Titrimetric Determination of Nitrogen in NitrocelluIose and of Nitrocellulose in Double-Base Propellants RAYMOND H. PIERSON and ELMO C. JULIAN U. S. Naval Ordnance Test Sfation, China lake, Calif.
b A modified ferrous-titanous titrimetric method for determining the per cent of nitrogen in nitrocellulose and of nitrocellulose in double-base propellants appears to be superior to previously available methods in a number of ways. It yields satisfactorily accurate and precise results when applied to a wide variety of nitrocellulose samples. Nitrocellulose or propellant residues which are not completely soluble in sulfuric or acetic acids may be analyzed by this procedure. A mixture of n-pentane and n-hexane is used as a transfer liquid and a mixture of glacial acetic acid and n-butyl acetate is utilized as solvent during the redox reaction with ferrous solution.
M
for determining nitrate nitrogen in nitrocellulose have been recently reviewed (2, 6, 12). The nitrometer procedure ( 1 ) is the one most commonly used but i t has limited applicability and is more hazardous than titrimetric or gravimetric methods. Solid propellants which contain nitrocellulose and nitroglycerin as their major components are called double base. These propellants also cominonly contain various other ingredients in moderate amounts. The nitrocellulose contents of these doublebase propellants have frequently been determined by a gravimetric procedure; the nitrocellulose is weighed after being separated from the other constituents. Recently, the addition of such ingredients as cellulose acetate to propellant formulations has rendered the direct gravimetric method inoperative because of the difficulty in achieving the required separations. Both the nitrometer method and the Bowman and Scott (3) titration procedure depend on complete dissolution of the samples in sulfuric acid. Some nitrocellulose samples and some residues from propellants, consisting partly of nitrocellulose and partly of other ingredients, are not sufficiently soluble in sulfuric acid for the application of either of these methods. The Shaefer and Becker method (11) is adequate for lacquer grade nitrocellulose but is not satisfactory ETHODS
for propellant grade nitrocellulose. Grodzinski’s (8) adaptation of the Shaefer and Becker technique to propellant materials is complicated and has not had wide acceptance in the United States. Good reproducibility in results couId not be obtained with the Grodzinski method and several modifications of the apparatus and procedure were tried without success. This paper presents a simple titrimetric method which is believed to be superior to any previously available and to have wide applicability. The method has been successfully applied to all types of lacquer or propellant grade nitrocellulose, and to current propellant formulations containing cellulose acetate. I n the latter case, propellants are extracted with methylene chloride or 68 to 70% acetic acid to give residues suitable for the ferrous-titanous procedure. These residues are free from nitroglycerin and other soluble nitrates, but contain all the nitrocellulose and certain other ingredients. This new procedure has been evaluated in a cooperative test program of the Joint Army-Navy-Air Force Panel on Analytical Chemistry of Solid Propellants, and found to yield satisfactorily precise and accurate values when applied to four samples of propellant grade nitrocellulose representing two levels of nitration and two cellulose source types-namely, cotton linters and wood pulp. EXPERIMENTAL
This procedure is an adaptation of the commonly used titrimetric method for nitroglycerin and a modification of the Shaefer and Becker method which was designed for the analysis of lacquers. I n extending the Shaefer and Beckermethod to propellant grades of nitrocellulose having higher nitrogen contents than the lacquer grades, there were solubility difficulties. Heretofore, complete solution of the nitrocellulose before the reduction with ferrous salt was believed necessary. Grodzinski employed acetic anhydride to aid solubility of the nitrocellulose and to prevent precipitation of nitrocellulose when the ferrous solution was added. This latter objective was not achieved with certain grades of
nitrocellulose in experiments using the Grodzinski method. I n the procedure described it is unnecessary for all the nitrocellulose to be in solutioii before the beginning of reduction with the ferrous solution. I n the two-phase system which exists after introduction of the ferrous solution and during the refluxing operation, the nitrocellulose is in the upper layer and the ferrous solution (and most of the water) is in tl-e lower one, so that little, if any, precipitation of the nitrocellulose occurs, Ever though some of the nitrocellulose in the upper layer is not in solution, il, is gradually reacted as refluxing contnues. Early in the heating step the iron migrates from the lower layer io the upper one, and reaction beg;ins immediately. The upper layer carkens rapidly (to dark green) and the lower one becomes lighter in color. The two-layer effects are believed largely i*esponsiblefor the success of the modified method. The changes of color, from light to dark green, then lightening to yellow, are those which are associated with the desirable type of redox reactior s for the nitrate-ferrous systems. Whm the amount of hydrochloric acid iii increased considerably, the reaction can be made to go very rapidly to yellow, signifying completion, but if this is done, recovery values will be low and erratic. Equally poor results are obtziined in the Grodzinski procedure if the temperature is raised too rapidly a t the beginning of the reduction and indesirable side reactions take place when the process is accelerated to give the yellow in a very short period. Of about a dozen solvents examined, n-butyl acetatt: was the only one which possessed the nscessary solubility effects. It is highly soluble in glacial acetic acid, relatively insoluble in cold dater, and a good solvent for nitrocellulose. I n the two-phase system, the acetic acidn-butyl acetate layer has a lower density than the ferrous solution, which, together with i1,s solvent and wetting effects on nitrocellulose, enables the upper layer to ;ake all the nitrocellulose from the lower, thus preventing overheating of the nitrocellulose by contact with the bottom of the flask, even though some may not be in solution. VOL. 31,
