INDUSTRIAL a n d ENGINEERING CHEMISTRY ANALYTICAL EDITION

WALTER J. MURPHY, EDITOR ISSUED JUNE 17, 1943 VOL. 15, NO. 6 CONSECUTIVE NO. 12

Editorial Assistant: G. Gladys Gordon Manuscript Assistant: Stella Anderson Make-up Assistant: C harlotte C. Sayre Advisory Board

B. L. C l a r k e G. E. F. L u n d e l l R. H. M ü l l e r T . R. C u n n i n g h a m M . G. M e l l o n H. H. W i l l a r d

Determination of Precision of Analytical Control Determination of Iron in Presence of Chromium M e t h o d s ...... Raymond F. Moran 361 and Titanium with Jones Reductor ...... F. S. Grimaldi, R. E. Stevens, and M. K. Carron 387 Turbidimetric Determination of Small Amounts of C h lo rid es . E. N. Luce, E. C. Denice, and F. E. Akerlund 365 Extraction of Ascorbic Acid from Plant M aterials . . J. D. Ponting 389 Color Index. Light-Colored Petroleum Products . . I. M. Diller, J. C. Dean, R. J. DeGray, and J. W. Wilson, Jr. 367 Apparatus for Purification of Hydrocarbons by Re­ crystallization ...... John Lake Keays 391

Determination of Iodine in Tetraiodophenol- Filtration Cylinder . R. J. DeGray and E. P. Rittershausen 392 p h th a le in ...... Samuel Weiner, Byron E. Leach, and Mary Jane Bratz 373 MICROCHEMISTRY: Determination of Monoalkyl Ethers of Ethylene Semimicroanalysis of Saline Soil Solutions .... Glycol . . . Harold W. Werner and James L. Mitchell 375 R. F. Reitemeier 393

Collection and Estimation of Traces of Formalde­ Estimation of Sulfonamides...... hyde in Air . . F. H. Goldman and Herman Yagoda 377 S. W. Lee, N. B. Hannay, and W. C. Hand 403

Microdetermination of Mercury in Organic Com­ Furfural (Correspondence) . . Vanderveer Voorhees 378 p o u n d s ...... H. William Eckert 406

Determination of Small Amounts of Tellurium in Microdetermination of Arsenic in Biological M a­ H ig h -L ead a n d T in -B ase A llo y s ...... te ria l ...... James A. Sultzaberger 408 Ralph A. Schaefer 379

Mustard Gas in A ir ...... William Rieman III 411 Mixed Solvent Extraction James H. Wiegand 380

Microdetermination of Magnesium with Polaro- Determination of Halogens in Organic Compounds g r a p h ...... Christopher C arruthers 412 Robert R. Umhoefer 383 Reproducibility of Weighings Made on Micro­ Determination of Chloride in Bauxite-Supported chemical Balances (A. C. S. Committee Report) Anhydrous Aluminum Chloride Catalysts .... Clement J. Rodden et al. 415 W. A. La Lande, Jr., Heinz Heinemann, and W. S. W. McCarter 385 Detection of Gold in Plating .... Melvin Lerner 416

The American Chemical Society assumes no responsibility for the statements and opinions advanced by contributors to its publications. 29,500 copies of this issue printed. Copyright 1943 by American Chemical Society.

Publication Office: Easton, Pcnna. Editorial Oilicc: 1155 16th Street, N. W., W ashington, D. C. Advcrti«ing Departm ent: 332 West 42nd Street, New York, N. Y. Telephone: Republic 5301. Cable: Jicchcm (Washington) Telephone: Bryant 9-W30 Published by the American Chemical Society, Publication Office, 20th & copies: Industrial Edition, $0.75; Analytical Edition, $0.50. Special rates Northampton Sts., Easton, Penna. Entered as second-class matter at the to members. Post Office at Easton, Penna., under the Act of March 3, 1879, as 24 times a No claims can be allowed for copies of journals lost in the mails unless year. Industrial Edition monthly on the 1st; Analytical Edition monthly such claims are received within 60 days of the date of issue, and no claims on the 15th. Acceptance for mailing at special rate of postage provided for will be allowed for issues lost as a result of insufficient notice of change of in Section 1103, Act of October 3, 1917, authorized July 13, 1918. address. (Ten days’ advance notice required.) “Missing from files” Annual subscription rate, Industrial Edition and Analytical Edition sold cannot be accepted as the reason for honoring a claim. Address claims to only as a unit, members S3.00, others $4.00. Foreign postage to countries Charles L. Parsons, Business Manager, 1155 16th Street, N. W., Washington. not in the Pan American Union, 82.25; Canadian postage, S0.75. Single D. C., U. S. A. 4 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

TYPE B-2

VVTE enthusiastically introduce a new "Photelometer,” not ” to supplant but to be a companion instrument to No. 12335 Cenco-Sheard-Sanford "Photelometer.” The new type is smaller in size and is known as Industrial Type B-2. It is designed to take economic advantage of production in quantity and utmost simplicity of construction without sacrifice of desirable features. Its sensitivity and accuracy are sufficient for most laboratory determinations. Type B-2 is a compact filter photometer for chemical analyses in the routine or control laboratory. Molybdenum, titanium, vanadium, or manganese in steel; lead, copper, iron, or vita­ mins in foods are a few typical determinations to which this type lends itself.

It is a barrier-layer instrument consisting of basically a low The basic parts are mounted in an attractive plastic case for voltage light source, an adjustable light aperture, a three-color convenient manipulation and reading. A constant voltage filter holder, receptacles for tubular or rectangular absorption cells, a single photoelectric cell, and a sensitive current transformer to supply constant intensity of light when operated measuring instrument with a 2}/¡“ scale. The scale reads from on a controlled frequency power line is furnished with the 0-100 in 50 divisions. 115 volt AC instruments. No. 12346 "Photelometer” Industrial Type B-2 including a package of (12) No. 12344G Tubular Absorption Cells. N o ...... : ...... A C For v o lts...... 115 AC 6 DC P ric e ...... $ 1 1 0.00 $100.00 CENTRAL SCIENTIFIC COMPANY

SCIENTIFIC INSTRUMENTS ( j N ( Q LABORATORY APPARATUS Rtc u * pat off. NEW YORK TORONTO CHICAGO BOSTON SAN FRANCISCO June 15, 1943 ANALYTICAL EDITION

PIKIIIIICTIIIII LINIiSi

America over, the accent is on ever-increasing pro­ Baker & Adamson Reagents are the standard for duction! From sprawling, busy assembly lines in laboratory control. Laboratories know they can shipyards . .. plants ... factories, war material is rely upon the highly uniform quality of B&A moving off production lines in surging volume ... Reagents... count on their dependability and pur­ America is “delivering tlie goods!” Yet with all the ity ... because accuracy, high quality and uni­ pressing need for more and more war materials— formity are “built” into each B &A Reagent through quality still remains the important factor. 60 years of experience. Dependable reagents are vital ... becausc Whether your laboratory requires reagents they help maintain this quality and accuracy ! for special analyses or for routine testing . . . In many leading plants the country over, select B& A Reagents to do the job right!

SETTING THE PACE IN CHEMICAL PURITY SINCE 1882

SOUU-i.iv/~ - - rmdini! 1 Baker Adamson CÁ&wttca/sAND plating in plating wide aPP “ hearings, stor- alrplanc molor ctc. Sigiufi- Division of GENERAL CHEMICAL COMPANY, 40 Rector St., New York C ."R A id e/s oge battery d lahor Technical Service Offices: Atlanta • Baltimore • Boston • Bridgeport (Conn.) • Buffalo • Charlotte (N. C.) Chicago • Cleveland • Denver • Detroit • Houston • Kansas City • Milwaukee • Minneapolis result from its «- New York • Philadelphia • Pittsburgh • Providence (R. I.) • St. Louis • Utica (N. Y.) Pacific Coast Technical Service Offices: San Francisco • Los Angeles today '• Pacific Northwest Technical Service Offices: Wenatchee (Wash.) • Yakima (Wash.) In Canada: The Nichols Chemical Company. Limited • Montreal • Toronto • Vancouver INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

AQUEOUS PROCESSING SOLUTION

HEATING STEAM AND CONDENSATE

TO BOILER

AQUEOUS PROCESSING SOLUTION

WANT TO IMMOBILIZE THIS LEAK?

To keep contaminated condensate ure, continuously, the electrolytic from sneaking into a boiler’s tubes conductivity of the water, and open and sabotaging steam production, a dump valve in the line when a leak we can supply an instrument which is indicated by a rise in conductivity. will halt the potential damage at its The instrument employed may be source, and hold it immobilized while either of the following: a maintenance pipe-fitter repairs the leak and thus ends the contamina­ If the operation of a signal light tion. and the opening of the dump valve are all you require when a pipe The method employed is to meas- springs a leak, we recommend No. 4S50 Signalling Conductivity Con­ troller. This instrument has no moving parts except a relay; re­ quires practically no maintenance; gives protection at minimum cost.

If you want a continuous, auto­ Micromax Controlling Recorder used by the matic record of the purity of the Cros*ett (Ark.) Lumber Co., on a water line. condensate, in addition to an auto­ tion than does a temperature re­ matic signal and the operation of a corder. dump valve, we recommend No. 33111 Micromax Signalling Re­ For further information, see Cata­ corder. This instrument can be log N-95-163(l), on the Signalling read from a considerable distance, Controller, or Catalog N-95-163, on and its chart records are 106/s" the Micromax Controlling Re­ Signalling Conductivity Controller installed on a steam return from processing vat to feed-water diameter. It requires no more atten­ supply. corder.

LEEDS 4. NORTHRUP COMPANY, «20 STENTON AVE., PHI LA.,

A Slogan for All Americans LEEDS & NORTHRUP Jrl. Ad N-93-163(3) MEASURING INSTRUMENTS • TELEMETERS • AUTOMATIC CONTROLS • HEAT-TREATING FURNA June 15, 1943 ANALYTICAL EDITION 7

The United States Navy’s “battlewagons,” material used in building our mighty fleet. “tin cans,” and “pig boats”—familiar and Chemistry has helped to make America’s affectionate terms used by the fighting men of naval force second to none! our fighting sea forces— actually are precision machines, designed by skilled engineers, and We are proud of the part that Merck Labo­ painstakingly built of the finest materials. ratory Chemicals played in this great achievement. To produce the thousand fold of intricate, even delicate, parts that make an efficient Chemicals which are destined for use in such fighting ship, requires highly accurate and a painstaking task must, of necessity, possess scientific control of the raw materials which superb quality themselves. The rigid control go into its construction. Such control has its exercised over Merck Reagents in our Ana­ birth in research, analytical, and metallur­ lytical Laboratories makes certain that they gical laboratories, where skilled scientists will always be fine tools for precision mea­ test and regulate the quality of every raw surements.

MERCK & CO.«, InC. ty/lan ttfa ctur-iny S/emi4t6 R A H W AY, N. J.

Please send m e th e following charts: N am e...... ACS-6-43

□ Sensitivity of Q ualitative Reactions C om pany...... P osition......

□ Periodic C h art of th e Elem ents S tre e t......

□ Sensitivity C h art C ity...... S ta te ...... 8 INDUSTRIAL ÄND ENGINEERING CHEMISTRY Vol. 15, No. 6

ELEOT.R-ICV F U R N A C E /FUSED June 15, 1943 ANALYTICAL EDITION

PROVIDING ACCURACY ASSURANCE Every Parr Calorimetric Thermometer is especially designed, produced and calibrated under specifi­ cations of extreme exactness to incorporate the highest available guality of material and workman­ ship. Each thermometer furnished with a Parr Calorime­ ter has a complete correction chart drawn from test data. This is just one of the many provisions made by the Parr laboratory men to provide manufacturers with precision Parr apparatus. INSTRUMENTED. Send for further information. MOLINE ILL.

PARR BOMB CALORIMETERS and CHEMICAL TESTING APPARATUS 10 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 CONSOLIDATED Engineering Corporation announces a new instrument and techniques for the analysis of gas and liquid mixtures.

SPECIAL FEATURES > Analysis of 15-20 sam­ ples can be obtained with MASS SPECTROMETER one instrument in an eight- hour day. > Results are practically independent of variable factors due to the human element. > Results can be computed in such a way as to be self-checking. l HE Consolidated Mass Spectrometer and > 1 /10 c.c. of sample is usually adequate for an pertinent analysis techniques have been developed to provide an analysis. accurate/ rapid method of analyzing simple or complex mixtures. As > The instrument may be adjusted for operation a result of unique design features of this instrument and of techniques over a molecular weight developed by Consolidated, mixtures containing as many as 15 or range from 1 to 250. > Units are designed for more components can be analyzed with speeds unattainable before. convenience of installation and operation. >Many automatic fea­ FUNCTION The Mass Spectrometer USES This development provides a tures result in ease of breaks molecules of the material intro­ marked improvement in the analysis of operation. charge stocks and feed streams used in duced into charged fragments, or ions, in > Automatic protective the manufacture of synthetic rubber and a manner dependent on the structure and circuits insure against ac­ high octane aviation gasoline, thus pro­ cidental damage. composition of the molecule. The ions are viding a superior means of control per­ > Conservative electrical then separated into beams, each beam mitting more efficient plant operations. design assures continuous In Research it provides unlimited oppor­ containing ions of a certain mass number. operation. The beams in the sequence of their mass tunities for investigation of problems arising in development of new products. numbers strike a collector where they surrender their charges. The resultant cur­ The Consolidated Mass Spectrometer is made available to W ar Industries on a basis which rents are amplified, and a permanent provides ample protection in future developments in this field. . . Write for particulars. automatic record of the mass spectrum is obtained. The composition of a mixture CONSOLIDATED ENGINEERING CORPORATION may be quantitatively determined from Herbert Hoover, Jr., President its mass spectrum. 1255 EAST GREEN STREET • PASADENA, CALIFORNIA June 15, 1943 ANALYTICAL EDITION

In meeting the demands of war produc­ tion now and in anticipating the needs of civilian life after victory comes, In­ dustry’s laboratories are served by

KIMBLE LABORATORY« ■ es* GLASSWARE *KS K K < EXAX> <^N0RMAXS / \ B L U E X u. L 1 N T M M C - U S A* T.M. REG . U S A. Using standard items wherever possible may be important in minimizing delays. Consult leading Laboratory Supply Houses throughout the United States and Canada about the Kimble products you require.

(j~uayâ n ie.e o f In f is-lbie .{,) uali.t.y KIMBLE GLASS COMPANY------vin eland, n. j. NEW YORK • CHICAGO . PHILADELPHIA • DETROIT • BOSTON • INDIANAPOLIS • SAN FRANCISCO 12 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

USE SPECTROGRAPHIC EQUIPMENT TO ANALYZE YOUR WAR MATERIALS

A complete set of mated A.R.L.-DIETERT Spectrographic Equip­ ment as illustrated increa­ ses production and assures material specification control in minutes instead of hours. Send today for catalogue No. 128, MODERN ANALYSIS.

APPLIED RESEARCH LABORATORIES H A R R Y W. D I E T E RT~ ~C O . 4336 SAN FERNANDO RD., GLENDALE, CALIF. 9330 ROSELAWN AVE.. DETROIT, MICH. i m l . June 15, 1943 ANALYTICAL EDITION 13

Locked in each grain of wheat is a precious substance—nitrogen— which is essential to the health and well-being of the human race. Scientists in the laboratories of America’s food industries tirelessly seek the hidden secrets locked inside these health- giving grains so that we may have more nutritious flour and cereals. Laboratory research, too, makes possible better, faster- acting fertilizers to invigorate the earth, to speed plant growth and to increase the production of vital food-stuffs. The Pyrex brand Kjeldahl Flask, because of its low expan­ The Pyrex brand Kjeldahl Flask, fabri­ cated from Pyrcx brand Balanced Chemi­ sion and high chemical stability, successfully meets the se­ cal Glass No. 774, is but one of the more than 2700 Pyrex Laboratory Ware items vere conditions encountered in nitrogen digestion and distil­ contributing to laboratory, achievement. lation. Pyrex Kjeldahl Flasks have strong tool-finished necks of uniform taper to insure accurate stopper fit. This finish in­ creases the strength and minimizes the possibility of break­ age in handling and use, particularly when inserting stoppers. Corning Research pledges itself to supplying constantly improved laboratory glassware. Consult your laboratory “YOU HAVE DONE A GOOD JOB supply dealer for complete information on any requirement. OF SENDING GLASS TO WAR”

P YR E X*>*»°LAB ORATORY WARE PYREX C orning — — m e a n s — — "PYREX” and "VYCO R" are registered trade-marks and indicate manufacture by CORNING GLASS WORKS . CORNING, NEW YORK Research in Glass 14 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

A.H.T. CO. SPECIFICATION GAST PORTABLE ROTARY AIR BLAST AND SUCTION APPARATUS

ROTARY AIR BLAST AND SUCTION APPARATUS, GAST PORTABLE, A.H.T. CO. SPECIFICATION. A quiet, air cooled, motor driven unit complete with vacuum and pressure gauges, and thermal overload circuit breaker. Suitable for continuous operation at pressures not exceeding 20 lbs., or for intermittent use up to 30 lbs. Consisting of a pump with cast iron rotor fitted with four composition vanes revolving in a pre­ cision machined housing of special alloy iron which is connected directly to a 1/6 h. p., 1725 r. p. m. motor. The pump rotor is an extension of, and integral with, the shaft of the motor. The com­ plete apparatus is mounted on five rubber feet and is equipped with carrying handle, air filters and oil trap attached directly behind the tapered rubber tubing connections at the inlet and outlet, dial type vacuum gauge 2 inches diameter, graduated from 0 to 30 inches of mercury in intervals of 1 inch, pressure gauge 2 inches diameter, graduated from 0 to 50 lbs. in intervals of 1 lb., automatic pressure release valve adjusted at 30“ lbs., and bleeder petcocks for regulating the pressure and vac­ uum to the requirements of the work intended within the limits described below. The combined filter, muffle and trap on the pressure side is enclosed in cast iron and is supplied with a cartridge which can be removed for cleaning or replacement. The filter on the vacuum side is a combined oiling and air filtering device. It is enclosed in a screw neck glass reservoir in which the oil level can be observed at all times. For intermittent use the pump can be operated up to 30 lbs. pressure, or higher, if the safety valve is readjusted in the laboratory but, for continuous use, the maximum pressure must not exceed 20 lbs. Speed, r. p. m ...... 1725 M aximum pressure, lbs. per sq. in...... 20 to 30 Cu. ft. of free air (atmospheric pressure) per minute, approxim ately...... 1.3 Cu. ft. of free air at 30 lbs. pressure (for 10-minute periods)...... 0.95 Number of blast lamps (M .I.T. type or equal)...... 4 Maximum vacuum, inches of m ercury...... 27 Power consumption, w atts...... 250 N et weight, lbs...... 32 Code 1033-G. Gast Portable Rotary Air Blast and Suction Apparatus, A.H.T. Co. Specification, as above described, Word complete with pressure and vacuum gauges, and thermal overload circuit breaker, filters, carrying handle and 10 ft. cord w ith snap switch and plug. For 110 volts, 60 cycles, a.c...... 33.50 Abxes 1033-H. Ditto, but for 110 volte, d.c...... 44.50 Abxhm 1033-J. Ditto, but for 220volts, 60 cycles, a.c...... 34.75 Abxik 1033-K. Ditto, bu t for 220 volts, d.c...... 44.75 Abxji

Io% discount in lots of J, assorted ARTHUR H. THOMAS COMPANY RETAIL — WHOLESALE— EXPORT LABORATORY APPARATUS AND REAGENTS WEST WASHINGTON SQUARE, PHILADELPHIA, U. S. A. Cable Address, “Balance,” Philadelphia INDUSTRIAL a n d ENGINEERING CHEMISTRY

ANALYTICAL EDITION

PUBLISHED BY THE AMERICAN CHEMICAL SOCIETY • WALTER J. MURPHY, EDITOR

Determination of the Precision of Analytical Control Methods

RAYMOND F. MORAN, Westvaco Chlorine Products Corporation, South Charleston, W. Va.

For the intelligent control of plant opera­ of uncertainty) can then be calculated tions and product quality, it is essential within which routine analyses may be that the precision of the analytical control guaranteed. The average of duplicate methods be quantitatively known. The use determinations made at the same time does of statistical reasoning based on the stand­ not result in as much improvement as may ard deviation has been found applicable. be theoretically calculated, evidently be­ The analytical method is first tested under cause the results are not truly random. ideal conditions to find the highest preci­ Normal control methods were found to have sion of which the method is capable. If this 1.5 to 2.5 as much variation under routine precision is judged high enough, the conditions as the same method under the method is then tested in routine practice for best conditions. The use of this method of a year to discover the variability under rou­ criticism has proved a valuable tool in the tine laboratory conditions. An LJJ2 (lim it author’s laboratories.

T IS a primary concept of nature that no one physical because the precision of the analytical method is not known, measurement is exact. Only those values that are ac­ There also exists a tendency on the part of many engineers, Icepted by definition are free from deviations in the last sales personnel, and even chemists to treat a single analytical significant figure. The determination of the composition result as an exact quantity and to make decisions therefrom of any sample even by the best known technique is similarly that would be unjustifiable if the significance of the result influenced by the inability to measure weights, volumes, were known. colors, chemical equilibria, etc., with exactitude. Practically After an extensive program of development and refinement all analytical methods contain enough small constant errors of analytical methods, this company was faced with the deter­ to make them somewhat empirical. It follows that some mination of how well the routine control laboratories were differences will be obtained between individual analytical following these new methods. Several years were spent in results from the same sample even if the best possible tech­ check sample work in which standard samples were run in nique and instrumentation are used. duplicate by one laboratory and attempts were made to check When the conditions that obtain in commercial control these results in another laboratory. Although considerable laboratories are considered, the above concept becomes much information was obtained, a great many disputable differ­ more important, since the variations between personal ana­ ences arose that could not be traced to any assignable cause. lytical techniques, solutions, apparatus, and surrounding One reason for this disagreement was found to be the use of conditions are certain to influence the precision of an ana­ the “average deviation” as a criterion, since any variations lytical method to a considerable extent. This is true even if greater than the criterion were treated as poor analyses. the method is followed exactly as written; a departure from In commercial analytical practice speed is often essential standard instructions or error in judgment would give further for good control, even if accuracy and precision are sacrificed. deviations from the truth. It is often expedient to employ nontechnical personnel as Since these variations are known to exist, they are a con­ analysts or testers for routine determinations. Any method stant threat to good commercial operation or product quality for determining the precision of routine results must be able until they are quantitatively determined. An attempt to to evaluate these personnel factors. control a plant operation within 0.05 per cent of a standard The procedure described in this article was developed in the value with a method precise to but =*=0.5 per cent would ob­ summer of 1940 and has been used throughout the company viously fail, yet similar situations frequently occur in industry with satisfaction since that time. 361 362 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

such as the use of standard samples, synthetic knowns, checks T a b l e I. Precision' of M ethod under B est Conditions with standard methods, etc., and if the limit of uncertainty Method: SC-11-3 of 6-27-41 for specific gravity of carbon tctrachloride by pycnomcter) of the average (described below) bracketed the known result, Specific Gravity the method was said to be “accurate” in a qualitative sense. T e st No. at 25°/4° C. d d* (X 10»*) 1 1.5849 -0.00016 256 2 1.5851 +0.00004 16 Limit of Uncertainty under the Best Conditions 3 1.5853 + 0.00024 576 4 1.5853 + 0.00024 576 (££/i) 5 1.5850 -0.00006 . 36 6 1.5849 -0.00016 256 The highest precision that can be expccted from an analyti­ 7 1.5847 - 0 .0 0 0 3 6 1296 8 1.5851 +0.00004 16 cal method is the precision as shown by the best available 9 1.5852 + 0 .0 0 0 1 4 196 technician working under the most closely controlled condi­ 10 1.5851 + 0.00004 16 Av. >=* X\ « 1.58506 , 2d* » 3240 X 10l° tions. The following procedure was prescribed as standard for obtaining a measure of this precision: JEÍ = 0.00018 *= on T 10 Prepare a synthetic mixture that exactly reproduces the com­ mercial material and contains a known amount of the desired in­ 3(0.00018) L U aw - 1.58506 * 1.58-189 to 1.58523 gredient, or select a representative homogeneous sample of the Vw commercial material. Store enough material for at least fifty analyses in containers that will prevent any change in composition 3(0.00018) LU i 0.00059 0.923 for at least one year. Select an analyst who is well acquainted with the method, pref­ erably the one who developed the method. Prepare fresh reagents, standardize all solutions, and calibrate Prior Concepts all apparatus before making the determinations. Run ten analyses as closely together as possible under the most A number of investigators (2,3,5,6) have indicated that the favorable conditions. Calculate all results to one more place than variations of an analytical method may be treated by statisti­ is generally reported. _ Calculate the arithmetical average, X t, of the ten analyses. cal calculations. Power (6) found that over 100 determina­ Calculate the standard deviation of the results by the use of the tions had to be made before the variability of a micromethod following equation: for carbon could be accurately estimated. Informative as /zd* such an extensive investigation might be, such a study in a trio \ 10 commercial laboratory would be difficult to justify for eco­ nomic reasons. The two routines given below use 10 samples where o-io = standard deviation from the average shown by the and 24 samples to obtain the precision and the results have 10 results been satisfactory from the author’s point of view. Al­ d = individual deviation of each result from the average though some of the calculations have the surface appearance Calculate the limit of uncertainty of the average, L U a,-., by the of complexity, the most complicated operation is the extrac­ following equation (1): tion of square roots. The criterion by which the precision of the method is to be LUa,. = Xl * 3

appeared satisfactory. The LUi oi tests 1, 3, 4, and 5 were Commercial Application low enough for the required closeness of control. Test 2, a determination by a modified Volhard method for chlorides, Because the study is founded on statistical reasoning from indicated an LUi that was 2.7 times as large as the LU\ and the comparatively small number of samples dictated by eco­ was considered too variable for good control. This method nomic reasoning, the results are open to some variations in was re-examined for possible improvement. Tests 6 and 7 an absolute sense. However, in practice this objection has were obtained by a refined Engler distillation technique, but been found academic, since the study has resulted in defi­ the results were found to be influenced by seasonal varia­ nite precision characteristics close enough for purposes of tions and superheating of vapors in the flask. These methods refinement of methods and techniques and to discover within were also improved to eliminate these assignable causes. what limits routine analytical results are significant. The

T a b l e III. Precision'TJeterm inations op R epresentative Commercial A nalytical M ethods Improvement No. Sam ple Analysis LU i L U t L U t/L U i Determinations 1 50% caustic soda NaOH, % ±0.078 ± 0 .1 4 7 1.9 1.2 2 50% caustic soda N aC l, % ± 0 .0 1 0 2 ± 0 .0 2 8 0 2.7 1.0 3 50% caustic soda Fe, p. p. m. ± 0 .7 7 ± 1 .0 0 1.3 1.1 4 Carbon tetrachloride Sp. gr. at 25°/4° C. ±0.0 0 0 5 9 ± 0 .0 0 0 6 8 1.15 1.05 5 Carbon bisulfide Nonvolatile residue, % ±0.0004 ±0.0008 2.0 1.0 6 Trichloroethylene Boiling range, ° C. ± 0 11 ± 0 .3 3 3 .0 1.07 7 Trichloroethylene First drop, ° C. ± 0 .0 5 ± 0 .3 0 6 .0 1.07

In the study of some 40 basic analytical methods during procedure described is offered as a tool for such investigation the last two years, the following relationships were found and as a standard method for describing precision. from the LU»/LUi ratio: The use of standard samples in the control laboratory has the additional advantages of providing periodical checks Ratios of 1.0 to 1.5 indicate that the variations result mainly on personal techniques and for the education of new labora­ from limitations of instrumentation. tory personnel. Since the basis of comparison is fair, the Ratios of 1.5 to 2.5 indicate normal relationship between analy­ ses made under the best conditions by one man and analyses made individual technician is more likely to cooperate than if he is under routine laboratory conditions. expected to duplicate exactly a single analysis from some out­ Ratios over 2.5 generally indicate considerable personal and side source. seasonal variations that may be reduced without modification of These techniques have been in use throughout this com­ the basic analytical method. pany for over two years and have formed the first common Duplicate determinations run at the same time are generally not truly random and the averaging of such results does not im­ bases for describing and comparing precision. They have prove the precision appreciably. The only justification for run­ led to the discovery and subsequent correction of many varia­ ning analyses in duplicate is the additional safeguard afforded tions previously not suspected and the quantitative evalua­ against an outright mistake, such as an error in calculation or an tion of other deviations that were known to exist. By the erroneous weighing. judicious use of L[/2 figures, the men having supervision of plant operation and shipping know within what limits routine Most commercial analyses are made on products of ap­ analyses are to be trusted. Some similar procedure could proximately constant composition, so that a measure of pre­ well be adopted as standard by industrial analytical chemists cision may be expressed in terms of the actual percentage to provide a mutual basis of understanding when describing variations on the original sample basis. Whenever several the precision of analytical methods. widely different levels of composition exist, it has been found desirable to make complete studies at each level to discover Acknowledgment whether or not the extent of the variations is proportional The writer gratefully acknowledges the assistance of Dwight to the amount of ingredient. Williams in the development of the described procedures and the preparation of this paper, and the permission of Westvaco Procedure for Decomposable Samples Chlorine Products Corporation for publication. Several cases have been encountered where the year’s Literature Cited study was rendered impossible because of the instability of the (1) Am. Soc. Testing Materials, "1933 Manual on Presentation of samples or attack on the containers. In this case either one Data", second printing, 1937. of two techniques has been used to approximate the true (2) Benedetti-Piehler, A. A., I n d . E n o . C h e m ., A n a l . E d ., 8, 373 LU2: (1936). (3) Crumpler, T. B., and Yoe, J. H., “Chemical Computations and Errors", New York, John Wiley & Sons, 1910. 1. If the sample decomposes at a constant rate, the results (4) International Critical Tables, 1st ed., Vol. 3, p. 28, Now York, may be plotted against time and an average curve drawn through McGraw-Hill Publishing Co., 1928. the points. The deviations of the individual results from this (5) Power, F. W., Ind. Eno. Chem., A nal. Ed., 11, 660 (1939). curve are used for the calculation. (6) Power, F. W., “Application of Modern Statistical Methods in 2. If the sample decomposes in an inconstant manner or at­ Chemical Analysis”, presented before Division of Analytical tacks its container appreciably, or if an approximate determina­ and Micro Chemistry, American Chem ical Society, Memphis, tion of the routine precision is desired rapidly, a speedup of the Tenn. technique may be used by making the L U i tests during one day (7) Shewhart, “Economic Control of Quality of Manufactured Prod­ by as many routine analysts as possible. This gives only an ucts", New York, D. Van Nostrand Co., 1931. approximation of the true LU i, because variations encountered during a yearly study may not be observed. It does yield an P r e s e n t e d before the Division of Analytical and Micro Chemistry at the approximation of personal differences. lOoth Meeting of the A m e r ic a n C h e m ic a l S o c i e t y , Detroit, Mich. Turbidimetric Determination of Small Amounts of Chlorides

E. N. LUCE, E. C. DENICE, AND F. E. AKERLUND The Dow Chcmical Company, Midland, Mich.

A turbidimetric method, based on the Tyn­ advantages of reference standards. The dall effect, is given for the determination of method is rapid, requiring less than 15 small amounts of chloride. The method is minutes’ working time per determination, essentially one in which the nephelometric and is readily adaptable to routine work. method has been adapted to the Hcllige The accuracy is comparable with that ob­ turbidimeter in order to eliminate the dis­ tained with the nephelometer.

HE first accurate measurements on silver chloride sus­ beam of light appears to the observer as a circular spot in the pensions were made nephelometrically by Richards and center of the Tyndall effect of the illuminated liquid, which is TWells (S). Lamb, Carleton, and Meldrum (2) later modified seen as lighter or darker than the Tyndall light, depending on the earlier procedure for the preparation of the suspensions, the size of the opening of a precision slit. By matching the and the method which they adapted to routine analysis has brightness of the two fields, the apparatus may be calibrated been put into general use (6). Kolthoff and Yutzy (i) made over a complete range of chloride concentrations. The con­ a complete investigation of this method and a systematic struction and operation of this instrument for measuring the study of the various factors affecting the results of nephelo­ turbidity of barium sulfate suspensions have been given in metric chloride determinations. From the results obtained complete detail by Sheen, Kahler, and Ross (4). by these and other workers, a tried and tested nephelometric method has been evolved. R eagents The nephelometer was designed originally to surmount the Stock chloride solution, 4.12 grams of c. p. sodium chloride per difficulties incurred in gravimetric and volumetric methods liter, in distilled water. Diluted chloride solution, 10 ml. of the stock solution diluted for small quantities of precipitate. When applied to small to 1 liter with distilled water. This solution contains 25 mg. of amounts of chloride, the ncphelometric method is admittedly accurate; but the preparation of fresh standards for each determination has been found inconvenient and time-con- suming. Certain permanent, artificial standards such as / kieselguhr, ground or etched glass, etc., have been used, but / only with limited success. Since the advent of the nephelom­ / eter, micro and semimicro techniques have been developed; / and principally because of the difficulties involved in the prep­ / aration of standards, low chlorides have most often been de­ termined by either microgravimetric or microvolumetric / methods. Both methods are time-consuming and require V considerable equipment and special technique; as a result t/ they are not readily adaptable to rapid routine work. / It is the purpose of this paper to show how the nephelo­ / metric method may be used with the Hellige turbidimeter to / give a rapid and reliable method for the determination of small amounts of chloride. The method is accurate, and / simple enough to be used by one unskilled in the use of elabo­ / rate and expensive equipment. It is rapid, requiring less than / / / 15 minutes’ working time per determination. A standard S' curve (Figure 1), covering the full range of the instrument, / may be prepared in a few hours, and eliminates the need of / reference standards. The instrument is easy to operate, and / V B unique inasmuch as it combines the principles of the turbidime­ / X ter and the nephelometer. Its range is greater than that of y the turbidimeter and equal to that of the nephelometer, and / / / its use does not necessitate the preparation of fresh standards y' for each determination. This greatly simplifies the proce­ A dure by eliminating possible errors due to faulty preparation C of individual standards, and also reduces the working time to / f S less than one-half that required for the nephelometer. / The turbidimeter used in this investigation compares a 0 10 10 30 40 SO «O 10 80 90 beam of light with the Tyndall effect produced from a lateral Dial Readings illumination of the specimen by the same light source. The F ig u r e 1. T y p ic a l C a l ib r a t io n C u r v e s 365 366 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 chloride per liter; 1 ml. diluted to 50 ml., as in the procedure, found that dial settings may easily be made to within ± 1 gives a solution corresponding to 0.5 p. p. m. of chloride. Nitric acid-silver nitrate reagent. À solution approximately unit, by either the same or different operators. This limits 0.2 N in nitric acid and 0.01 N in silver nitrate is prepared by the maximum deviation to less than 3 per cent for all concen­ diluting 12.5 ml. of concentrated nitric acid to 1 liter and adding trations of chloride, which is comparable with the accuracy 1.70 gram s of silver nitrate. obtained with the nephelometer. Absolute ethanol. (Standard curves have also been prepared using 95 per cent and Formula 30 alcohols.) Kolthoff and Yutzy {1) determined the effect of certain A blank determination must be made in order to eliminate the foreign ions on the magnitude of opalescence of silver chloride possibility of chloride contamination from the reagents. suspensions. They found that different ions have different effects, and these changes should be taken into account w'hen preparing standard curves. If the instrument is to be used for routine work, standard curves should be prepared by taking a T a b l e I . S tandardization ó f I n s t r u m e n t given volume of the diluted standard chloride solution, adding (20-mm. optical cell) an amount of purified electrolyte equivalent to that in the C hloride Solution No filter ^Milk filter Gray filter sample to be investigated, diluting to 20 ml., and then follow­ P. p. m. M l. ing the procedure given above. In this way accurate curves 5 10 72-73 may be prepared for each compound in which the chloride 4 8 62-63 3 6 51-52 content is being determined. In many instances, however, 2 4 36-37 80-82 chloride concentrations of 0.005 per cent or less are of interest 1 2 20-21 45-46 0 .7 5 1 .5 15-16 35-36 to an accuracy of only one significant figure, and a special 0 .5 0 1.0 12 26 70 0 .2 5 0 .5 7 16 48 reference curve is unnecessary. B lank on The volumes of the reagents used must be the same as those reagents 8 21 given in the procedure, because the alcohol serves to decrease the intensity of the opalescence as well as to stabilize the sus­ pension. The final 10 ml. of alcohol which are added have no effect on the suspension other than dilution, but are added to Procedure adjust the solution to the proper depth in the 20-mm. optical Transfer the sample to a 50-ml. volumetric flask, adjust the pH cell. Other factors which affect the opalescence are heating, with either 1 N nitric acid or 1 N sodium hydroxide until the solu­ time of standing, mixing, and relative concentrations. Lamb tion is just neutral to phenolphthalein, and add distilled water and co-workers (2) found that the intensity approached a until the volume approximates 20 =*= 1 ml. Add 20 ml. of ab­ solute ethanol and then, dropwise from a pipet, 5 ml. of the nitric maximum upon heating the suspension for 30 minutes at acid-silver nitrate reagent while swirling the contents of the flask. 40 ° C., and at that point it remained stable for an hour. Higher Make up to the mark with absolute ethanol and shake. If any concentrations are stable for about 0.5 hour. The suspension initial turbidity is found before the addition of the nitric acid- should be mixed just before it is placed in the water bath and silver nitrate reagent, the sample must be rejected, because tur­ bidity at this point indicates contamination. just before it is read, as continuous mixing will cause the silver Place the volumetric flask in a water bath at 40° C. for 30 chloride to coagulate. minutes, and then cool the sample rapidly to room temperature. At present this instrument is being used in conjunction Pour into a 20-mm. optical cell (a manufacturer’s accessory for with the lamp combustion method for the routine analysis of the turbidimeter), stir in an additional 10 ml. of absolute ethanol, and determine the turbidity as soon as the bubbles cease to form organic compounds for small amounts of chloride and sulfur. in the cell. The sample should be read within 30 minutes after It may also be used to determine the chloride content of wash it has been cooled to room temperature. The reading is com­ liquors, and for the routine checking of many inorganic com­ pared with the standard curve, and the amount of chloride cal­ pounds which contain traces of chloride. culated. The limited solubility of certain compounds in 1 to 1 water- alcohol mixtures makes the use of acetic acid solutions neces­ Calculation sary. Good results may be obtained when acetic acid is sub­ t> , c r,, p. p. m. of Cl (from the curve) X 0.005 stituted for alcohol, and the standard curves may be prepared Per cent of Cl = — ------=-r r~-j------i— :------— weight of sample, in grams by the same procedure as given above.

