control work on the chlorination process, where relative values only are required, the great rapidity and ease of manipulation of the glacial acetic acid method makes i t the most desirable.

The solution of the many-sided problem of nitrogen fixation required the carrying out of a number of researches, and a t the request of the Nitrate Division of the Ordnance Department, the Geophysical Laboratory of the Carnegie Institute undertook, with t h e cooperation of t h a t Division, several of these investigations, one of which was a study of the Bucher process. I n this work the Nitrate Division was represented by Mr. Manning. The so-called Bucher process* consists essentially in heating a mixture of sodium carbonate, carbon, and iron in an atmosphere of nitrogen. The r6le of the iron is generally assumed t o be purely catalytic, a n d according t o Bucher the equation for the reaction may be written 2Na2COs 4C NzZ2NaCN 3CO. (I) I n his original article, Bucher states t h a t a t moderate temperatures, such as gooo t o 950' C., the process may be operated using producer gas instead of pure nitrogen without loss in efficiency, provided t h a t t h e product be not allowed t o cool in the producer gas. The validity of this assertion seemed t o be questioned by several results obtained during the operation of semi-commercial plants a n d for this reason we were asked t o determine, if possible, the influence of carbon monoxide in the furnace gases upon the yield of sodium cyanide. The character of this influence may be seen from t h e following theoretical considerachns: Assuming, as Bucher has done, t h a t the reaction involves only t h e two salts, sodium carbonate and sodium cyanide, and knowing t h a t these salts when melted together form a homogeneous liquid,2 and also t h a t a t temperatures a t or above gjoo C. metallic iron is the stable phase3 in the presence of equilibrium mixtures of carbon monoxide, carbon dioxide, and carbon, then the phases present a t equilibrium a t these temperatures will be iron, carbon, a liquid phase and a gas phase. The components are iron, carbon, nitrogen, sodium, and oxygen, and hence from the application of t h e phase rule in its simplest form4 the degrees of freedom are three in number. This means t h a t if the temperature and pressure be fixed, variations in the composition of the gas phase must be accompanied by corresponding variations in the composition of the liquid phase and vice versa. Thus the yield of cyanide will be directly dependent upon the composition of the gas phase present during the operation of the process. T h e magnitude of the variations in the composition

+

1 2

8

+

+

THISJOURNAL, 9 (1917), 233. Unpublished results of Dr. E. Posnjak of this laboratory. Hilpert and Dreckmann, Bey., 48 (1915), 1281-6.

r P + F = C + 2 .

\oCt.,

1919

T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y

947

fl

FIG. I-CROSS SECTION OF FURNACE USEDFOR HEATING CARBONATE-CYANIDE-IRON-CARBON CHARQEIN CURRENT OF CO-COa-Nn MIXTURE

of one phase caused by given variations in the composition of the other phase is not indicated by the phase rule and Bucher evidently thought that a t temperatures as high as 950' C. the deleterious effect of t h e presence of considerable percentages of carbon monoxide in the furnace gases was inappreciable. But the complicated relations which exist a t equilibrium are in no wise indicated by Equation I . I n addition t o nitrogen and carbon monoxide, the vapor phase must contain carbon dioxide,' sodium or an oxide of sodium, and sodium cyanide,2 and possibly other substances as well. Similarly, in addition to sodium carbonate and sodium cyanide, t h e liquid phase must contain carbon dioxide, carbon monoxide, nitrogen, and some excess sodium;3 the exact form of combination of the sodium is unknown, but probably most of i t is present as oxide. The calculation of t h e equilibrium constant for the gaseous or for the liquid phase can only be made if we know the concentration of all of the reacting constituents of the phase in question. Thus if we assume that the reaction in the vapor phase is represented by the equation zNa zC Ns'zNaCN (2) the equilibrium constant will be defined by the equation