NO. 4, APRIL 1959
589
A number of experiments on drying nitrocellulose samples were conducted; the method outlined under Preparation of Samples was selected from several whicli mere effective and convenient. Drying a t room temperature in a vacuum desiccator with fresli Drierite is nearly as effective. Vacuum desiccator drying over phosphorus pentoxide thoroughly dries nitrocellulose but requires a somewhat longer time and is inconvenient because of the relatively low capacity of the phosphorus pentoside and its tendency to crust over upon exposure to moisture. Nitrocellulose samples that have been airdried for a day or so m-ill usually be completely dried after 2 hours in the vacuum oven a t 60" to 70" C. The experiments showed, however, that dried samples not in covered weighing bottles would gain moisture rapidly when stored in a desiccator and exposed to moisture of the room air by an occasional removal of the desiccator lid. M7hen a number of determinations are to be made, the follon~ingprocedure is advantageous. Using tared weighing bottles with good covers, dry all the samples a t one time in the vacuum oven, remove them to the desiccator, and when partly cooled, place the covers on the bottles. When thoroughly cool, weigh the bottles as rapidly as possible and return them to desiccator storage. They can then be used as needed and any change in weight due t o moisture absorption will have no influence. I n previously reported work, nitrocellulose has been dried, weighed, and transferred in the dry state. During this investigation it became desirable to use some kind of liquid stream to transfer the dry fluffy samples of nitrocellulose or the nitrocellulose residues from propellants to the special reduction flasks. Neither glacial acetic acid nor n-butyl acetate is suitable, because either mill cause the residue to become gummy, and to stick to the glassware. Water is undesirable, one reason being that the water content in the reduction step must be held a t a low level, A light hydrocarbon solvent was therefore chosen. n-Pentane and n-hexane a t about a 3 to 1 ratio is advantageous because transfer without loss due to air currents, or clinging to the crucible is facilitated, and in the succeeding step where glacial acetic acid is added, the light hydrocarbon prevents the agglomeration of the nitrocellulose. When glacial acetic acid is added to nitrocellulose without the presence of the light hydrocarbon, the nitrocellulose agglomerates into a single gummy layer on the bottom of the vessel and cannot be dissolved without the risk of overheating or considerable loss of time. After the nitrocellulose 590
ANALYTICAL CHEMISTRY
has been maslir tl into the reduction flask through a powder funnel with the hydrocarbon solvent, glacial acetic acid is added through the same funnel, to ensure the retiioval of all the nitrocellulose. The hydrocarbons are rapidly evaporated by hmting on a steam bath; the particles renain discrete and solution in the glatial acetic acid and nbutyl acetate is greatly expedited. Complete remoid of the hydrocarbon solvent is not d t i c a l but most of it must be remored before the redox operation to kwp the liquid volume in the flask a t Llie proper level and to provide a mixed solvent with appropriate boiling point and solvent effects. %-Pentane alone can be used as transfer liquid but the addition of the n-hexane causes less clinging of the nitrocellulose to the lower edg;e of the powder funnel, because of its ower evaporation rate, and gives a s noother boiling action during removal of the transfer liquid on the steam bath A petroleum ether of suitable boiling range and purity should also be satisfactory; the pentanehexane mixturc was selected because the two solvents mere available in a high degree of p .irity. The amount of hydrochloric acid usually specifietl (9, 11) for preparation of the titanouf solution is 200 ml. of hydrochloric actid per liter, while this method specifics 350 ml. With the smaller amou ?t, white precipitates (probably titanium oxide) are deposited in the buret, :;torage container, and connecting tuhing when the solution stands for a few weeks. With the larger amount, the solution mill not form such deposits, even on standing for many months. I n laboratories where the consumpticii of the reagent is high, the white depcEit will not be a serious nuisance, but r h e r e the solution is used only occasionally, the higher acid content in the reagmt is advantageous. This method includes a preliminary boiling step which gives improved blanks by removing twces of oxygen. Boiling for more than 2 minutes is undesirable because of tht! possibility that there might be sorro decomposition of the nitrocellulose. Most samples appear to dissolve completely or nearly so in this treatment, but even if there is considerable turbidity, valid results will still be obtained. Baniples of high nitrogen content are likely to be less soluble and are more turbid than those of lower nitrogen contecit. DEVELOPED METHOD
OF ANALYSIS
Apparatus. Reduction-titration flask. Stand.trd-taper Florence flask of 300-ml. cspacity, with sealed-in side arm for introducing a stream of carbon dioxide which mill impinge of the surface of the liquid in the flask. Standard-taper Graham-type condenser of 5O-ciii. jacket length.