D iscussion The standard curves were prepared from values obtained T a b l e I I . L im i t s o p A c c u r a c y o f I n s t r u m e n t ■Filters when known volumes of the standard chloride solution were C hloride N one M ilk G ray treated in the same way as the sample. The dial readings P . p. m. P . p. m. P . p. m. P . p. in. were taken and these values plotted. Table I gives the results 5 ± 0 .1 1 4 ± 0 .1 0 obtained, and Figure 1 shows a set of curves obtained in a 3 ± 0 .0 8 typical calibration. Curve A represents the complete range 2 ±0.06 ±0.04 1 ±0.0-1 of the instrument, 0 to 250 micrograms of chloride, and was 0 .7 5 * * • • ± 0 .0 4 ± 0:01 0 .5 0 ± 0.01 obtained without the use of a filter. Suspensions of silver 0 .2 5 ± 0.01 chloride made by the above procedure, which contain more than 250 micrograms of chloride, are too opaque to be read accurately with the instrument. Sample sizes taken must be such that the upper limit of the chloride concentration is not Literature Cited exceeded. Curve B was obtained by using the milk-glass filter, and curve C with the gray glass filter (both filters sup­ (1) Kolthoff, I. M., and Yutzy, H., J . Am. Chem. Soc., 55, 1915 (1933). plied as manufacturer’s accessories). From the curves it fol­ (2) Lamb, A. B., Carleton, P. W., and Meldrum, W. B., Ibid., 42, lows that suspensions of less than 2 p. p. m. can be read more 251 (1920). accurately on curve B than on curve A. Curve C may be (3) Richards, T. W., and Wells, R. C., Am. Chem. J., 31, 235 (1904). used for concentrations of less than 1 p. p. m. (4) Sheen, R. T ., Kahler, H. L., and Ross, E. M., I n d . E ng. C h e m ., A n a l . E d., 7, 262 (1935). The accuracy to which dial readings may be reproduced (5) Yoe, J. H., “Photometric Chemical Analysis”, Vol. II, p. 137, determines, in part, the accuracy of the method. It has been “Nephelometry”, New York, John Wiley & Sons, 1929. Color Index Light-Colored Petroleum Products

I. M. DILLER, 218 Linden Blvd., Brooklyn, N. Y.

J . C . DEAN, R. J. D eGRAY, a n d J. W. WILSON, JR. Socony-Vacuuni Oil Co., Inc., General Laboratories, Brooklyn, N. Y.

The applicability of the system tentatively known I. C. I. data may be calculated from the color in­ as “photoelectric color” to the evaluation of light- dex if desired, although this is usually unnecessary colorcd petroleum products has been demon­ in petroleum technology. strated, thus making possible a continuous color The Saybolt chromometer has been shown to be scale for all petroleum products. unreliable, and to be importantly affected by the This color index consists of two parameters which surface tension, refractive index, and specific dis­ are read directly on a special photoelectric instru­ persion of the oil being tested. ment. The two parameters completely designate This color index may be converted to Saybolt the color of materials such as oils in terms which color by the establishment of suitable curves for are of direct technological significance and which the types of product in question. The conversion express directly the appearance values. is prepared statistically to allow for the unrelia­ Light-colored oils are read with a violet filter in­ bility of the Saybolt method. stead of North Sky and red and converted to this The required colorimeter can be adjusted during color index by a suitable equation, so that no manufacture to the specified standard, so that the change in sample depth is required. For light readings obtained are independent of one’s par­ oils, the second parameter is zero. ticular instrument.

HE color of petroleum products is of importance in the colors of dark- and light-colored products. Several their processing and marketing. Hitherto, for its attempts of this nature have been made prior to and since Tmeasurement, the petroleum industry has employed several the development of the Saybolt instrument, leading to the systems, all of which are visual, and depend upon matching development of the Stammer colorimeter (17, 25), methods the color of the sample with that of a standard glass. Com­ of using the Lovibond glasses (6, 16, IS, SO, SI), and color mittee D-2 of the A. S. T. M. on Petroleum Products and Lu­ systems known as “true color” (S3, S/h 26, 28), and “optical bricants specifies two such methods, the Union colorimeter density” (IS). (Ą) for lubricating oils and the Saybolt chromometer (6) All these methods failed in their objective for one reason for light-colored products. The color of oils customarily or another. None of them is direct!}' related to accepted measured with the Union colorimeter has been the subject of a systems of color definition (14, 19, 22), which, with several previous paper (11), which described a system for color others, have been described by Gardner (IS). All are not designation and measurement. It is the purpose of this only subject to peculiarities of the individual instrument, paper to discuss in a similar manner the colors of oils usually but lean heavily upon the personal equation. Furthermore, determined with the Saybolt chromometer, and to propose for light-colored oils, these systems require cell depths ap­ means for handling them by the method already described preciably greater than used for darker oils, which prevents 01). use of a continuous color scale. The Stammer and the The Saybolt chromometer consists essentially of two glass Duboscq type colorimeters (15), in which the depth of oil is tubes 50 cm. (20 inches) in length, which are illuminated from the varied until a standard is matched, suffer from the same bottom by light reflected from a mirror. A prismatic optical faults, including the variations in hue and, in addition, head provides a circular field of vision, one half of which is frequent failure of Bouguer’s law to apply to petroleum illuminated by the light from the sample, and the other by light passing through the color standard. The level of the oil in the products. “True color” has certain advantages, but is sample tube is adjusted so that its color matches that of the frequently invalidated by comparison with dilutions of oils standard disk in the other tube. The height of oil is read and is of different hue, or because Beer’s and Bouguer’s laws do not converted to Saybolt chromometer color by an arbitrary table. hold for such dilutions. (According to Bouguer’s law the The instrument, invented by George Saybolt, has been brought to its present stage of standardization largely as the result of logarithm of the reciprocal transmittance is proportional to A. S. T. M. activity (1, 2, 3, 7). depth: logio(Vr) = kd.) Story and Kalichevsky (27) proposed the use of a photo­ Both the Union colorimeter and the Saybolt chromometer electric colorimeter, primarily as a substitute for the eye. are arbitrarily standardized and are not related to one This eliminated the subjective nature of color measurement another by any fundamental means. Furthermore, the of petroleum products. specifications of each of these instruments are so drawn that A new color index for the designation and determination of sufficient latitude is allowed in the selection of color disks color, previously called “photoelectric color” has been to permit an appreciable variation of results from instrument proposed (11), and the designation of the colors of lubricating to instrument. For these reasons, among many others, oils and darker colored petroleum products by means of this the industry has needed one system of color measurement index has been described. “Photoelectric color” was orig­ which will be applicable to all types of petroleum products, inally chosen for want of a better name to distinguish this and which will provide a continuous relationship between system of color designation and measurement from Lovibond, 367 368 INDUSTRIAL ÄND ENGINEERING CHEMISTRY Vol. 15, No. 6

Saybolt, or Union, but it is probable that the industry will find one which is more appropriate as this color index be­ o / comes accepted. / This color index conveys full color information concerning y the oil. In addition to presenting this information in terms y which are of immediate and direct significance in petroleum / technology, advantage is taken of the substantially invariant ./I color properties of these series of products—that is, dark- o colored petroleum products, light-colored petroleum products, / fatty oils (9), etc. By so doing, all the desired information can be given despite the use of less than three parameters, in the present instance, only one parameter. o The color index consists of two terms, the second being zero when the color is “normal” or when it is that of light- colored petroleum products. The first term is the North Sky reading (or its computed equivalent in the case of light products as shown below), and the second is the deviation / of the red reading from the normal at that North Sky reading. » / This normal is obtained from Figure 3 of another paper of this / series (11). By reporting deviation from the normal red 78 8 0 8 2 84 8 6 88 reading rather than the red reading itself, one can tell at a X - T R IS T I MULUS VALUE X IOO glance whether the oil is greenish or reddish and to what extent. For example, a color index of 90 + 3 would represent Figure 2. Conversion C hart, V iolet Readings to X a North Sky reading of 90 and a red reading of 100 and the oil is on the reddish side of normal. An index of 90 — 3 represents a North Sky reading of 90 and a red reading of consequent increases in complexity, cost, and likelihood of 94 and the oil is greenish to the extent of 3 units. When the error. The use of two instruments, as is present practice, second term is zero, it will often be omitted, as in Figure 6, a would be even more objectionable. chart for the conversion of this color index to Saybolt color. In the case of light-colored petroleum products, the eye This system expresses and measures directly what the “sees” and is influenced almost exclusively by light trans­ eye sees. It does so by measuring the equivalent of the mission of the oil in the violet range. Accordingly, magni­ energy received by the average eye looking through the fication of the reading, without change in sample depth, sample toward C illuminant, and it also measures variations can be accomplished by the use of the violet filter. Readings in hue. While it serves for darker oils, modification was of light-colored petroleum oils with this filter have a direct necessary in the case of light-colored products, to magnify relationship with those made with the North Sky filter. the readings without changing (increasing) sample depth. Thus, the disturbing effects of turbidity, bubbles, and fluorescence are not equally magnified, and no reliance need be placed upon the applicability of Bouguer’s law. One instrument, therefore, provides a continuous color scale for all petroleum products. A further advantage lies in the fact that the color of these products can be obtained with less than 10 ml. of sample, instead of the 50 ml. or more required for other methods. In the range of darker oils, no monochromatic filter could be found which was the equivalent of the North Sky filter. However, with lighter colored oils, the changes in the spectral curves are confined to the portion below 550 m/i with virtually no absorption above this wave length. Curves for these oils are smooth in shape and show greater transmittance as the wave length increases. The ultimate is a curve for a water-white oil which consists substantially of a straight line at 90 per cent transmission. As these water-white oils oxidize and darken in color, their spectral curves deviate from this line, but only in the region below 550 mju. This is illustrated by Figure 1, showing the progressive darkening of a water-white oil. Thus, the average eye, in viewing C

WAVE LENGTH, mp illuminant through these light-colored oils, is influenced only by changes in the blue and violet region, so that a meas­ F i g u r e 1. T y p ic a l S pectrophotometric C u r v e s f o r S o l v e n t s ure of the absorption below 550 m/t is all that is required. The instrument with the violet filter in position has an excellent response in this region, and thereby performs the One way to magnify the reading would be to increase same function for light-colored oils that the North Sky the sample depth appreciably above the present 1.8 cm. filter does with darker ones. Indeed, the authors’ ex­ However, this increase would be objectionable because (1) perimental work has confirmed that the relationship be­ errors due to bubbles and turbidity would be equally mag­ tween violet and North Sky filter readings in this range is nified; (2) the energy loss due to fluorescence would be in­ linear. creased; (3) Bouguer’s law would not hold rigidly, owing to A further peculiarity of light-colored petroleum products, the effect of fluorescence and other factors; and (4) it would as distinguished from darker ones, is that because of the high entail redesign of a successfully simplified instrument with degree of refining given them there is a negligible light June 15, 1943 ANALYTICAL EDITION 369 absorption at wave lengths greater than 550 m/i. Also, for same filters were used. However, since such oils have practically oils of the same color index, there is no appreciable difference identical spectral curves above 550 mn, and there are but slight differences in their hues, the response of the red filter was sub­ in the steepness of their spectral curves. lienee, only one stantially the same in all cases; so that this filter reading is not parameter suffices for a full description of the color and the required. The same reference standard used previously (11)—• red reading for determination of variations of hue is not namely, distilled water—was employed throughout, and the required. colorimeter was set by this standard to a scale reading of 100 prior to all determinations. Definitions The terminology used by the petroleum industry in dealing with light-colored products is not rigidly standardized. Therefore, for the purpose of this discussion, the following definitions have been established: /

R e f in e d O i l . Any oil which has been rendered light in color / (straw color to colorless). K e r o s e n e . Any refined oil suitable for use as an illuminant / in a wick lamp. S o l v e n t . Any refined oil of narrow boiling range suitable for solvent purposes. o U s e d O i l . Any oil suitable as a lubricant which has been subjected to service conditions. / Description of the Method o / As has been described (11), the color index of dark oils is y taken with the North Sky filter, and optionally with the red / filter. For the reasons already given, refined oils must be / / measured with the violet filter, which gives readings related / to those found with the North Sky filter, and thus provides / a continuation of the scale. This filter was so chosen that, o with paraffinic kerosene and paraffinic solvents, it gives / / exactly ten times the deflection obtained with the North r Sky filter. Thus, the equivalent North Sky reading of these o products is computed, using the equation:

40 ------L- i\j g _ 90 _j_ violet filter reading 5 0 60 70 80 9 0 100 Z-TRISTIMULUS VALUE X 100

For values up to 96.0 the North Sky filter is used, but above F ig u r e 3. C o n v e r s io n C h a r t , V io l e t R e a d in g s t o Z 96.0 the violet filter and the above equation are employed. Readings are taken to the nearest half scale division, and the first term of the color index is reported to the nearest 0.05 unit. Thus, an oil with a violet filter reading of 78.5 As has been previously emphasized, the instrument as a will have a color index of 97.85. whole, and not merely the filters alone, must be designed and manufactured to yield the standard readings. The tempera­ ture of the light source, characteristics of the photocell, T a b l e I. C o l o r D a ta o n R e f in e d O il s optical system, etc., as well as the filter determine the re­ Spectrophotometric Color Index sponse. Of particular importance to this work is the fact Oil N atu re X Y Z N . S. V that the instrument is so designed as to minimize the effects 1 Paraffinic 0.880 0.899 1.056 100.0 100.0 of variations in refractive index and fluorescence of the 2 Aromatic 0.884 0.903 1.051 102.0 101.0 3 Aromatic 0.880 0.903 1.026 100.5 96.0 sample. 4 Paraffinic 0.873 0.899 1.022 100.0 9 5 .0 Correlation with I. C. I. Values. As in the previous 5 Aromatic 0.859 0 .895 0.941 100.0 88.0 6 Paraffinic 0.856 0.892 0.938 9 9 .0 8 4 .0 investigation, the readings of the colorimeter can be cor­ 7 Paraffinic 0.832 0.898 0.801 9 8 .0 7 5 .0 8 Paraffinic 0.820 0.890 0.752 9 7 .5 7 2 .0 related directly with I. C. I. data. This was accomplished by 9 Aromatic 0.814 0.884 0.709 9 8 .0 7 2 .0 having eleven representative oils evaluated by the Electrical 10 Aromatic 0.783 0.879 0.527 9 6 .0 5 6 .0 11 Paraffinic 0.778 0.876 0.526 94.0 4 8 .0 Testing Laboratories with a Hardy spectrophotometer, and obtaining data simultaneously with the photoelectric color­ imeter. These samples represented two series of refined Since the colors of certain types of refined oils are highly oils having widely differing physical characteristics, one unstable, and are markedly affected by exposure to light, being of paraffinic and the other of aromatic nature. The it is essential that measurements be taken as quickly as spectrophotometric data were converted to tristimulus values accuracy permits. Because of the bleaching action of the by the 30 selected ordinates method described by Hardy (14). light beam, the color of the sample is lightened appreciably These values and readings taken with the two filters are in the vicinity of the focal point of this beam. The oil in shown in Table I. the sample tube should, therefore, be agitated between The conversion of North Sky filter readings to Y tristimulus check readings to disperse the bleached material. The values has already been established (11), and the conversion instrument enables measurements to be made at the rate of factor of 0.90 was used here also. This factor results from six samples per minute. the use of distilled water, which has a transmission of 90 per cent as the standard for the color index. Experimental In order to obtain conversions to X and Z, the values ob­ tained spectrally were plotted against violet filter readings Colorim eter a n d F i l t e r s . For this work the colorimeter already described (11) and based on Diller’s (8,10) was employed. as shown in Figures 2 and 3. Figure 2 is the more sensitive In order to extend the proposed system to light-colored oils, the curve mentioned in the previous paper (11) for obtaining 370 INDUSTRIAL AND EN INEERING CHEMISTRY Vol. 15, No. 6

the X tristimulus value for light-colored oils. The most representative curves for these points were straight lines. T a b l e I I I . T richromatic C oefficients It is logical to expect that such correlations exist, since, as x______y______z already shown, the only variations in the spectropliotometric Color Color Color Oil S pectral index Spectral index Spectral index curves occur at wave lengths below about 550 mp., In this 1 0 .3 1 0 0 .310 0.317 0.316 0.373 0.374 range, readings with the violet filter, as well as the Z function 2 0.3U 0.307 0.318 0.318 0.371 0.375 3 0.313 0.312 0.322 0.323 0.365 0.365 and a portion of the X function, are affected. This relationship 4 0.312 0.313 0.322 0.323 0.366 0.364 is further confirmed by Figure 4 wherein for these eleven 5 0.319 0.319 0.332 0.334 0 .349 0.347 6 0.319 0.323 0.332 0.339 0.349 0.338 representative samples X is plotted against Z, which results 7 0.329 0.332 0.355 0.353 0 .316 0.315 8 0.333 0.335 0.362 0.357 0 .305 0.308 in a straight line. Using Figures 2 and 3, X and Z tristimu­ 9 0.338 0.335 0.367 0.358 0.295 0.307 lus values were obtained from the violet filter readings. 10 0.358 0.353 0.401 0.387 0.241 0.260 11 0.357 0.366 0.402 0.401 0.241 0.233 These and the Y tristimulus values obtained as above are Mean deviation from spectral =*=0.004 given in Table II. The mean deviation of the authors’ values obtained spectrophotometrically is shown to be =*=0.012. The trichromatic coefficients were calculated from the tristimulus values, and are shown in Table III. A mean Saybolt Chromometer deviation of ±0.004 is indicated. Finally, from Y, x, and y It was advisable to establish a correlation between this the luminous transmission, dominant wave length, and purity new system and the one now in general use. However, early were obtained. These data are shown in Table IV, wherein in the work with the Saybolt instrument, it was found that mean deviations from the true values of =*=0.9 per cent in that instrument did not yield comparable results for the transmission, ±3 m/i in dominant wave length, and =*=1.5 different classes of oils and no one correlation curve could per cent in purity are found. be drawn. Dependent upon other physical characteristics, each class of oils required an individual curve. In Table V, data obtained on two series of solvents are

T a b l e I I . T r is t im u l u s V a l u e s given. These include the samples tested spectrophoto­ X X 100 Y X 100 Z X 100 metrically together with others evaluated photoelectrically. Color Color Color It is evident from this table that the Saybolt chromometer is Oil Spectral“ index t* Spectral“ index & Spectral“ index & primarily measuring purity, and only secondarily luminous 1 8 8 .0 8 8 .4 8 9 .9 9 0 .0 105.6 106.5 2 8 8 .4 8 8 .6 90.3 9 1 .8 105.1 108.0 transmission. As is indicated, the dominant wave length 3 8 8 .0 8 7 .5 90.3 90.5 102.3 102.5 remains constant throughout both series. Saybolt colors 4 87.3 87.3 89.9 90.0 102.2 101.4 5 85.9 85.8 89.5 90.0 94.1 9 3 .5 and purity are plotted graphically in Figure 5. Two curves 6 8 5 .G 8 5 .0 8 9 .2 89.1 93.8 89.0 7 8 3 .2 8 3 .0 8 9 .8 8 8 .2 80.1 7 8 .8 result which, even though slightly irregular in shape, il­ 8 82.0 82.4 89.0 87.8 75.2 7 5 .6 lustrate that basically, purity is being measured. 9 81.4 82.4 88.4 S8.2 70.9 75.6 10 78.3 78.9 87.9 86.4 52.7 58.0 11 7 7 .8 7 7 .2 87.6 84.6 52.6 49.0 Mean deviation from spectra * 1 .2 ° Computed from Hardy spectrophotometric curves. T a b l e IV . L u m in o u s T ransmission , D o m in a n t b Converted from readings of photoelectric colorimeter. W a v e L e n g t h , a n d P u r it y Luminous Dominant Transmission, % Wave Length, m/i P u rity , % Color Color Color S pectral index Spectral index Spectral index The above may be summarized as indicating that these Oil 1 89.9 9 0 .0 569 570 0 .2 0.1 color indices for refined oils may be converted to I. C. I. 2 90.3 91.8 569 560 0.8 0 .6 3 9 0 .3 90.5 569 563 2.2 2 .2 values with reasonable accuracy. This correlation is only for 4 8 9 .9 90.0 568 567 2.0 2 .6 petroleum oils, and is possible because their spectral curves 5 89.5 90.0 569 568 6.5 7.1 6 89.2 89.1 569 570 6.5 9 .6 have the same generic shape. 7 8 9 .8 8 8 .2 569 571 15.2 15.2 8 8 9 .0 8 7 .8 569 572 18.1 17.5 9 8 8 .4 8 8 .2 569 571 2 2 .3 18.0 10 8 7 .9 8 6 .4 569.5 571 3 5 .8 3 0 .5 11 8 7 .6 8 4 .6 570 572 35.5 37.5 Mean deviation ± 0 .9 % ± 3 m/j * 1 .5 %

Purity of a color may be considered as that percentage of a pure spectral color which when mixed with C illuminant of the I. C. I. system will match the color (14). Since Saybolt chromometer colors are related to purity, the instrument is substantially measuring the concentration of colored bodies of the same dominant wave length. This is proved by correlating violet filter readings with “modified Saybolt” colors, which are calculated by the following equation:

Modified Saybolt color = (20) (number of disks) depth of oil m inches This equation is based upon the validity of Bouguer’s and Beer’s laws in this range. If modified Saybolt colors of any one family of oils are plotted graphically against the logarithms of the violet filter transmissions, a linear relation­ ship is found. Each family of oils is represented by a separate X-TRISTlMULUS VALUE X IOO curve, but all are straight lines over the range in which these Figure 4. Relationship op X a n d Z laws hold. June 15, 1943 ANALYTIC L E D I T I O N 371

Obviously, therefore, some other factor is involved which affects the optical properties of the oils. In Table VI, other data obtained with the series of oils mentioned above are shown, which indicate that despite identical color indices, the Saybolt colors vary with surface tension, refractive index, and specific dispersion. In this particular instance, the variations are due entirely to differences in chemical composition. The oils studied here are all darker than those most frequently evaluated by the Saybolt chromometer, but were chosen primarily because they illustrate the dis­ crepancies which may be obtained with this instrument. Similar discrepancies exist throughout the Saybolt range, as indicated by Figure 6, although they are not so pro­ nounced.

T a b l e V I. S a y b o l t C o l o r , R e f r a c t iv e I n d e x , S p e c if ic S AYBOLT COLO*? D is p e r s io n , a n d S u r f a c e T e n s io n F ig u r e 5. R elationship o f P u r it y a n d S a y b o l t C o lo r (All samples gave violet filter readings of 70.0) Saybolt Refractive Specific Surface Oil Color Index Dispersion Tension D ynes/cm . T a b l e V . C o r r e l a t io n o f S a y b o l t C o l o r s w it h I. C. I. D a ta 1 - 4 1.3994 103 19.5 (Computed from DiUer colorimeter readings and Hardy spectrophotometric 2 - 6 1.4100 102 2 0 .5 curves) 3 - 9 1.4765 145 2 6 .8 4 - 1 3 1.4864 158 27.3 Saybolt Luminous D o m in an t 5 - 1 7 1.5010 185 2 7 .8 Oil Color Transmission Wave Length P u rity % Ï7IJ1 % A. Paraffinic 1« +30+ 89.9 569 0.2 If the optical system of the Saybolt chromometer is an­ 2 +30 90.0 579 0 .2 3 + 2 4 9 0 .0 569 1 .5 alyzed, it becomes apparent that the physical characteristics 4 “ + 2 1 8 9 .9 568 2 .0 listed in Table VI affect the results obtained. A constant 5 + 2 0 8 9 .0 569 3 .1 0 + 15 89.1 572 6 .8 source of light exists, and the oil level in one tube is so ad­ 7 + 10 89.1 570 8 .7 8“ + 8 89.2 569 6 .5 justed that the short-wave (violet) portion of the light 9 + 6 88.7 571 10.0 energy reaching the eye is substantially equal to that passing 10 0 88.2 572 12.7 11» - 3 8 9 .8 569 15.2 through the standard disk and the empty tube. The energy 12 -3 87.8 572 15.7 loss in the sample tube is substantially equal to that lost 13 -6 87.8 571 16.8 14“ -7 89.0 569 18.1 because of refraction, reflection from the side walls, and 15 - 1 0 8 7 .3 571 2 0 .0 16 - 1 9 8 5 .5 571 3 6 .0 absorption by the oil layer. Surface tension, in a sense, 17“ Too d ark 87 .6 570 3 5 .5 determines the contour of the meniscus which is formed at B. Aromatic the liquid-air interface. Specific dispersion and refractive 1“ + 30 + 90.3 569 0 .8 index are closely related to the energy loss at the interface, 2 + 2 9 9 0 .0 566 1.0 3“ +27 90.3 569 2.2 and also to the angle at which light leaves the oil surface. 4 + 2 3 9 0 .0 568 3 .8 Thus, all three determine the average angle a t which light 5 + 19 9 0 .0 568 6 .5 0“ + 18 8 9 .5 569 6 .5 from the interface is directed toward the side walls of the 7 + 15 89.1 570 10.8 8 + 9 88.2 571 16.0 tube. The more nearly the light strikes the side walls at 9“ + 5 8 8 .4 569 2 2 .3 right angles to them, the greater will be the energy loss. 10 + 3 87.8 571 20.0 11 0 87.3 571 24.5 Higher surface tensions, refractive indices, and specific 12 - 4 8 6 .9 571 27 .8 13“ -7 87.9 569.5 35 .8 14 - 8 8 6 .0 571 32 .2 15 - 1 2 8 5 .5 571 3 8 .0 ° I. C. I. data obtained from Hardy spectrophotometer curves.

However, despite the apparent relationship between Saybolt color and purity, several anomalies have been en­ countered. In one experiment, a solution of off-colored kerosene in carbon tetrachloride was prepared and found to match a standard disk at a height of 23.75 cm. (9.5 inches). Various amounts of water were introduced above this solution in the tube, and the color was measured again. Matches were obtained when the total heights were 9.5 inches in each case. Apparently then, even though the instrument measures color concentration, it indicated the same concentration when the colored bodies were in the ratio 9.5:8.5:7.5. A second experiment was conducted, wherein mixtures of re­ fined oils were prepared so that all had the same violet reading, and were closely matched visually. However, when evaluated by the Saybolt chromometer, their colors ranged from —4 to —17. Thus, even though all had the same purity, or color con­ SAY0OLT COLOR centration, based upon 1. C. I. data, their Saybolt colors indicated concentrations over a range directly proportional to the heights F i g u r e 6. Conversion Chart, Color Index to Saybolt in the tubes of from 5 to 9.13 cm. (2 to 3.625 inches). C o l o r 372 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

dispersions individually result in greater energy losses. In It will be noted that all these curves and also those of order to compensate for this added loss above the oil layer, Figure 5 exhibit points of inflection. These deviations from its depth must be reduced to decrease the loss due to absorp­ smooth curves are the results of the arbitrary table used in tion. Hence, a low Saybolt color reading results. converting depths of oil to Saybolt colors (6), and the peculiar To illustrate the effect of surface tension, a potassium di- relationship between Saybolt colors and modified Saybolt chromate solution having a Saybolt color of +10 was prepared colors. If these are plotted graphically, one against the and measured in a sample tube of the chromometer coated other, an irregular line with the same inflection points will with a thin film of highly viscous light-colored oil. A value result. These inflections will appear in any curve obtained of +15 was obtained, -but when a few drops of wetting by plotting Saybolt values against any fundamental value agent solution were added, the color was found to be +6, such as color index or purity. despite the fact that the color index was unchanged. Figure 6 was prepared by evaluating over a thousand The personal equation involved in color measurements separate samples. However, similar curves can be obtained with the Saybolt instrument also cannot be disregarded. by a simpler procedure involving the linear relationship of It has been established that there are certain individuals modified Saybolt colors and the logarithms of the violet incapable of reproducible results. These operators fre­ filter readings—for example, a more limited number of sam­ quently state that because of a different "shade” of color ples could be tested with both instruments. By computing they are unable to match it with the standard. This is not modified Saybolt colors from the heights in the sample tube, the case with color-blind observers, or those less sensitive and recording the violet transmission on the logarithmic to minor variations in hue, who are able to obtain accurate (concentration) scale, values will be obtained through which readings even with abnormal oils. straight lines may be drawn. These lines may then be A. S. T. M. Method D156-38 specifies the standard disks converted to Saybolt colors and violet filter readings, yielding as follows: curves such as those shown in Figure 6. (For any given W hole D isk H alf Disk sample, instrument, and operator, deviations from these - 0.860 to 0.865 0.888 to 0.801 curves should be no greater than one Saybolt unit.) - 0.342 to 0.350 0.327 to 0.331 - 0.367 to 0.378 0.3« to 0.350 Standardization and Reproducibility Using the graphs prepared by Hardy (14), these limits are Means for checking the setting of the instrument for equivalent to the following: the production of standard readings have been described (11). The same standard aqueous solutions were used in H alf Disk W hole Disk this study. Luminous trans­ Lum inous tra n s­ mission, % 80.0 to 80.5 mission, % 88.8 to 89.1 Since the colors of some off-test solvents and kerosenes are Dominant wave Dominant wave length, m*t 509 to 574 length, m/i 509 to 574 highly unstable, it was obviously impossible to check the P u rity , % 22.2 to 27.2 P u rity , % 12.0 to 14.5 reproducibility of the instrument on the samples after a period of storage. However, the colors of the series of oils It has been shown that paraffinic and aromatic refined listed in Table I were determined simultaneously with two oils have the same dominant wave length of about 570 m¡x, instruments to demonstrate the concordance between them which is within the limits of the disks. Since the dominant (Table VII). One instrument was used at the Electrical wave lengths of the oil colors do not match that of the disk, Testing Laboratories, and the second at the authors’ labora­ they must have been varied before the light reached the tory. By means of telephone communication, it was eye, and that variation may be caused by differences in sur­ possible to have both readings taken within 5 minutes of face tension and refractive index. Since the energy loss is each other. The differences between the results are no disproportionate for light of certain wave lengths, and greater than the limits of error previously established (11). because of the shape of the spectral curves, this results in a slight variation in dominant wave length that is easily de­ tected by operators with keen color perception. T a b l e V II. Reproducibility of R esults The effects of refractive index and surface tension are North Sky Filter Violet Filter thus shown to invalidate certain Saybolt color measurements. Oil AB AB These factors are also significant in photoelectric measure­ 1 9 9 .5 100.5 100.0 100.0 2 102.0 102.0 101.0 101.0 ments. However, in the photoelectric colorimeter used in 3 100.5 100.5 96.0 9 6 .0 this work, measurements are not made through a liquid-air 4 100.0 100.0 94.5 9 5 .5 5 100.0 100.0 88.0 88.0 interface, so that surface tension has no effect. Refractive 0 99.0 99.0 84.0 84.0 7 9 8 .0 9 7 .5 7 4 .5 7 5 .5 index also applies to the photoelectric measurement, but the 8 9 7 .5 9 7 .5 7 2 .0 7 2 .0 use of a round sample tube in conjunction with the con- 9 98.0 98.0 72.0 72.0 10 9 6 .0 9 6 .0 56.0 5 6 .0 verging-diverging light beam of Diller’s instrument (S, 10) 11 94 .0 94.0 48.0 48.0 minimizes the refractive error, because the light enters the Average deviation ±0.24 scale division. tube at substantially zero incidence. Provision is also made for reducing as far as possible the effect of fluorescence. Correlation of Color Index with Saybolt Values C onclusion Since, as stated above, other physical characteristics mark­ The photoelectric colorimeter may be used for measuring edly affect Saybolt values but not color index, no rigid rela­ colors of refined oils, thus providing a continuous scale for tionship between them can be established, and each family all petroleum products. This has been made possible by the of oils must be evaluated to fix its own correlation curve. use of a violet filter, which is directly related to the North Figure 6 shows typical correlation curves of four types of Sky filter, and which has its response in the blue-violet region oil, prepared by evaluating paraffinic solvents, aromatic in which all variations in spectral absorption occur. Read­ solvents, solutions of used oils in paraffinic solvents, and ings obtained with these two filters may be converted to highly refined paraffinic spindle oils. The divergence of these I. C. I. data if desired. curves is in accordance with the explanation already ad­ The Saybolt chromometer, customarily used for measuring vanced. the colors of these oils, has been studied and found to be June 15, 1943 ANALYTICAL EDITION 373

unreliable in that erroneous and anomalous results may be (11) Diller, I. M., DcGray, II. J., and Wilson, J. W., Jr., I n d . E n g . obtained. Variations in physical characteristics of the oils, C h e m ., A n a l . E d ., 14, 607-14 (1942). (12) Ferris, F. W., and Mcllvain, J. M„ Ibid., 6, 23-9 (1934). such as surface tension and refractive index, are probably (13) Gardner, II. A., “Physical and Chemical Examination of chiefly responsible for these errors. The color index de­ Paints, Varnishes, Lacquers and Color”, 9th ed., Washington, scribed above can be correlated with Saybolt values by the D. C., Institute of Paint and Varnish Research, 1939. use of suitable conversion curves. (14) Hardy, A. C., “Handbook of Colorimetry” , Cambridge, Mass., Technology Press, 1936. Acknowledgment (15) Heilige, F., Petroleum Z., 10, 725 (1914). (16) Herbrich, J., J. Itist. Petroleum Tech., 18, 140 (1932); Ann. Acknowledgment is made to D. B. Judd of the National chim. anal. chim. appl., 14, 193-201 (1932). Bureau of Standards, who has carefully reviewed this paper (17) Holde, D., “Examination of Hydrocarbon Oils and of Saponifi­ able Fats and Waxes”, 2nd ed. rev., New York, John Wiley & and offered suggestions. Frequent references to a private Sons, 1923. report by V. A. Kalichevsky and B. W. Story were of great (18) Inst. Petroleum Technologists, Standard Methods K. Z., P. S. 2, assistance in this work. The instrument used in this in­ P. S. 2a. vestigation is known as the Hellige-Diller photoelectric (19) Judd, D. B., J. Optical Soc. Am., 23, 359-74 (1933). (20) Lovibond, F. E., Proc. Optical Convention, 1926, I, 211-14. colorimeter, Model 405-A. (21) Lovibond, J. W., “Light and Color Theories and Their Relation to Light and Color Standardization”, London, E. and F. N. Literature Cited Spon, 1916. (1) Am. Soc. Testing Materials, Proc. Am. Soc. Testing Materials, (22) Munsell, A. H., “A Color Notation” and “A Color Atlas", 23, 352 (1923). Baltimore, Md., Munsell Color Co., 1933. (2) Ibid., 24, 524 (1924). (23) Nelson, W. L„ Oil Gas J., 37, No. 3, 74 (Juno 2, 1938). (3) Ibid., 34, 895 (1934). (24) Parsons, L. W„ and Wilson, R. E ., J. In d . E n g . Chem., 14, 269- {4) Am. Soc. Testing Materials, Standards on Petroleum Products 7S (1922). and Lubricants, A. S. T. M. Designation D155-39T. (25) Redwood, B., “Treatise on Petroleum”, 3rd ed., Part II, pp. (5) Ibid., A. S. T. M. Designation D156-38. 214-15, London, C. Griffin and Co., 1913. (G) Campbell, A., "Petroleum Refining”, 2nd ed., pp. 74-85, New (26) Rogers, T. H., Grimm, F. V., and Lemmon, N. E., Ind. Eng. York, Petroleum Ago, 1922. Chem., 18, 1G4-9 (1926). (7) Delbridge, T. G., Proc. Am. Soc. Testing Materials, 22 I, 425-9 (27) Story, B. W., and Kalichevsky, V. A., Ind. Eng. Chem., A n a l . (1922). E d ., 5, 214-17 (1933). (8) Diller, I. M., /. Biol. Chem., 115, 315-22 (193G). (28) Vinock, H., Refiner Natural Gasoline Mfr., 16, 601 (1937). (9) Diller, I. M., paper presented before tho American Oil Chem­ ists’ Society, fall meeting, 1941. P r e s e n t e d before the Division of Petroleum Chemistry at the 104th Meeting (10) Diller, I. M„ U. S. Patent 2,232,109 (Feb. 18, 1941). of th e A m e r ic a n C h e m ic a l S o c ie t y , Buffalo, N. Y.