+

+

(3)

and no matter how small the values of C N ~ C Nor C N may ~ be, unless they are constant they cannot be neglected. I n the liquid phase the relations are much more complex and here also the concentration of the various constituents, regardless of their magnitude, cannot be neglected unless they are constant. The determination of the equilibrium constants is, therefore, probably unattainable experimentally. I n view of this situation our problem resolves itself into t h e determination of certain empirical relations which may be considered a measure of the actual equilibrium conditions, and we therefore set out t o determine the variations in the amount of the alkali in a charge which was converted a t equilibrium to cyanide, with known variatiqns in the composition of the gas stream passing over the charge. For this purpose boat experiments were thought best suited and the apparatus used is shown in Fig. I . 1 Carbon monoxide dissociates t o form carbon dioxide according to C. t h e equation 2C0,C ' Oz 2 Bucher and others have shown t h a t sodium cyanide is volatile at temperatures in the neighborhood of 1000° C. 8 Sodium carbonate is supposed t o disqociate-according to the equation E z C O a 3-)NazO COz, but the relations which ohtain at small COa pressures have not been thoroughly investigated.

+

+

The furnace was one especially designed for the production of uniform temperature in a central region 1 2 t o 15 cm. in length, and has been previously described elsewhere.' The temperatures were measured by means of a differential platinum-platinrhodium thermoelement with a suitable potentiometer set-up. The outer tube of silica was protected by an inner tube of copper or iron from the fluxing action of the furnace vapors.2 For the quant'tative studies the copper tube waq used. The boat was made of iron, I z cm. in length, divided transversely into two compartments, and was welded to a small, steel rod which served as a means of inserting or withdrawing it. The baffles on the rod and the porcelain plug in the other end of t h e furnace tube all served t o reduce the end heat-losses from the furnace. I n front of the boat in the metal tube was placed some pure charcoal which was intended t o reduce any carbon dioxide formed in the gas mixtures a t the lower temperatures and t o ensure t h a t the gas mixture passing over the charge contained carbon monoxide and carbon dioxide in equilibrium proportions. The charges were made up by mixing together sodium cyanide or sodium carbonate (Squibb's reagents) with carbon and iron, a t first in varying amounts and finally in the proportions 3 of carbonate (or cyanide) : I of carbon : I of iron, by weight. The carbon used was pure untreated gas-mask charcoal furnished us from the supply a t the American University Experiment Station of the Chemical Warfare Service. The iron was a hydrogen-reduced product obtained from Merck and was preheatec! at 800' C. in a vacuum for several hours t o remove hydrogen. The gas mixtures were prepared from commercial (Linde) nitrogen from which the oxygen had been removed, and pure carbon monoxide prepared by the sulfuric acid-formic acid method. Before such mixtures entered the furnace they were carefully dried with sulfuric acid and phosphorus pentox;de. T h e y were kept in gasometers over mercury. The procedure was as follows: ( I ) The furnace was brought up to temperature with the empty boat inserted and a slow stream of nitrogenflowing through. (2) The boat was pulled into the cold portion of the tube and when cooled, removed. (3) The boat was filled with the charge or charges as desired, and slowly inserted into the furnace, a gradual heating reducing the danger of the charge being carried out of the boat by the gases released a t the high temperatures. 1 2

J. E. Ferguson, Phys. Rea., 12 (1918). 81. E. Posnjak and H. E. Merwin, J. Wash. Acad. S c i , 9 (1919), 28.

948

I

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

(4)When the boat was in position, the nitrogen was shut off and the gas mixture allowed to flow slowly through the furnace for some hours (usually 3 to j hours) during which time the furnace was again brought up to and maintained a t the desired temperature. (j) The experiment having run for a sufficient time, the boat was pulled into the cold part of the tube and when cold, removed from the furnace. (6) The charges were taken from the boat, hottled, placed in a desiccator, and as soon as possible after removal examined microscopically. Later, chemical analyses of the charges were made, for which purpose aqueous extracts of the charges were used. These analyses consisted in most cases in determining the cyanide by the Lundelll method in which nickel ammonium sulfate and di-methyl-glyoxime are used and the total alkali by titration with standard sulfuric acid, using methyl orange as an indicator. Cyanates were sometimes tested for by boiling the solution with an excess of sulfuric acid and determining this excess with standard alkali, using methyl orange as an indicator. The difference between this latter titration and the direct titration with acid should represent approximately twice the amount of cyanate present.* In several experiments tests were made for free alkali. This was done by determining, in addition to the . cyanate and cyanide, the carbonate, and then getting by difference from the total alkali the alkali not present in the form of these salts. The carbonate was determined by precipitation with barium chloride with suitable precautions.