System for providing carbon dioxide protection for titration flasks, solutions in storage, and in burets. Both the titanous chloride and ferrous ammonium sulfate solutions should be protected by carbon dioxide a t all times, and the titanous chloride solution in storage from light. The carbon dioxide system should have an automatic pressure relief control to prevent excessive pressure in solutions, and the system be arranged so that pressure on burets and on storage containers will be approximately equal. Each outlet leading to a reflux condenser should have a small bubbler to indicate the rate of gas flow. Vacuum oven which can be niaintained a t 60' to 70" C. and a pressure of 2 to 5 em. of mercury. Reagents. Titanous chloride soluPrepare as described tions, 0.2" (2, 9, 11), but use 350 ml. of hydrochloric acid per liter. Ferrous ammonium sulfate solution, 0.7N. Prepare a solution containing 1100 grams of ferrous ammonium sulfate hexahydrate, FeS04(NH4)2S04. 61&0, and 560 ml. of concentrated sulfuric acid in 4000 ml. of solution (use oxygenfree distilled water). Reduce tlie ferric iron by treating the solution with powdered iron (reduced by hydrogen) until a drop produces no immediate pink with ammonium thiocyanate solution. Filter through a large folded filter paper. It is advisable to feed a current of carbon dioxide into the filtrate during filtration. Agitate the reagent with carbon dioxide for a few minutes and store under carbon dioxide. Reduce and refilter whenever a 25ml. portion requires more than 0.2 ml. of 0.214' titanous cliloridc solution to reduce the ferric iron present. Acetic acid, 68 to 70% by weight (sparged with carbon dioxide to make oxygen-free). wButvl acetate. Eastman Kodnk Co. white label. n-Pentane, Phillips technical grade. n-Hexane, Phillips technical grade. Sodium diphenyl benzidine sulfonate indicator, 0.5 gram in 100 ml. of distilled water. Standardization of Titanous Chloride. Standardize the titanous chloride [daily before each series of determinations by titrating the solution against a 0.2000N potassium dichromate solution as described (IO)]. The calculations for the iron content correction given in the reference can be greatly simplified to yield the following equations:
where vd = nii1lilitei-s of standard potassium dichromate solution. A i d = normality of standard potassium dichromate solution. V I = milliliters of impure titanous solution used in the direct titration against standard potassium dichromate using sodium diphenyl benzidine sulfonate indicator. V 2 = milliliters of sample of impure
titanous solution taken for oxidation with potassium permanganate. V 3 = milliliters of impure titanous solution required to titrate the ferric iron from VZ. A still greater simplification for the correction for iron content of the titanous solution may be made as follows: When the green end point of the titration of standard potassium dichromate with titanous solution (sulfonate indicator) is reached, add 5 ml. of ammonium thiocyanate indicator "and continue the titration with titanous solution to discharge the red color produced by ferric iron and the latter indicator. The final color will be the same green of the previous end point (due to chromic ion). Calculate the desired corrected normality, N , as follows : 1' T