Determination of Iodine in Tetraiodo- phenolphthalein

SAMUEL WEINER, BYRON E. LEACH1, AND MARY JANE BRATZ Paul-Lewis Laboratories, Inc., Milwaukee, Wis.

T ) UTLER and Burdette (I) in 1939 proposed a method for and 7.5, preferably 6.0. On the other hand, Kolthoff’s (4) I ) determining iodine in tetraiodophenolphthalein that starch-iodine complex will give sharp end points at pH’s as was much more rapid and accurate than the procedure of the low as 0.5. U. S. Pharmacopoeia of that time. It consisted of three Place 0.15 to 0.20 gram, accurately weighed, of the tetraiodo­ steps: (1) digestion with alkaline permanganate solution to phenolphthalein or its sodium salt in a 500-ml. Erlenmeyer flask, decompose the material and release the iodine as iodate, (2) add 10 ml. of 5 per cent sodium hydroxide solution, and when the acidification and treatment with sodium bisulfite to reduce sample has dissolved on the water bath or steam bath add 25 ml. of saturated potassium permanganate solution. Digest 45 min­ the manganese and iodine to their lower oxidation states, and utes on the water bath, cool, add 75 ml. of water and 10 ml. of 1 to (3) partial neutralization with ammonium carbonate and 1 sulfuric acid, and then add 3 M sodium bisulfite from a buret titration of the iodide with silver nitrate, using eosin or diiodo- with constant swirling of the flask until the solution is colorless. fluorescein as an adsorption indicator. Add dilute potassium permanganate solution drop by drop till a faint permanent yellow appears. Add 10 ml. of M ammonium carbonate solution and 4 drops of dilute starch solution and titrate Adsorption Indicator Methods with 0.1 N silver nitrate solution. N o t e s . The end point wiU be the disappearance of the last The clear advantages of the Butler-Burdette method led to trace of blue, green, or gray, leaving the solution and the precipi­ tate with a clear canary yellow color. If the silver iodide coagu­ its adoption by the U. S. Pharmacopoeia XII (5) with certain lates, the end point is best observed on the precipitate; if, as modifications. The U. S. P. XII method is almost identical sometimes happens, it fails to coagulate, the end point is observed with one that has been successfully used in this laboratory in the body of the solution. The end point is usually very sharp. since October, 1941, but in which starch-iodine is used as the From 10 to 18 ml. of sodium bisulfite will be needed to reduce the iodate and permanganate. After one has observed the adsorption indicator. amount needed in a few titrations, it is preferable to add a slight If eosin or diiodofluorescein is used as adsorption indicator excess at once from a graduate cylinder, so as to avoid the loss by in the titration of the iodide, the amounts of acid and alkali volatilization of the iodine during the reduction from iodate to added after the digestion with permanganate must be ex­ iodide: 15 ml. of 3 M sodium bisulfite would be a convenient am ount. actly such that the pH at the time of titration is between 4.5 On rare occasions, the silver iodide fails to coagulate and the 1 Present address, University of Illinois, Urbana, 111. starch complex gives a very vague end point. 374 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

The procedure was checked against potassium iodide and potassium iodate, both reagent grade, and dried 1 hour at T a b l e I I . D etermination o p I o d in e 105° C. W eight of Thiosulfate Used Sam ple Volum e N orm ality I Found I Calcd. KI I found, 76.3 *±=0.1; calculated, 76.4^ Gram M l. % % KIOj I found, 59.1 =±=0.1; calculated, 5 9 .3 9 0.1488 3 6.15 0.1000 51.4 51.2 0.1514 3 6.75 0.1000 51.3 The influence of the organic material was studied by trying 0.1505 3 6.50 0.1000 5 1 .3 this procedure on mixtures of 0.1000 gram each of phenol- phthalein and the potassium halogen salt. T a b l e I I I . D etermination o p I o d in e KI I found, 70.6 0.3; calculated, 76.4 °, KIOi I found,59.6 0.4; calculated, 59.3 > W eight of Thiosulfate Used Sample Volume N orm ality I Found I Calcd. A comparison of the results obtained by using starch and by Gram M l. %% using eosin as indicators in the determination of iodine in KI 0.1000 36.10 0.1000 76.35 76.4 sodium tetraiodophenolphthalein is given in Table I. 0.1000 36.05 0.1000 7 6.25 K lOa 0.1000 26.25 0.1064 59.1 59.3 0.1000 26.25 0.1064 59.1 0.1000 26.25 0.1064 59.1 T a b l e I. Samples and Iodine C ontents Found P ro d u ct N o. Weight I (Eosin) Weight I (Starch) Gram % Gram % 14 0.2334 53.3 0.2463 52.8 cator. One must shake the solution vigorously in a stoppered 0.2173 5 3 .0 0.2039 5 2 .7 Erlenmeyer flask after each addition of thiosulfate as the end 0.2074 53.0 14a 0.1978 5 5 .5 0.2024 5 5 .6 point is approached. The end point, of course, is the disappear­ 0.2006 5 5 .7 0.2100 55.4 ance of the pink color in the carbon tetrachloride. 0.1962 5 5 .2 0.2136 5 5 .7 15 0.1446 52.0 0.1478 52.7 0.1529 51.3 0.1570 52.4 0.2051 52.5 0.2047 52.9 The procedure was checked against o-iodobenzoic acid 0.2051 52.6 0.2040 52.6 (Eastman) which melted at 161.5-162.5° C. and at 163— 163.5° C. (corrected). Titration with 0.1 N sodium hy­ droxide and phenolphthalein gave an apparent equivalent weight of 247.8 =*= 0.6; the calculated equivalent weight is Reduction Method 248.0. Because of the stability of the I—C bond in iodo- Clark and Jones (2) in August, 1942, published a method benzoic acid it was necessary to replace the 45-minute diges­ for determining iodine in organic compounds, based on tion in alkaline permanganate by 8 hours of refiuxing over a Groak’s method (S). Clark and Jones did not report on the free flame (Table II). applicability of their method to tetraiodophenolphthalein, The procedure was also checked against potassium iodide but a method similar to it has been in use by the authors for and iodate mixed with an equal weight of phenolphthalein this purpose since October, 1941. Their method consists of U. S. P. (Table III). To effect complete oxidation of the four steps: (1) boiling with alkaline permanganate to de­ phenolphthalein it was necessary to use 45 ml. of saturated compose the sample, if it is of an acid-insoluble alkali-soluble permanganate and to carry out the digestion for 1 hour. compound; (2) reduction of the permanganate with nitrite in Part of the agreement between the observed and calculated acid solution; (3) destruction of the excess nitrite with sulf­ values is due to a balancing of errors; if the “blank” of 0.30 amic acid; and (4) titration of the iodate with potassium ml. is subtracted, the observed values are uniformly low. iodide, thiosulfate, and starch. As an example of the reproducibility of this method one As used by the authors it is as follows: can consider the results of the six routine determinations made on and prior to February 23, 1943, on samples of com­ Place 0.15 gram, accurately weighed, of the sample in a 500-ml. mercial tetraiodophenolphthalein sodium of various origins Erlenmeyer flask, add 15 ml. of 5 per cent sodium hydroxide, let it digest on the steam bath till completely dissolved, add 25 ml. of (Table IV). saturated potassium permanganate solution, let it digest 45 min­ utes, remove, add immediately 10 ml. of glacial acetic acid, and then add 5 per cent sodium nitrite solution drop by drop from a buret with rapid swirling until the red color of the permanganate T a b l e IV. Reproducibility has been replaced by the brown color of the manganese dioxide. Add 5 ml. of concentrated sulfuric acid slowly and with swirling to Thiosulfate Used prevent loss by spattering. Once more add 5 per cent sodium Sam ple Weight Volume Normality I Found nitrite solution drop by drop, and when the solution is colorless Gram M l. % and free of manganese dioxide add 10 ml. of 5 per cent sulfamic A 0.1500 32.25 0.1064 48.4 0.1500 32.35 0.1064 48.5 acid solution and cool under the tap. Add 10 ml. of 33 per B 0.1503 35.00 0.1064 52.4 cent sodium hydroxide solution (5 grams of sodium hydroxide 0.1500 35.05 0.1064 52.6 dissolved in 10 ml. of water), cool again, add 10 ml. of M potassium C 0.0500 11.40 0.1064 51.3 0.0500 11.30 0.1064 50.9 iodide solution, and titrate with 0.1 N thiosulfate and starch. D 0.0500 11.30 0.1064 49.9 0.0500 11.10 0.1064 50.9 6 equivalents of thiosulfate = E 0.1500 32.00 0.1064 48.0 1 mole of iodine = 127 grams of iodine 0.1500 3 1.85 0.1064 4 7 .8 F 0.1500 33.10 0.1064 49.7 0.1500 33.20 0.1064 4 9 .8 N o t e s . Sometimes a few particles of manganese dioxide will persist after the solution has been cleared by the sodium nitrite. Warming the solution will dissolve these; this should be done before the sulfamic acid is added. The 5 ml. of concentrated sulfuric acid can be replaced by 10 Literature Cited ml. of 1 to 1 sulfuric acid. The nitrate ion will not oxidize the iodide ion rapidly enough to (1) Butler and Burdette, Ind. Eng. Chem., A nal. Ed., 11, 237 (1939). interfere, if the final pH is above 2 and the solution is cool. (2) Clark and Jones, J . Assoc. O ffic ia l Agr. Chcm., 25, 755 (1942). Occasionally the solution becomes yellow after addition of the (3) Groak, Biochem. Z., 270, 291 (1934). sodium hydroxide. This usually does not affect the final result. (4) Kolthoff and Furman, “Volumetric Analysis”, Vol. II, p. 236, A less convenient but often more precise method of titration is New York, John Wiley & Sons, 1929. to use 5 ml. of carbon tetrachloride instead of starch as the indi­ (5) U. S. Pharmacopoeia X II, p. 251, 1942. Determination of Monoalkyl Ethers of Etliylene Glycol

[IIAROLD W. WERNER AND JAMES L. MITCHELL Division of Industrial Hygiene, National Institute of Health, Bethesda, Md.

HEN toxicological work on the monoalkyl ethers of reactions consumed 15.7, 13.8, 19.9, and 26 equivalents of ethylene glycol was undertaken (6) no suitable oxygen, and they required 15, 15, 240, and 120 minutes, re­ methodW for determining these compounds was available. spectively, at water bath temperature, 90° to 100° C. As is Thus an immediate need in toxicological studies as well as the illustrated by the curves, heating beyond these minimal likelihood of a future need in attempts to evaluate and prevent periods did not significantly increase the extent of oxidation. any possible hazards connected with the use of these solvents From the number of equivalents consumed in these reac­ in industry led to a study of methods of estimation. tions it appears that, in all cases, the glycol portion of the Preliminary experiments with the methyl, ethyl, n-propyl, molecule is completely oxidized to carbon dioxide and water. and n-butyl derivatives, using specimens for which constants This oxidation theoretically requires 10 oxygen equivalents. have been described (£?), indicated that these derivatives Considering the equivalents used in excess of 10 for each com­ could be oxidized quantitatively with potassium dichromate. pound, it appears that the methyl radical is oxidized to carbon (Methyl Cellosolve, Cellosolve, and Butyl Cellosolve are dioxide and water, the ethyl to water and acetic acid, and the trade names for the methyl, ethyl, and butyl derivatives.) propyl and butyl to carbon dioxide, water, and acetic acid. These experiments indicated, however, that progress of the These oxidations of alkyl radicals are in agreement with the oxidations was influenced by acid concentration, period of results of Polonovski (5) for propionic and butyric acids, heating, order of adding reagents, and quantities of reagents Nicloux (4) for ethyl alcohol, and Chapman and Thorp (1) used. Thus, the first part of the experimental work neces­ for the methyl radical in esters. sarily is concerned with the definition of conditions yielding With an oxidizing mixture containing 55 per cent sulfuric precise quantitative reactions, while the second part is an acid, reactions requiring 20.4 instead of the theoretical 20 investigation of the suitability of these reactions for estimat­ equivalents for the propyl derivative, occur within 60 minutes ing various concentrations in water and in air. (fifth curve of Figure 1). Additional heating results in further oxidation, and at the end of 480 minutes 21.1 equivalents have Experimental Method been consumed. Reactions of this type are less suited to determinative work than the ones previously described; In general, the method of conducting oxidations and determin­ nevertheless, there are instances where a saving in time ing amounts of oxygen consumed is similar to a modification of justifies the use of less accurate methods. the Nicloux (4) method for ethyl alcohol described by Muehl- berger (S). With the glycol ethers, however, the results were The theoretical considerations and results already presented found to be most consistent when standard aqueous solutions of suggest that theoretical factors, based on reactions requiring glycol derivatives (1, 2, or 3 ml.) were added to cooled mixtures of 16, 14, 20, and 26 equivalents, respectively, for the methyl, 5 ml. of 0.33 N potassium dichromate and an amount of concen­ ethyl, propyl, and butyl derivatives, should be satisfactory for trated sulfuric acid equal to the combined volume of potassium dichromate and sample solutions. This 50 per cent proportion of determining these glycol ethers. These factors as well as acid was used with all compounds. Other concentrations of conditions producing corresponding reactions are shown in sulfuric acid were investigated, but with the exception of 55 per Table I. cent acid employed in determining the propyl., derivative, the The suitability of these factors and conditions for deter­ other concentrations offer no advantages. The oxidations of the aqueous samples were conducted in 25 X mining various samples in aqueous solution is indicated by 200 mm. test tubes which were heated in a boiling water bath for recovery experiments summarized in Table II. definite periods. At the termination of heat­ ing, the tubes were cooled and the contents were transferred into 500-ml. wide-mouthed Erlenmeyer flasks. Volumes were made up to approximately 300 ml., about 3 grams of potassium iodide were added, and the liberated iodine was titrated with 0.05 N sodium thio- sulfate. Starch solution was added, tow'ard the end of the titration, as an indicator.

R esults and Discussion. Curves illus­ trating the effect of the period of heating on the extent of oxidations are shown in Figure 1. Each point on these curves is based on four or more determinations, and in each determination the sample (2 ml. of standard solution) was of such size that approximately a 100 per cent excess of potassium dichromate, based on complete reactions, was present. From the four curves for oxidation with 50 per cent acid mixtures it appears that F ig u r e 1. E f f e c t o f P e r io d o f H e a t in g in a W a t e r B a t h o n C o m p l e t io n stoichiometrical reactions occur. In the order o f R e a c t io n s Complete reactions are theoretical ones requiring 16, 14, 20, and 26 oxygen equivalents, re­ methyl, ethyl, propyl, and butyl, these spectively, for the methyl, ethyl, n-propyl, and n-butyl derivatives. 376 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

Elkins and co-workers (8) have recently described a method T a b l e I. F a c t o r s a n d C o n d it io n s U s e d i n D e t e r m i n i n g for the estimation of Methyl Cellosolve (ethylene glycol C e l i .o s o l v e s monomethyl ether) in the presence of certain other solvents. F actor, Reaction Heating 0.33 N As a result of several determinations using 0.5 N potassium D erivative M ixture Period E^CrjOi dichromate and 33 per cent sulfuric acid and gently refluxing % acid M in. for 4 hours, they obtained an oxygen equivalent of 15.5 for M ethyl 50 30 1.584 E th y l 50 30 2.141 Methyl Cellosolve. In a single similar determination of n-Propyl 50 240 1.736 Cellosolve (ethylene glycol monoethyl ether) they ob­ n-Propyl 55 GO 1.736 n-B utyl 50 120 1.514 tained a value of 13.2. The first equivalent is 3.1 per cent and the second is 5.7 per cent below theoretical. Thus, it appears that when these solvents alone are being determined, T a b l e II. D etermination o f C e l i .o s o l v e s i n A q u e o u s stronger acid solutions than those employed by Elkins can be S o l u t io n used advantageously. With 50 per cent acid, reactions for No. of 0.33 N Deter­ KîCrjO'r Average the methyl and ethyl glycol ethers are complete within 15 D erivative Sam ple m inations Used Deviation Recovery minutes at boiling water bath temperature, while similar Mg. M l. %% reactions require over 4 hours’ refluxing with 33 per cent 50 Per Cent Sulfuric Acid sulfuric acid media. M ethyl 2.013 4 . 1.263 0 .2 99.4 4.026 4 2.491 0.1 98.2 6.039 4 3.732 0 .2 9 8 .0 E th y l 2.513 4 1.181 0.8 100.7 5.026 4 2.309 0 .3 98.4 7.539 4 3.492 0.5 99.3 T a b l e III. D etermination o f C e l l o s o l v e V a p o h s «-Propyl 2.005 4 1.152 0 .3 100.0 4.010 4 2.292 0.7 99.3 Average Concentration 6.015 4 3.338 0 .5 96.3 Derivatives C alculated Found Difference n-Butyl 1.000 C 0 .660 0 .8 100.0 M a./l. M g./I. % 2.000 6 1.318 0.7 99.8 4.000 G 2.642 0.5 100.0 Methyl 0.93 0.88 - 5 . 4 6.000 G 3.879 1 .0 97.9 2 .8 8 2 .9 2 + 1.4 •1.03 4 .1 2 + 2 .2 55 Per Cent Sulfuric Acid 5.19 5.15 - 0 . 8 6.59 6.54 - 0 . 7 n-Propyl 2 .005 4 1.211 0.7 104.6 6.74 6 .1 9 - 8 . 1 4.010 4 2.351 0.1 101.8 7 .0 3 7 .1 6 + 1 .8 6.015 4 3.374 0 .3 97.4 10.50 9.40 - 1 0 . 4 21.08 20.36 - 3 . 4 E th y l 1.30 1.34 - 1 . 5 4 .1 5 4 .3 2 + 4 .1 5.82 5.95 + 2.2 In general, with 50 per cent acid mixtures, recoveries are 6.40 6.64 + 3 .8 6.74 7 .1 4 + 5.9 not significantly affected by the size of the sample used, al­ S . 14 7 .9 2 - 2 . 7 though there is an indication that large samples and minimal 10.30 11.27 + 9 .4 n-Propyl 1.82 2 .0 0 + 9 .9 heating periods give low recoveries in determinations of the 5.54 5.81 + 4 .9 propyl derivative. 7 .0 4 7 .1 5 + 1 .6 9.94 10.11 With the 55 per cent acid mixtures, recoveries of propyl + 1.7 n-B utyl 1.87 1.94 + 3.6 glycol ether show considerable spread. They are 2.6 per cent 2.71 2 .6 6 - 1 . 8 3 .2 2 3 .2 8 + 1.9 low from large and 4.6 per cent high from small samples. 3.72 3.43 - 7 . 8 4 .4 6 4.47 + 0 .2 Vapors The results presented indicated that oxidation with potas­ sium dichromate is satisfactory for determining the glycol Sum m ary ethers in water solution; therefore, the suitability of the Conditions for the oxidimetric determination of four method for vapors was investigated. monoalkyl glycol ethers have been investigated, and stoichio- Standard vapor-air mixtures were prepared by a method de­ metrical reactions are described. In general, these reactions scribed by the writers (

R esults and Discussion. The agreement between aver­ Literature Cited age concentrations found in different chemical determinations made at intervals and concentrations calculated from weight (1) Chapman, E. T., and Thorp, W., J. Chem. Soc., 19, 477 (1866). (2) Elkins, H. B., Storlazzi, E. D., and Hammond, J. W., J. Ind. loss, during several hours, in the saturating bubblers and total Ilyg. Toxicol., 24, 229 (1942). air flow used is shown in Table III. In view of the well- (3) McNally, W. D., “Toxicology", p. 64S, Chicago, Industrial known difficulties in preparing and sampling standard vapor- Medicine, 1937. air mixtures, the agreement shown in Table III is as good as (4) Nidoux, M., Compt. rend. soc. biol., 48, 841 (1S96). (5) Polonovski, M., Compt. rend., 178, 576 (1924). could be expected. Apparently the reactions employed are (6) Werner, H. W., Mitchell, J. L., Miller, J. W., and von Oettingen, about as suitable for determining gaseous as aqueous samples. W. F„ J. Ind. Hyg. Toxicol., 25, 157 (1943). Collection and Estimation of Traces of Formaldehyde in Air

F. H . GOLDM AN A N D HERMAN YAGODA Division of Industrial Hygiene, National Institute of Health, Bethesda, Md.

ORMALDEHYDE is one of the common contaminants composed in media containing a low concentration of sodium of the industrial environment. Its presence in the air bicarbonate, whereas the formaldehyde-bisulfite is disso­ Fconstitutes a potential health hazard. Maximal allowable ciated only when the solution is made alkaline with sodium concentrations for its presence in air continue to be formulated carbonate. and this, of course, necessitates accurate methods for its estimation. Method of Analysis In general, small amounts of gaseous contaminants in the R e a g e n t s . 1 per cent sodium bisulfite, NaHSOj. atmosphere are most efficiently collected for analysis by 0.1 N iodine solution. This reagent need not be standardized. 0.01 N iodine, prepared by dilution of 0.1 N iodine solution. bubbling through a medium that reacts with the vapor to This solution must be standardized daily by titration against form a nonvolatile compound. This fixation mechanism 0.05 N sodium thiosulfate. prevents the expulsion of the gas during the sampling by the 1 per cent starch solution. aerating action of the unabsorbed constituents. As is well Sodium carbonate buffer prepared by dissolving 80 grams of sodium carbonate in about 500 ml. of water, adding 20 ml. of known, formaldehyde gas reacts rapidly with aqueous solu­ glacial acetic acid, and diluting to 1 liter. tions of sodium bisulfite to form the nonvolatile sodium M e t h o d . The air is drawn through a midget impinger (8) formaldehyde-bisulfite compound: containing 10 ml. of 1 per cent sodium bisulfite solution at a rate of 1 to 3 liters per minute. At the termination of the sampling OH the bisulfite solution and washings are transferred to a 300-ml. cone flask. When sampling at 28 liters per minute the large im­ IICHO + NaHSOa — I h C ^ pinger (8) is filled with 100 ml. of bisulfite solution and a 10-ml. aliquot is removed for analysis. The solution is titrated with the Xx Osdtsia 0.1 N iodine solution to a dark blue end point, using 1 ml. of starch as an indicator. Excess iodine is destroyed by the addition This derivative is stable in slightly acid and neutral solu­ of 1 or 2 drops of 0.05 N sodium thiosulfate, and the solution is tions and can be decomposed only when the solution is made equilibrated to a faint blue color by the dropwise addition of the distinctly alkaline. These properties permit the direct 0.01 Ar iodine solution. The excess inorganic bisulfite is thereby completely oxidized to sulfate, and the solution is now ready for estimation of small amounts of formaldehyde by destroying assay of the formaldehyde-bisulfite compound. excess bisulfite with iodine at pH 6 to 7, and subsequently Twenty-five milliliters of the sodium carbonate reagent are liberating the sulfite combined as sulfoxylate by proper ad­ added and the liberated sulfite is titrated with 0.01 N iodine solu­ justment of the hydrogen-ion concentration. The direct tion to a faint blue end point. If acetone is present, the solution is treated with 2 ml. of 5 per cent sodium bicarbonate after the titration of this dissociated bisulfite with a standard iodine destruction of the free inorganic sulfite in the collection medium. solution affords an accurate measure of the formaldehyde The sulfite liberated by the bicarbonate is removed by addition of collected, which is independent of change in the bisulfite-ion 0.01 N iodine solution and the pH is finally adjusted by means of concentration caused by atmospheric oxidation of the col­ the sodium carbonate solution for the final titration of the form­ aldehyde-bisulfite. When the volume of solution used for the lection medium. titration is less than 1 ml. it is desirable to run a blank on 10 ml. Analytical methods for different aldehydes based on the of 1 per cent sodium bisulfite. This blank correction does not formation of sulfoxylate have been described by Clausen exceed 0.10 ml. and is negligible in most assays. One milliliter of (8), Donnally (4), Edwards (5), and Friedemann (6). These 0.01 N iodine solution is equivalent to 0.15 mg. of formaldehyde. methods were devised chiefly for the determination of the The accuracy of the method was tested by adding measured volumes of a standardized formaldehyde solution to 10-ml. por­ acetaldehyde-bisulfite derivative formed in the assay of tions of 1 per cent sodium sulfite and analyzing the mixture by the lactic acid. The principal difficulty in the titration of the described procedure. The standard formaldehyde solution pre­ formaldehyde sulfoxylate is its slow rate of decomposition in pared by diluting the 37 per cent commercial solution to a con­ slightly alkaline solutions. While readily decomposed by centration of about 1 mg. per ml. was assayed by the method of Romijn (7). The results shown in Table I exhibit good agree­ caustic solutions, the latter interfere with the iodometric ment between the formaldehyde taken and recovered. determination of the liberated sulfite. Investigation revealed that the formaldehyde complex could be hydrolyzed in­ By using a microburet reading to 0.01 ml. and delivering stantaneously at room temperature by the addition of a droplets of about 0.03 ml. it is possible to establish the two solution of sodium carbonate and sodium acetate, and that end points with a maximum error of 0.0G ml. and this intro­ the resulting solution could be titrated to a stable stoichio­ duces a possible error of 1 per cent in the volumetric esti­ metric end point. mation of 1 mg. of formaldehyde. With smaller quantities The formation of bisulfite addition compounds is a char­ of formaldehyde, the percentage error increases rapidly, and acteristic reaction of aldehydes and ketones. A method it is desirable to sample a sufficient volume of air to provide based on this reaction is not subject to interference by sulfur at least 0.5 mg. of formaldehyde for the analysis. In the dioxide, alcohol, the halogen gases, and volatile organic acids case of atmospheres that contain formaldehyde concentra­ which interfere with the determination of formaldehyde using tions of about 20 parts per million, approximately 20 liters Schiff’s reagent (1). The method, as proposed, assays total of air must be sampled for each 10 ml. of bisulfite solution in aldehyde vapor. In the examination of plant atmospheres the impinger. for toxic constituents, the identity of the vapors is usually The formation of the formaldehyde-bisulfite is not inter­ available to the analyst from knowledge of the manufactur­ fered with by the concomitant presence of small amounts of ing process. methyl alcohol, bromine, or acetic acid, which may on occa­ On rare occasions where the formaldehyde is accompanied sion be collected in the sodium bisulfite solution. This was by acetone, a separation can be effected by means of sodium tested by adding the substances shown in Table II to 10 bicarbonate. The acetone derivative is completely de­ ml. of 1 per cent sodium bisulfite containing 1.03 mg. of 378 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

in the formaldehyde vapor concentration. The 95 per cent T a b l e I. A n a l y s e s o f F ormaldehyde S o l u t io n s absorption in 1 per cent sodium bisulfite is in good agreement Formaldehyde Taken Formaldehyde Recovered Difference with the value of 93.6 per cent reported by Barnes and Mg. Mg. Mg. Speicher (2) using 1.25 per cent potassium hydroxide as the 0.021 0.024 + 0 .0 0 3 0.053 0.058 + 0 .0 0 5 collection medium, and the same air flow of 28 liters per 0.106 0.110 + 0 .0 0 4 minute. 0.216 0.220 + 0 .0 0 4 0.426 0.432 + 0.006 0.638 0.625 -0.013 Sum m ary 1.064 1.085 +0.021 1.276 1.270 -0.006 1.49 1.46 - 0 . 0 3 Formaldehyde vapors can be collected quantitatively in a 2.13 2.04 -0.09 single impinger tube containing 1 per cent sodium bisulfite. 3.19 3.01 - 0 . 1 8 5.32 5.24 • - 0 . 0 8 The resultant solution is analyzed volumetrically by titrating the formaldehyde-bisulfite by means of a standard iodine solution. The simplicity of the procedure provides a con­ T a b l e I I. E f f e c t o f O t h e r V a p o r s o n E s t im a t io n o f venient method for field work. F ormaldehyde Form aldehyde T aken Foreign Vapor------» Formaldehyde Recovered M g. Mg. Mg. 1.03 N one 1.02 1.03 N one 1.04 T a b l e III. C o m p a r a t iv e E f f ic ie n c y o f S o d iu m B is u l f it e 1.03 20 M pthyl alcohol 1.05 a n d W a t e r a s M e d ia f o r t h e C o l l e c t io n o f F ormaldehyde 1.03 50 Acetic acid 1.05 1.03 10 B rom ine 1.02 1.03 50 Bromine 1.00 Air Flow Concentration 1% NaHSOa Water 1.03 10 Acetone® 1.01 L ./m in . P . p. m. %% a Acetone-bisulfite dissociated by means of sodium bicarbonate, prior to 1 hydrolysis of formaldehyde sulfoxylate. 7 100 77 1 21 97 79 1 101 98 87 3 7 96 68 3 20 97 74 3 78 98 81 formaldehyde and analyzing the mixtures by the proposed 28 7 95 69 28 14 95 72 method. 28 41 95 76 In the absence of other aldehydes the use of bisulfite for the collection medium affords a simple and rapid method for the estimation of the formaldehyde. Literature Cited Efficiency of Bisulfite as a Collection Medium (1) Ackerbauer, C. F., and Lebowich, R. J., J. Lab. Clin. Med.i 28, The retentive power of sodium bisulfite was studied by 372 (1942). blowing air through dilute solutions of formaldehyde and (2) Barnes, E . C., and Speicher, H . W ., J. Ind. Hyg. Tox., 24, 10 (1912), passing the vapors, freed from spray by means of a glass wool (3) Clausen, S. W., J. Biol. Chem., 52, 263 (1922). plug, through a train of two impingers. Both impingers con­ (4) Donnally, L . H ., I n d . E n g . Chem., A n a l. E d ., 5, 91 (1933). tained sodium bisulfite, and in each case a similar experiment (5) Edwards, H . T., J . Biol. Chemi., 125, 571 (1938). was conducted in which the vapor was first bubbled through (6) Friedemann, T . E ., Contonio, M ., and Shaffer, P. A., Ibid., 73, 335 (1927). water and then through a second impinger containing 1 per (7) Romijn, G., Z. anal. Chem., 36, 19 (1897). cent sodium bisulfite solution. After analysis of the four (8) Jacobs, M. B., “Analytical Chemistry of Industrial Poisons, solutions it was possible to compare the relative collecting Hazards and Solvents”, New York, Interscience Publishers efficiencies of water and the bisulfite solution under similar 1941. conditions of air flow and formaldehyde vapor concentration. Preliminary tests in which known amounts of dilute form­ aldehyde solutions were completely volatilized by a stream of dry warm air and collected in bisulfite revealed that the vapor was recovered quantitatively, and that at a flow of 3 liters per minute, 97 per cent was caught in the first impinger. Furfural On the basis of these experiments, it is safe to calculate the efficiency as the ratio of the weight of formaldehyde collected ( Correspondence) in the first impinger to the total weight found in the train of collecting tubes. Reproducible rates of air flow were secured S i r : The article on furfural by Ira J. Duncan [In d . E n g . in the case of the slow speed measurements with the aid of a C h e m ., A n a l . E d ., 15, 162 (1943)] brought to mind some work I large Mariotte flask. It was observed that formaldehyde once did on preparation of furfural from ground corncobs. I also vapor was strongly absorbed by rubber tubing. To avoid observed the decomposition of furfural by hot dilute acid and loss of formaldehyde from this source, impingers were as­ reasoned that considerable furfural is probably destroyed during sembled for the testing work that have the entrance and exit the hydrolysis of the pentosans before it is liberated from the hot tubes in a straight line with abutting joints. This permits acid. If the concentration of furfural in the acid can be reduced, glass-to-glass contact between adjoining impingers, using yield should increase. little or no exposed rubber surface. One way to accomplish a reduction in furfural concentration in The results of these experiments (Table III) show that at the reaction mixture is by the addition of sodium chloride. slow rates of air flow (1 to 3 liters per minute) about 98 per When large quantities of salt were added to the acid, it was found cent of the formaldehyde is collected in a single tube of bi­ that the yield of furfural could be increased from about 10 per sulfite solution, and that even at a flow of 28 liters per minute cent to about 15 per cent of the weight of the cobs, indicating a the 95 per cent recovery makes the use of a single impinger considerable destruction of furfural in ordinary hydrolysis. Per­ tube permissible. haps some use could be made of this technique in analytical de­ Under the same conditions water retains between 69 and terminations. 87 per cent of the formaldehyde vapor. The results also V a n d e r v e e r V o o r h e e s Standard Oil Co. (Indiana) reveal that the collection efficiency increases with increase Chicago, I1L Determination of Small Amounts of Tellurium in High-Lead and Tin-Base Alloys

RALPH A. SCHAEFER The Cleveland Graphite Bronze Co., Cleveland, Ohio

A simple, rapid, and accurate volumetric method for the quantitative determination of tellurium in high-lead and tin-base babbitts has been developed. The analytical de­ tails for quantitative determination of small quantities of tellurium arc given.