Experiments at 950' t o 1 0 0 0 C. ~ with concentrations of carbon monoxide as high as 80 per cent did not cause any change in t h e iron boat, thus confirming t h e o b servations of Hilpert and Dreckmann who found t h a t iron was stable in t h e presence of equilibrium mixtures of carbon monoxide and carbon dioxide at these temperatures. After several experiments i t was found necessary t o remove the protecting metal tube and clean it, as t h e sublimate which condensed in the colder portions of t h e tube interfered withthe movement of the boat. This sublimate, when gas mixtures containing carbon monoxide were used, was always mainly sodium carbonate, but when pure nitrogen was used was sodium cyanide. The amount of this sublimate appeared t o be greater when sodium carbonate was in the initial charge than when cyanide only was present, and this indicates that the alkali present in the vapor phase is only partly there in the form of cyanide. I n some experiments evidence of t h e presence of sodium vapor was obtained b u t the evidence is not conclusive, as a n equilibrium condition may not have existed in these experiments. These observations indicate t h a t appreciable amounts of cyanide and alkali are probably present in the vapor phase a t equilibrium. Posnjak and Merwin3 found t h a t when the process was carried out in a copper tube in the laboratory the product was essentially a mixture of carbonate and cyanide, but t h a t the sinters prepared commercially contained mainly an unknown compound. The former observation using pure nitrogen has been confirmed by us a t 950' and 1000' C. and also with gas Lundell and Rridgman, THISJ O U R N A L , 6 (1914), 554. Cyanates are decomposed by stiong acids like hydrochloric or sulfuric t o form ammonia and carbon dioxide. The small amount of cyanic acid which may form when sulfuric acid is used would not interfere appreciably with thiq test. 8 J . Wash A c a d . Sci., 9 (1919), 28. 1

2

Vol.

11,

No.

IO

mixtures which contained initially from j t o 80 per cent carbon monoxide. However, when we used an iron tube we were able t o prepare small amounts of the unknown compound mixed with cyanide. The only reasonable explanation of this, in view of t h e fact t h a t an iron boat was used in all experiments, seemed t o be t h a t t h e iron had a t low temperatures taken t h e oxygen from the incoming gases and so changed their composition t h a t the unknown compound could be formed. T o test this hypothesis one experiment at 950' C. using a copper tube and an initial gas containing but 1.9 per cent of carbon monoxide wa4 made. The initial mixture placed in the boat was a pure cyanide mixture and after four hours was found t o contain definite traces of t h e unknown compound. This compound would, therefore, appear t o form very slowly in t h e presence of low concentrations of carbon monoxide and carbon dioxide. The presence of cyanates in the liquid phase could not be proved, even when high concentrations of carbon monoxide were used, b y either the microscopic or chemical methods, but charges containing carbonate always contain also some free alkali. One charge was found t o contain as high as 5 per cent free alkali and the commercial sinters often contain even higher percentages. TABLE I-RESULTS OR COPPER TUBEEXPERIMENTS AT loooo c. B EXPT. No.