where V d is the volume in milliliters of standard potassium dichromate solution and V Tis the total in milliliters of titanous solution. Preparation of Samples. Nitrocellulose. Dry nitrocellulose samples t o constant weight in shallow weighing bottles in a vacuum oven at 60" t o 70" C. and a pressure of 2 t o 5 cm. of mercury. Before weighing, cool in a desiccator containing fresh Drierite either with covers on the bottles or under vacuum. Propellant. Extract propellant samples (previously ground in a Wiley mill t o pass 20-mesh) with a suitable solvent-methylene chloride or 68 to 70% acetic acid-to remove nitroglycerin or other soluble nitrate esters. Filter the residue consisting chiefly of nitrocellulose on a fritted-glass or Selas crucible of medium porosity. Dry this residue to constant weight in a vacuum oven a t 60" to 70" C. and a pressure of 2 to 5 cm. of mercury. Cool before weighing as described above. Transferring Samples to Reduction Titration Flask. Use a sample of about 0.20 t o 0.25 gram. Transfer a nitrocellulose sample from a tared weighing bottle, or a propellant residue from a filtration crucible t o the 300-nil. reduction flask by using a powder funnel and a stream of npentane-n-hexane solvent (ratio of about 3 t o 1) from a n all-glass or polyethylene wash bottle. Wash the funnel by pouring 45 ml. of glacial acetic acid through it into the reduction flask. Add a few glass beads or Carborundum chips to the flask. Remove most of the pentane-hexane solvent b y heating for a few minutes on a steam bath. Solution and Titration of Sample. T o the sample transferred t o the titration flask as described (containing 45 ml. of glacial acetic acid), add 25 ml. of n-butyl acetate. Connect the carbon dioxide tube t o the flask, and the flask to the reflux condenser, and gently boil the solution for 1 t o 2 minutes.
Bring the flask and contents to approsimately room temperature in a water bath. Add rapidly from a buret (with a coarse tip and protected b y carbon dioxide) 25 ml. of the 0.7N ferrous amnionium sulfate, and from a dispensing buret, 8 to 10 ml. of concentrated hydrochloric acid, (This amount is critical and should be held within the prescribed limits.) Replace the flask under thc reflux condenser and heat gently until boiling begins. (Boiling should start in about 10 minutes.) Two layers are readily apparent in the early stages of heating. The upper will turn light green, then deepen to a very dark green. On continued boiling, the two phases will take on the appearance of a single dark phase (although two phases are actually present), and will then begin to grow lighter. Boil until the color changes to yellow and then for 10 minutes or more after no further change is observable which usually requires 30 to 40 minutes. Throughout the boiling period agitate the flask contents by hand shaking a t about &minute intervals to wash down any particles which niiaht adhere to the flask above the li ;id level. Pncrease the stream of carbon dioside and cool the flask and contents in a water bath to approximately room temperature. Loosen the condenser joint slightly and wash down the condenser and the ioint with 30 ml. of oxvrzen-free 68 to "7Oyoacetic acid. Tirhten the flask and condenser and draiGthe latter for a few minutes. Disconnect the flask and titrate its contents with 0.2147 titanous chloride, adding 5 ml. of ammonium thiocyanate near the end. Titrate slowly (dropwise) near the end point, allowing a t least 10 seconds with good agitation for the color to reach equilibrium between drops, and titrate until the red coloration is discharged. AIake blank determinations on the reagents in the same amounts and including all the steps as used for the nitrogen determination. COMPUTATIONS
For per cent nitrogen in nitrocellulose. Per cent N
=
0.46603 (V
- B)Nr
JV
where V = volume in milliliters of titanous solution used in the determination, B = volume in milliliters of titanous solution used in the blank determination, ATt = normality of titanous solution, and TV = weight of sample, grams. For per cent nitrocellulose in propellant. Per cent nitrocellulose =
F (V
-
w
B) N
t
(2) where F is a factor depending on the nominal or known nitrogen content of the nitrocellulose used in manufacturing the propellant, and V , B , Nt,.and W are identical with those of Equation 1.