N THE course of an extensive research program on the im­ cool, add 80 ml. of water and 10 ml. of concentrated hydrochloric provement of tin- and lead-base babbitts for bearing pur­ acid, stir the solution well to dissolve the tellurium completely, I filter into a 400-ml. beaker, and wash. Add 80 ml. of concen­ poses, the addition of tellurium was studied thoroughly. trated hydrochloric acid, dilute the solution to 300 ml. with water, The analytical determination of these tellurium additions was and pass a rapid stream of sulfur dioxide through the hot solu­ very important. The various analytical methods recorded in tion; then proceed as in the case of tin-base alloys. the literature were tried and found to be either too long and tedious or of insufficient accuracy (1, 8-9). D iscussion The method developed in this laboratory is based upon dis­ solving the alloy in concentrated hydrochloric acid with the In order to check the accuracy of this method chemically aid of several grams of potassium chlorate, diluting the solu­ pure tellurium dioxide was purchased from Eimer and Amend. tion with water to approximately an acid concentration of 10 With this material as a source of the tellurium the following per cent by volume, and precipitating the tellurium from solu­ data were obtained by using aliquot parts: tion by saturating with sulfur dioxide. The solution is fil­ tered and the precipitate is dissolved in dilute nitric acid Tellurium Tellurium Found (2 to 1), which oxidizes the tellurium only to the valence of Taken Titration method Gravimetric method 4 (2). Potassium iodide is added in excess and the liberated Gram Gram Gram 0 .0080 0.0080 0.0081 iodine is titrated with sodium thiosulfate. 0 .0112 0.0112 0.0114 The following equations are involved: 0.0160 0.0159 0.0162 0.0189 0.0189 0.0191 HjTeOj + 4KI + 4HNO, = Te + 2h + 4KNO, + 3H.0

Is + 2NaJSJOs = 2NaI + Na,S«06 The addition of various amounts of metallic tellurium to standard alloys also gave very satisfactory results with the titration method. Gravimetric results show a greater devia­ Procedure tion, probably due to contamination. These samples were Tellurium -Bearing Tin-Base Alloys. Weigh a 5.00-gram taken in the same manner as in the case of the pure tellurium sample of the tin-base alloy containing the tellurium into a 400- dioxide. ml. beaker. Add approximately 5 grams of potassium chlorate The following table shows the results when various amounts and 50 ml. of concentrated hydrochloric acid (22° B 6 .) and heat to of tellurium were added to a standard tin-base babbitt, alloy the boiling point; if the solution is not complete add more potas­ sium chlorate and repeat the process. No. 54a, purchased from the National Bureau of Standards: For every volume of solution add 2 volumes of water and boil for at least 5 minutes. Add 10 ml. of a 1 per cent aqueous solu­ tion of hydrazine sulfate to the sample containing the tellurium Tellurium Tellurium Found Taken Titration method Gravimetric method and heat to the boiling point. Precipitate the tellurium by bub­ Gram Gram Gram bling the sulfur dioxide through this hot solution for at least 15 minutes. After precipitation is complete allow to settle for at 0.0100 0.0101 0.0103 0.0186 0.0184 0.0190 least one hour, filter on a platinum cone with suction, using 0.0216 0.0216 0.0221 W hatm an 4111 filter paper or a similar grade of paper and wash the 0.0268 0.0268 0.0273 tellurium with hot water until the chloride ions are completely removed. Dissolve the precipitate from the filter paper with suc­ cessive portions of warm dilute nitric acid (2 to 1), to the extent Literature Cited of 25 ml. Wash and dilute the filtrate to approximately 200 ml. Add about 2 grams of urea to eliminate any small amount of (1) Bersin, T., Z. atial. Chem., 91, 170 (1923). nitrous acid which may be present, cool to room temperature, and (2) Guthier, T., Z. anorg. Chem., 32, 31 (1902). add potassium iodide in excess. After 1 minute titrate the liber­ (3) Hillebrand, W. F., and Lundell, G. E. F., "Applied Inorganic ated iodine with sodium thiosulfate solution, using starch as an Analysis”, pp. 266-7, New York, John Wiley & Sons, 1929. indicator. The end point is permanent for several minutes. (4) Jilek, A., and Kata, J., Collection Czechoslov. Chem. Commun., 6, In alloys containing less than 1 per cent of tellurium the use of 398-407 (1934). 0.01 N sodium thiosulfate gives an accuracy of 0.02 per cent. (5) Lenker, V., and Wakefield, H., J. Am. Chem. Soc., 45, 1423-5 1 Iowever, for higher percentages of tellurium satisfactory results (1923). can be obtained with more concentrated sodium thiosulfate solu­ (6) Norris, J. F., and Fay, H„ Ibid., 20, 278-83 (1898). tions. (7) Schrenck, W., and Browning, B., Ibid., 48, 139-40 (1926). Tellurium -B earing Lead-Base A lloys. Weigh a 5.00-gram (8) Scott, W. W., and Furman, N. H ., "Standard Methods of Chem­ sample of the lead-base alloy containing tellurium into a 500-ml. ical Analysis”, 5th ed., Vol. 1, p. 793, New York, D. Yan Noa- flask, add 30 ml. of concentrated sulfuric acid (66° B6.) and 10 trand Co., 1939. grams of potassium hydrogen sulfate, and heat until all the metal (9) Willard, H. H., and Young, F., J. Am. Chem. Soc., 42, 2982-92 is dissolved and the precipitated salts appear white. Allow to (1921). Mixed Solvent Extraction Notes on an Analytical Method

JAMES II. WIEGAND, Traverse City, Mich.

Hunter recently presented a graphical method of (2) attack for mixed solvent extraction problems in­ volving four components. The equilateral tetra­ Equations 1 and 2 together define the line of intersection, RT. hedron used by Hunter is modified in this articIc It is desired to find the intersection of line RT with the straight to a rectangular type of tetrahedron, three of line joining corresponding points on the two equilibrium curves. Except for the case of RT parallel to the x axis, the intersection whose edges coincide with the'three axes of a can be found by intersecting plane rectangular coordinate system, and three of whose faecs are isosceles right triangles. With rectangu­ L (3) lar coordinates, the methods of analytical geome­ V* + r* = 1 try are readily applied. The problem given by with the planes defined by Equations i and 2. For the final Ilunter is solved analytically and the solution solution, corresponding values of x, y*, and z* can then be presented in the form of generalized equations in obtained from the equilibrium data of the systems involved. Equations 1, 2, and 3 are solved simultaneously by first terms of values obtainable from the characteristics eliminating x between Equations 1 and 2, then solving the result of the system and from the statement of the with Equation 3 to obtain z, giving problem. The numerical example given by Hunter (fe - fa) - (k, - k¡/a)y* is solved by this method. The results compare (4) well with the graphical method, and there is a - fa) - (fa - k¡/a)y*/z* large saving in the time required for solution of the problem. This value of z is then placedpla in Equation 3 and the equation solved for y, giving'g y* - (fa - h)y*/z* RECENT article by Hunter (1) presented a graphical y = (!-) (5) A method of computation for four-component solvent ex­ kl/a)y*/z* traction problems. This article shows the application of (j - k') ~ {kl analytical geometry methods to such a problem, with large These values for y and z are then placed in Equation 2 and savings in the time and the difficulty of solution. the equation, is solved for x, giving x _ (1/ca - l)y* - (1/c - 1) - (1/a - 1 )y*/z* . . Four-Component System on Rectangular —fafa (fa/c — fa) — (ki — kl/a)y*/z* Coordinates The system of representation used by Hunter consisted of a regular equilateral tetrahedron, but such a system does not lend itself readily to the use of analytical geometry methods. Kinney (3) pointed out that the equilateral method of representation for three components can easily be converted to a right-triangle system, using the ordinary x-y coordinates. In such a case, the third component is defined by the other two, since the total of the weight fractions must equal unity. This method is easily extended to four components, using the ordinary x-y-z coordinates of analyti­ cal geometry. In this case, three edges of the tetrahedron coincide with the three axes, and x, y, and z become the weight fractions of three of the components. The fourth component is thus defined, since the total of the weight fractions of the four components must equal unity. Figure 1 shows the problem given by Hunter for the system acetone-chloroform-water-acetic acid, where x, y, and z represent the weight fractions of water, acetic acid, and acetone, respectively. For convenience in comparison with the Hunter article, the apexes of the tetrahedron are labeled A, B, S\, and S2. The line RT is that given by Hunter and is the intersection of two planes, each of which is deter­ mined by the tie line in one face and the opposite apex.

The plane containing the tie line in the x-y plane and the z apex has the equation

- + y + Z- = 1 1 fa a 1 1 ( )

Similarly, the plane containing the tie line in the x-z plane F ig u r e 1. F o u r -C o m p o n e n t S y s t e m R e p r e s e n t e d o n and the y apex is represented by Equation 2. R e c t a n g u l a r C o o r d in a t e s 380 June 15, 1943 ANALYTICAL EDITION 381

Also

OM _ zm — z0 (14) OSi zq so th a t

Z il Zo (15) RmXs + 1 From the similarity of right triangles in Figure 2,

C Zo h kj — xo (16) Combining Equations 13, 15, and 16 and simplifying, gives

Zü/fcl RiaXs(ki — 1) hi (17) Similarly

wt. of S wt. of jS wt. of M R m X w t. of B w t. of M wt. of B “ 1 — z u (18) and from Figure 3, F ig u r e 2. R e l a t io n o p T i e L in e L o c a t io n in x- z P l a n e wt. of S OB zo t o R e l a t iv e A m o u n t s o p M ix t u r e a n d S o l v e n t (19) wt. of B OS x s — Xo so

_ Knowledge of y*, z*, hi, k2, a, and c allows solution of Equa­ R m xs tions 4, 5, and 6. Tarasenkov (£?) has pointed out that the tie Xo = (20) lines of many systems intersect at a point on the extension of a R m + 1 — zm side of the phase triangle. For such a case, Also from Figure 3, OB _ ___ (7) 2 /0 (21) fci — x i k \ — x2 OS ys - I/o z* 1 _ Z*2 (8) fo — Xj JCi — Xi R mVs y» (22) where subscripts 1 and 2 on x, y, and z refer to the two layers, R m + 1 — zm the B-rich and the

= R m Xs (12) . 1 — X0 and RmXs So = (13) Rmxs + 1 382 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

For systems whose tie lines do not intersect in a point on the base line, the particular tie lines specified by the problem in question can be used to determine the values of fa and kt for solution of the equations. Tbe solution is also valid for fa and k2 both positive, both negative, or one negative and the other positive.

T a b l e I. C o m p a r is o n o p V a l u e s O b t a in e d by t h e S e v e r a l M e t h o d s

Phase X y z Upper layer Experimental 0.574 0.316 0.075 G raphical 0 .578 0.318 0.066 Analytical 0.565 0.323 0.069 2. Lower layer Experimental 0 .028 0.096 0.203 G raphical 0 .033 0.100 0.202 A nalytical 0.024 0.095 0.207

E xam ple Taking the example used by Hunter (1): A mixture, consist­ ing of 27.8 per cent by weight of acetone and 72.2 per cent of chloroform is to be batch-extracted at 25° C. in a single stage with a mixed solvent, consisting of 58.5 per cent of water and 41.5 per cent of acetic acid. The ratio of the weight of mixed solvent to the weight of treated mixture is to be 0.93. Then R m is 0.93, xs is 0.585, and zm is 0.278. From the experimental data given by Hunter, three values F ig u r e 4. E q u il ib r iu m R elationships f o r I llustrative of ki may be calculated from Equation 7: —0.239, —0.273, and P r o b l e m f r o m D a ta o f H u n t e r ( i ) f o r S y s t e m W a t e r - —0.264, with a mean of —0.259. Three values of k2 m ay be A c e t o n e -A c e t ic A c id -C h l o r o f o r m calculated from Equation 8: 1.44, 1.56, and 2.04, with a mean of 1.68. The third of these values of fe differs from the mean by 3.57 times the standard deviation for the three values. On a plot of Equation 9, for the data from which these three values Table I shows a comparison of values calculated by this of k-i were calculated, the data giving the value of fe as 2.04 gave a point differing widely from other determinations on this method with both the graphical and experimental values system (5), while the two groups giving ki values of 1.44 and of Hunter (1). The comparison is satisfactory. 1.56 gave good agreement. For these reasons, the value of 2.04 for ki has been discarded, and a mean of 1.50 used in this example for fcj. Solvent Refining of Oils Substituting these values in Equation 25 gives The method described above for four-component systems 0.93(1 - 0.585) is suitable for use in calculations on the solvent refining of 0.103 0.93 - 1 - 0.278 + 0.93(0.585/0.259) oils by two immiscible solvents, provided that the assump­ tions of Hunter (1) apply. These assumptions are (1) that and the line of intersection, RT, in Figure 1 is the desired tie line 0.278 X 1-50 0.235 and (2) that the boundary line running from z = z* to y = y* 0.93(1.50 - 1)0.585 + 1.50 is a straight line. A test of these assumptions is to calculate the compositions on the basis that the assumptions are valid, Substituting the known and calculated quantities in Equations 4, 5, and 6 gives and then check against the experimental results. Two changes in the method are necessary for solvent- 15.7y* - 3A 0y*/z* - 1.27 (26) refining calculations. The first is to change the x and y 6.64 - 4.02y7z* axes from weight fractions to volume fractions of the solvents. 6.64y* - 1.759y*/z* This also requires all calculations to be on the basis of volumes (27) V = 6.64 - 4.02y*/z* rather than weights. The second change is to place an aux­ iliary scale on the z axis graduated in terms of the physical 1.759 - 4.02y* (28) property, such as density, refractive index, or viscosity 6.64 - 4.02y*/z* gravity constant, which is used to characterize the stock to be treated. This method of plotting has been described Figure 4 shows data plotted from Hunter (1) for the by Hunter and Nash (2) and by Kurtz (4). Values of this systems water-acetic acid-chloroform and water-acetone- property as a function of volume fraction of solvent Si are chloroform. The values of y* versus x shown in Figure 4 plotted in the x-z plane. The scale used for the physical are for the first of these systems, and the values of z* versus property should be such as to give a suitably sized equilibrium x are for the latter. Values of the ratio y*/z* are also plotted curve (2). Once plotted, values of z* for use in the calcula­ versus x in Figure 4 for each of these systems to aid in the tions should not be read from the property scale, but rather calculations. from the normal z scale which is graduated from 0 to 1. Equation 26 is solved as follows: From this point on, the calculations are identical with those A value of x is assumed, the corresponding values of y* and described above for four-component systems. y*/z* are read from Figure 4, and these values are used to calculate x from Equation 26. This process is repeated until the calculated value of x equals the assumed value. The value Conclusions of * so obtained is used to find the corresponding values of y* and y*/z* from Figure 4 to use in solving Equations 27 and 28 The methods of analytical geometry have been applied for y and z. This process is repeated for the other layer. to the solution of four-component system problems with June 15, 1943 ANALYTICAL EDITION 383 satisfactory results. To facilitate use of the method, the £ = mixed solvent equilateral tetrahedron has been distorted to a tetrahedron Si, S 2 = components of mixed solvent a = intercept of tie line in the x -y plane with y axis of three isosceles right-angle triangular faces, which tetra­ b = weight fraction of component B in the treated mixture hedron is placed on an rectangular coordinate system. c = intercept of tie line in the x -z plane with the z axis Equations.have been derived in terms of general properties k = intercept of tie line on x axis n = constant, Equation 9 of liquid-liquid equilibrium systems. The introduction of x = weight fraction of component Si the use of the intersection point of the tie lines on the ex­ x s = weight fraction of component Si in mixed solvent S tension of the base line of the phase triangle allows generalized y = weight fraction of component S i equations to be derived for the solution of varied problems on ys = weight fraction of component S i in mixed solvent S y* = value of y from the equilibrium curve in the x -y plane the same system. Additional equations have been derived at a given value of x which relate the above equations to the conditions of a z = weight fraction of component A specific problem. The resultant saving in time by this ana­ zm = weight fraction of component A in treated mixture M lytical method is large over the graphical method of Hunter z* = value of z from the equilibrium curve in the x -z plane a t (1), especially if a number of problems must be worked on a given value of x the same system. The accuracy of the analytical method has been shown to be satisfactory. Its application to solvent refining has also been shown. Literature Cited (1) Hunter, T. G,, Ind. Eng. Chem., 34, 963-72 (1942). Nomenclature (2) Hunter, T. G., and Nash, A. W., Ibid., 27, 836-45 (1935). A , B — components of treated mixture (3) Kinney, G. F „ Ibid., 34, 1102-3 (1942). C = constant, Equation 9 (4) Kurtz, S. S., Jr., Ibid., 27, 845-6 (1935). M = treated mixture (5) Othmer, D. F„ and Tobias, P. E„ Ibid., 34, 693-6 (1942). R m = ratio of weight of mixed solvent to weight of treated (6) Tarasenkov, D. N., J . Phys. Chem. (U. S. S. R.), 14, 589-97 mixture (1940).

Determination of Halogens in Organic Compounds

ROBERT R. UMHOEFER The General Electric Co., Pittsfield, Mass.

HE decomposition of organic halogen compounds by the elude increasing the amount of sodium used (1, IS), and sub­ sodium-alcohol method (12) has advantages over most stituting higher boiling alcohols for the ethyl alcohol. The otherT procedures in the simplicity of both the technique and ap­ alcohols that have been used are butyl, amyl (9), isoamyl (4), paratus required. However, the usefulness of the method has and benzyl (6). Mixtures of a high boiling, inert solvent and suffered not only from inherent limitations, such as the in­ an alcohol, such as kerosene and amyl alcohol (8), have been ability to decompose effectively gaseous or very volatile com­ tried. With the exception of butyl alcohol, the use of these pounds, but also from some degree of unreliability. Uncer­ higher alcohols has the disadvantage of requiring a water ex­ tainties in the analyses have been experienced mostly with traction of the halide ion before its determination. Further­ the aromatic compounds in which the halogens are firmly more, these alcohols have only slightly greater usefulness bound. This unreliability is evident from the number of than ethyl alcohol, as results which have been reported show papers that have appeared on this method of analysis, and the difficulty in obtaining complete decomposition of some which list in detail the conditions that are necessary for good of the compounds analyzed. Monoethanolamine-dioxane results (1, 2, 8). mixtures have been successfully used as a solvent in this A number of modifications have been tried in order to im­ method of analysis (10). prove the reliability of the original procedure. These in- Improvement in Method Secondary alcohols, notably isopropyl and sec-butyl alco­ hol, have been found to be particularly effective solvents for T a b l e I. A nalysis of H alogen Compounds the decomposition of halogen compounds with sodium. Iso­ Halogen Found Halogen propyl alcohol is soluble in water and sec-butyl alcohol is suffi­ C om pound Isopropyl alcohol sec-Butyl alcohol Calculated ciently soluble so that extraction of the halide ion is not neces­ %% % sary before its determination. 1,2,4,5-Tetrachloro- 65.6 6 5 .7 6 5 .7 benzene All the chlorine and bromine compounds, which have been Hexachlorobenzene 74.7 74.7 7 4 .7 Hexachlorodiphenyl0 6 0 .3 60.3 5 9 .0 analyzed with these solvents, have been easily and com­ GO.2 60.2 pletely decomposed. Results for a few of the more refractory Heptachlorodiphenyl 6 0 .3 60.2 6 0 .3 oxide0 60.3 60.2 type compounds are given in Table I. The applicability of Bromobenzene 5 0 .9 5 1 .0 50 .9 o-Fluorodiphenyl 10.7 10.8 11.0 this method to the analysis of chlorine compounds, and also 11.0 to the less stable bromine and iodine compounds, is further a Commercial mixtures. indicated by the favorable results obtained in the analysis of an aromatic fluorine compound, o-fluorodiphenyl. In this 384 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 case, a longer period of heating was necessary to secure com­ an absorption-indicator. However, this may be an advantage plete decomposition. with the Volhard procedure for determining chlorides. Analysis of o-Fluorodiphenyl. Better conditions for de­ D iscussion composing o-fluorodiphenyl, m. p. 7 4 -7 5 ° , were obtained by us­ ing a 20 X 130 mm. test tube provided with a ground-glass joint. Some of the advantages which isopropyl and sec-butyl In this case, 10 ml. of sec-butyl or isopropyl alcohol and 1.5 grams alcohol have over primary alcohols in this method of analysis of sodium were used. The sec-butyl alcohol was refluxed for 5 to 6 hours. The isopropyl alcohol was refluxed for 4.5 hours, are due to their slower rates of reaction with sodium. When which resulted in the consumption of most of the sodium and the ethyl alcohol is used as the solvent, the evolved hydrogen precipitation of considerable quantities of sodium isopropoxide. covers the surface of the sodium and prevents effective de­ Fluoride ion was determined as lead chlorofluoride, essentially composition of the halogen compound (8). This kind of in­ by the method of Stark (11). terference is considerably reduced with the secondary alcohols. Also, the slow rate of formation of the sodium secondary al- Discussion of Analytical Procedure coholates is a further advantage, as no interfering precipitates The use of the secondary alcohols in the decomposition of are formed during the decomposition. The precipitation of stable, slightly soluble organic halides results in a greater sodium ethoxide from ethyl alcohol is detrimental to the precision and accuracy than are obtained by using ethyl al­ analysis of insoluble compounds. cohol in accordance with published procedures. sec-Butyl alcohol has been refluxed from 5 to 6 hours with In the interests of simplicity and convenience, no attempt sodium without the precipitation of sodium secondary bu- has been made to proportion the relative amounts of alcohol, toxide. Isopropyl alcohol starts to precipitate sodium iso- sodium, and sample. This has been done according to em­ propoxide in 3 to 4 hours, depending somewhat on the condi­ pirical formulas in the case of ethyl alcohol (1,3). tions. As all the chlorine and bromine compounds, which The amount of sodium recommended here is in slight ex­ have been analyzed, were completely decomposed in isopropyl cess of that which is consumed in the 2- to 2.5-hour period of alcohol before the appearance of sodium isopropoxide, this reflux used for the decomposition of chlorine and bromine would not be a factor in the analysis of these compounds. compounds. This amount should thus be adequate for most For this reason, sec-butyl alcohol would be the preferred sol­ types of compounds and for variations in sample size. It vent for compounds that may require periods of decomposi­ may be necessary to increase the amount of sodium used by tion longer than 4 to 5 hours. Outside of this consideration, 0.2 to 0.3 gram for periods of reflux longer than 3 to 4 hours, there appears to be no choice between the two solvents. depending on conditions such as rate of reflux, purity of al­ Isopropyl and sec-butyl alcohols are better solvents for the cohol, and the reaction vessel. chlorinated hydrocarbon compounds listed in Table I than is The 2- to 2.5-hour period of reflux has proved adequate to ethyl alcohol. This is an aid to the decomposition of these decompose the more refractory type of chlorine compounds. compounds. Difficulty has been experienced in analyzing In many cases this time can be reduced as 1,2,4,5-tetrachloro- 1,2,4,5-tetrachlorobenzene because of its low solubility in benzene has been completely decomposed after 1 hour of re­ ethyl alcohol (2), whereas no difficulty of that nature has been flux, using either isopropyl or sec-butyl alcohol as the solvent. experienced with either isopropyl or sec-butyl alcohol. Additional analyses are required on different types of No formation of color has been noticed during the decompo­ fluorine compounds in order to establish more definitely the sition of the samples in these alcohols, even when “technical” best conditions for their decomposition. As many fluorine grades have been used. Thus, the inconvenience of specially compounds show great stability, longer periods of reflux will purifying the solvent is avoided. probably be necessary to obtain complete decomposition by feri-Butyl alcohol proved to be an unsatisfactory solvent this method. In this case, sec-butyl alcohol would be the because of the low solubility of its sodium salt. better solvent to use because of the higher temperature and the longer reflux period that is possible before the separation Procedure for Analysis of sodium sec-butoxide. W it h I s o p r o p y l A l c o h o l . About 0.15 to 0.3 gram of sample, depending on the halogen content, was weighed in a short glass Sum m ary vial and placed in a 200-ml. Erlenmeyer flask provided with a ground-glass joint, and 25 ml. of isopropyl alcohol and 1.8 to 2.0 Isopropyl and sec-butyl alcohols are very effective solvents grams of sodium, cut into 4 to 5 pieces, were added. The iso- for the determination of halogen in nonvolatile compounds propyl alcohol was Merck’s, 98 per cent. It was chloride-free, and was used without further purification. The flask was by the sodium-alcohol method. shaken in order to tip the sample vial onto its side, and then con­ The application of these alcohols to the analysis of stable nected to a reflux condenser. The alcohol was gently refluxed fluorine compounds has been indicated. for 2 to 2.5 hours with occasional shaking. Excess sodium was then decomposed by cautiously adding water, a few drops at a Literature Cited time, through the condenser. Finally, 60 to 80 ml. of water were added, the flask was cooled, 2 to 3 drops of phenolphthalein solu­ (1) Bacon, C. W., J. Am. Chem. Soc., 31, 49 (1909). tion were added, and the contents were neutralized by adding (2) Cook, W . A ., and Cook, K. H., Ind. Eng. Chem., A nal. Ed., approximately 6 N nitric acid drop by drop. To determine 5, 1S6 (1933). chlorides, the solution was then titrated with 0.1 N silver nitrate, (3) Drogin, I., and Rosanoff, M. A., J. Am. Chem. Soc., 38, 711 using 4 to 5 drops of a 1 per cent solution of dichlorofluorescein (1916). as an absorption indicator. Eosin was used as the indicator (4) Favrel, G., and Bucher, Ann. chim. anal. chim. appl., 9, 321 when bromides were determined (5, 7). (1927). W it h see-BurYL A l c o h o l . The procedure was essentially (5) Feldman, H . B., and Powell, A. L., Ind. Eng. Chem., A nal. the same as for isopropyl alcohol. The sec-butyl alcohol was E d.. 11, 89 (1939). Eastman Kodak “technical”, chloride-free, and was used (6) Kimura, Wasaburo, J. Soc. Chem. Ind. Japan, 37, Suppl. Bind­ without further purification. Twenty milliliters of alcohol and ing, 589 (1934). 1.5 to 1.7 grams of sodium were used. At the end of 2 to 2.5 (7) Kolthoff, I. M., Lauer, W. M., and Sunde, C. J., J. Am. hours, excess sodium was usually decomposed by adding 15 to 20 Chem. Soc., 51, 3273 (1929). ml. of 95 per cent ethyl alcohol through the condenser. Sixty to (8) Landis, Q., and Wichmann, H. J., Ind. Eng. Chem., A nal. Ed., 80 ml. of water were then added, cautiously at first. Before the 2, 394 (1930). titration for chlorides or bromides, 5 to 10 ml. of 95 per cent (9) Palfray, L., and Sontag, D., Bull. soc. chim., 47, 118 (1930). ethyl alcohol were added, if the solution was not homogeneous (10) Rauscher, W. H., Ind. Eng. Chem., A nal. Ed., 9, 296 (1937). owing to the insolubility of part of the sec-butyl alcohol. (11) Stark, G., Z. anorg. Chem., 70, 173-7 (1911). The presence of undissolved sec-butyl alcohol causes a (12) Stepanow, A., Ber., 39, 4056 (1906). coagulation of the silver halide, which interferes with the use of (13) Van Duin, C. F., Rec. trav. chim., 45, 363 (1926). Determination of Chloride in Bauxite-Supported Anhydrous Aluminum Chloride Catalysts

W. A. L a. LANDE, J r ., IIEINZ IIEINEMANN, a n d W. S. W. McCARTER Porocel Corporation, 260 South road St., Philadelphia, Penna.

UMEROUS patents describe the use of catalysts made denser. The 500-ml. absorption flask shown is convenient, by impregnating bauxite and other granular adsorbents although other types of absorption vessels may be used. N Approximately 20 grams of a representative sample are rapidly with anhydrous aluminum chloride. The growing large- transferred to a 30-ml. weighing bottle and weighed to the scale application of this type of catalyst makes necessary a nearest milligram. The contents of the weighing bottle are dependable method of analysis for control during manu­ added to the flask through a wide glass tube, flared at the top, facture and for assaying shipments and spent material. and extending a few centimeters below the bottom of the neck of the flask. The addition tube is withdrawn, the flask is The purpose of this paper is twofold: to describe satisfactory clamped in position, 200 ml. of 18 N sulfuric acid are added practical procedures for the determination of the chloride rapidly, and the condenser and absorption vessel (containing content of the catalyst, and to discuss the experience gained 200 ml. of water) are quickly fitted into place. from several thousand analyses. For a routine determination of the aluminum chloride content of the catalyst the total chloride may be calculated to aluminum chloride. The total chloride is derived not only from anhydrous aluminum chloride, but also from partially hydrated aiuminum chloride, probably basic aluminum chloride, small amounts of hydrogen chloride and ferric chloride, and possibly traces of other metallic chlorides produced by the action of aluminum chloride, hydrogen chloride, and water on the carrier during the impregnation. Of all the methods investigated, extraction with water and distillation with concentrated sulfuric acid, followed by the volumetric or gravimetric estimation of chloride in the ex­ tract or distillate, gave the most reproducible results after the procedures were carefully standardized. Both methods give low results, since it is impossible to wash the bauxite free of chloride in the extraction procedure, and impractical to ensure complete volatilization of hydrogen chloride by the hot concentrated sulfuric acid because of the coarse mesh size of the sample and the amounts of sample and sulfuric acid it is feasible to employ in the analysis.

T a b l e I. Comparison op Extraction and D istillation M e t h o d s ✓------—A lum inum C hloride------Distillation Method Extraction Method Difference Sam ples0 1 2 A v. (a) 1 2 A v. (6; (o) - (6) % % % % %% % U nim ­ pregnated bauxite 0 .0 0 0 .0 0 0 .0 0 0 .0 0 1 5 .6 7 5 .9 9 5 .8 3 5 .7 7 5 .8 3 5 .8 0 0^03 The mixture is then heated carefully, with occasional shaking, 2 6 .2 8 6 .2 8 6.28 6.20 6.15 6 .1 8 0 .1 0 to control foaming. The distillation is continued for 5 minutes 3 11.06 11.22 11.14 11.54 11.60 11.57 - 0 . 4 3 4 12.38 12.27 12.33 12.49 12.18 12.34 - 0 . 0 1 following the first appearance (usually after 20 to 25 minutes) 5 12.78 12.70 12.74 12.22 12.57 12.40 0 .3 4 of an oily condensate (sulfuric acid) in the upper part of the 6 14.52 14.46 14.49 13.97 14.24 14.11 0 .3 8 condenser. During the period of heating approximately 100 ml. 7 14.40 14.67 14.54 14.53 14.63 14.58 - 0 . 0 4 8 16.79 16.56 16.68 16.48 16.52 16.50 0 .1 8 of distillate collect in the receiver. The condenser is then 9 16.63 16.58 16.61 16.56 16.16 16.37 0.24 removed from the flask and washed down into the absorption 10 15.70 15.63 15.67 14.74 14.75 14.75 0.92 vessel with two 25-ml. portions of water. The contents of the 11 16.12 16.26 16.19 15.86 15.74 15.80 0.39 12 18.25 18.27 18.26 17.83 17.84 17.84 0.42 absorption vessel are washed into a 1-liter volumetric flask and 13 18.10 17.98 18.04 17.48 17.49 17.49 0 .5 5 the volume is adjusted to the mark. A 25-ml. aliquot is trans­ ° Samples 1 to 9, inclusive, were 4/10 mesh; samples 10 to 13, inclusive, ferred to a 250-ml. Erlenmeyer flask, 1.5 ml. of concentrated were 6/14 mesh. nitric acid, 25 ml. of 0.1 N silver nitrate, and 1 ml. of saturated ferric alum solution are added, and the mixture is titrated with 0.1 JV ammonium thiocyanate according to the directions of Kolthoff (/). Analytical Procedures Extraction M ethod. Approximately 50 grams of sample are weighed into a 60-ml. low-form weighing bottle. The The two recommended methods are given in detail in weighing bottle is then opened under 300 ml. of cold distilled the following sections. Typical results are shown in Table I. water in a 800-ml. tall-form beaker (or the sample may be The analyses were made on commercial preparations of rapidly poured into the water). The bottle and stopper are removed from the mixture, which is then heated to the boiling Isocel, a catalyst containing 15 to 20 per cent of aluminum point with mechanical stirring at a rate which keeps the granules chloride on low-iron bauxite, and on samples taken from the in suspension. The boiling is maintained for exactly 1 minute. impregnators at intermediate stages of manufacture. The mixture is then allowed to settle for 1 minute and the super­ natant liquid (bearing some finely divided bauxite in suspension) D istillation M ethod. The apparatus illustrated in Figure 1 is decanted into a 1-liter volumetric flask. Three hundred is constructed from a 500-ml. Kjeldahl flask and a suitable con- milliliters of water are added to the residue in the beaker and the 386 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 heating, stirring, boiling, and décantation are repeated. The ously low results (due to variation in the amount of chloride residue in the beaker is washed into a mortar and the lumps are retained by the bauxite) are encountered when samples crushed to a coarse powder. The material is then washed back quantitatively into the beaker, the volume adjusted to approxi­ substantially larger than 20 grams are used. mately 200 ml., and the extraction procedure again repeated. Extraction M ethod. The amount of chloride found by The contents of the volumetric flask are refrigerated to room this method varies with the number of extractions and the temperature, and made up to 1000 ml. The chloride content is determined as described under the distillation method. state of subdivision of the sample. Higher values are ob­ tained by extracting a finely ground sample, but serious The data in Table I show that the distillation method errors are introduced if the sample is ground prior to the usually gives higher results than the extraction method, first extraction; however, after one or more extractions the irrespective of the chloride content. The difference between granules may be reduced to powder without loss of chloride. the results obtained by the two methods is greater for the There is no detectable loss of hydrogen chloride during the preparations of finer mesh size. The mean variation between boilings. The volumes of water recommended for the ex­ duplicate determinations on each of a series of 13 samples tractions are optimum; larger amounts are unnecessary and was 0.12 for the distillation method and 0.14 for the ex­ inconvenient, while the use of smaller portions results in traction method. lower values for the chloride content. The data in Table III show that two extractions followed by crushing and a third extraction give higher values than one to three extrac­ T a b l e II. C h l o r id e C o n t e n t i n R e l a t io n t o S iz e o f S a m p l e tions without crushing, or one extraction with subsequent (Distillation method) crushing and one re-extraction. The extracted residues ------A lum inum Chloride------. were shown by the distillation analysis to contain residual 0g a a l 2-r m a l 5-r m sm pea ple No. 10-gram sam ple 20-gram sam ple 50-gram sam pleSam % % % chloride equivalent to 8 to 9 per cent of the total chloride 21 12.18 11.48 10.12 content—i. e., 1.0 to 1.6 per cent of aluminum chloride for 22 10.03 15.55 14.37 samples containing 13.5 to 19.5 per cent of (total) aluminum chloride (Table III). Although the distillation method and the extraction method D iscussion as applied directly to the sample give low results, the total D istillation M ethod. In determining the conditions chloride content can be calculated from the sum of the for the distillation method it was necessary to consider the values obtained by the standard extraction of the original size of the sample, the amount of acid used for the digestion, sample and the standard distillation procedure applied to the the mesh size of the sample, and the means for detecting ground residue from the extraction. The following com­ when the evolution of hydrogen chloride could be considered pilation compares the results obtained by the standard complete. methods and the combination procedure. The numbers in It was found that reproducible results could be obtained parentheses indicate the percentage of chloride detected by heating the mixture for 5 minutes following the first by the different techniques. While a combination of the appearance of a few drops of oily distillate (sulfuric acid) two methods will indicate the total chloride content of the in the upper part of the condenser. At this point the super­ catalyst, it is too time-consuming to be recommended for natant liquid in the flask is chloride-free, but the granules routine plant control procedure: of catalyst remaining undisintegrated contain chloride which can be detected after pro­ -Per Cent Aluminum Chloride- longed soaking in water. The chloride can be Sam ple 14 Sam ple 15 Sam ple 16 Sam ple 17 entirely expelled from the unground catalyst Standard extraction plus distillation of extracted by using 20 ml. of 18 N sulfuric acid for each residue 17.84 (100) 13.52 (100) 16.37 (100) 19.49 (100) gram of sample and continuing the digestion Standard distillation 16.08 (93.5) 12.49 (92.3) 15.67 (96.4) 18.26 (93.5) Standard extraction 16.5S (92.7) 12.42 (91.9) 14.75 (90.5) 17.84 (91.5) until the bauxite is completely decomposed. This procedure is impractical for a routine analysis, since a 20-gram sample would require 400 ml. of Soluble Iron Content of Catalyst. The data in sulfuric acid and several hours’ digestion at fuming temper­ Table IV show that 12 to 15 per cent of the iron content of ature. It follows that higher results should be obtained by the carrier is converted into soluble form (probably ferrous decreasing the size of the sample without changing the and ferric chlorides) by the action of the aluminum chloride amount of sulfuric acid used for the decomposition. The and hydrogen chloride during impregnation. The iron con­ data in Table II show the average values obtained for the tributed by the aluminum chloride may be neglected, since chloride content by varying the size of the analytical sample the chemical used in the manufacture of these samples con­ while otherwise following the recommended procedure. tained only 0.1 per cent iron (as FejOs). The bauxite used Experience indicates, however, that samples smaller than as carrier contained not more than 2.5 per cent FejOs (vola­ 20 grams should not be used because of excessive variation tile-free basis). The samples were processed by the extraction due to sampling difficulties; lack of reproducibility and seri- method and the extract and extracted residue were analyzed for iron. The iron values ______are calculated on the basis of the “as received” catalyst T a b l e I I I . V a r ia t io n o f C h l o r id e C o n t e n t w it h V a r io u s E x t r a c t io n P r o c e d u r e s sample. T reatm en t S a m p l i n g . Extrae- Extrac­ -Aluminum Chloride Content- The sample tions tions Sample 14a sample iou sample 1(5° Sam ple 17 a taken in the plant for before after Original Extracted & Original Extracted & Original Extracted*» O riginal E x tracted & No. crushing crushing sample residue sample residue sample residue sam ple residue analysis should be as large % % % %% % % % and as representative as I 1 6 .5S 1.26 12.42 1.10 14.75 1.52 17.84 1.65 16.39 1.09 12.19 0 .95 possible. It is delivered to 16.20 1.35 11.87 1.10 the laboratory in a tightly- i 16.17 1.48 12.00 1.25 ° Samples 14 and 15 were 4/8 mesh; samples 16 and 17 were 6/14 mesh. stoppered glass container and b Determined by distillation method. subdivided with a 16 to 1 splitter contained i n a June 15, 1943 ANALYTICAL EDITION 387

subsequent argentiometric estimation of chloride, give re­ T a b l e IV. D istribution o f I r o n in t u b E x t r a c t io n M e t h o d producible results representing 92 to 96 per cent of the total o f A n a l y s is chloride content. % Fe aa FejOa Total Fe as % FejOj (6) D irect Reproducible but low results for the chloride content of f t ? In extracted analysis of bauxite-supported aluminum chloride catalysts may be Sam ple ex tract residue (a) + (6) cataly st 18 0.24 1.74 1.98 1.87 obtained by water-extracting the catalysts by a standardized 19 0.24 1.66 1.90 1.90 procedure. The method indicates 90 to 93 per cent of the 20 0 .3 5 2 .0 5 2 .4 0 2 .2 3 total chloride content of the sample. The total chloride content of the catalyst may be deter­ mined by a combination of the two methods: standard ex­ “ dry box” . The small fraction is transferred to a glass traction of the original sample plus standard distillation of the container for analysis. The analytical sample is taken by ground residue from the extraction. inserting a test tube (holding 20 or 50 grams under the sam­ Either the extraction or the distillation method is proposed pling conditions) into the bottle to the bottom, thereby re­ as a satisfactory routine procedure. The combination moving a reasonably representative portion of the material method may be used when it is essential to know the ab­ for transfer to the weighing bottle. solute chloride content. Sum m ary Literature Cited Decomposition of the catalyst with sulfuric acid and re­ (1) Kolthoff and Furman, ‘'Volumetric Analysis”, Vol. 2, pp. 218, moval of the liberated hydrochloric acid by distillation, with 244, New York, John Wiley & Sons, 1929.