Initial Charge Carbonate Cyanide Carbonate Carbonate Carbonate Carbonate Cyanid e Carbonate

7

Carbonate Cyanide

Initial Gas Per cent co = 5 Nz= 95 co = 5 Ns = 95 CO = 7 . 2 Nz = 92.8 CO = 15.3 Nz = 84.7 co = 1 9 . 1 Nz 80.9 CO = 48 Nz = 52 co = 48 Nz = 52 co = 47

= S.1 CO = 8 1 . 5 Nz = 18.5 co = 8 1 . 5 Nz = 18.5

A Cyanide Titration cc. 8.2 9.9 9.2 10.8

Total Alkali Titration cc. 2 x 5.2 2 X 6.1

Conversion Cyanide

= 100

x

:cent $; 79 81

12

77

15.4

70

5.8

9

64

3.7

6.4

58

8.1

13.6

60

2.4

4.0

60

3.4

14.1

24

1.5

8.3

18

Nn

TABLE11-RESULTS OF COPPERTUBE EXPERIMENTS AT 946' C. B Cvanide ~ ~ A Total Conversion Initial Cyanide Alkali = 100 X EXPT. Initial Gas Titration Titration A/B No. Charge Per cent cc. Cc. Percent 1 Cyanide co = 4 . 5 8.8 11.8 75 Nz 95.5 2 Cvanide co = 9 . 5 6.4 9.8 65 NI = 90.5 6.6 59 3 Cyanide CO = 16.9 3.9 Nz = 83.1 5.6 52 Carbonate CO = 16.9 2.9 Nz = 83.1 CO = 48 3.5 12.3 28 4 Cyanide Nz = 52 7.7 27 Carbonate CO = 48 2.1 Nz = 52

The same liquid phase could be obtained with a given gas mixture, but different initial solids. Thus when the boat was filled with two mixtures, one in each section (a pure carbonate and a pure cyanide mixture) and t h e gas stream allowed t o flow over the charges for a sufficient time (about 5 hrs.), t h e charges appeared t o be identical both by microscopic and by chemical tests. Were it not for the volatile nature of some of t h e constituents, this result might be taken as absolute proof t h a t an equilibrium condition had been reached. It

OCt., 1919

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

“i;-I T

.p /o

c,,

3

946-C

:\

40

K 2



p f‘ 0

10 CV

W

40 S O 10 70 do SO /OO

0

/b

2;

sb





,

IO0

-

70

60

-- 5so 40 4

--. .-. ..\.\-‘. 60 &

40 80

;O

80

90

-do

20

/o

;OO

I n Fig. 2 the percentage conversions given in Tables I and I1 are plotted against the carbon monoxide cont e n t of the initial gas upon the assumption t h a t no complications occur a t very high or very low concentrations of carbon monoxide. This assumption must involve some error a t low concentrations, since under these conditions some of the unknown compound is formed, a n d may also not be justified a t the highconcentrations, although t h e error here is probably less t h a n a t low concentrations. The following interesting deductions relative t o commercial cyanide processes’ may be made from the location of these curves: ( I ) The effect of temperature upon the possible yield of cyanide (yield a t equilibrium) is relatively small a t small concentrations of carbon monoxide, but if producer gas containing say 30 per cent carbon monoxide were used, would be a n important factor. ( 2 ) A t IOOO’ C. producer gas can be used with a possible conversion of 63 per cent. A higher temperature would probably increase this conversion2 but even a t 1000’ the advantages gained through the use of producer gas might more t h a n offset the smaller conversion due t o the use of producer gas in place of pure nitrogen. ( 3 ) The presence of 1 5 per cent carbon monoxide 1 If a process involves components in addition to those considered b y us, this fact must be borne in mind in the application of our results t o such .cases, in which quite different yields might be obtained. 2 Bucher‘s ideas regarding the use of producer gas were probably based upon experiments in which the tempeiatures of the charge were not ac.curately known.