46.693 F = __ per tent nitrogen in the nitrocellulose
Tempeiature corrections for buret readings. If the ambient tempcrature changes over a range greater than about 2" 0. during a series of determinations and standardizations, apply tcmpcrature corrections to all readings of the titanous sdution or dichromate solution burets. For moderate changes in temperature, lake the cubic expansion of the titanous solution as 0.00028, and that of dichromate solution as 0.00022. RESULTS AND DISCUSSION
This method has been tried on it variety of nitrocellulose samples. The method vas suitable for all of the samples and, therefore, has much promise of being universally applicable. Table I gives typical data for three lacquer arid six propellant grade samples, The nitrogen content for the lacquer gi-ades ranged from 10.05 to 12,2370,and the range for the propellant grades m-aE 12.2 to 13.4%. In the latter case both cotton linters and wood-pulp types weix examined. The over-all precision of the method for per cent nitrogen in nitrocellulose is reflccted by a pooled standard deviation of 0.021770 absolute. The values shown in Table I for per cent nitrogen by nitrometer f x samples l, 2, and 3 were average values supplied by the manufacturer. The nitrometer values for samples 5 7, 8, and 9 were grand averages f 'om a cooperative test program of thc Joint Army-Navy-Air Force Panel on A nalytical Chemistry of Solid Propellantri. Each grand average was obtained by averaging the averages from seven laboratorics. An extensive statistical analysis was made on the data for srmples 5, 7, 8, and 9, using the techniques supplied by Cochran (4). There was no evidence of bias in the titrimetric method when compared with the nitrometer data. Average results by b e two procedures were not significant\$ different a t the 5% level. Both the Shaefer and Becker (11) and the Grodz nski (8) reports indicated that a smE11 quantity of hydrobromic acid was lecessary for complete recovery values in the determination of nitrogen in nitroccllulose. Experiments were conduzted to test the effect on this procedure of adding hydrobromic acid. The results shown in Table I1 were obtained on the sample identified as No. 7 in Table 1. The amount of hydrobromic acid used for the determinations reported in the second column was about twice the amount recommended b y Shaefer and Becker. All the data for this table w,:re obtained in a randomized order in a single day of testing. The average value obtained with hydrobromic acid present is slightly lower than the average by the unmodified VOL. 31, NO. 4, APRIL 1959
591
Table 1. Per Cent Nitrogen by Titration for Nine Samples 0 :Nitrocellulose Raniple No. 1 2 3 4 5 6 7 8 9 nd iiso T, P P P P P P By nitrometer 10.95 11.00 12.23 . . . 12.499 . . . 13.418 12.458 13.414 By tit,ration 10.99 11.01 12.25 12.33 12.54 13.26 13.4!) 12.49 13.45 (replicates)
10.98 10.99 12.23 12.31 10.97 10.99 12.23 12.27 10.97 10.99 12.22 12.27 10.96 10.98 12.21 12.24 10.974 10.992 12.228 12.284 Mean, 3 5 Est. std. dev. 0.011 0.011 0.015 0.036 % ’ coef. var. 0.100 0.100 0.123 0.293 Pooled std. dev. for titrations, sp = 0.019 Av. % coef. var. = 0.135.
12.52 12.52 12.50 12.48 12.512 0.023 0.184
13.25 13.24 13.24 13.22 13.242 0.015 0.113
13.11‘7 12.48 13.a1.5 12.47 13.,15 12.47 13:14 12.47 13.,160 12.476 0.020 0.009 0.149 0.072
13.44 13.44 13.43 13.42 13.436 0.011 0.082
L. Lacquer grade. P. Propellant grade. CL. Cotton linters. WP. Wood pulp. N. Nitrogen.
the three tables were based on techniques described in the literature (6,7). SUMMARY
The ferrous-titanous titration procedure described for determining the nitrogen content of nitrocellulose and the nitrocellulose content of solid propellants offers the following advantages in addition to those previously mentioned. It utilizes equipment and solutions already available in propellants laboratories for titrimetric nitroglycerin and gravimetric nitrocellulose determinations and it requires a relatively short time. About 10 replicate analyses may be made by one analyst in an 8hour day, excluding the time required for sample preparation, extraction, and weighing of extraction residues. ACKNOWLEDGMENT
Table II. Comparison of Methods with and without Addition of Hydrobromic Acid Unmodified Modified (without (with Method HBr) HBr) yo N by titration, redicates 13.47 13.47 13.47 13.44 13.44 13.43 13.44 13.38 13.44 13.24 13.452 13.392 Mean, Xa 0.016 0.091 Est. std. dev., s 0.119 0.680 % coef. of var. Student 1 value, 8 degrees of freedom = 1.463 (not significant at 570 level).