Determination of Iron

In the Presence of Chromium and Titanium with the Jones Reductor

F. S. GRIMALDI, R. E. STEVENS, a n d M. K. CARRON Geological Survey, U. S. Department of the Interior, Washington, D. C.

Sulfuric acid solutions of titanous and and is titrated with permanganate after chromous sulfates, obtained by passage aeration . through the Jones reductor, are oxidized Best results are obtained by using 0.0003 by aeration for from 5 to 10 minutes in the millimole of copper sulfate in about 300 ml. presence of a trace of copper sulfate as a of solution. Larger quantities of copper catalyst. sulfate lead to slightly low results when Ferrous sulfate is essentially unoxidized both chromium and titanium are present.

HE Jones (6) reductor was originally proposed as a preferential oxidation of the titanium. Their method requires rapid and convenient device for the reduction of ferric to the removal of the excess of bismuth trioxide before the estima­ T tion of iron. Brandt (3) used titanium trichloride as the reduc­ ferrous salts, prior to titration with a standardized solution ing agent for ferric ion, the excess titanium trichloride being de­ of an oxidizing agent. The solution to be reduced is prefer­ stroyed by copper sulfate. In this procedure the cupric ion is re­ ably a sulfuric acid solution, because side reactions that may duced to metal by the titanium trichloride and the precipitated copper is filtered off before titrating the iron. occur in the presence of hydrochloric acid are thereby avoided That simple air-oxidation of a titanous solution is not depend­ in the ensuing titration. able is shown by MeNabb and Skolnik (7). Their results corrob­ Although the Jones reductor is convenient, several elements orate the experience of Margaret D. Foster of this laboratory, other than iron are also reduced by zinc, and various methods who found that in a solution containing titanous salt equivalent to 0.04 gram of titanium dioxide less than three fourths of the ti­ have been proposed to eliminate the effect of these interfering tanium was reoxidized to the quadrivalent state after 10 hours of elements. Previous studies have dealt mainly with the inter­ aeration; a solution containing 0.09 gram of titanium dioxide ference of titanium. was less than nine-tenths converted in the same time. Axt and Leroy (1) increased the rate of oxidation by using a perforated The stability to air-oxidation of ferrous sulfate in sulfuric acid plate for supplying the air. solution was studied by Baskerville and Stevenson {2), who McNabb and Skolnik (7) found that the addition of 50 ml. of showed that practically no oxidation of the ferrous sulfate oc­ saturated mercuric chloride solution greatly increased the rate curred after 3 hours of aeration. They observed also that the of oxidation of titanous sulfate by aeration. presence of cobalt; chromium, copper, and titanium salts had no effect upon the air-oxidation of the ferrous sulfate. Thornton The method described below is based on the discovery by and Roseman (S) studied the preferential oxidation by air of ti­ Zintl and Wattenberg (10) that copper in solution catalyzes tanous sulfate in the presence of ferrous sulfate. Their results air-oxidation of titanous ion. This method is applicable for were good, but they suggest that “the procedure is most apt to succeed when the iron is equal to, or preponderates over, the ti­ all proportions of iron, titanium, and chromium. Molyb­ tanium”. Gooch and Newton (4) used bismuth trioxide for the denum and vanadium should be absent. 388 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

of the iron standard was established T a b l e I . D etermination o f I r o n i n t h e P r e s e n c e o f C h r o m iu m a n d T it a n iu m by reduction in the Jones reductor (3 ml. of 0.0001 M C 11SO4 used) and titration with permanganate. E xperi­ Fe TiO j C r,0 . Fe Colorimetric tests with hydrogen m en t No. T aken T ak en Taken Aeration Fo u n d E rro r peroxide showed titanium to be es­ Gram Gram Gram M in. Gram Gram 1 0 .0315 N one N one 10 0.0317 + 0.0002 sentially absent in the iron solution. 2 0.1576 N one N one 10 0.1576 None The titanium standard was prepared 3 0.0314 0.5000 N one 15 0.0316 + 0 .0 0 0 2 4 0.0314 0.0030 None 5 0.0313 -0.0001 from reagent grade potassium ti­ 5 0.0942 0.0030 N one 5 0.0940 -0.0002 6 0.2512 0,0030 N one 5 0.2511 -0.0001 tanium fluoride (the fluorine being 7 0.2512 0.5000 N one 15 0.2513 + 0.0001 expelled by repeated fuming with sul­ S 0.0314 None 0.0030 10 0.0316 + 0.0002 9 0.0314 N one 0.5000 10 0.0314 None furic acid) and the titanium content 10 0 .1256 N one , 0.0030 10 0.1255 -0.0001 11 0.1256 None 0.5000 10 0.1254 -0 .0 0 0 2 established gravimetrically. Colori­ 12 0.0630 0.0010 0.0010 5 0.0630 None metric tests with thiocyanate showed 13 0.0314 0.0030 0.0030 5 0.0314 None 14 0.1570 0.0030 0.1200 10 0.1566 -0.0004 no iron in the titanium solution. 15 0.0314 0.3000 0.0400 10 0.0313 -0.0001 16 0.0315 0.1400 0.1400 10 0.0312 -0.0003 Chromium was added as potassium 17 0.0315 0.0400 0.3000 10 0.0313 -0 .0 0 0 2 dichromate solution made from the 18 0.1260 0.1400 0.0400 10 0.1259 -0 .0 0 0 1 19 0.1577 0.0400 0.0100 10 0.1573 -0.0004 reagent grade salt by direct weight. 20 0.0630 0.0400 0 -1400 10 0.062S -0 .0 0 0 2 21 0.1265 0.0400 0.3000 10 0.1265 N one Standardized pipets and burets were 22 0.0316 0.2000 0.0030 10 0.0316 N one employed. 23 0.1256 0.0030 0.0030 20 0.1256 N one Preliminary studies were made to establish the optimum con­ centration of copper sulfate to catalyze the oxidation. A pparatus Large concentrations of copper sulfate, up to 1 mole, caused some oxidation of ferrous sulfate when both titanium and chro­ Jones Reductor, as described and illustrated by Hillebrand and Lundell (5). The 20- to 30-mesh zinc is amalgamated with mium were present, giving a maximum error in ten experi­ 3 per cent mercury by shaking with a solution of mercuric ments of —0.0010 gram of iron. However, nine experiments chloride. in which only chromium or only titanium was present with the iron showed a maximum error of ± 0 . 0 0 0 1 gram of iron. R eagents Table I shows the results obtained by adding 3 ml. of 0.0001 molar copper sulfate as catalyst. The results are seen to be Copper Sulfate C atalyst. Solution A, 0.25 gram of cupric sulfate pentahydrate in 500 ml. of water. Solution B, 0.0001 accurate in all the experiments made and throughout ex­ molar cupric sulfate, 5 ml. of solution A in 100 ml. of water. treme ranges of composition. Experiments using still less Solution B should be made up fresh occasionally. copper sulfate showed somewhat incomplete oxidation of ti­ Potassium Perm anganate Solution, 0.05 N, aged at least a tanous sulfite after 10 minutes’ aeration. In the last experi­ week and filtered through asbestos. Standardized against Bu­ ment in Table I the aeration period was increased to 20 min­ reau of Standards sodium oxalate. o-P henanthroline I n d ic a t o r (5), 0.75 gram of o-phenanthro- utes and showed no loss in ferrous content through oxidation. line monohydrate and 25 ml. of 0.05 N ferrous sulfate diluted to In experiments 3 and 7, in which 0.5 gram of titanium dioxide 50 m l. was taken, the violet titanous sulfate color persisted for 9 minutes; consequently the aeration period was extended to 15 Procedure minutes. Pass the iron solution at room temperature (about 200 ml., Acknowledgments containing 10 ml. of concentratcd sulfuric acid) through the Jones reductor at a rate sufficiently slow to ensure complete The writers gratefully acknowledge the cooperation of reduction (about 100 ml. per minute). Add to the reduced solu­ several members of the Chemical Laboratory of the U. S. tion 3 ml. of 0.0001 molar copper sulfate and pass through the solution as rapid a stream of air as is possible without loss by Geological Survey. The work was done under the supervision splashing. Continue the aeration for 5 minutes after the violet of R . C. Wells, chief chemist, whose review of the manuscript color of titanous sulfate has disappeared. If the violet color is led to many improvements. Margaret D. Foster made avail­ masked by the green of chromium salts, aerate the solution for 10 minutes. If the solution is essentially colorless after reduction able to the writers her experiences with air oxidation of titan­ aerate for only 5 minutes. ous salts. Joseph M. Axelrod contributed many useful sug­ After aeration, titrate the solution of ferrous sulfate in the fol­ gestions during the progress of the investigation. lowing manner: First prepare an indicator solution by adding 0.05 N potassium permanganate, a portion of a drop at a time, to a solution con­ Literature Cited taining 1 to 4 drops of the o-phenanthroline indicator in about 10 ml. of 10 per cent sulfuric acid, until the red color just turns to (1) Axt, M., and Leroy, M., Ing. chim., 24, 28 (1940). blue. Pour into this indicator solution a portion of the solution (2) Baskerville, C., and Stevenson, R., J. Am. Chem. Soc., 33, 1104 to be titrated, and set aside. Titrate the main portion of the (1911). solution to the appearance of the permanganate purple color (3) Brandt, L., Chem.-Zlg., 42, 433, 450 (1918). (if much chromium is present the solution becomes gray-green), (4) Gooch, F. A., and Newton, H. O., Am. J. Sci., 23, 365 (1907). add the reserved portion, and titrate to the disappearance of the (5) Hillebrand, W. F., and Lundell, G. E. F., “Applied Inorganic red color of the indicator. Calculate the iron after subtracting Analysis”, p. 101, New York, John Wiley & Sons, 1929. a blank determination on the reagents used. (6) Jones, Clemens, Trans. Am. Inst. Mining Engrs., 17, 411 (1888- In the absence of much chromium the solutions may be titrated, 89). but with less accuracy, without the o-phenanthroline indicator. (7) McNabb, TV. M., and Skolnik, H., I n d . E n o . Chem., A n a l . E d ., 14, 711 (1942). (8) Thornton, W. M., Jr., and Roseman, R., J . Am. Chem. Soc., 57, Experimental 619 (1935). (9) Walden, G. H., Jr., Hammett, L. P., and Chapman, R. 1?.,Ibid., 53, 3908 (1931). For the experimental work solutions were made con­ (10) Zintl, E., and Wattenberg, H., Ber., 56, 472 (1923). taining known quantities of iron, titanium, and chromium by mixing various volumes of standard solutions. The strength P u blish ed by permisaion of the Director, U. S. Geological Survey. Extraction of Ascorbic Acid from Plant Materials Relative Suitability of Various Acids

J. D. PONTING Western Regional Research Laboratory, Bureau of Agricultural and Industrial Chemistry, U. S. Department of Agriculture, Albany, Calif.

T HE stabilization of ascorbic acid during its extraction introduce an appreciable error due to bleaching of the dye. and determination has long been a problem. Many This optimum concentration was determined for each acid as different extracting acids have been recommended, such as follows: trichloroacetic acid by Birch, Harris, and Ray (S), acetic by The rate of drift in galvanometer reading of the standard 1 + 9 Bessey and King {2), metaphosphoric by Fujita and Iwatake mixture of acid and dye was measured (with the Evelyn color­ (£>), oxalic by Watanabe {14), sulfosalicylic by Okrent and imeter) in the absence of ascorbic acid, using a series of different Wachholder {18), metaphosphoric plus trichloroacetic by acid concentrations. That concentration of acid was selected Musulin and King {12), metaphosphoric plus sulfuric by which caused a drift in galvanometer reading of not over '/a Mack and Tressler {11), and various other mixtures. Of division per minute (1 division = 1 per cent transmission), since in the determination the readings are made in 15 seconds to the these extractants metaphosphoric acid appears to be in most nearest 1/i division. Therefore, there was no error caused by general use at present, and has been found in this laboratory fading of the dye. The acid concentrations selected are listed in by Loeffier and Ponting {9) to be satisfactory for extracting a Table I. These concentrations are only approximately correct, wide range of plant materials. However, there is little in­ since a variation of about 5 per cent is not important. They are for use with pure ascorbic acid solutions and do not apply to the formation in the literature on the relative ability of the extraction of plant tissue. In the latter case the acid concentra­ various acids to stabilize ascorbic acid solutions, and there is tions can be at least doubled without causing bleaching of the dye. some difference of opinion as to which acid is most satis­ factory for the purpose. With the above criterion of suitable acid concentration, sulfuric acid can be ruled out at once as unsatisfactory, since W atanabe {14) first recommended oxalic acid as the best ex­ it has the peculiar property of causing rapid bleaching of the tractant for ascorbic acid, but in a later paper {15) he advised the use of a mixture of metaphosphoric and oxalic acids. Lyman, dye even at concentrations too low to turn the dye red, and Schultze, and King {10) compared metaphosphoric, ortbophos- far too low to stabilize ascorbic acid. phoric, sulfuric, and hydrochloric acids as extractants in the In determining the stabilizing effect of each acid on as­ Eresence of added copper and found metaphosphoric acid the corbic acid the following procedure was employed: est of this group. Willberg {16) compared the stability of ascor­ bic acid in oxalic, acetic, citric, and tartaric acids and found Approximately S mg. of crystalline ascorbic acid were weighed oxalic acid to be the best of the group. Krishnamurthy and and dissolved in acid of suitable concentration as determined Giri (S) compared the stability of ascorbic acid in m any carboxylic above, to give 250 ml. of solution containing about 3.2 mg. of and hydroxy acids in the presence of acetate buffer and copper, ascorbic acid per 100 ml. Two 50-ml. portions were set aside in and found the stability greatest in oxalic acid. Metaphosphoric stoppered 250-ml. flasks at room temperature (about 23° C.). acid was not included in this comparison. Krishnamurthy (7) Since it is well known that copper catalyzes the oxidation of studied further the retardation by oxalic acid of ascorbic acid ascorbic acid, copper sulfate pentahydrate was added to another oxidation (copper, iron, and enzyme-catalyzed), also employing 100-ml. portion to give a cupric-ion concentration of 10“ * M . acetate-buffered solutions of pH 5.6. He stated that the degree This solution also was divided into 50-ml. portions and set aside. of retardation produced by oxalic acid is much higher than that The remaining 50-ml. portion of the original solution was used to produced by either metaphosphoric acid or sodium pyrophos­ determine the pH and initial concentration of ascorbic acid. phate at corresponding concentrations, but gave no supporting After 24 hours the residual ascorbic acid was measured in the data. The experiments herein described were undertaken in order to establish the relative merits of most of the recom­ T a b l e I. Loss op Ascorbic Acid from Pure Solutions in mended acids and also other common acids as extractants of V a r io u s A c id s ascorbic acid, with the possibility of finding a suitable substi­ N o C opper 10 M C opper tute for metaphosphoric acid. Another readily available A dded A dded In itial Ascorbic Ascorbic acid would be desirable because of the present difficulty in Ascorbic acid after Av. acid after Av. obtaining metaphosphoric acid and because of its expense. Acid pH Acid 24 hours loss 24 hours loss M g. % M g. % % M g. % % Experimental 0.5% > metaphos- 2 .1 0 3 .2 0 3.1 6 2 .8 2 .7 9 14.4 pnoric 3 .1 8 2.7 9 0.2% oxalic 1.86 3 .2 4 3 .1 4 2 .8 2.92 9 .9 The method used for the ascorbic acid determinations in 3 .1 6 2.9 2 making these comparisons was that of Loeffler and Ponting 0.07% perchloric 1.84 3 .2 4 2 .7 6 14.5 0 .1 7 9 5 .7 2 .7 8 0 .1 0 {9), in which the ascorbic acid is determined in dilute un­ 1% citric 2 .1 7 3 .3 2 2 .5 8 24.1 0 .0 8 9 7 .6 2 .4 6 0 .0 8 buffered acid by measuring the reduction of the dye 2 ,6 - 0.5% tartaric 2.21 3 .2 6 2 .4 2 2 4 .5 0 .1 7 9 4 .8 dichlorophenolindophenol with a photoelectric colorimeter. 2 .4 6 0 .1 7 0.07% nitric 1.88 3.23 2.50 25.7 0.18 94.4 Since the method employs a pure acid extractant without 2 .5 0 0 .1 8 5% acetic 2 .3 0 3.2 8 2 .1 6 31.4 0.03 99.5 added buffer, the direct comparison of acids is simple and less 2 .3 4 0 .0 0 interpretation of results is necessary than in the case of 0.2% maleic 1.97 3.32 2 .0 5 3 5 .6 0.0 0 9 9 .4 2 .2 2 0 .0 4 buffered extractants. 0.5% sulfosalicylic 1.64 3.26 1.79 45.4 0.71 7 8 .3 1.77 0.71 Too high a concentration of any of the acids causes bleach­ 0.2% trichloro­ 2.07 3.22 1.71 4 6 .6 0 .0 8 9 7 .5 ing of the dye (in the absence of ascorbic acid) and therefore acetic 1.72 0 .0 8 1% lactic 2.01 3.26 1.39 58.3 0.65 8 1 .6 makes the determination difficult and subject to error because 1.33 0 .5 5 0.07% hydrochloric 1.68 3 .2 4 1.34 61.1 0 .0 2 9 9 .4 of the resulting drift in the galvanometer reading. Since the 0.2% orthophos- 1.17 0 .0 2 effectiveness of the various acids in stabilizing ascorbic acid pnoric 1.87 3 .2 6 0 .4 5 86.2 0.02 99.7 0.4 5 0 .0 0 in solution tends to increase with their concentration, it is W ater 4 .2 7 3 .2 9 0 100 0 100 desirable to use the greatest concentration that does not 389 390 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 solutions which had been set aside. The solutions without orthophosphoric acid with a pH of 1.87 shows a loss of 86.2 added copper were tested for copper with sodium diethyldithio- per cent under similar conditions. carbamate and found to give negative results in all cases. The solutions were made up in water twice distilled from glass and Oxidation of pure ascorbic acid in water solution was com­ giving no test for copper even when concentrated 1000 to 1. plete at pH 4.27, although the water contained a copper con­ centration of less than 5 X 10~s M. One hundred milligrams The results of these experiments are listed in Table I. of the ascorbic acid in 1 0 ml. of water gave a negative test for The only acids showing low losses with pure solutions of copper; this was about 300 times the concentration used in ascorbic acid are seen to be metaphosphoric and oxalic. the experiments. These acids therefore were used to extract the ascorbic acid From the results listed in Table I, it is evident that the from a series of vegetables and fruits in order'to check the only acids of the group listed which are satisfactory for the consistency with which the acids exerted a stabilizing effect extraction of ascorbic acid are oxalic and metaphosphoric. in different plant extracts and in the presence of ascorbic Oxalic acid seems to be the more effective of the two in in­ acid oxidase. hibiting oxidation in the presence of copper, whether ascorbic acid oxidase is present or not (Tables I and II). With some Twenty-five to 50-gram samples of plant tissue, depending on its ascorbic acid content, were blended in a Waring Blendor for 5 vegetable extracts in oxalic acid (Table II) the loss of ascorbic minutes in 450 ml. of extracting acid of concentration shown in acid in the presence of added copper was less than in its Table II and suction-filtered. About 125 ml. of filtrate were absence. No explanation is offered for this, but it appears obtained and two 50-ml. portions were set aside in 250-ml. glass- to be a rather consistent phenomenon. In the absence of stoppered flasks. The pH and original ascorbic acid were deter­ mined using the remaining 25 ml. of filtrate. In the case of copper the losses of ascorbic acid in plant extracts were cabbage, broccoli, and strawberries more filtrate was obtained generally a little less in metaphosphoric than in oxalic and two portions also were set aside with added cupric ion at a acid. concentration of 10“' M. After 24 hours the ascorbic acid was In a series of tests with plant extracts in 0.5 per cent oxalic measured in the flasks which had been set aside. The results of these experiments are shown in Table II. acid and 1 per cent metaphosphoric acid, the losses were generally a little less in oxalic than in metaphosphoric acid. The concentration of either acid can be markedly increased in the presence of plant extracts without causing bleaching of T a b l e II. Loss of Ascorbic Acid from E xtracts of P lant the dye, but even in vegetables having a high pH it has not M aterials in Oxalic and M etaphosphoric Acids been found necessary to use a concentration greater than Ascorbic Acid 0 . 5 2 p H of A fter 24 Av. per cent for oxalic acid or per cent for metaphosphoric Plant Material Extracting Acid Filtrate Initial hours Loss acid to prevent oxidation of ascorbic acid during blending. Mg./lOO ml.a % Ordinarily a concentration of 0.3 per cent for oxalic or 1 per Freah cabbage 0.4% oxalic 1.79 3.56 3.14,3.16 11.5 cent for metaphosphoric acid is satisfactory, used in a ratio 0.4% oxalic + Cu 1.79 3.56 3.18,3.20 10.4 1% metaphosphoric 1.91 3 .3 3 3 .2 0 ,3 .2 2 3 .6 of 7 volumes of acid to 1 of plant material, or higher. This 1% metaphosphoric + Cu 1.91 3.33 2.95,2.95 11.4 ratio has been found necessary to obtain extraction of the Frozen broccoli 0.4% oxalic 1.87 3 .5 2 3.12,3.14 11.1 ascorbic acid to equilibrium between the solid and liquid 0.4% oxalic + Cu 1.87 3 .5 2 3 .1 8 ,3 .1 6 9 .9 1% metaphosphoric 1.97 3 .3 7 3.08,3.06 8.9 phases (Loeffler and Ponting, 9). 1% metaphosphoric ■+• Cu 1.97 3 .3 7 2 .8 8 ,2 .9 6 13.4 Both oxalic and metaphosphoric acids prevent enzymic Frozen lima 0.4% oxalic 2 .3 7 1.67 1 .4 7 ,1 .4 2 13.5 beans 1% metaphosphoric 2.41 1.76 1.39,1.39 21.0 oxidation of ascorbic acid, as shown by an even lower loss in Frozen peas0.4% oxalic 2.02 1.65 1.08,1.09 34 .2 fresh cabbage extract, which contains a relatively large 1 % metaphosphoric 2 .1 7 1.52 1.11,1.17 25.0 Frozen straw­ 0.4% oxalic 1.68 3.55 3.04,3.03 14.5 amount of ascorbic acid oxidase, than in extracts of peas or berries 0.4% oxalic + Cu 1.68 3.55 2.97,2.97 16.3 1 % metaphosphoric 1.88 3.1 2 2.90,2.90 7.1 strawberries, which contain practically none. (The frozen 1% metaphosphoric vegetables listed in Table II were all blanched prior to freez­ 4" Cu 1.88 3 .1 2 2.55,2.55 18.3 ing.) This is to be expected, since Ebihara (4) found no ° Me. of ascorbic acid per 100 crams of plant tissue * mg. per 100 ml. of filtrate X 18.9 for cabbage, broccoli, and strawberries; and mg. per 100 ml. of activity of ascorbic acid oxidase below pH 4 and Engelhardt filtrate X 9.8 for lima beans and peas. and Bukin (6) found none below a pH of about 3.5. As Lyman, Schultze, and King (10) have shown, meta­ phosphoric acid buffers have no value in stabilizing ascorbic acid above a pH of about 3.3. This is in marked contrast to D iscussion the stability at a high pH in buffers containing oxalic acid, As can be seen from the duplicate values listed in Table I, which has been pointed out by Krishnamurthy (7). In the reproducibility of results is satisfactory under the condi­ confirmation of Krishnamurthy, the author has found almost tions employed and the values are comparable. With plant the same rate of oxidation in a 0.01 M ammonium oxalate extracts (Table II) the losses in each acid are reproducible buffer at pH 5.6 as in the acid itself at pH 1.86—namely, and the initial and final values are comparable, but the about 3 per cent in 24 hours under the conditions described initial ascorbic acid values in one acid are not to be compared above—and for the measurement of ascorbic acid oxidase, a with those in the other acid because no effort was made to 0.01 M oxalate buffer of pH 6.0 is employed, in which no obtain completely uniform samples of plant tissue. With measurable nonenzymatic oxidation of ascorbic acid occurs uniform samples the ascorbic acid values are the same with in 2 0 minutes, even though considerable copper (up to 2 or either metaphosphoric or oxalic acid. In either acid there is 3 p. p. m.) is present. no loss of ascorbic acid during blending for the usual 5 There is often a difference in turbidity of oxalic acid and minutes; in fact, at the start of this work a 1 0 -minute blend­ metaphosphoric acid extracts of fruit or vegetable tissue. ing period was used to promote oxidation, but this method With some materials oxalic acid gives the clearer filtrate and was abandoned because the losses in metaphosphoric and with some the metaphosphoric acid extract is clearer. In oxalic acids were within the experimental error, even though either case the turbidities do not interfere in the colorimetric the temperature reached 45° C. method used. It is obvious that the stability of ascorbic acid solutions is Because of the instability of metaphosphoric acid, pointed not merely a function of pH but depends also on the nature out by Bessey (1), solutions of this acid must be made fresh of the acid (Table I). Thus metaphosphoric acid with a pH every few days. Oxalic acid solutions, on the other hand, of 2 . 1 shows a 2 . 8 per cent loss, in the absence of copper, and can be made in large quantities and stored indefinitely. June 15, 1943 ANALYTICAL EDITION 391

Summary and Conclusion (5) Engelhardt, W. A., and Bukin, B. N., Biokhimia, 2, 274 (1937). (6) Fujita, A., and Iwatake, D., Biochem. Z., 300, 136 (1938-9). Of the 13 acids compared as to their stabilizing effect on (7) Krishnamurthy, P. V., J. Indian Chem. Soc., 18, 201 (1941). ascorbic acid solutions under conditions favorable to oxida­ (8) Krishnamurthy, P. V., and Giri, K. V., Ibid., 18, 191 tion, only metaphosphoric and oxalic acids appeared suitable, (1941). (9) Loeffler, H. J., and Ponting, J. D., Ind. E ng. C h e m ., Anal. these two acids being far superior to any of the others and E d., 14, 846 (1942). about equally satisfactory. It is concluded that oxalic acid (10) Lyman, C. M., Schultzo, M. D., and King, C. G., J . Biol. Chem., may be safely substituted for metaphosphoric acid in the 118, 757 (1937). determination of ascorbic acid, thus providing a more stable, (11) Mack, G. L„ and Tressler, D. K„ Ibid., 118, 735 (1937). (12) Musulin, R. R., and King, C. G.,Ibid., 116, 409 (1936). more easily obtainable, and less expensive extractant. (13) Okrent, A., and Wachholder, K., Biochem. Z., 306, 6 (1940). (14) Watanabe, K., J. Soc. Trop. Agr. Taihoku Im p. Univ., 8, 381 Literature Cited (1937). (1) Bessey, O., J. Biol. Chcm., 126, 771 (1938). (15) Ibid., 9, 162 (1937). (2) Bessey, O., and King, C. G„ Ibid., 103, 687 (1933), (16) Willborg, B., Z. Untersuch. Lebensm., 76, 128 (1938). (3) Birch, T. W., Harris, L. J., and Ray, S. N., Biochem. J., 27, 303 (1933). B u r e a u of Agricultural and Industrial Chemistry, U. S. Department of (4) Ebihara, T., J. Biochcm. {Japan), 29, 199 (1939). Agriculture, Outside Publication Series No. 390S.

Apparatus for Purification of Hydrocarbons by Recrystallization

JOHN LAKE KEAYS1, University of British Columbia, Vancouver, Canada

URING the course of an investigation into the properties In this way it was possible to purify by recrystallization up to 15 D grams of the hydrocarbon at one time. of normal paraffin straight-chain hydrocarbons, it was When the acid was allowed to cool to room temperature, the necessary to synthesize various homologs and to obtain hydrocarbon crystallized out and collected as a diffuse, white, them in the highest possible state of purity. Purification cloudlike layer on top of the acid. was accomplished principally by repeated recrystallization Section F consisted of 10 cm. of 2.5-cm. bore Pyrex, joined to 2-mm. bore tubing. This tubing was connected through stopcock from glacial acetic acid, since the higher paraffin homo­ D to flask 1. The expanded section, F, was filled with acid- logs are slightly soluble in this acid at boiling temperatures washed glass wool and so packed as to provide a filter for the and quite insoluble in the acid at room temperatures. A hydrocarbon crystals and whatever solid or insoluble impurity might be included with the sample being purified. With stop­ description of the apparatus used may be of particular inter­ cocks C and A closed, suction was applied through B, transmitted est to those wishing to prepare pure hydrocarbons. through stopcock D, through E and F to the bottom of flask 2. In practice, the impure hydrocarbon was dissolved by When all the mother liquor had been drawn into flask 1 from heating to the boiling point 1 to 3 grams of the hydrocarbon per liter of acid. Upon cooling, the hydrocarbon crystallized out in a mass of small, white needles. This procedure was repeated until the sample gave a constant melting point. In some cases it was found necessary to repeat the crystallizations as many as twenty times, the final melting point being approached asymptotically. In order to facilitate the operation, which is rather cumbersome and tedious when carried out by the usual method, and to reduce the possibility of contamination from outside sources, the apparatus shown in Figure 1 was devised.

Two 5-liter flasks were connected across a condenser, as shown. The hydrocarbon to be purified was dissolved in pure glacial acetic acid and poured into flaslc 2 through the condenser, vacuum being applied to the system through A and C. W hen this method was not practical, as in the case of hydro­ carbons with high melting points, flask 2 was filled with the acid and the solid hy­ drocarbon was dropped down through the condenser. Flask 2 was then heated and the mixture allowed to reflux for several hours.

1 Present address, Powell River, B. C. 392 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 flask 2, stopcocks B and D were closed, C was opened, and heat 2 was then refilled with fresh acetic acid, heated to boiling, and was applied to flask 1. Acid was distilled into flask 2 until 2.5 the entire contents of the flask were drawn over into flask 1. or 5 cm. (1 or 2 inches) of impure acid residue remained in the From there they were drawn or blown directly into a receiving bottom of flask 1. flask. Alternatively, the purified hydrocarbon could have been With successive cycles of operation, the hydrocarbon in flask 2 drawn directly from flask 2. became purer, and the acid in flask 1 became contaminated with All passages were cleared of hydrocarbon by drawing hot acid increasing amounts of soluble impurities. At regular intervals, washings down through G, up through F, E , and D, and out at samples of hydrocarbon were withdrawn from flask 2. This was A . In order to decrease the possibility of contamination from accomplished as follows: When all but 50 to 100 ml. of acid had the stopcock grease, several liters of hot acetic acid were drawn been drawn into flask 1, a long glass tube was inserted into flask through all stopcocks before the hydrocarbon was introduced into 2 through the mouth of the condenser at G, and through this the system. tube several milliliters of acid and crystals were drawn off. The crystals were filtered, washed thoroughly with distilled water, The apparatus was used only for the purification of hydro­ and dried, and a melting point determination was made. When the hydrocarbon was considered sufficiently pure, as in­ carbons, but it is possible that the same apparatus, or a dicated by constant melting point, stopcocks C and D were modification in which the glass wool, F, is replaced by a closed, pressure was applied through B and the impure acid was selected glass filter, could be utilized for a variety of laboratory discharged through A. Flask 1 was thoroughly cleaned by re­ crystallizations. It is particularly convenient where the peated washing with hot acetic acid. The cold acid, which by this time contained negligible soluble impurity, was drawn from amount of solvent is large and a number of recrystallizations flask 2 into flask 1, and, as before, blown out through A. Flask are necessary.

Filtration Cylinder

It. J. D eGRAY a n d E. P. RITTERSHAUSEN Socony-Vacuum Oil Company, Inc., New York, N. Y.

N QUANTITATIVE analysis there are times when the I precipitate must be filtered and washed and the filtrate brought to a required volume. At other times, the precipi­ tate collected on the filter may require resolution, and the solution must be made up to a known volume. This is especially true in colorimetric work (1, 2, S). The use of an ordinary suction flask for either of these operations requires rinsing from the suction flask into a volumetric flask, with the possibility that thorough rinsing will give a volume larger than desired. A review of the literature showed that this problem has been evident to many. The apparatus most nearly approach­ ing the authors’ needs was one described by Yagoda (4), but this was designed for microanalysis, and no commercial source of the apparatus was indicated.