919

in the initial mixture causes a reduction of 30 per cent in the conversion a t 1000’ C. The presence of 60 per cent carbon monoxide reduces the conversion t o 50 per cent. Therefore, if the initial gas contains carbon monoxide in small amounts, efforts should be made t o keep this content as small as possible, but if the initial gas contains relatively large amounts of carbon monoxide, such as 2 5 or 30 per cent, slight variations in this content will have little effect upon the yield. The mechanism of the reaction was conceived by Bucher t o consist first in the reduction of the sodium carbonate by carbon t o form sodium and, second, the reaction of this sodium with carbon and nitrogen t o form cyanide. The first stage may well be the controlling one. Now the r81e played by the carbon in this stage may be merely t o control the carbon dioxide pressure and ensure t h a t it is kept well below the dissociation pressure of the carbonate. The equilibrium between carbon monoxide. carbon dioxide, and carbon is very dependent upon both the temperature and the total pressure, so if our speculations even approximate the t r u t h one would expect the yield of cyanide t o be less dependent upon the temperature when referred t o a carbon dioxide basis than when referred t o a carbon monoxide basis. I n Fig. 3 are curves showing the variation of t h e cyanide conversion with the carbon dioxide content of the gas phase. The results of Rhead and Wheeler1 were used t o calculate the carbon dioxide content from the carbon monoxide content, but their work did not extend t o the low pressures and the curves are, therefore, only approximately correct. They do bear out, t o a considerable extent, the speculations in which we have just indulged.

70

P€ECENT CO,

FIG. 3-RELATION CARBON

OF

GA5

BETWEEN

Pr?sS/#G

OVER

CHARGE

C Y A N I D E CONVERSION AND PERCENTAGE O F

DIOXIDEI N GAS PASSING OVER C H A R G E (GAS CONSISTING

EQUILIBRIUM MIXTUREOF CO WITH

A N D COS TOGETHER NITROGEN)

The microscopic examinations were made by Dr. H. E. Merwin of this laboratory, and thanks are due him for his generous assistance. SUMMARY

Experiments were made on t h e Bucher process using pure chemicals and mixtures of pure nitrogen and carbon monoxide in known proportions. Curves have been obtained showing ( I ) the relation between the carbon monoxide content of the furnace gases a n d the yield of cyanide, and ( 2 ) the relation between t h e carbon dioxide content of the furnace gases and the yield of cyanide, both a t two temperatures. T h e 1

J . Chem. SOC.,99 (1911), 1140.

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

950

curves indicate t h a t under certain conditions producer gas may be used in the process and t h a t the dissociation of sodium carbonate is probably one of the controlling chemical reactions. CARNEGIE INSTITUTION OF WASHINGTON GEOPHYSICAL LABORATORY WASHINGTON, D . C.

A STUDY OF THE OIL FROM SUMAC (RHUS GLABRA) By H. W. BRUBAKER Received April 24, 1919

Since the demand for fats has increased so greatly and their price has reached such a high level i t has become imperative t h a t we make use of all the available sources of this most important material. A great deal of the rocky wasteland of Kansas and other states is covered with the common sumac ( R h u s Glabva). It occurred t o the author t o make a chemical study of the oil from the sumac seed t o determine its fitness as a food or for industrial purposes and the amount available. The berries from which this oil was obtained were gathered a t Manhattan, Kansas, in February 1919. The husks were removed from the berries by rubbing gently in a mortar and sending the material through a small fanning mill. The clean, air-dried seeds were ground in a mill and the fat extracted with dry ether in a continuous extraction apparatus large enough t o hold 2 or 3 lbs. of the material. Two determinations gave an average of 11.71 per cent of oil in the ground seeds. Table I summarizes the results of the physical and chemical examination of the oil.

SAMPLESp. Gr. at 15' C. No. l . . . . . . . 0,92568 Z . . . . . . . 0.92587 3 ..... Av 0,92577

....... ......

...