variance modified = 32.35 (sigvariance unmodified nificant at 1% level),
F-ratio,
method, but the two averages are not statistically different a t the 5% level of significance (by two-tail t test), (5). The variability by the modified method is greater than that by the unmodified method and this difference is statistically significant a t the 1% level (by two-tail F test) (5). It is concluded that the presence of hydrobromic acid with the reagents mould impair the proposed procedure. The proposed method has been used at the authors’ activity for more than a year for determining nitrocellulose in double-base propellants which contain ingredients which interfere with the gravimetric method. For this purpose, the degree of nitration of the nitrocellulose used in manufacturing the propellant must be known with reasonable certainty or a sample of the nitrocellulose must be available and its nitrogen content determined. The results on these determinations of nitrocellulose have been satisfactory. Typical replicate analyses are shown in 592
ANALYTICAL CHEMISTRY
Table 111. Typical Results for Per Cent Nitrocellulose in Do Jble-Base Propellant by Titratiwi Procedure Method of 3-Hr. Extraction (68 to 70% 30-Min. Steam Acetic Acid) Reflux Bath % nitrocellulose, replicates 41g.40 48.47
Mean, 2, Est. std. dev., s % coef. of var. Pooled std. dev., SP
43,32 413.26 41.20 4:s.12 413.260 1). 109 11,226
48.37 48.29 48.19 48.12 48.288 0,139 0.288
0.112
0.257 Av. Yo coef. of var. Student 1 value, 8 degrees of freedom = 0.355 (not significant a t 5% level) variance, steam bat.h - 1.626 (not variance, r2flux significant at 5% le {el)
Table 111. The tr,ble also compares the two heating techniques for making the initial extraction of nitroglycerin from the propellan; using 68 to 70% acetic acid. I n both cases the samples are placed in Erleumeyer flasks (250 or 300 ml.) and acetic acid is added. I n one case, the flasks are attached to a standard taper con:lenser and refluxed for 30 minutes on a n electric hot plate. I n the other, they tire loosely stoppered and heated for 3 ho .irs on a steam bath. This comparison n its made because a newly proposed military standard (9) allows the use of either the reflux or steam bath heating steps. No significant difference in rileans or variability is evident. Typicrtl precision for the method for per cent nitrocellulose in propellant is given by the pooled standard deviation of 0.112% absolute, corresponding to 2111 average per cent coefficient of variat Lon of 0.257. Statistical evalurttions summarized in
The authors are indebted to the Joint Army-Navy-Air Force Panel on Analytical Chemistry of Solid Propellants for assistance in evaluating the proposed procedure in a cooperative test program. Two members of the group, Radford Arsenal and Naval Powder Factory, supplied some of the samples. The lacquer grade nitrocellulose samples were supplied by Hercules Powder Co., Parlin, N. J. J. J. Lamond, a visiting Panel representative from Scotland, tested the procedure a t the Chemical Inspectorate, Bishopton, United Kingdom, and contributed a number of helpful suggestions. LITERATURE CITED
(1) Am. SOC.Testing Materials, Philadelphia, Pa., D 301-50, Part 4, p. 362, 1952. (2) Becker, W. W Shaefer, W. E., L‘OrganicAnalysis.,”;:”Vol. 11, p. 97ff, Interscience, New York, 1954. (3) Bowman, F. C., Scott W. W., J . Ind. Eng. Chent. 7,766 (1915). (4) . . Cochran, 11’. G., Biomehics 10, 101 (1954). . (5) Crow, E. L Davis, F. A., Maxfield, M. w., StaXistics illanual,” pp. 57fJ 74f, U. S. Naval Ordnance Test Station, China Lake, Calif., 1955. (6) Easterbrook, W. C., Mathew, R. H.,
Imperial Chemical Industries, Ltd., Nobel Division, Rept. RI-5470 (July
26,1956). (7) Fisher, R. A., “Statistical Methods for Research Workers,” 10th ed., Hafner Publishing Co., New York, 1948. (8) Grodzinski, Joseph, ANAL. CHEW. 29, 150 (1957). (9) Military Standard 286, in reparation. (10) Pierson, R. H., Gantz, S., ANAL. CHEhI. 26,1809 (1954). (11) Shaefer, W. E., Becker, W. W., Ibid., 25,1226 (1953). (12) Verschragen, P., Anal. Chim. Acta 12,227 (March 1955).
8.
RECEIVEDfor review June 2 1958. Accepted October 17, 1958. Divhon of Analytical Chemistry, 134th Meeting ACS, Chicago, Ill., September 1958.