A piece of apparatus most suitable to the authors’ needs was made, and this standard volume filtration cylinder may now be obtained from Emil Greiner, 161 Sixth Ave., New York, N. Y. The cylinder was designed to accommodate a 100-ml. volumetric flask, or a 120-ml. (4-ounce) sample bottle. Figure 1 shows the apparatus in this 100-ml. form. This size could be made to meet almost any need, or a 200-ml. volumetric flask could be shaped to fit the present cylinder. The letters in Figure 1 refer to the following parts: 1. A No. 3 Gooch crucible. 2. An ordinary Gooch crucible adapter for a suction flask. The tip of the glass tube must be drawn out to give a fine stream. F i g u r e 1 This prevents the stream from blocking the air being displaced from the volumetric flask, causing bubbling and loss of sample. 3. A special heavy rubber stopper, with a hole in the center to hold the adapter (2). The shoulder of the stopper rests on is turned off, the flask removed, and the volume made exactly 100 the top of the cylinder, making an air-tight seal when suction is ml. For work where the volume is not important, a 20-ml. (4- applied. This construction prevents tne cylinder from being ounce) sample bottle may be used to receive the filtrate. split by the wedge effect of a tapered stopper. 4. Heavy glass cylinder, 22.5 cm. (9 inches) tall and 7.5 cm. (3 inches) in inside diameter. Literature Cited 5. A three-way stopcock permitting suction to be applied or air introduced into the cylinder without disconnecting the suction (1) DeGray, R. J., and Price, W. E., “Colorimetric Determination line. of Lead” , unpublished paper. In using the apparatus, the rubber stopper and adapter are re­ (2) Rittershausen, E. P., and DeGray, R. J., "Colorimetric Deter­ moved from the cylinder. The 100-ml. volumetric flask is mination of Iron”, unpublished paper. placed in the cylinder, the rubber stopper replaced, and the (3) Rittershausen, E. P., and DeGray, R. J., ‘‘Colorimetric Deter­ adapter put into position, so that its stem is in the neck of the mination of Sulfates”, unpublished paper. flask. A prepared Gooch crucible is placed on the adapter, the (4) Yagoda, H., Chemist Analyst, 3, 20 (July, 1935). stopcock is turned to connect the suction line to the cylinder, then the filtration is made, and the precipitate is washed. When the P r e s e n t e d before the Division of Petroleum Chemistry at the 101st Meet­ Tolume of the filtrate and washings is nearly 100 ml., the suction ing of the A m e r i c a s C h e m i c a l S o c i e t y , St. Louis, Mo. Semimicro analysis of Saline Soil Solutions

R. F. REITEMEIER U. S. Regional Salinity Laboratory, U. S. Department of Agriculture, Riverside, Calif.

A system of photometric and volumetric methods applicable to small samples of soil solution. In analytical semimicromethods for ions that addition to the small amount of sample required, the methods outlined here involve a saving of time and reagent, a consider­ contribute to soil salinity is described. ation which might be of even greater importance to some These methods involve a considerable re­ analysts. There is an expanding interest in the application of duction in the quantity of soil solution re- microanalytical methods to problems of agricultural chemis­ (fuired, which is an important considera­ try. Peech (24) recently published a scheme for the micro- tion in the extraction of such solutions. determination of exchangeable bases in soils. Wall (36) has developed a set of microprocedures for the determination of In addition, they involve a saving of re­ some inorganic constituents of plant ash. agents and time. The precision and ac­ This article presents photometric and volumetric methods curacy of the methods are considered ade­ for the semimicrodetermination of calcium, magnesium, so­ quate for most soil analyses. dium, potassium, ammonium, carbonate, bicarbonate, chloride, sulfate, and nitrate ions. These methods generally represent adaptations of other methods previously published for the analysis of soils, waters, plants, and clinical specimens. The HE detailed analysis of soil solutions is rendered difficult methods necessarily vary as to convenience and accuracy. T by the large volume of sample required by the standard The aim has been to develop simple procedures whose preci­ analytical methods. The method of extraction, size of appa­ sion would not be seriously less than that of corresponding ratus, size of soil sample, necessity for repetition of the extrac­ macromethods. tion, and length of time required for extraction have been in­ The methods described apply primarily to saline alkaline fluenced by the necessity of securing sufficient solution for soils in which salts of alkali and alkaline earth metals pre­ complete analysis. (In this paper, the term “soil solution” dominate. For use under other conditions, where additional refers to the aqueous solution occurring in the soil at field interfering substances might occur, appropriate modifications moisture; the term “soil extract” refers to the solution ob­ might be necessary. If the soil solution is not analyzed im­ tained from a soil that has been mixed with an artificially high mediately, the concentration of some ions may be appreciably quantity of water—e. g., at soil-water weight ratios of 1 to 2 altered by the activity of microorganisms. Treatments to and 1 to 5.) In a discussion of these factors, Eaton and minimize the direct effect of such processes on the nutrient Sokoloff (10) pointed out that "a material reduction in the ions and indirect effects on other ions are usually not reliable. quantity of solution required in the laboratory would mini­ In addition, calcium carbonate and calcium sulfate precipitate mize some of the difficulties” . Anderson, Keyes, and Cromer from some solutions after extraction. For these reasons, soil (4) recently mentioned the necessity of altering analytical solutions should be analyzed as soon as possible. conditions in the direction of microchemistry. The centrifuge procedures involve the use of an 8-place cen­ The staff of this laboratory has been engaged in the ex­ trifuge head rotating at 3000 r. p. m, in a No. 2 International amination and development of methods for the extraction of centrifuge. Heavy-duty 12-ml. conical centrifuge tubes are soil solutions, particularly of saline and irrigated soils. The necessary at this higli speed in place of the ordinary 15-ml. tubes. A 16-place head rotating at 2000 r. p. m. also provides acceptable pressure-membrane method, described by Richards (28), is an results, and may even be more practical for a large number of effective means of obtaining solutions from soils covering wide samples. An angle head rotating at 3000 r. p. m. was tested, but ranges of moisture content, texture, structure, and salt con­ in general the results were inferior with respect to precipitate com­ tent. This method appears to be especially well adapted to paction and over-all accuracy. A photoelectric photometer is very satisfactory for the colori­ soils at low moisture contents—e. g., near the wilting range metric measurements because of its speed and relative precision. (87). The advantages of the pressure-membrane method An Aminco Type F double photocell photometer (manufactured would be largely lost if it were necessary to apply the conven­ by the American Instrument Company, Silver Spring, Md.) was tional analytical methods to the limited volumes of solution used in this work. It has a permanent, mounting of six pairs of matched color filters and permits the use of both optical cells and obtainable from comparatively dry soils. photometer test tubes. A constant-voltage transformer in the Consequently, the development of the pressure-membrane llo-volt power line eliminates fluctuations in light transmission method has emphasized the need for semimicroanalytical due to voltage variations. Other color-measuring instruments, 393 394 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

T a b l e I. D escription of Soils and Their E xtracts -Extract Characteristics- E lcctrical Accession c, c, conductivity,_o~:i. Soil: w ater No. Soil Type L ocation pH % % p H K X 10‘ © 25° C. ratio Color 56° Imperial clay Meloland, Calif. 7 .8 0 .3 4 0.00094 7 .8 125 1:2 Very pale yellow 57 Imperial clay Imperial, Calif. 7 .2 0.33 7 .0 803 1 :5 Colorless 62 Oasis clay loam Delta, Utah 7 .7 1.07 7 .6 998 1:5 Yellow 63° Oasis clay subsoil Delta, Utah 8 .0 0 .5 8 8 .1 557 1:5 Yellow 68° Vale silty clay loam V ale, Ore. 10.0 0 .2 7 0 .6 0 3 2 b 10.2 555 1:5 B row n 79 Cajon silty clay loam Glendale, Ariz. 7 .7 1.10 0.0044 7 .3 75 1:2 Yellow 84 Gila adobe clay Las Cruces, N. Mex. 8.0 0 .9 2 0.0034 8 .1 85 1:2 Pale yellow 85 Regan clay loam Roswell, N. Mex. 7 .8 0.93 0.0032 7 .6 353 1:2 Pale yellow 86 Fort Collins loam Laramie, Wyo. 8 .0 0 .9 9 0.0053 7 .6 544 1:2 Pale yellow 248a Indio very fine sandy loam Coachella, Calif. 9.1 0.4 9 0.0047 9 .0 12 1:2 Reddish brown 314 Merced clay loam Buttonwillow, Calif. 7.8 2 .3 4 0.0047 b 7 .6 625 1:5 Dark yellow ° Soils used only for organio matter investigation. b Values corrected for medium chloride contents and probably less accurate than those for other soils.

such as spectrophotometers, gradation photometers, neutral Removal of Organic Matter wedge photometers, and visual color comparators, can also be used. For accurate photometric work it is usually hazardous to The possible interference of organic matter in the analysis rely on permanent photometer calibration curves, because of the of soil solutions often raises questions concerning the necessity variable conditions that affect colorimetric procedures. In this laboratory it is a matter of routine to take a series of standards for its removal. It may interfere in such ways as color mask­ through the analytical procedure each time a group of samples is ing, reducing action, mechanical contamination of precipi­ analyzed. tates, and in other direct and indirect ways. The magnitude The calibration of microburets and small pipets is recom­ of these effects is usually unknown. In some systems of mended. A 2-liter beaker covered by a 20-hole perforated brass plate, which holds the centrifuge tubes vertical, makes an ade­ analysis, all samples are treated to remove organic matter, quate water bath. regardless of the amount and composition. In other cases the solutions are analyzed without prior separation of the organic Ionic concentrations are calculated in terms of milliequiva- fraction. lents per liter (m. e./l.). Attention is called to the increasing As the time involved in the preliminary removal of organic use among water chemists of the term “ equivalent per mil­ matter represents an appreciable fraction of the total time lion”, e. p. m. (S). This unit of concentration is numerically required for analysis, information as to the feasibility of the same as milliequivalents per liter if the specific gravity of omitting this operation is important, especially in the routine the solution is unity. analysis of a large number of soil samples. It is also possible In addition to the determination of ionic concentrations, the that some or all of the methods for removing organic matter analysis of soil solutions usually includes the pH value and the may actually introduce errors into the analytical results. electrical conductivity as a measure of the total electrolyte These considerations may affect both macro- and micro- concentration. For conductivity measurements on small analytical methods. samples, a micromodification of the common pipet type of The eleven water extracts of Table I were subjected to four conductivity cell, which holds approximately 5 ml., is very different treatments: ignition, oxidation by hydrogen per­ convenient. For pH measurements a Beckman “one-drop” oxide, oxidation by bromine, and adsorption by carbon. glass electrode (manufactured by National Technical Labora­ Other possible treatments were not investigated systemati­ tories, South Pasadena, Calif.) is satisfactory. Capillary cally because they would definitely introduce various kinds of glass microelectrodes, which require even less sample, are also interference. The treated and untreated samples were available. analyzed for the ions mentioned except nitrate and ammo­ nium. Sodium was determined by a gravimetric uranyl zinc Description of Soils and Extracts acetate procedure [89, Sect. 70 (b), p. 42] instead of the colori­ The systematic investigations of organic matter and pre­ metric method. cision and accuracy reported here were made on extracts of Bromine removed the color from all samples, but the ana­ eleven soil samples of different soil types from various locali­ lytical values agreed with those of the untreated solutions. ties. Pertinent characteristics of these soils and extracts are Consequently, there would be no advantage in the use of this presented in Table I. Two of these soils, 6 8 and 248, are oxidant in the scheme of analysis described here. “black alkali” soils. Carbon not only adsorbed the colored constituents but The organic carbon contents of the soils and extracts were significantly reduced the concentrations of most of the ions, determined by the chromic acid oxidation method of Schollen- especially calcium and magnesium, and lowered the bicar­ berger (SO), involving the modified phosphoric acid reagent of bonate-carbonate value of every sample. However, it did not Purvis and Higson (26). The experimental values were affect the chloride values, which indicates that carbon treat­ multiplied by the factor 1.15, according to Allison (1), which ment may be useful in removing color that interferes with the corrects for incomplete oxidation of the organic matter. The chloride titration. carbon contents of the extracts were determined on the Hydrogen peroxide treatment at a temperature not exceed­ evaporation residue of 25-ml. aliquots. Chloride reduces ing 1 0 0 ° C. caused a general decrease of ions in most samples, chromic acid, and appropriate corrections are included in the particularly of sodium and chloride. The loss of chloride is eight extract values reported. The soil carbon contents of probably a result of oxidation to chlorine. The cause of the the more saline soils also include corrections for chloride, losses of the other ions remains somewhat obscure. These which are very slight compared to those for the corresponding results, coupled wdth the resistance of some organic matter to extracts. oxidation by peroxide and the possibility of the catalytic de­ The pH values were determined by a glass electrode assem­ composition of peroxide by soil constituents, make this treat­ bly. The soil pH measurements were made on saturated soil ment unsatisfactory. pastes. The ignition treatments were made in porcelain casseroles June 15, 1943 ANALYTICAL EDITION 395

at 600° C. The more resistant organic constituents did not relative precision of the two methods. The per cent error repre­ decompose completely over extended periods at lower tem­ sents the algebraic percentile deviation of the mean semimicro value from the mean macro value; this calculation assumes that peratures. This has been observed also on base-exchange the macromethod usually provides the more correct result. The residue ignitions. The results of ignition were variable. average per cent error is the arithmetical average of the percentile Sodium and chloride were lost from every sample, to about the errors for the entire group of semimicrodeterminations; this same extent. Decreases in magnesium and sulfate occurred figure provides a general index of the over-all accuracy of the in several samples. Calcium showed no significant change method. except an increase in one “black alkali” sample, 6 8 . Potas­ VOLUMETRIC DETERMINATION OF CALCIUM sium was extremely high in ten ignited samples, upwards to 500 per cent of the correct value. Fresh samples of four solu­ Calcium is determined by a method involving precipitation tions that were extremely erratic in this respect were ignited as the oxalate, centrifugal washing, and direct titration in per­ in platinum dishes, and these yielded the correct values for chloric acid solution with ammonium hexanitrato cerate, with potassium. The excess potassium evidently originated in nitro-ferroin as indicator. (The ammonium hexanitrato ce­ the material of the casseroles; the results indicate an ex­ rate and nitro-ferroin can be obtained from the G. Frederick change of sodium for potassium. Smith Chemical Company, Columbus, Ohio.) The precipita­ The two black alkali samples were also treated with nitric tion and washing technique represents a combination of acid and boiled, to precipitate the colored humates. Analysis modified Clark and Collip (9) and Blasdale (8) procedures, of the filtrates showed no appreciable deviation from the un­ while the titration technique follows the procedure of Smith treated samples. and Getz (82). The use of cerate and nitro-ferroin permits a In this investigation, carbonate determinations were made direct titration at room temperature with a very sharp end­ only on the untreated and the carbon-treated samples, as the point change from red to pale blue, and a low blank correction. other treatments precluded this determination. Chlorides R e a g e n t s . (Keep reagents B, C, D, and E in Pyrex bottles.) could not be determined on the bromine-treated sample. A. Methyl orange, 0.01 per cent in water. The results of this investigation suggest the following rec­ B. 1 to 1 hydrochloric acid. C. IN oxalic acid. ommendations and possible conclusions. For titration of D. 1 to 1 ammonium hydroxide. carbonate species in dark solutions, a potentiometric titration E. 1 to 50 ammonium hydroxide. can be substituted for the indicator procedure. Purified de­ F. 4 N perchloric acid. Dilute 340 ml. of 70 per cent per­ colorizing carbon can safely be used to treat dark solutions chloric acid or 430 ml. of 60 per cent perchloric acid to 1 liter. G. 0.01 N ammonium hexanitrato cerate in 1 jV perchloric prior to the chloride titration. Carbon can evidently be used acid. Dissolve 5.76 grams of “standard or reference purity” to remove the color of solutions prior to the determination of ammonium hexanitrato cerate in 250 ml. of 4 N perchloric acid sodium and potassium. Ignition may cause appreciable loss and dilute to 1 liter. The reagent should be standardized in the of many ions common to saline soils, especially sodium and following manner: Pipet 5 or 10 ml. of fresh standard 0.01 N sodium oxalate into a small beaker containing 5 ml. of 4 N per­ chloride. Ignitions should not be made in porcelain ware. chloric acid, add 0.2 ml. of nitro-ferroin indicator and titrate Oxidation of organic matter by bromine and hydrogen per­ with the cerate solution to the pale blue end point. Determine a oxide accomplishes no apparent beneficial results. With blank titration correction on a similar sample minus the oxalate especial regard to the inorganic composition of black alkali solution. The milliliters of oxalate used divided by the corrected solutions, the inclusions of ions such as calcium and mag­ milliliters of cerate and times 0.01 provide the normality of the cerate. Do not attempt to adjust the solution to exactly 0.01 N, nesium that may be combined with the humates may not and restandardize whenever the reagent is used several days or always be desirable. more apart. Keep in a dark bottle away from light. H. Nitro-ferroin indicator (nitro-orthophenanthroline ferrous Precision, and Accuracy sulfate). Dilute the stock 0.025 M indicator solution 1 to 20. Use 0.1 ml. in analyses and 0.2 ml. in standardizations. To demonstrate the possible ranges of precision and accu­ Procedure. Pipet an aliquot containing 0.005 to 0.08 m. e. of racy that can be expected from the various methods, water calcium into a clean 12-ml. conical centrifuge tube, dilute or evaporate to 5 ml., and add 1 drop of (A), 2 drops of (B), and 1 extracts of seven soils of Table I were systematically analyzed ml. of (C). Heat to the boiling point in a water bath. While in duplicate by the semimicromethods and by the correspond­ twirling the tube, add (D) dropwise until the solution just turns ing macro- or standard methods in use at this laboratory. yellow. Replace in the bath, and after 30 minutes cool the tube The results for each particular ion are presented in the section n air or in water. If necessary add more (D) to keep the solution just yellow. devoted to the d'scussion of that method. Centrifuge at 3000 r. p. m. for 10 minutes. Carefully decant the Organic matter was not removed from these extracts prior supernatant liquid into a 25-, 50-, or 100-ml. volumetric flask. to their analysis by either the macro- or microprocedures. Stir the precipitate, and rinse the sides of the tube with a stream As the extracts vary considerably in composition, some re­ of 5 ml. of (E) blown from a pipet. Centrifuge at 3000 r. p. m. for 10 minutes. Decant the washings into the same flask. ported determinations may involve quantities of ions that do Drain the tube by inversion on filter paper for 10 minutes. Wipe not represent favorable conditions for the evaluation of the the mouth of the tube with a clean towel or lintless filter paper. accuracy of a method. This applies also to the macro­ Blow into the tube 3 ml. of (F) from a pipet. When the pre­ cipitate is dissolved, add 0.1 ml. of (H). Titrate with (G) from a methods. 10-ml. microburet to the pale blue end point. If more than 5 ml. In addition to the comparisons reported here, many other of (G) is required, transfer the sample to a small beaker and com­ similar studies have been made on soil solutions and soil ex­ plete the titration. Determine the blank correction in the same tracts, waters, plant nutrient culture solutions, and plant ash manner; it usually is about 0.03 ml. Dilute the supernatant extracts. These studies have yielded results as satisfactory liquids in the volumetric flask to volume and save for the magne­ sium determination. as those presented in this paper. C alculation. M. e. of Ca per liter = (corrected ml. of cerate In the succeeding tables, several symbols and terms are used solution X normality of cerate X 1000) -4- ml. in sample aliquot. that possibly require brief explanations. The letters A and B in­ dicate duplicate determinations. The mean is the average of the Precision and Accuracy duplicate values, reported to the same decimal point. The per cent deviation represents the average deviation of the duplicates The calcium concentrations of the seven soil extracts indi­ from the mean divided by the mean value, and times 100; this cated in Table I were determined by this procedure and by a figure is an index of precision or reproducibility. The average calcium oxalate-potassium permanganate volumetric macro­ per cent deviation is the arithmetical average of the per cent devi­ method outlined by Wilcox (39, pp. 38-9) and based on the ation values for the entire group of extracts; comparison of the two values obtained for two methods provides a measure of the calcium-magnesium separation technique of Blasdale (8). 396 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

is best prepared by dilution of a more T a b l e II. Comparison of M acro- and Semimicrometiiods for Calcium concentrated solution of magnesium sulfate that has been standardized by ------Macromethod------—Semimicromethod------gravimetric determination of magne­ Soil Calcium C alcium D evi­ sium. No. A liquot A B Mean ation Aliquot A B Mean ation Error G. IN sulfuric acid. ML e. /Ufa % ML M. e./liter % % H. Ammonium molybdate reagent. 50 39.05 39.09 39.07 0 .0 5 1 39 .9 39.9 39.9 0.00 +2.1 Dissolve 40 grams of ammonium 62 200 3.53 3.54 3 .5 4 0.14 10 3.57 3.58 3.5 8 0.14 + 1.1 molybdate in 400 ml. of water at 60° C., 79 200 2 19 2 20 2 .2 0 0 .2 3 10 2.16 2.18 2 .1 7 0 .4 6 - 1 . 4 84 200 2.3 4 2.34 2 .3 4 0 .0 0 10 2.32 2.34 2.33 0.43 - 0 . 4 add 456 ml. of arsenic-free concentrated 50 28.35 28.45 28.40 0 .1 8 o 28 .6 28.9 2 8 .8 0.5 2 + 1.4 sulfuric acid to 1000 ml. of water, and 2 86 50 24.35 24.51 24.43 0 .3 3 24 .8 2 4 .9 24.9 0 .2 0 + 1.9 cool both solutions. Stir the molyb­ 314 200 7 71 7 .8 0 7 .7 6 0 .5 8 5 • 7 .7 5 7 .8 0 7 .7 8 0.3 2 + 0 .3 date solution into the acid solution Av. 0 .2 2 0 .3 0 1.2 and dilute to 2000 ml. when cool. The ______reagent is a 2 per cent solution of ammonium molybdate in 8 N sulfuric acid. It keeps indefinitely in a brown bottle. The results, presented in Table II, indicate highly satis­ I. Stannous chloride reagent. Place 0.300 gram of c. p. stannous chloride dihydrate in a 100-ml. volumetric flask. Dis­ factory precision and accuracy for the semimicromethod. solve rapidly in water, dilute to the mark, and mix. Any turbidity The reproducibility data demonstrate that it usually is un­ will be removed on mixing with reagent H. Prepare fresh daily. necessary to replicate analytical samples. Procedure. From the volumetric flask containing the cal­ It has been known that the clinical calcium methods in­ cium-free sample pipet an aliquot containing 0.0005 to 0.003 m. e. of magnesium into a 12-ml. conical centrifuge tube and dilute or volve a negative error due to loss of calcium oxalate on decant­ evaporate to 5 ml. Add 1 ml. each of (A) and (B) and 1 drop of ing and a positive error resulting from incomplete washing of (C). Heat to 90° C. in a water bath and while twirling the tube the precipitate; Wang (87) indicates that these two errors are add (D) dropwise until pink. Cool, add 2 ml. of (D), and stir very evenly balanced in most analyses. The present results with a thin glass rod. Withdraw the rod, stopper the tube, and let stand overnight. support this view and show that the net resultant error is of Centrifuge at 3000 r. p. m. for 10 minutes, decant carefully, slight magnitude. drain on filter paper for 10 minutes, and wipe the mouth of the tube with a clean towel or lintless filter paper. Wash the pre­ cipitate and sides of the tube with a stream of 5 ml. of (E) from a COLORIMETRIC DETERMINATION OF MAGNESIUM pipet equipped with a rubber aspirator bulb or by a similar ar­ rangement. Centrifuge at 3000 r. p. m. for 5 minutes, decant, Magnesium is determined on calcium-free solutions by drain for 5 minutes, and wipe the mouth of the tube. Repeat this precipitation as magnesium ammonium phosphate hexa- washing procedure once. hydrate, centrifugal washing, and colorimetric estimation of Pipet 2 ml. of (G) into the tube and dilute to about 10 ml. After 5 minutes, wash the contents into a 100-ml. volumetric flask the phosphate content by the ceruleomolybdate reaction. to which exactly 5 ml. of (H) have previously been added. Dilute This standard clinical procedure, recently described for plant to about 60 ml. and pipet in 1 ml. of (I) while rapidly twirling the ash by Wall (86), has been modified in some details. No flask. Dilute to the mark and mix. At exactly 10 minutes method for the precise determination of small amounts of after adding (I) measure the light transmission of the blue solu­ tion in a photometer test tube through the 650-millimicron filter magnesium in soils appears to have been advanced. Although versus that of water in a similar tube. Previously, the pho­ the method described does not involve a high degree of pre­ tometer is balanced at 100 per cent transmission with water in both cision, the use of duplicate analytical samples usually pro­ tubes. The accuracy is increased somewhat by the use of the vides satisfactory accuracy. same test tube for all samples and standards. Prepare a photometer calibration curve on semilogarithmic Because of the sensitivity of the colorimetric phosphate graph paper by taking a series of 0, 0.5, 1, 2, and 3 ml. of (F) measurement, usually only a fraction of the filtrate from the through the same entire procedure. A typical calibration is calcium determination is used for the magnesium determina­ shown in Figure 1. The amount of magnesium in the sample is tion. This practice is also influenced by the inhibition of obtained by simple interpolation on the curve. Because of the effect of oxalate, the magnesium sample should precipitation of magnesium by high concentrations of oxalate not represent more than one fifth of the calcium sample. If the ion (14,24), which must be reduced to a safe value. magnesium concentration is so low that it cannot be accurately

R e a g e n t s . The concentrations of molybdate and sulfuric acid used in the development of ’the blue color are those recommended by Truog and Meyer (84), but the strength of the reagent has been modified slightly. The stannous chloride reagent is prepared daily and not acidified, according to Zinzadze (41). Because of the effect of time on the color, especially of darker solutions, photometer read­ ings are made at exactly 10 minutes after addition of the stannous chloride. The ammoniacal wash liquid is similar to that recommended by Wang (37) for the washing of calcium oxalate precipitates. Reagents A, B, D, E, and F should be kept in Pyrex bottles and replaced if the precipitation blank color becomes too intense. A. 30 per cent ammonium chloride solution. Dissolve 30 grams of recrystallized ammonium chlo­ ride in water and dilute to 100 ml. Filter before use. B. 5 per cent ammonium dihydrogen phosphate solution. Dissolve 25 grams of ammonium dihydro­ gen phosphate in water and dilute to 500 ml. Filter before use. C. Phenolphthalein, 1 per cent in 60 per cent ethanol. D. Concentrated ammonium hydroxide. E. Ammoniacal wash liquid. Mix 20 ml. of concentrated ammonium hydroxide with SO ml. of water, 100 ml. of ethanol, and 100 ml. of ether. F. Standard 0.001 N magnesium sulfate. This F ig u r e 1. P h o t o m e t e r C a l ib r a t io n C u r v e f o r M a g n e s iu m June 15, 1943 ANALYTICAL EDITION 397

determined in this manner, a larger fraction of the calcium porcelain crucible. Wash with several portions of glacial acetic filtrate or even all of it can be used, provided an equivalent acid, then likewise with ether. Dry in a desiccator over calcium amount of oxalic acid is added to each standard sample before chloride for one hour. precipitation of the magnesium. Results obtained under these B. Uranyl zinc acetate reagent. Solution 1. Stir 80 grams conditions should not be expected to be as satisfactory as those of uranyl acetate dihydrate into a mixture of 14 ml. of glacial obtained by the routine procedure. acetic acid and 427 ml. of water. C a l c u l a t io n . M. e. of Mg per liter = (m. e. of Mg as found Solution 2. Stir 220 grams of zinc acetate dihydrate into a by interpolation X 1000) -s- (ml. in Ca aliquot X fraction of Ca mixture of 7 ml. of glacial acetic acid and 294 ml. of water. aliquot used for Mg determination). Heat the two solutions separately on a water bath and stir until the salts are dissolved. Mix while hot, and when cool ------add 0.2 gram of (A). Let stand overnight. Keep in a T a b l e III. C o m p a r is o n o p M a c ro - a n d S emimicromethods f o r M a g n e s iu m dark bottle and filter before ------M acrom ethod------Sem limieromethod— use. [For preparation of (A), flnil M agnesium D evi­ M agnesium this reagent does not have to No. A liquot AB M ean ation A liquot AB C D Mean ation Error be saturated with (A). ] M l. M. e./liter % M l. M. e./liter %% C. Acetic acid-ethanol 57 50 13.34 13.30 13.35 0.08 o.a i a .5 13.1 13.3 13.4 13.1 1.91 - 1 . 9 wash liquid. Mix 75 ml. of fi2 aoo 3 .3 3 3 .3 5 3.34 0.30 0.8 3.41 3.46 3.30 3.38 3 .3 8 3 .4 6 + 0.6 glacial acetic acid with 425 ml. 79 aoo 1.08 1.09 1.09 0.46 1.6 1.11 1.11 1.13 1.11 1.11 0 .2 3 + 1.8 84 aoo 0 .5 8 0 .5 8 0 .5 8 0 .0 0 3 0 .5 8 0 .6 0 0.56 0.57 0.58 2.16 0.0 of 95 per cent ethanol. Shake 85 50 13.4fi ia .5 0 13.48 0.16 o.a ia.9 13.9 ia.3 13.7 13.7 1.57 + 1 .8 with an excess of (A). Keep 8fi 50 33.48 33.83 33.65 0 .5 3 0.1 33.1 3 3 .7 33.1 3 3 .5 3 3 .4 0 .7 7 - 0 . 8 in a dark bottle and filter be­ 314 aoo 6 .3 3 0.3 8 6 .3 6 0.39 0.4 6.10 6.48 6.50 6.45 6.38 2.24 + 0.3 fore use. Av. 0 .3 7 1.62 1 .0 D. 0.1 N ammonium thio- ______cyanate. Dissolve 3.81 grams o f c . p . ammonium thio­ cyanate in water and dilute to 500 ml. Precision and Accuracy Prepare a sufficient quantity fresh each time. E. Ether, c. p ., anhydrous. The magnesium concentrations of the seven soil extracts F. Standard 0.005 N sodium chloride. Dissolve 0.2923 were determined by this procedure and by a gravimetric grams of dry recrystallized sodium chloride in water and dilute to exactly 1 liter in a volumetric flask. method described by Wilcox (89, pp. 39-40), except that the magnesium ammonium phosphate hexahydrate precipitates were collected on porous-bottomed ceramic Gooch crucibles, washed with ammonium hydroxide, ethanol, and ether, and weighed as the hexahydrate. Colorimetric magnesium deter­ minations were made on duplicate fractions of each of the two calcium-free semimicroanalytical samples. This arrangement provides an opportunity to decide whether errors possibly arising in the separation of magnesium from calcium con­ tribute significantly to the over-all errors. In Table III, colorimetric samples A and B are from one calcium-free ali­ quot, and C and D from the other. The precision of the colorimetric method for most samples as shown by variations among four replicates is not very high. However, the use of average values from duplicate aliquots usually results in acceptable accuracy. Four ex­ tracts indicate that errors in the separation of calcium may influence the results, but this is hardly significant and is not upheld by previous data. The interference of other ions, such as sodium and potassium, is always possible when a double F ig u r e 2. P h o t o m e t e r C a l ib r a t io n C u r v e f o r S o d iu m precipitation is not made. Hillebrand and Lundell (14) dis­ cuss these and other interferences in detail. Aside from the

various theoretical aspects of the method, it provides satis­ P r o c e d u r e . Pipet an aliquot containing 0.002 to 0.012 m. e. factory results if too much faith is not placed in a single of sodium into a clean 12-ml. conical centrifuge tube. Evaporate determination. in a water bath to 0.2 ml. Cool, add 8 ml. of (B), stopper, and mix immediately by repeated inversions for one minute. Let stand one hour. Remove the stopper and centrifuge at 3000 COLORIMETRIC DETERMINATION OF SODIUM r. p. m. for 10 minutes. Drain on filter paper for 10 minutes. Wipe the mouth of the tube with a clean towel or lintless filter Sodium is determined by a procedure which follows closely paper. Stir the precipitate and wash the sides of the tube with 4 the clinical method of Hoffman and Osgood (17). Sodium ml. of (C) in a stream from a pipet equipped with a rubber aspi­ uranyl zinc acetate is precipitated, centrifugally washed, and rator bulb. Centrifuge at 3000 r. p. m. for 10 minutes, decant, dissolved, and the yellow color is compared with those of and drain for 10 minutes. Wipe the mouth of the tube. Wash with 5 ml. of ether, but centrifuge for only 5 minutes and decant sodium standards. Dissolving the precipitates in ammonium carefully without draining. (If the tube is drained after the thiocyanate solution helps to stabilize the color against tem­ ether wash, portions of a precipitate may drop from the tube.) perature changes. Because of the great sensitivity of the Repeat the ether washing and decanting once. gravimetric uranyl zinc acetate method, the colorimetric When the ether is completely evaporated, pipet into the tube exactly 10 ml. of (D), mix by inversion until the precipitate is dis­ technique sometimes may not extend the analytical range solved, and centrifuge at 3000 r. p. m. for 5 minutes to remove any to the same extent as for the determination of some other phosphate precipitate. Pour into a 1-inch optical cell and ions. Nevertheless, the photometric method does increase measure the light transmission of the solution through the 420- the precision of estimation of small quantities of sodium. millimicron filter versus th a t of (D) in a similar cell. Previously, balance the photometer at 100 per cent transmission with (D) R e a g e n t s . A. Sodium uranyl zinc acetate crystals. Add in both cells. A calibration curve is prepared by taking a series of 125 ml. of (B) to 5 ml. of 2 per cent sodium chloride solution, stir, 0, 0.5, 1, 1.5, 2, and 2.5 ml. of (F) through the same procedure and after 15 minutes collect the precipitate in a porous-bottomed and plotting the results on ordinary graph paper. Figure 2 398 INDUSTRIAL AND ENGINEERING’ CHEMISTRY Vol. 15, No. 6 shows a typical curve. The amount of sodium in the sample is P r o c e d u r e . Pipet an aliquot containing 0.0005 to 0.005 m. e. obtained by simple interpolation on the curve. of potassium into a clean 12-ml. graduated conical centrifuge tube, C a l c u l a t io n . M. e. of Na per liter = (m . e. of N a as found b y dilute or evaporate to 1 ml., blow in 2 ml. of (A) from a pipet, and interpolation X 1000) -*■ ml. in aliquot. twirl the tube for a few seconds. Let stand 3 hours in a refriger­ ator at about 5° C., and mix by twirling several times during this interval. Wash the upper walls of the tube with a stream of 0.5 Precision and Accuracy ml. of water, but do not mix it with the precipitation liquid. Centrifuge at 3000 r. p. m. for 10 minutes and drain on filter The macromethod used for comparison was the original paper for 5 minutes. Wipe the mouth of the tube with a clean gravimetric uranyl zinc acetate method of Barber and Kolt- towel oi- lintless filter paper. Stir the precipitate and wash the tube walls with a stream of 3 ml. of (B). Centrifuge for 5 minutes and drain. Repeat this washing once. (Ammonia should be absent from the atmosphere T a b l e IV. C o m p a r is o n ok M a cro- a n d S emimicromethods f o r S od iu m during the preceding operations.) ------Macromethod------Semiumcromethod----- Vigorously stir the precipitate with a Sodium Sodium Soil D evi­ D evi­ stream of 5 ml. of water from a pipet. A liquot A B M ean ation Aliquot A B Mean ation E rro r No. Immediately place the tube in a boiling M. e./liter ML M. e./liter ML % % % water bath and keep it there until the 57 5 29.71 29'. 83 29.77 0 .2 0 0 .3 30.0 30.3 30.2 0 .5 0 + 1.4 precipitate is completely dissolved. 62 2 86.15 86.35 86 .2 5 0.12 0.1 86.5 87.5 87.0 0 .5 8 + 0.9 79 50 3.40 3 .4 4 3 .4 2 0.5 9 2 3.4 7 3.57 3.5 2 1.42 + 2.9 Cool, add exactly 2 ml. of (C), and then 84 50 5.16 5.16 5.16 0.00 2 5 .0 5 5 .1 5 5 .1 0 0 .9 8 - 1 . 2 blow in exactly 2 ml. of (D). Dilute to 85 50 3 .4 8 3 .4 8 3 .4 8 0 .0 0 2 3 .5 3 3 .6 3 3 .5 8 1.40 + 2.9 the 12-ml. mark and mix by inver­ 86 10 21.71 21.71 21.71 0.00 0.4 21.3 21.3 21.3 0 .0 0 - 1 . 9 314 5 52.84 53.02 52.93 0.17 0 .2 52.5 53.0 52.8 0.47 - 0 . 2 sion. If turbid, centrifuge for 5 minutes. After an interval of 15 to 30 minutes, Av. 0.15 0.76 1.6 compare the light transmission in a 1-inch optical cell through the 580- millimicron filter with that of water in a similar cell. hoff (6) as modified by Wilcox [39, Sect. 70 (b), p. 42], The Previously, balance the photometer at 100 per cent transmission w ith w ater in both cells. results are presented in Table IV. Prepare a calibration curve for each set of samples by carrying a The colorimetric results are similar to those found for mag­ series of 0, 0.1, 0.2, 0.3, 0.4, and 0.5 ml. of 0.01 N potassium nesium in that while the precision of the method is not great chloride through the same operations. The amount of potassium the accuracy obtained by using duplicate samples is satis­ in the sample is found by interpolation on this curve. A straight- factory. Since the precipitation procedures of the colori­ line semilogarithmic calibration curve is obtained for the rangejO to 0.005 m. e. of potassium, as shown in Figure 3. metric and gravimetric methods are very similar, the source of any errors must lie mainly in the centrifuge or color- measuring technique. Iioffman and Osgood (17) point out that the color intensity is affected appreciably by acid, which must be carefully washed out of the centrifuge tube by the ether. Also, the low sensitivity of photometers and color­ imeters to yellow colors tends to reduce the attainable precision.