SaponiAcetyl fication Iodine Value Value N o . 9.27 193.2 126.55 9 . 2 0 193.8 126.98 . . . . 190.8 9.235 192.6 l i k : ? 6

InsolSoluble uble Fatty Fatty Acids Acids Per Per cent cent 0.85 92.68 0.67 93.55 0 . 7 8 94.38 0 . 7 6 6 93.54

Table I1 gives the characteristics of the insoluble fatty acids. TABLE11 ,Melting Point Den. C. i7

Solidification Temperature Deg. C. 6

Index of Refraction 1.470

Iodine Value 121.8

The oil of sumac has a mild odor, pleasant taste, and a deep yellow color. It is quite viscid a t ordinary room temperature. Upon being cooled i t thickens gradually until a t -16' C. i t has the consistency of soft vaseline. T h e oil was not cooled t o its freezing point; G. B. Frankforter and A. W. Martin give the freezing point of the oil from Rhus Glabra gathered in C.l Minnesota as -24' TABLEI11 RISE IN TEMP. ON TREATMENT Percentage Increase in Weight WITH CONCD. H&Oi Oil in 7 Days of a Initial Temp. Highest Temp. Tested Thin Film of Oil Deg. C. Deg. C. 20 94 Linseed oil. 9.30 20 70 1.66 Sumac o i l . . 20 55 Cottonseed oil.. 0.65

.. .. .... .. .. .. .. .. .. ...

These authors also found a n iodine value of 87 which differs materially from t h a t found for the Kansas oil, 1

A m . J . Pharm., 76 (1904), 1 5 1 .

11,

KO.I O

126.76. The high iodine value would indicate t h a t the oil should have fairly good drying qualities. This conclusion is substantiated by the results of comparative tests shown in Table 111. A small amount of the oil mixed into a paste of t h e consistency of paint with sublimed white lead and spread on a plate of glass dried completely in three days. T h e oil saponifies readily, giving a sodium soap of semisolid consistency. It seems fair t o conclude from the above study t h a t sumac oil compares favorably in properties with other vegetable oils such as cottonseed oil and corn oil. It might readily find a use as an edible oil or in the soapmaking industry or as a semidrying oil in the paint industry, if it can be put on the market a t a reasonable cost. The amount which might be made available can only be estimated. The author believes a conservative estimate of the amount of sumac seed in the state of Kansas alone t o be 60,000,000 lbs. containing 6,000,ooo lbs. of oil. Whether sumac can be made a practical source of oil or not can be determined only by some manufacturer situated so as t o be able t o handle the extraction of the oil. Those companies which extract the coloring matter from sumac or extractors of other vegetable oils are probably best situated t o work out the problem. DEPARTMENT OR CHEMISTRY KANSASSTATEAGRICULTURAL COLLEGE MANHATTAN, KANSAS

COLOR STANDARDS FOR COTTONSEED OIL

TABLEI Index of Refraction at 20' C. AbbB's ReAcid fractometer Value 1.4710 0.9 1,4710 1 4710 1:4710 6:9

Vol.

B y H . V. ARNY,CHARLOTTE KISH AND FRANCES NEWMARK Received April 21, 1919

As is commonly known, the commercial grading of cottonseed oil is largely a matter of color, and much work has been done in attempting t o find an ideal standard for the color of this commodity. The glasses of the Lovibond tintometer have been largely used for this purpose, but I. G. Priest of t h e Bureau of Standards has shown1 t h a t out of 219 glasses borrowed from cotton oil concerns and tested by him,, 9 per cent of the red glasses between 0.1 and 3.0 were not matches against the Bureau of Standard sets; 5 1 per cent of the red glasses between 4.0 and 2 0 . 0 were not matches; 14 per cent of the yellow glasses between 0 . 1 and 3.0 were not matched; and 40 per cent of the yellow glasses between 4.0 and 2 0 . 0 were n o t matched. This report shows t h a t the Lovibond apparatus is not the ideal standard upon which t o base a countrywide valuation of cottonseed oil. Priest, in turn, attempted t o solve the problem of authentic samples of cottonseed oil enclosed in sealed vacuum cells of the proper shape and dimensions t o be examined in a colorimeter. While i t is known t h a t the color of cottonseed oil is susceptible t o change when exposed t o the air, Priest's preliminary experiments led him t o the conclusion t h a t a sample inclosed in a sealed vacuum cell would not be thus altered. The test of time, however, showed t h a t such changes did take place a n d 1

Proc SOL.Cotton Products Analysts, 1913, p. 6.