COLORIMETRIC DETERMINATION OF POTASSIUM Potassium is determined by a photometric method in which the cobalt content of the centrifuged cobaltinitrite precipitate is estimated by treatment with ferrocyanide and choline hydrochloride, according to a method of Jacobs and Hoffman (18), later modified for the photoelectric colorim­ eter by Hoffman (16). Morris and Gerdel (22) describe a similar method for plant samples. The technique of precipitation of the potassium sodium cobaltinitrite involves some suggestions of Volk (35) applied to the clinical method of Kramer and Tisdall (21). Pro­ vision is included for the volatilization of ammonia from samples containing amounts that would cause positive errors F ig u r e 3. P h o t o m e t e r C a l ib r a t io n C u r v e f o r in the potassium value. P o t a ss iu m R e a g e n t s . A. 30 per cent sodium cobaltinitrite reagent. Dissolve 30 grams of c. p . sodium cobaltinitrite in water, add 2 ml. of glacial acetic acid, and dilute to 100 ml. Prepare a sufficient If a qualitative or quantitative test for ammonia indicates quantity fresh daily. Filter before use. sufficient to interfere, it can be removed by the following pre­ B. 70 per cent ethanol. liminary treatment. Pipet the potassium aliquot into a 15-ml. C. 0.4 per cent choline hydrochloride. Recrystallize choline beaker, dilute to 5 ml., and add 0.2 ml. of (F). Boil slowly until hydrochloride as follows: Dissolve in a minimum quantity of just dry, add 5 ml. of water, and boil again until the residue is absolute ethanol, filter, and precipitate by addition of excess barely moist. Dissolve in a minimum volume of water, not over ether. Collect on a suction funnel, wash with ether, and dry 0.5 ml. and decant into a graduated 12-ml. centrifuge tube. in a desiccator. Dissolve 0.2 gram in 50 ml. of water. Prepare Rinse the beaker with several 0.2-ml. portions of water, and add a sufficient quantity of the solution fresh daily. these to the centrifuge tube. Add 1 drop of (H) and neutralize D. 0.8 per cent potassium ferrocyanide. Recrystallize with (G). Evaporate in a water bath to a volume of 1 ml. Eotassium ferrocyanide from a boiling saturated aqueous solution Cool and proceed in the regular manner. The series of standards y cooling. Collect on a suction funnel and pull air through until should be treated in the same manner because of the possible dry. Dissolve 0.4 gram in 50 ml. of water. Prepare a sufficient addition of potassium by way of the extra reagents. fresh quantity of the solution daily. C a l c u l a t io n . M. e. of K per liter = (m. e. of K in aliquot as E. Standard 0.01 N potassium chloride solution. Dissolve found by interpolation X 1000) -*■ ml. in aliquot. 0.7456 gram of dry recrystallized potassium chloride in water and dilute to exactly 1 liter. Précision and Accuracy F. 1 N sodium hydroxide. G. 1 N acetic acid. Quantitative comparisons with macromethods are pre­ H. Methyl orange, 0.01 per cent in water. sented in Table V. Samples 62. 8-5, and 314 were analyzed June 15, 1943 ANALYTICAL EDITION 399 by the gravimetric cobaltinitrite method of Wilcox (40). The other titratable anions, such as silicate, phosphate, and remainder were analyzed by a volumetric cobaltinitrite pro­ borate (2). cedure, in which the potassium was precipitated according In spite of these difficulties, no other method approaches to Wilcox (40) and filtered and titrated according to Hibbard it in convenience. The indicated method of calculation is and Stout (18). the customary one and assumes the absence of other titrat­ None of the three methods shows a high degree of pre­ able ions. cision on these samples. Cobaltinitrite methods usually are not the most accurate methods for potassium, but are widely R e a g e n t s . A. Phenolphthalein, 1 per cent in 60 per cent ethanol. B. Methyl orange, 0.01 per cent in water. C. used in soil analysis because of their convenience and sensi­ Standard 0.01 N sulfuric acid. tivity. The precipitation in both the macro- and semi­ Procedure. Pipet an aliquot containing 0.005 to 0.04 m. e. of microprocedures is probably affected by the same factors. chloride into a 15-ml. wide-mouthed porcelain crucible or a small Interferences in saline soils may occur because of the low pro­ porcelain casserole. Add one drop of (A). If the solution turns pink, add (C) from a 10-ml. microburet dropwise at 5-second portion of potassium to other ions, such as sodium and cal­ intervals until the color just disappears. Record the buret read­ cium. In gypsiferous soils, the deposition of calcium sulfate ing. Add 2 drops of (B) and titrate to the first orange color. sometimes occurs on evaporation, and the double salt CaSO.t. - Save the titrated sample for the chloride determination. K2SO4 may be formed. The predictable accuracy of the An indicator correction blank in boiled water should be deter­ mined, and applied if it is not negligible. The lighting should be colorimetric method is very satisfactory when compared to adequate for the recognition of the various colors. The use of the macroprocedures. comparison color standards at the correct end points is helpful. C alculation. If A is the milliliters of (C ) to the phenol- phthalein end point and B the milliliters to the methyl orange end COLORIMETRIC DETERMINATION OF AMMONIUM point, In samples containing appreciable quantities of ammonium M. e. of COj— per liter = (2A X 0.01 X 1000) +• ml. in aliquot. ion resulting from ammonification, fertilization, or other M. e. of HCO3- per liter = [(B - 2A) X 0.01 X 1000] -4- ml. source, it may be desirable to determine its concentration. in aliquot. This can be accomplished by the procedure outlined under the description of the colorimetric determination of nitrate. Précision and Accuracy A similar method employing 0.05 N sulfuric acid was used for com­ parison (5, p. 535; 89, p. 18). The T a b l e V. Comparison of M acro- and Semimicromethods fo r Potassium results on the seven extracts are Macromethod------> /■------Semimicromethod' presented in Table VI. As the Soil Potassium D evi­ Potassium D vi- samples were neutral to phenol­ No. Aliquot A B M ean ation A liquot A B M ean atio n E rro r Ml. M. c./liter % M l. M. e./liter % % phthalein, the listed bicarbonate 57 200 0.417 0 .4 2 5 0.421 0 .9 5 10 0 .4 2 8 0.437 0.433 1.04 + 2.9 values consist of methyl orange alka- 62 100 3.10 3.16 3.13 0 .9 6 1 3 .1 8 3 .2 2 3 .2 0 0 .6 3 + 2 .2 linities. 79 100 0 .7 0 4 0.704 0 .7 0 4 0 .0 0 5 0.673 0.677 0.675 0 .3 0 - 4 . 1 84 100 0.335 0.339 0.337 0.59 10 0.340 0.346 0.343 0.S7 + 1 .8 The precision of the semimicro­ 85 100 3.56 3.61 3 .5 9 0 .7 0 1 3 .5 3 3 .6 0 3 .5 7 0 .9 8 - 0 . 6 86 100 1.086 1.106 1.096 0.91 5 1.114 1.128 1.121 0 .6 2 + 2 .3 method corresponds favorably to 314 100 0 .9 8 1.02 1.00 2.00 5 0 .9 8 1.00 0.99 1.01 - 1 . 0 that of the official method. This Av. 0 .8 7 0 .7 8 2.1 should be expected of a method in­ volving merely a straightforward ti­ tration. It is somewhat more dif­ ficult to estimate the accuracy of the VOLUMETRIC DETERMINATION OF CARBONATE AND method; although by comparison with the official method it BICARBONATE is satisfactory, the various factors previously discussed affect These ions are determined by a micromodification of the both methods. Comparisons made on other samples by Warder (38) alkalimetric titration to the phenolphthalein potentiometric and indicator procedures, not reported here, and methyl orange end points, respectively. Titration meth­ provided similar accuracy. ods are rapid in operation, but can be highly inaccurate under the ordinary conditions of sampling and analysis. These factors were stressed by Johnston (19). More recently, VOLUMETRIC DETERMINATION OF CHLORIDE Benedetti-Pichler et al. (7) discussed the various errors of the Chloride is determined by a micromodification of Mohr’s ordinary Warder procedure. The estimation of a small volumetric silver nitrate-potassium chromate method (33). amount of carbonate in the presence of a much larger quan­ The determination is made on the carbonate-bicarbonate tity of bicarbonate, a condition frequently encountered in sample, which has been neutralized to methyl orange in that alkaline soil extracts, usually involves con­ siderable error. [Hirsch (15) recently de­ scribed the construction of a slide rule for the calculation of these ions from the pH T a b l e VI. Comparison of M acro- and Semimicromethods for value and the methyl orange alkalinity. His B ic a r b o n a t e charts, based on earlier calculations involv­ — Macromethod —Semimicromethod Soil Bicarbonate D evi­ Bicarbonate D evi­ ing carbonate-bicarbonate equilibria, show No. A liquot A B M ean ation A liquot A B M ean atio n E rro r that no titration method, potentiometric or Ml. M. e./liter % ML M. e./liter % % colorimetric, can be theoretically exact. The 57 100 0.38 0.39 0.39 1.28 10 0.37 0.37 0.37 0.00 - 5 . 1 use of such a slide rule should provide more 62 100 1.73 1.73 1.73 0.00 10 1.76 1.78 1.77 0.57 + 2.3 79 100 2.67 2.69 2 .6 8 0 .3 7 10 2.67 2.70 2.69 0 .5 6 + 0 .4 accurate results than those obtained from 84 100 2.29 2.32 2.31 0 .6 5 10 2.27 2.27 2.27 0.00 - 1 . 7 85 100 1.80 1.81 1.81 0 .2 8 10 1.79 1.81 1.80 0.56 - 0 . 6 the usual calculation, especially where the 86 100 2.85 2.86 2.86 0 .1 7 10 2 .7 8 2.81 2 .8 0 0 .5 4 - 2 . 1 carbonate value represents a small frac­ 314 50 1.07 1.08 1.08 0 .4 6 10 1.11 1.13 1.12 0.89 + 3.7 tion of the total alkalinity.] Additional Av. 0 .4 0 0 .4 5 2 .3 errors may result from the presence of 400 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

procedure. With adequate lighting and appropriate blank brown bottle. Suspend 4 grams in 150 ml. of water in a 250-ml. volumetric flask, add 7 ml. of 1 N hydrochloric acid, shake until corrections, satisfactory accuracy can be obtained. dissolved, and dilute to the mark. Filter before use. D. 95 per cent acetone. R e a g e n t s . A. 5 per cent potassium chromate indicator. E. Phenol red. Prepare a 0.05 per cent aqueous solution. Dissolve 5 grams of potassium chromate in 50 ml. of water and To reduce the titration blank caused by the indicator, add suffi­ add 1 N silver nitrate dropwise until a slight permanent red pre­ cient 0.05 N sodium hydroxide so that when diluted in the titra­ cipitate is produced. Filter and dilute to 100 ml. tion sample, the blank will be no greater than 0.05 ml. of 0.01 N B. Standard 0.005 N silver nitrate solution. Dissolve 0.8495 sodium hydroxide. As an illustration, 3 parts of phenol red gram of c. p. silver nitrate in water and dilute to exactly 1 liter. solution are treated with 1 part of 0.05 N sodium hydroxide, when Keep in a brown bottle away from light. 0 . 2 ml. of indicator is used in the titration of unknowns and Procedure. To the sample preserved from the carbonate- standard phtlialate. bicarbonate determination, add 4 drops of (A). While stirring, F. Standard 0.01 N sodium hydroxide. This should be titrate under a bright light with (B) from a 1 0 -ml. microburet to standardized against 5 ml. of 0.01 N potassium acid phthalate at the first permanent light brown color. The titration blank cor­ the boiling point with phenol red. The end point is the deep rection varies with the volume of the sample at the end point, and purplish-red color that persists on boiling. The total volume of usually increases regularly from about 0.03 to 0.20 ml. as the titrated sample at the end point should be 10 ml. Do not at­ volume increases from 2 to 1 2 ml. tempt to make exactly 0 . 0 1 N. C alculation. M. e. of Cl- per liter = (ml. of AgNOj — ml. Procedure. Pipet an aliquot containing 0.005 to 0.08 m. e. of of AgN03 for blank) X 0.005 X 1000 ml. in aliquot. sulfate into a clean 12-ml. conical centrifuge tube. Dilute or evaporate to 5 ml. Add 2 drops of (A) and then (B) dropwise until yellow. Place in ice water, and after 5 minutes blow in 2 ml. of (C) from a T a b l e VII. Comparison of M acro- and Semimicromethods fo r C hloride pipet, and mix well by twirling. Let Macromethod------—* «■------Semimicromethod- stand 2 0 minutes in ice water, and Soil Chloride D evi­ Chloride D evi­ centrifuge at 3000 r. p. m. for 10 min­ No. A liquot A B M ean atio n A liquot AB M ean atio n E rro r utes. Decant carefully without drain­ M l. M. e./liter % M l. M. e./liter % % ing. Wash the tube walls and stir 57 20 63.3 6 3 .5 6 3 .4 0.1 6 0 .5 6 3 .4 63 .6 6 3 .5 0 .1 6 + 0 .2 the precipitate with a stream of 5 ml. 62 20 8 0 .2 8 0 .3 80 .3 0 .0 6 0 .5 80.4 80.5 80.5 0 .0 6 + 0 .2 of (D) from a pipet. Centrifuge at 79 100 1.30 1.30 1.30 0.00 10 1.25 1.26 1.26 0.40 -3.1 84 100 1.04 1.05 1.05 0 .4 8 10 0 .9 9 1 .0 0 1.00 0 .5 0 - 4 . 8 3000 r. p. m. for 5 minutes, decant, 85 100 1.52 1.52 1.52 0 .0 0 10 1.49 1.51 1.50 0.67 -1.3 and wash twice again in the same 86 100 1.77 1.78 1.78 0 .2 8 10 1.76 1.79 1.78 0.84 0.0 manner. 314 50 17.93 17.94 17.94 0 .0 3 2 17.93 17.98 17.96 0 .1 4 + 0.1 Wash the precipitate into a 50- or Av. 0.14 0.40 1.4 1 0 0 -ml. beaker with a 1 0 -ml. stream of water, add 0.2 ml. of (E), and titrate boiling hot with (F) from a 10-ml. microburet. During the titration, pour the hot solution back and forth Precision and Accuracy from the beaker to the centrifuge tube, to remove any adhering precipitate. Titrate to the same permanent end-point tint used The seven soil extracts were analyzed by this procedure in the standardization. The total volume of titrated sample at and by the official Mohr method (5, p. 528) as outlined by the end point should be 10 ml. Determine the titration blank by exactly the same procedure. Wilcox (39, p. 18). The results are presented in Table VII. C alculation. M.e. of SO< per liter = (ml. of NaOH — ml. of The reproducibility of the semimicromethod is seen to be NaOH for blank) X normality of NaOH X 1000 -s- ml. in aliquot. very satisfactory, comparable to that of the macromethod. Both the precision and accuracy improve with increasing Precision and Accuracy quantities of chloride. A gravimetric barium sulfate method (39, p. 19) was used The end point of the chloride titration sometimes is not for comparison. Table VIII contains the results. The semi- very distinct, which is probably the main source of error. micromethod shows a high degree of precision and a pre­ Because of the low bicarbonate content of several extracts, dictable accuracy of about 2 per cent. the aliquots used for the bicarbonate determination were It is important that all titrations, including unknowns, larger than those for the chloride determination. blanks, and standardizations, be made to the same phenol red end point in equal final volumes of sample. Precipitation in VOLUMETRIC DETERMINATION OF SULFATE an acid solution precludes the interfering precipitation of Sulfate is determined by precipitation as benzidine sulfate, phosphate. centrifugation, and direct titration of the liberated sulfuric COLORIMETRIC DETERMINATION OF NITRATE acid with dilute standard base. The reagents and precipita­ tion procedure are based on Fiske’s modification (11) of the Nitrate is determined by a distillation method, because Rosenheim-Drummond method (29). Precipitation in an the well-known phenoldisulfonic acid method is seriously ice bath and centrifugal washing appear to increase the affected by chloride. The procedure involves reduction of accuracy. A source of error in benzidine methods has been nitrate to ammonia, distillation into dilute acid, nesslerization, the coprecipitation of benzidine hydrochloride, which is and photometric comparison with standards similarly treated. difficult to wash out of the precipitate (12). It is believed Any nitrite is included in the nitrate value. Ammonium ion that the method described here reduces this error. can be determined by preliminary distillation in the absence of Devarda’s alloy. Reagents. A. Bromophenol blue, 0.04 pèr cent in 95 per Various nitrogenous organic compounds are hydrolyzed on cent ethanol. boiling with sodium hydroxide, sodium carbonate, and mag­ B. IN hydrochloric acid. nesium oxide with the formation of ammonia (23, 25, 31). C. Benzidine hydrochloride reagent. Purify benzidine hy­ drochloride and prepare the reagent as follows: Dissolve 10 If ammonia is distilled prior to the nitrate reduction, possible grams of benzidine hydrochloride in 400 ml. of 1 N hydrochloric errors arising from the hydrolysis of such compounds in soil acid by warming to 50° C. Filter, add 40 ml. of concentrated solutions or extracts are included in the ammonium value. hydrochloric acid with stirring, cool in ice water for 30 minutes, According to Nichols and Foote (23) and Shrikhande (31), and collect crystals on a Büchner funnel. Wash with cold 1 N hydrochloric acid, then with two 25-ml. portions of cold 95 per this error in the estimation of free ammonia can be eliminated cent ethanol and four portions of ether. When dry, transfer to a by distilling the ammonia from a solution buffered at pH 7.4. June 15, 1943 ANALYTICAL EDITION 401 optical cell against that of water in T a b l e VIII. Comparison op M acro- and Semimicromethods fo r S ulfate a similar cell. Previously balance the photometer at 1 0 0 per cent trans­ ------Macromethod------' —Semimicrometr lod------mission with water in both cells. Soil Sulfate D evi­ S ulfate D evi­ No. A liquot A B M ean ation A liquot A B Mean ation Error Distill and treat a series of 0, 0.2, ML M. e./litcr % Ml. M. e./liter % % 0.5, 1, and 1.5 ml. of (D) in the 57 100 14.85 14.94 14.90 0.30 3 15.0 15.0 15.0 0.00 + 0 .7 same manner. From the photometer 62 100 13.76 13.78 13.77 0.07 3 14.0 14.1 14.1 0 .3 5 + 2 .4 readings plot a calibration curve on 70 200 2 .4 9 2.49 2.49 0.00 10 2.40 2.45 2.43 1.03 - 2 . 4 84 200 4.38 4.39 4 .3 9 0.11 10 4 .5 0 4 .5 3 4 .5 2 0 .3 3 + 3 .0 semilogarithmic graph paper. The 85 50 41 .5 4 41.60 41.57 0.07 1 42.5 42.6 42.6 0 .1 2 + 2 .5 milliequivalents of nitrate in the ali­ 86 50 71.70 7 1 .8 0 71.75 0.07 1 72.9 73.0 73.0 0.07 + 1.7 4 7.90 4 7.74 0 .3 4 1 48.5 48.8 48.7 0.31 + 2.0 quot are determined by interpolation 314 50 4 7 .5 8 on this curve. A typical curve is Av. 0 .1 4 0 .3 2 2 .1 shown in Figure 4...... - If a qualitative nesslerization test indicates a measurable amount of ammonia, both ammonium and ni­ The distillation equipment includes microburners, 100-ml. trate can be determined on the same aliquot by the following Kjeldahl flasks, and “inverted U” air condensers of 1.9-cm. modification. Distill,nesslerize, and measure light transmis­ (0.75-inch) diameter. (The glassware is obtainable from the sion in the manner described, omitting (A) and substituting Ilengar Company, Philadelphia, Penna.) A distillation rack a boiling stone. Now immerse the tip of the rinsed delivery tube in a fresh mixture of 3 ml. of (C) and 20 ml. of water in of six units is convenient. The Nessler reagent is prepared a clean beaker; add to the cooled Kjeldahl flask 0.50 gram of according to Koch and McKeekin (SO). The distilled water (A) and 15 ml. of water. Immediately connect to the condenser, should be practically free of nitrogen compounds. distill, and treat in the same manner. R e a g e n t s . A. Devarda's alloy. Boil a quantity in 0.2 N To obtain the ammonium calibration curve, distill, without sodium hydroxide for a few minutes to reduce the nitrogen con­ Devarda’s alloy, a series of 0, 0.2, 0.5, 1, and 1.5 ml. of (E), tent. Wash and dry. B. 2 N sodium hydroxide. Dissolve 80 grams of nitrogen-free sodium hydroxide in 1 liter of water. C. 0.01 N sulfuric acid. D. Standard 0.01 N potassium nitrate. Dis­ solve 1 . 0 1 1 grams of dry recrystallized potassium nitrate in water and dilute to exactly 1 liter. E. Standard 0.01 N ammonium sulfate. Dis­ solve 0.6607 gram of pyridine-free ammonium sulfate in water and dilute to exactly 1 liter. F. Nessler reagent (19). “Dissolve 22.5 grams of iodine in 20 cc. of water containing 30 grams of potassium iodide. After the solution is com­ pleted, add 30 grams of pure metallic mercury, and shake the mixture well, keeping it from be­ coming hot by immersing in tap water from time to time. Continue this until the supernatant liquid has lost all of the yellow color due to iodine. Decant the supernatant aqueous solution and test a portion by adding a few drops thereof to 1 • cc. of a 1 per cent soluble starch solution. Unless the starch test for iodine is obtained the solution may contain mercurous compounds. To the remaining solution add a few drops of an iodine solution of the same concentration as em­ F i g u r e 4. Photom eter C alibration Curve for N itrate ployed above, until a faint excess of free iodine can be detected by adding a few drops thereof to 1 cc. of the starch solution. Dilute to 200 cc. and mix well. To 975 cc. of an accurately prepared 10 per cent sodium nesslerize, and measure the light transmission. Prepare the hydroxide solution now add the entire solution of potassium nitrate calibration curve as previously outlined. Determine the mercuric iodide prepared above. Mix thoroughly and allow to milliequivalents of ammonium and nitrate ions in the aliquot by clear by standing.” Keep in a brown bottle. interpolation on their respective curves. Procedure. Pipet an aliquot containing 0.002 to 0.015 m. e. of nitrate into a 100-ml. Kjeldahl flask. (If the sample contains If it is suspected that hydrolyzable nitrogenous organic com­ carbonate, barely neutralize the aliquot with dilute sulfuric acid.) pounds are increasing the ammonium value, an additional dis­ Add 0.50 gram of (A) and dilute to 30 ml. Immerse the tip of the tillation should be made in which the Kjeldahl flask contents are delivery tube in a mixture of 3 ml. of (C) and 20 ml. of water in a buffered at pH 7.4, according to Shrikhande (31). A series of 1 0 0 -ml. beaker. ammonium standards can be treated in the same manner. Carefully run 2 ml. of (B) down the neck of the flask. Connect the flask to the air condenser by the r— ...... rubber sleeve and twirl it, to mix T a b l e IX. Comparison of M acro- and Semimicrometiiods for N itrate

the contents. Heat the flask with Î a microbumer until 15 ml. have been i

C alculation. M. e. of N 0 3 per liter = (m. e. of N 0 3 in aliquot (3) Am. Soc. Testing Materials, P art III, p. 541, 1940. as found by interpolation on NOj curve X 1000) -5- ml. in aliquot. (4) Anderson, M. S., Keyes, M. G., and Cromer, G. W., U. S. Dept. M. e. of NH< per liter = (m. e. of NTI< in aliquot as found by Agr., Tech. Bull. 813 (June, 1942). interpolation on NH( curve X 1000) -f- ml. in aliquot. (5) Assoc. Official Agr. Chem., "Official and Tentative Methods of Analysis”, 5th ed., 1940. Precision and Accuracy (6) Barber, H. H., and Kolthoff, I. M., J. Am. Chem. Soc., 50, 1625 (1928). The seven soil extracts were analyzed for nitrate by this (?) Benedetti-Pichler, A . A ., Cefola, M., and Waldman, B., I n d . method and by a regular Devarda procedure (89, p. 21) in E n g . Chem., A nal. Ed., 11, 327 (1939). (8) Blasdale, W. C., J. Am. Chem. Soc., 31, 917 (1909). which the ammonia is collected in boric acid solution and (9) Clark, E. P., and Collip, J. B„ J. Biol. Chem., 63, 461 (1925). titrated with 0.05 N sulfuric acid. For this purpose, no pre­ (10) Eaton, F. M., and SokolofT, V. P., Soil Sci., 40, 237 (1935). liminary separation of ammonia was made, and the values in (11) Fiske, C. H„ J. Biol. Chem., 47, 59 (1921). Table IX include all nitrogen that would be liberated under (12) Hibbard, P. L., Soil Sci., 8, 61 (1919). (13) Hibbard, P. L.t and Stout, P. II., J. Assoc. Official Agr. Chem., the analytical conditions.. 16, 137 (1933). The extracts of soils S4 and 8 6 contained so little nitrate (14) Hillebrand, W. F., and Lundell, G. E. F., “Applied Inorganic that the results by the titration method are probably in­ Analysis” , p. 510, Now York, John Wiley & Sons, 1929. accurate. By excluding these from the accuracy compari­ (15) Hirsch, A. A., Ind. Enq. Chem., A nal. Ed., 14, 943 (1942). sons, the average “error” of the semimicromethod is reduced (16) Hoffman, W. S., J. Biol. Chem., 120, 57 (1937). (17) Hoffman, W. S., and Osgood, B„ Ibid., 124, 347 (1938). from 4.0 to 1.5 per cent. As for some other ions, the semi­ (18) Jacobs, H. E. D., and Hoffman, W. S., Ibid., 93, 685 (1931). micromethod sometimes may actually provide more accurate (19) Johnston, J., J . Am. Chem. Soc., 38, 947 (1916). results than the comparison method. The colorimetric (20) Koch, F. C., and McKeekin, T. L., Ibid., 46, 2066 (1924). method shows a satisfactory degree of precision. (21) Kramer, B., and Tisdall, F. F., J. Biol. Chem., 46, 339 (1921). (22) Morris, V. II., and Gerdel, R. W., Plant Physiol., 8, 315 (1933). (23) Nichols, M. S., and Foote, M. E., Ind. Eng. Chem., A n a l . E d ., D iscussion 3, 311 (1931). (24) Peech, M., Ibid., 13, 436 (1941). The volume of sample used in the semimicroanalysis of the (25) Pucher, G. W., Vickery, H. B„ and Leavenworth, C. S., Ibid., 7, seven soil extracts comprised Vis to '/36 of that used in the 152 (1935). comparison methods, with an average of ‘/V This repre­ (26) Purvis, E. R., and Higson, G. E., Jr., Ibid., 11, 19 (1939). (27) Reitemeier, R. F., and Richards, L. A., unpublished manuscript. sents a considerable reduction in sample requirements. In (28) Richards, L. A., Soil Sci., 51, 377 (1941). general, semimicroanalysis requires less time, although no (29) Rosenheim, O., and Drummond, J. C , Biochem. J., 8, 143 quantitative comparisons have been made. The economy (1914). effected in the analytical reagents is often important. (30) Schollenberger, C. J., Soil Sci., 24, 65 (1927). (31) Shrikhande, J. G., I n d . E n g . Chem., A nal. Ed., 13, 187 (1941). The accuracy obtainable under these conditions is not (32) Smith, G. F., and Getz, C. A., Ibid., 10, 304 (1938). seriously reduced, and is adequate for most soil analyses. (33) Treadwell, F. P., “Analytical Chemistry. Volume II. Quanti­ The average predictable error of the methods is about 2 per tative Analysis”, tr. and rev. by W. T. Hall, 7th ed., p. 604, cent, based on comparisons with methods involving much New York, John Wiley & Sons, 1930. (34) Truog, E ., and Meyer, A . H., I n d . E n g . Chem., A nal. Ed., 1, larger samples. Many of the methods are sufficiently precise 136 (1929). that replication of determinations usually is unnecessary. (35) Volk, N. J., J. Am. Soc. Agron., 33, 685 (1941). The quantitative reproducibility data presented for the (36) Wall, M. E„ Plant Physiol., 15, 537 (1940). various methods should assist in determining the desirability (37) Wang, C. C., J. Biol. Chem., I l l , 443 (1935). (38) Warder, R. B., Chem. News, 43, 22S (18S1). of replication, based on particular requirements. (39) Wilcox, L. V., “Methods of Analysis Used in the Rubidoux Laboratory, Riverside, Calif.”, mimeographed, Division of Acknowledgment Irrigation Agriculture, Bur. Plant Industry, U. S. Dept. Agr., 3rd ed. (January, 1941). The assistance of Betty Mabry, Barbara Pederson, K. R. (40) Wilcox, L. V., Ind. E ng. Chem., A nal. E d., 9, 136 (1937). Goodwin, L. W. Healton, L. R. Weaver, and A. F. Wendel (41) Zinzadze, Ch., Ibid., 7, 227 (1935). in developing and testing these methods is gratefully acknowl­ edged. C ontribution from the U. S. Regional Salinity Laboratory, Bureaus of Plant Industry, Soils, and Agricultural Engineering, Agricultural Research Literature Cited Administration, U. S. Department of Agriculture, Riverside, Calif., in co­ (1) Allison, L. E., Soil Sci., 40. 311 (1935). operation with the eleven western states and the Territory of Hawaii. This (2) Am. Pub. Health Assoc., "Standard Methods for Examination article is a revised edition of a mimeographed publication given limited dis­ of Water and Sewage”, 8th ed., pp. 64-8, New York, 1936. tribution since November, 1941.

r m T k gB C « _ it it A M it- __ o il T fL affflW lt Estimation of the Sulfonamides A Rapid and Accurate Micromethod

S. W. LEE, N. B. IIANNAY, AND W. C. HAND Wallace Laboratories, Inc., New Brunswick, N. J.

ECENT work in the sulfonamide field, in which at­ furic acids (“acid mixture”). A small amount of sulfuric acid R tempts are being made to correlate low blood levels was found to aid materially in a quick and complete precipi­ with efficacy, has made exact determinations of these drugs tation of the protein, and was present in sufficient amount for more important than ever. This need, and the desire to ob­ hydrolysis purposes, for use in the “total” sulfonamide de­ tain the advantages of working on a micro scale, have led to a terminations. Laking the blood prior to precipitation is un­ method which has most of the advantages of the methods necessary (Table I). now in wide use, and practically none of their shortcomings. The precipitated protein is filtered off, the diazotization is carried out, ethyl alcohol is added, and the naphthylethylene- diamine dihydrochloride is added immediately. The color T a b l e I. Free Sulfathiazole Levels attains its maximum intensity in 15 seconds. In the presence Regular Brat­ Micromethod of alcohol, it is not necessary to destroy the excess nitrous Subject and Sulfathia- ton and Mar­ Blood laked Blood pptd. acid, for possible products formed by its reaction with the dye zole Dose shall Method before pptn. directly are likely to be of the same color and soluble in the medium. Mo. % M g. % Mg. % Rabbit, 0.5 gram orally, No nitrogen bubbles are formed, because the sulfamate- blood taken after one nitrous acid reaction is eliminated, and little or no inter­ hour 3.0 (trip.) 3.5 (trip.) Man, 2 grams taken or­ ference from bubbles (7) was noticed. Parallel experiments ally, blood taken after showed that the sulfamate addition is not necessary, the 2 hours 2.9 (dupl.) 3.6 (dupl.) Man, 1 gram taken or­ colors being even more stable in its absence. The recoveries ally, blood taken after 2 hours 2 . 8 3.3 (dupl.) 3.3 (dupl.) obtained by different means are shown in Tables I, IV, and V. Rabbit, 0.5 gram taken orally, sample after 1 hour 3.5 (quad.) 4.0 (dupl.) 4.0 (dupl.) In the tables, results are expressed in different terms for the Man, 2 grams orally, sake of clearness. In some cases milligram per cent figures are sample after 1 hour 2.3 (quad.) 2.6 (quad.) 2.6 (trip.) Man, 2 grams orally, given; in others, actual density readings from the drum of the sam ple a fte r 4 hours 2.7 (trip.) 3.1 (quad.) 3.1 (quad.) Coleman spectrophotometer are given. Density is defined as the logarithm of the reciprocal of the fraction of transmitted light. If Beer’s law is obeyed, as it is in this case, the density is a linear function of the concentration. All results have been checked. The methods most commonly used are those of Bratton Elapsed time was measured with a stop watch. All blood was and Marshall (2) and Werner {8), or modifications or adapta­ added at the ratio 1 to 20. All blood samples were whole and oxalated, and usually less than 0.5 hour old. It was found that tions of them. In the former procedure, the sulfonamide is blood which contained sulfathiazole increased in apparent sulfa diazotized, and the diazonium salt coupled with JV(l-naph- drug content by about 10 per cent after standing at room temper­ thyl)ethylenediamine dihydrochloride. The azo dye which is ature for 20 hours. Appropriate concentrations of trichloroacetic formed is determined colorimetrically. The undesirable fea­ acid were present in all diazotizations. All density readings were made against reagent blanks of the same age. tures of this procedure have been mentioned in a preliminary note on the method under discussion (5). In the Werner method, p-dimethylaminobenzaldehyde is used to form a yel­ Accurate analysis was possible for a period of 24 hours. low anil with the “free” sulfonamides, and the intensity of The stability of the colors in the various methods is shown in this color is measured. It is difficult to use this method for Table II. accurate micro work because of the low tinctorial value of the Total Sulfa Drugs. In the determination of total sulfa yellow dye. It has also been pointed out that “total” values drug concentration, the Bratton and Marshall method is obtained with these methods are subject to error resulting much to be preferred to the shorter method of Werner (<9), from the change in color intensity with small changes in pH primarily because of the negligible effect of mineral acids on (6). the intensities of the azo colors formed. In the micromethod The time required for a single analysis is greatly reduced 0.1 ml. of 4 iV sulfuric acid is used in the hydrolysis, although using the micromethod described below; an analysis may be 0.4 ml. of the acid does not seriously alter the density read- completed in 7 or 8 minutes. This is even less than the time . necessary for the Werner method, which is about 12 minutes (/). Reasonable amounts of sodium chloride do not inter­ fere. Fifty mg. per cent or less of potassium thiocyanate (6) T a b l e II. Stability op Colors Formed in Various M e t h o d s did not interfere in the determination of 1 0 mg. per cent of D en sity R eading of a 10-/zg. Sam ple sulfathiazole by the Bratton and Marshall or the micro­ of Sulfathiazole after: 5 1 2 3 4 24 method. This concentration of thiocyanate is much higher M ethod min. hour hours hours hours hours than is ever obtained in the blood. Blank readings on normal Aqueous sulfathiazole solu­ blood (human, horse, and rabbit) were found to be zero as a tion, Bratton and Mar­ shall procedure 0.25 0.25 0.20 0 .1 5 rule, and never exceeded 0 . 2 mg. per cent. Aqueous sulfathiazole solu­ tion, micromethod with­ out sulfuric acid 0 .2 5 0.2 5 5 0 .2 6 0 .2 6 0 .2 6 Determination of Sulfa Drugs Aqueous sulfathiazole solu­ tion, micromethod. 0 .2 5 0 .2 6 0 .2 6 0 .2 6 0 .2 5 Sulfath azole in rabbit blood Free Sulfa Drxjgs. In the micromethod, the blood is filtrate, micromethod 0 .2 5 0.245 0.2 3 5 precipitated directly in a mixture of trichloroacetic and sul­ 403 404 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

added to 2 . 0 0 ml. of solution, both in the reaction with T a b l e III. Recoveries of Sulfathiazole Added to U rine and without added sulfuric acid. The indicated presence of sodium chloride means that 1.0 ml. of 0.7 per cent saline was Recovered by Brat­ Recovered by Mi- croinethod added. Final Dilution ton and Marshall The experimental conditions for diazotizations at 35° C. were % % the same as those in Table IV. 1 : 1 0 90 90 1:20 92 98 1:30 96 99 1:60 99 100 The addition of a small amount of sulfuric acid to the usual trichloroacetic acid decreased the recovery very slightly. The recovery of sulfathiazole added to whole blood, using the micromethod, is almost quantitative (95 to 100 per cent). T a b l e IV . R e c o v e r ie s o f S ulfathiazole f r o m H o r s e B l o o d A comparison of recoveries is given in Table IV. (Regular procedures as outlined were followed. Sulfathiazole was added ■ I i 1. A 1 1 - _ . . i . t i \ ■ ■ 1 x] am m 4- r\ ■ n O A O 1* 1 A M rt f 1n The variation in recoveries among the sulfa drugs has been drug.) explained on a solubility basis: the more insoluble the sul­ Sulfathiazole Recovered fonamide, the greater the error in the determination due to 2 mg. 4 mg. 6 mg. 8 mg. 10 mg. % . loss in preparing the blood filtrate (4) • This explanation seems M ethod adcfed added adcfed added added Milligram per cent unlikely when one considers that only micrograms are present in solutions in which tenths or hundredths of grams will dis­ Bratton and Marshall 1.85 5.27 8 .5 Micromethod 1.80 3.76 5.8 7.75 9 .5 solve, so that solutions are often one ten-thousandth satu­ Micromethod without sul­ rated. The sulfa drugs vary in solubility by factors of 10 or furic acid 1.95 4.0 5.8 7.84 9 .6 20. Solubility would thus be expected to exert little influence on the recoveries. Adsorption of the drug on the precipitated protein can play ings. If the acid mixture of trichloroacetic and sulfuric acids no important role in producing the low results, for complete is used for precipitation of the blood, the filtrate may be used (98 to 100 per cent) recovery can be obtained by the micro­ directly for the total sulfa drug determination. method after blood has been precipitated in the presence of Sulfa Drugs in Urine. The recovery of sulfathiazole the sulfa drug at 1 to 20 dilution. The Bratton and Marshall added to various dilutions of normal urine is shown in Table procedure gives 85 per cent recoveries under these conditions. III, compared with results obtained on the same samples by On the blood of subjects who received sulfathiazole orally, the Bratton and Marshall procedure. the micromethod gives results that are about 15 per cent higher than those given by the Bratton and Marshall method. Recovery of Sulfa Drugs from Whole Blood This was taken to indicate complete recovery by the modified method, in conjunction with the results given in Table V. Much has been written (7) about the recovery of sulfa Table I gives typical blood analyses by the two procedures, drugs from blood. It is the general opinion that, using the and shows that laking the blood before precipitation is not Bratton and Marshall method, added sulfathiazole is re­ necessary. covered from whole blood to the extent of 85 to 90 per cent, at the dilution of 1 to 20. This was =_ _ _ confirmed, as shown in Tables IV and V. Work previously reported (6) and experiments carried T a b l e V . R e c o v e r i e s o f S ulfathiazole A d d e d t o V a r io u s M e d ia out in this laboratory indicate that the filtrates Diazotization Sulfathia­ Sulfathiazole Temper­ zole Added Found from precipitated blood (with added sulfathiazole Method Time ature M edium 1 2 1 2 and at 1 to 2 0 dilution) contain practically all the M in . ° C. M g. % M g. % sulfathiazole (95 to 100 per cent). Low recoveries B. & M. 3 ca. 20 Water and CCIiCOjH 4 8 3 .2 6 .0 B. & M. 3 ca. 20 Water and CCUCO- 4 8 4.0 8.0 were obtained with the Bratton and Marshall (in presence of NaCl) OH B. & M . 10 ca. 25 Water and CCI1CO2H 4 8 4 .0 8 .0 procedure when the sulfathiazole was added to B.

These results indicate that the low recoveries obtained by the Bratton and Marshall method may be due in some cases T a b l e VII. R ates of Diazotization of Several Sulfa to incomplete diazotization. The rate of diazotization is in­ D rugs in Various M edia fluenced by the temperature, sodium chloride and other Per Cent Diazotized 0 .5 1 2 3 5 catalysts, mineral acid content, and possibly by retarding Medium and Method min. min. min. min. min. substances in the blood filtrates. If the routine time for di­ Aqueous sulfacetamide solution, B. & M . procedure 100 100 100 100 100 azotization is to be 3 minutes, careful note must be made of Aqueous sulfanilamide solution, these conditions. Higher recoveries (by the Bratton and B. & M . procedure 46 62 79 92 96 Aqueous sulfapyridino solution, Marshall method) were frequently obtained when diazotiza­ B. & M . procedure 58 75 92 96 100 Aqueous sulfacetamide solution, tion was carried out for longer than 3 minutes, or when the micromethod 100 100 100 100 100 temperature was markedly higher than 25° C. Addition of Aqueous sulfanilamide solution, micromethod 79 98 100 100 100 sodium chloride to one of the reagents in the regular Bratton Aqueous sulfapyridine solution, micromethod 100 100 100 100 100 and Marshall procedure would preclude the chance of in­ Sulfathiazole in 2 ml. of distilled complete diazotization, and reduce the time for this step to 1 water containing CCUCO:li minute. Cooper, Gross, and Hogan (3) recommend diluting and 1 ml. of alcohol 50 the blood with normal saline. Using this method they ob­ tained high recoveries, which they attributed to the preven­ tion of hemolysis on diluting the blood. It might be possible that they are, in part, due to the catalytic effect" of the salt on the diazotization. in 10 ml. flat-bottomed vials. These were found to be convenient, since they do not require racks, and allow complete mixing of the reagents by shaking. To each of a series of vials were added 0.47 ml. of water, 0.43 ml. of 15 per cent trichloroacetic acid, 0.10 ml. of 4 AT sulfuric acid, and 1.00 ml. of water or one of the various concentrations of sulfathiazole solutions. The procedure gave solutions comparable to 2 ml. of blood filtrate. T a b l e VI. Rates of Diazotization of Sulfathiazole in The sulfathiazole solutions were made up as follows: 0.1000 Various M edia gram of c. p. sulfathiazole was dissolved in 1 liter of distilled Per Cent Reaction after Diazotizing water. Various amounts of this solution (2, 4, 6 , 8 , and 10 ml.) for: were diluted to 100 ml. The resulting dilutions contained 2 , 4, 6 , 0 .5 1 2 3 5 8 Medium and Method min. min. min. min. min. , and 10 micrograms of sulfathiazole per ml. The color was de­ veloped as in the micromethod, using 0 . 1 0 ml. of nitrite, waiting A t 25° C. 3 minutes, and then adding 1.00 ml. of alcohol and 0.10 ml. of Aqueous sulfathiazole solution, Bratton and Marshall’s reagent. By plotting the densities (or B. & M. procedure 4G 62 87 90 100 Rabbit blood filtrate, B. Sc M . the per cent transmission) of the colors against the concentration procedure 58 79 92 95 100 (2, 4, 6 , 8 , and 10 micrograms per determination), a very useful Aqueous sulfathiazole solution, calibration curve was obtained. If the outlined procedure for added NaCI, B. & M . proce­ dure 96 100 100 100 100 blood is followed, the same scale (2, 4, 6 , 8 , and 10) will correspond Aqueous sulfathiazole solution, to the milligram per cent of sulfathiazole in the sample taken. In micromethod 83 100 100 100 100 other words, milligram per cent is the same as micrograms per Rabbit blood filtrate, micro­ m ethod 83 100 100 100 100 0.1 ml. Beer’s law is exactly obeyed to a concentration of at least Rabbit blood filtrate, micro- 1 2 micrograms per determination. method without sulfuric acid 83 96 97 . 100 100 A t 35° C. Acmeous sulfathiazole solution, B. & M. procedure 67 83 100 100 100 Procedure for Blood and Urine Rabbit blood filtrate, B. & M. procedure 83 100 100 100 100 Precipitation of the Blood. Whole blood (0.30 ml.) is Aqueous sulfathiazole solution, micromethod 100 100 100 100 100 added dropwise to 5.70 cc. of trichloroacetic acid, or to the same Aqueous sulfathiazole solution, volume of “acid mixture”. Each drop of blood is broken up by micromethod, without sulfurio vigorous stirring with the end of the pipet, and the tube is agi­ acid 93 100 100 100 100 tated by hand after all the blood is added. The mixture is allowed to stand until the precipitated protein settles and is then filtered through either No. 1 or 42 Whatman paper of about 6 -cm. di­ ameter. The filtrates are uniformly water-clear. Slightly more than 4 ml. are collected, sufficient for the determination of both “free” and “total” sulfonamides. Experimental Procedure Determination of Free Sulfa Drug in W hole Blood. R e a g e n t s . Trichloroacetic acid, 3.33 per cent. Exactly 2 ml. of the filtrate (from either precipitating acid) are Trichloroacetic and sulfuric acid (“acid mixture”). Sulfuric pipetted into 1 0 ml. flat-bottomed vials, sodium nitrite (0 . 1 0 ml. acid (56 ml. of 4 AT) is added to 1 liter of 3.33 per cent trichloro­ of 0.1 per cent) is added, and at least 3 minutes are allowed for acetic acid. the diazotization. Alcohol (1.00 ml.) is added, and the tube is Sodium nitrite. An aqueous solution of c. p. sodium nitrite swirled. On mixing Bratton and Marshall’s reagent (0.10 ml.) (0.1 per cent) is used. This was found to be very stable (more with the solution, the characteristic color is produced. The for­ than a month in summer weather). It was renewed when low mation of the color is complete within 15 scconds, and the densi­ readings were obtained from known amounts of sulfa drugs. ties of the solutions are reasonably constant for 24 hours. Read­ Bratton and Marshall’s reagent. iV(l-naphthyl)ethylenedia- ing in all cases should be made against reagent blanks, and values mine dihydrochloride (0.1 per cent in water) was used. It was obtained from a calibration chart. found possible to use this for a matter of months also, providing Determ ination of T otal Sulfa Drugs in W hole Blood. the determinations are made against reagent blanks. Exactly 2 ml. of the filtrate from the “acid mixture” precipita­ Ethyl alcohol, undenatured 95 per cent ethanol. tion, or 2 ml. from the trichloroacetic acid precipitation with 0 . 1 0 Instrum ent. All determinations wrere made with a Coleman ml. of 4 AT sulfuric acid added, are heated in a boiling water bath Universal spectrophotometer, with the wave length dial set at for about one hour. The volume is adjusted to 2.00 ml. with dis­ 550 m/j. All colors were compared against reagent blanks of the tilled water, and the procedure for the free determination is fol­ same age, in microcuvettes of 2.5-ml. volume. The similarity in lowed. the absorption curves of the dyes formed in the two methods Determ ination of Sulfa Drugs in Urine. Urine (0.10, would indicate the method to be equally useful with other colori­ 0.20, or 0.30 ml.) is added to the mixed acid blood-precipitating metric instruments. reagent (5.90, 5.80, or 5.70 ml., respectively), representing dilu­ Calibration Curve. Calibration curves were run in dis­ tions of 1 to 60, 1 to 30, or 1 to 20. The pipet is rinsed with the tilled water containing the concentration of acids which were solution. The sample is filtered if any turbidity develops. The present in the regular analyses. All the final colors w'ere formed solution or filtrate (2 . 0 0 ml.) is analyzed according to the method 406 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 outlined for blood. The colors formed in the presence of diluted and Marshall procedure. The micromethod has proved very urine are stable for only an hour, with both Bratton and Mar­ shall’s and the micromethods. reliable and useful for large-scale work.

Sum m ary Literature Cited (1) Andrews, M. C., and Strausa, A. F., Lab. Clin. Med., 26, 888 A micromethod for the estimation of sulfonanvdes is based (1941). on the coupling of completely diazotized sulfonamides to (2) Bratton, A. C., Marshall, E. K., Jr., Babbitt, Dorothea, and N (l-naphthyl)ethylenediamine. Changes in the procedure Hendrickson, A. R., J. Biol. Chem., 128, 537 (1939). have removed most of the sources of objection to the Bratton (3) Cooper, F. B., Gross, Paul, and Hogan, M. L., Am. J. Clin. Path., and Marshall method. The time for a single complete analy­ 12, 149 (1942). (4) Hoffman, W. S., “Photelometric Clinical Chemistry”, p. 224, sis has been reduced to about 8 minutes. The colors formed Now York, William Morrow and Co., 1941. are stable for 24 hours. -Recovery of sulfathiazole added to (5) Lee, S. W., Hannay, N. B., and Hand, W. C., Science, 97, 359 whole blood has been shown to be essentially complete at a (1943). dilution of 1 to 20. On blood from subjects receiving the drug (6) Morris, C. J. 0., Biochem. J., 35, 952 (1941). (7) Sunderman, W. F., and Pepper, D. S., Am. J. Med. Sei., 200, 792 orally, the micromethod gives results which are uniformly (1940) about 15 per cent higher than those obtained by the Bratton (8) Werner, A. E. A., Lancet, 1, 18 (1939).

Microdetermination of Mercury in Organic Compounds

H. WILLIAM ECKERT Division of Laboratories and Research, New York State Department of Health, Albany, N. Y.

D ESPITE the desirability of a quantitative method for Standardization of the titration technique also contributes the determination of mercury in biologic products that greatly to the accuracy of the determination,-as described have been preserved with phenyl mercuric acetate, Merthio- below. late, or other organic mercury compounds, no entirely satis­ Under the conditions described, no evidence of oxidation factory micromethod has been published. In the author’s of the dithizone has been encountered. experience the Gettler and Lehman modification (2) of the Winkler method (6, 7) for the determination of mercuric Apparatus. Pyrex reflux apparatus consisting of a 50- or salts, although satisfactory for mercuric ions, has given very 1 0 0 -ml. round-bottomed flask and Liebig-type condenser with a low results with organic mercury compounds such as those 25-cm. outside water-cooled jacket. R e a g e n t s . Dithizone Solution. Dissolve 12.5 to 13.0 mg. of mentioned, except under certain circumstances when phenyl diphenylthiocarbazone (Eastman) in 500 ml. of carbon tetra­ mercuric acetate may be titrated with dithizone as a monova­ chloride (Baker, analyzed, in bottles), and allow to stand for a lent ion. The Gettler and Lehman method consists, briefly, day in the dark. It is important to use the best carbon tetra­ of a nitric acid and permanganate digestion followed by de­ chloride obtainable. Filter through paper and store in a dark brown bottle in the dark. struction of excess permanganate by nitrite, removal of ex­ The titration value of this reagent remains constant for at least cess nitrous acid by hydroxylamine sulfate, and titration of a month. the metal ion by dithizone. Hydroxylamine Sulfate Solution. Dissolve 20 grams of hy­ droxylamine sulfate (Eastman) in 100 ml. of water. Experimental Since the losses mentioned above might be due either to incomplete digestion or to volatilization, many modifications T a b l e I. V ariations in T itration of M ercury w ith of the digestion conditions were made in an attempt to avoid D i t h i z o n e these errors. Two methods gave satisfactory results: Carius (0.109 mg. mercury in the presence of different concentrations of acids and salts) digestion in a microbomb and treatment of the material with Amounts of Salts and Concentrated Dithizone in aluminum in a neutral or slightly alkaline medium at 75° to Acids in 100 Ml. of Solution Titrations A verage 80° C. The disadvantages of the first were the difficulty in M l. M l. M l. M l. 500 mg. of Tyrode's salts0 manipulation and the formation of chlorine oxides which had 0,5 ml. of HsSOj 11.4 11.3 11.3 11.3 to be removed. The reaction with aluminum provided a 0.5 ml. of H3SO< + 1 ml. of HNOj 11.4 11.4 11.3 U . 4 0 .5 ml. of HjSCU + 2 m l. of H N O j 11.2 11.2 11.3 11.2 practical digestion method and its application to the deter­ 0.5 ml. of HiSO* 4* 3 ml. of HNOj 11.3 11.3 11.4 11.3 mination of organic mercurials is the subject of this report. 0.5 ml. of HiSOi + 5 ml. of HNOi 11.3 11.2 11.2 11.2 N o added salts The reaction between metallic aluminum and the organic 0.5 ml. of H:SO( 10.2 9.7 9 .9 9 .9 mercurials studied ran smoothly and quantitatively at 75° to 0.5 ml. of H iSOj + 1 ml. of HNO j 9 .4 9 .9 10.0 9 .8 0.5 ml. of H 2SO4 + 2 ml. of HNOj 9 .7 10.1 9 .9 9 .9 80° C. at an initial pH range between 7.8 and S.4. 0.5 ml. of HsSOi + 3 ml. of HNOj 10.3 10.0 9 .7 10.0 In order to obtain consistent end points in the dithizone 0.5 ml. of H jSO* + 5 ml. of H N O j 9 .5 11.2 10.3 10.3 0.5 ml. of HiSO* + 5 ml. of HNOj titration, a certain minimum concentration of salts in the 0.5 gram of NaCl 11.3 11.4 11.3 11.3 aqueous phase is essential. Amounts exceeding the mini­ 1.0 gram of NaCl 11.4 11.3 11.2 11.3 2.0 grams of NaCl 11.2 11.3 11.4 11.3 mum have no detrimental effect. Varying the concentration a Composition of Tyrode’s salts (4): NaCl 8.0, KC1 0.2, CaClj 0.2, MgClj of acids over the range studied has no effect on the titration 0.1, NaliiPOi 0.05, NaHCOj 1.0, and d-glucose 1.0 gram. (Table I). June 15, 1943 ANALYTIC L E D I T I O N 407

Mercury Standard Solution. Dissolve 500 mg. of mercury in concentrated nitric acid and dilute to 500 ml. Add 10 ml. of T a b l e II. Recovery op M ercury from M ixtures Contain­ concentrated nitric acid to 10 ml. of this solution and dilute to ing Phenyl M ercuric A cetate and M erthiolate 1000 ml. This solution contains 0.0100 mg. of mercury per 1 ml. Additional Substances Theory Found R ecovery Concentrated Nitric Acid (Baker, c. p., analyzed, or Grasselli). M g. M g. % Concentrated Sulfuric Acid (Baker, c. p., special). + + Aluminum Powder (Merck), best quality. Phenyl mercuric acetate equivalent to 0.200 mg. of Hg 500 mg. of Tyrode’s salts 0.200 0.196 98.0 The actual reagents used are specified, since they are known 0.200 0.195 9 7 .5 to be satisfactory. Equivalent products of other manufacture 0 .200 0.198 9 9 .0 0 .200 0.194 9 7 .0 may be substituted. 0 .200 0.194 9 7 .0 Procedure. Transfer the sample containing 0.10 to 0.20 mg. 0 .200 0.196 9 8 .0 of mercury to a round-bottomed flask, and adjust with 0.1 N 500 mg. of Tyrode’s salts and 6 drops of ethyl alcohol 0.200 0.204 102.0 sodium hydroxide and 0.1 N hydrochloric acid to pH 7.8 to 8.4, 10 ml. of antipneumococcus horse using phenolphthalein as indicator. Concentrate the solution serum, type 1 0 .200 0.201 101.5 to 2 to 3 ml. on a boiling water bath unless volatile mercury com­ 0.200 0.203 101.5 pounds, such as Merthiolate, are present, in which case take an 1 gram of dried egg yolk (powder) 0 .200 0.212 106.0 aliquot without attempting to concentrate. Add 100 to 200 mg. Merthiolate equivalent to 0.167 mg. of Hg + * of aluminum powder and reflux on a water bath at 75° to 80° C. 500 mg. of Tyrode’s salts 0.167 0.159 9 5 .2 0.167 0.169 101.2 overnight. An ordinary constant-level water bath may be ad­ 10 ml. of antipneumococcus horse justed to hold this temperature with sufficient accuracy by serum, type 1 0.167 0.169 101.2 simply regulating the height of the burner. Creeping of the aluminum along the walls of the flask, which reduces the surface exposed to the solution, may be prevented by adding a few drops of ethyl alcohol. D iscussion Cool the flask, add 0.5 ml. of concentrated sulfuric acid through the condenser, and replace on the water bath (75° to 80° C.) The recovery experiments demonstrate that the method until nearly all the aluminum has dissolved. Add 5 ml. of con­ presented is useful in the presence of considerable amounts of centrated nitric acid by way of the condenser, wash down the con­ denser with a few milliliters of water, and continue heating until salts and organic material for the determination of mercury the solution clarifies. Cool the flask and thoroughly rinse the in phenyl mercuric acetate and Merthiolate. The fact that condenser with water into the flask. Transfer the mixture to a these compounds contain mercury in two different types of pear-shaped 150-ml. separatory funnel, filtering if necessary, linkage suggests the possible application of the method in and dilute to 100 ml. Glycerol is a satisfactory lubricant for the glass stopper and stopcock of the separatory funnel. Cool the determining mercury in other organic mercury compounds. solution, add 5 ml. of hydroxylamine sulfate solution, shake the Preliminary investigation of the reaction between aluminum mixture, and allow to stand a few minutes. Unless salts are and organic mercury compounds as a possible step in deter­ known to be present in the original sample, add 0.5 to 1.0 gram mining much smaller amounts of organic mercury compounds of c. p. sodium chloride and titrate. In performing the titration, the solution must not be exposed colorimetrically with dithizone has shown definite promise. to direct sunlight, because this will promote destruction of dithi- The metals most likely to interfere with the determination zone. Add 2 to 3 ml. of dithizone solution to the separatory are copper, silver, gold, and palladium. Of these only copper funnel, stopper the funnel, and shake until the carbon tetrachlo­ may be present in significant amounts in biologic products. ride layer is bright orange. Unless the amount of mercury is very small, about 20 shakes are sufficient. The shaking of the Methods exist for the separation of copper from mercury separatory funnel should bo uniform, counting the forward and {6, 7, 8) and prevention of interference of copper (5). In ac­ backward stroke as one shake. Should the carbon tetrachloride tual experience, the blanks have been very low on practically layer remain green, add a known amount of mercury standard, all materials examined. It is advisable, however, to have as a correction for which must, of course, be made in the final calcu­ lations. Allow the carbon tetrachloride layer to settle, then control some of the material before the addition of the pre­ draw it off, and discard. Add smaller amounts of dithizone and servative. repeat the operation until the time required for the change from It is reconmiended by Laug and Nelson (3) that during the green to orange color is lengthened to 30 to 40 vigorous shakes. extraction of mercury with dithizone the acidity of the mix­ A d d 0.1- to 0.2-ml. amounts of the dithizone and repeat until the green color remains on 40 shakes and then changes to orange on ture should not be greater than 1.5 N. In the procedure here an additional 20 shakes. When the addition of 0.1 ml. of dithi­ described, the total amount of acid does not exceed 0.9 N at zone gives a green color which remains after 60 shakes, the titra­ the time of titration. tion is finished. A little practice may be necessary to obtain ac­ The author lias been unable to find any previous report on curately reproducible end points but the technique is easily ac­ quired. Titrations should check at least within 0.2 ml. of dithi­ the effect of salts on the extraction of the mercury. zone reagent (±0.001 mg. of mercury). The dithizone solution used here is approximately 0.0001 N When the end point is reached, draw off the dithizone solution and, theoretically (1), two molecules of dithizone are equiva­ remaining in the separatory funnel and add a known amount of lent to one atom of mercury. The theoretical and actual mercury standard—for instance, 10 ml. of the dilute standard titration values are in good agreement. equivalent to 0.10 mg. of mercury. Shake the solution and allow it to stand a few minutes, then titrate exactly as above. This last step need be carried out only occasionally to standardize the Acknowled gmcnt dithizone solution. The standard is introduced at this step only as a matter of convenience; it may be titrated separately if The author washes to express his appreciation of the criti­ desired. A blank determination should be carried out, using all the cisms and suggestions of Mary C. Pangborn. reagents. When the reagents specified are employed, the blank is rarely greater than 0.1 ml. of dithizone reagent. Literature Cited C alculations . The titration usually requires 10 to 15 ml. of the dithizone solution for 0 .1 0 mg. of mercury. The amount of (1) Fischer, H., Z. anrjew. Chem., 42, 1025-7 (1929). mercury equivalent to 1 ml. of dithizone solution is calculated (2) Gettler, A. O., and Lehman, R. A., Am. J. Clin. Path., 8 (Tech. from the titration of the standard given above and the titration Suppl., 2), 161-4 (1938). of the sample, corrected for the blank, is multiplied by this figure (3) Laug, E. P., and Nelson, K. W., J. Assoc. Official Agr. Chcm., 25, to give the amount of mercury present in the sample. 399-403(1942). R e s u l t s . To determine the reliability of the method under (4) Tyrode, M. V., Arch. Intern. Pharmacodynamic, 20, 205-23 the conditions described, 2 ml. of a 1 to 6000 solution of either (1910). phenyl mercuric acetate or Merthiolate were added to various (5) Von Stein, P., “Organic Reagents in Inorganic Analysis”, pp. substances such as Tyrode’s salts, powdered egg yolk, and anti­ 125-6, New York, Chemical Publishing Co., 1942. pneumococcus horse serum and the mixtures were analyzed ex­ (6) Winkler, W. O., J . Assoc. Official Agr. Chem., 18, 638-44 (1935). actly as described above. Table II shows that the recovery of (7) Ibid., 19, 233-6 (1936). mercury from these mixtures is quantitative. (8) Ibid., 21, 220-8 (1938). Microdetermination of Arsenic in Biological M aterial

JAMES A. SULTZABERGER, Research and Riological Laboratories, Parke, Davis & Co., Detroit, Mich.

N THIS laboratory, many satisfactory arsenic deter­ hydroxide, dilute; and adjust with N hydrochloric acid to pH 6 .., I minations have been made by distilling arsenic tri­ to 6 .8 . Add sodium bicarbonate to about pH 7.2 and dilute to 1 liter. This solution, containing 1 mg. of arsenic per ml. as sodium chloride and titrating with iodine according to Bang (1). arsenite, is stable in that form for many months. Ten milli­ Since the end point of this method is not sufficiently sensitive liters of this stock solution diluted to 1 liter contain 1 0 micro­ for minute quantities of arsenic, the method was modified grams per ml. Color-D eveloping Solution. Solution I. _ Dissolve 1 gram by trapping the arsenic trichloride in dilute nitric acid of ammonium molybdate in about 45 ml. of distilled water, add and determining as molybdenum blue according to Sobotka 50 ml. of approximately 10 N sulfuric acid, and dilute to 100 ml. (5), whose method was based on that of Morris and Calvery This solution is stable for several weeks. Larger crystals oi (S). Rodden U) determined arsenic as the trichloride in arpmonium molybdate appear to yield a lower blank. Solution II. Take 20 ml. of Solution I, dilute to about 90 ml., ferrous and nonferrous alloys by distilling it in a stream of and shake. Add 2 ml. of 0.15 per cent hydrazine sulfate solution, carbon dioxide into water, oxidizing with nitric acid, evaporat­ dilute to 100 ml., and mix well. This solution is freshly prepared ing and drying at 130° C., and reading the molybdenum blue daily. color photometrically. Hubbard (0) adapted the method to biological material. Procedure Preparation of the Sample. The sample or an aliquot is di­ gested with sufficient nitric acid and 5 ml. of sulfuric acid as in the procedure of Bang (1) or Morris and Calvery (S). Care is taken to maintain an excess of nitric acid until all halogens are removed and all organic matter is oxidized. About 20 ml. of distilled water are added to the cooled digest, which is heated until the sulfuric acid fumes strongly, in order to remove any re­ maining oxides of nitrogen. In the decomposition of bone, the oxidation is facilitated by dissolving the sample in reagent fuming nitric acid, specific grav­ ity 1.5, diluting with an equal volume of water, and extracting the fat twice with ether. The residue of the combined extracts is digested separately as above and added to the original nitric acid solution. The calcium sulfate is filtered off and washed well and the digestion of the filtrate is completed as usual. Any loss of the 5 ml. of sulfuric acid is replaced. D istillation. The Fresenius flask (Figure 1), employed as the receiver and containing 1 ml. of concentrated nitric acid and about 8 ml. of distilled water, is surrounded by an ice-water bath and attached to the connecting tube for distillation. (The amount of distilled water required depends on the construction of the Fresenius flask. It should be sufficient to trap the arsenic trichloride and yet not fill more than two thirds of the lower bulb before bubbles of gas are released when the nitric acid solution is F i g u r e 1 . F r e s e n i u s F l a s k under gradual pressure from the distillation.) The following salts are previously placed in a small covered beaker: 4 grams of potassium chloride, about 0.4 to 0.5 gram of potassium bromide, and 1 gram of ferrous ammonium sulfate. The digest, in 5 ml. of In the method described below, arsenic trichloride is dis­ concentrated sulfuric acid in the Kjeldahl flask, is diluted with 6 tilled rapidly in a stream of hydrogen chloride and trapped ml. of distilled water and chilled under the tap. The salts are added all at once through a short wide-stemmed funnel to the in dilute nitric acid. flask which is connected immediately for distillation. A strong flame is applied directly to the flask until boiling starts, when it is A pparatus reduced at once until the tip of the flame just touches the flask. Kjeldahl flasks, 300 ml., and a Fresenius flask (Figure 1 ), 25 ml. When the effervescence begins to subside and the salts are about Bent connecting tube with rubber stoppers; inside diameter of half dissolved, the flame is gradually increased until the salts are tube, 4.5 mm. The middle section of the tube slopes slightly all dissolved and the solution is boiling rapidly. Meanwhile, the toward the receiver. solution in the Fresenius flask is kept under strict observation. Erlenmeyer flasks, 25 ml., glass bulbs, electric hot plate, ther­ When the first drop of liquid has fallen from the connecting mostatically controlled (Fisher Scientific Co., Autemp heater), tube into the receiver, the distillation is stopped. The whole and an electric oven. time required for this distillation should not be more than 2 to 3 Cenco Photelometer (photoelectric colorimeter) with a 625- minutes. The receiver is removed and the nitric acid solution is millimicron (red) filter and 1 -cm. cells with a volume of about 8 transferred to a 25-ml. Erlenmeyer flask by pouring the solution ml. through the bulbs of the Fresenius flask. The flask is washed twice with 1 ml. of distilled water which is added to the distillate. An aliquot may be taken at this point, especially if the arsenic C hem icals content is entirely unknown. The nitric acid solution is trans­ Sulfuric acid, specific gravity 1.84, arsenic-free; nitric acid, ferred from the receiver to a 25-ml. glass-stoppered volumetric specific gravity 1.42, arsenic-free. flask and diluted to the mark, and an aliquot of 1 0 ml. or less is Potassium chloride, potassium bromide, ferrous ammonium taken for evaporation. If it becomes necessary to take another sulfate (Mohr’s salt), ammonium molybdate, and hydrazine aliquot for evaporation, this should be done within 24 hours, be­ sulfate (Merck’s reagent). cause this solution is somewhat unstable after that time. Other­ wise, the aliquot should be evaporated with sulfuric acid and re­ distilled. The size of the aliquot should be such that it contains Reagent Solutions less than 100 micrograms of arsenic. Spinal fluid and blood Standard Arsenic Solutions. Dissolve 1.32 grams of containing very small amounts of arsenic should be analyzed arsenic trioxide (Bureau of Standards) in 30 ml. of N sodium directly without taking an aliquot. In this case it is desirable to June 15, 1943 ANALYTICAL EDITION 409

obtain it sufficiently pure, the blank should be determined until experience has shown that it is unnecessary.

Calibration Curve It was found more practical to plot the calibration curve as shown in Figure 2 on the basis of the arsenic concentration versus the photoelectric colorimeter readings than against color density values which must be calculated from instru­ ment readings. In practice, two curves are employed, one from 0 to 15 micrograms and the other from 15 to 100 micro­ grams. The light filter first employed in the photoelectric colorim­ eter in this method to measure the absorption of arsenic color reaction had a maximum transmission band at about 625 millimicrons. The bands employed in other methods have been in the range of 567 to 725 millimicrons. Very 20 40 60 80 MiCROGRAMS ARSENIC recently the maximum absorption of the arsenic color re­ action was determined as shown by the curve in Figure 3. F ig u r e 2. C a l ib r a t io n C u r v e It is obvious that the maximum absorption is in the infrared at about 840 millimicrons. When a dark infrared gelatin filter was employed in the photoelectric colorimeter, the use sufficiently pure distilled water to reduce the blank to a sensitivity was doubled. A glass filter having the proper minimum. wave band is now being obtained. When it arrives, it will E v a p o r a t io n a n d D r y in g . The nitric acid solution in the be standardized. If it has the maximum transmission at the open Erlenmeyer flask is evaporated almost to dryness as soon as possible and not allowed to stand overnight, because it is easily proper wave band, a new calibration curve will be prepared. contaminated with laboratory vapors. This evaporation is done on a thermostatically controlled hot plate. A temperature of 120° to 130° C. is maintained as read on a thermometer suspended Test of the Method in oil in a 30-ml. beaker in the center of the hot plate, so that the bulb of the thermometer touches the bottom of the beaker. Arsenic was added to 10-ml. samples of whole blood (When the evaporating solution has mostly disappeared the tem­ (Table I) to 50-ml. samples of normal urine (Table II) perature may rise suddenly above 130° C. If the tem perature is and to normal mice (Table III). The recovery of arsenic adjusted to 120° to 125° C., it is not apt to rise above 130° at the in the three tables is 100 =*= 5 per cent except with 1 end of evaporation.) Drying is completed by placing the Erlen­ microgram, where the limits are ± 10 per cent. Hence, the m eyer flask in an electric oven at 120 0 C. for 1 hour. This residue is very stable. accuracy of the method is about 0 . 2 microgram in the lowest D e v e l o p m e n t o p C o l o r . Exactly 10 ml. of the color-de­ range. No arsenic was found in samples of blood to which veloping reagent solution II are added to the dry residue in the 25- none was added. As arsenic is usually found in normal ml. Erlenmeyer flask, which is covered with a glass bulb and urine, it appears in Table II as a sample blank, although heated in a water bath for 10 minutes at 80° to 90° C. After cooling under the tap, the colored solution is transferred the reagent blank lias been reduced to a minimum. In to a 1 -cm. cell and compared with water in the comparison cell at Table III, a sample blank is also found. In order to identify a reading of 100 in the photoelectric colorimeter with a 625-milli­ this as arsenic and not phosphorus, which also gives the micron (red) filter. To the reading thus obtained is added a reagent blank correction, which is determined with each set of molybdenum blue color, these tests were repeated and a analyses. From a curve, previously prepared by plotting the double distillation was performed in order to remove any corrected Photelometer readings from the dis­ tillation of known amounts of arsenic, the amount of arsenic in the sample is determined. T h e R e a g e n t B l a n k . The amount of the blank to be added is determined by placing 5 ml. of concentrated sulfuric acid and 6 ml. of distilled water in a Kjeldahl flask, distilling, evaporating, and reading as in other deter­ minations. The reading from the blank is subtracted from 1 0 0 to obtain the blank correction which is added to the readings from known or unknown amounts of arsenic. Several factors affect the value of the blank. The molybdic acid solution I is stable and will not produce any blank for at least 2 weeks. The nitric acid which is to be em­ ployed in the receiver should be stored in a separate reagent bottle, where it is pro­ tected from fumes evolved from digestion processes. Distilled water employed in the receiver and in preparing the color develop­ ing solutions should be pure and protected from laboratory fumes. Although these fumes do not usually contain arsenic, they increase the color in the blank. When these precautions were observed, a zero blank was obtained. However, since it is some­ times necessary to redistill the water to F ig u r e 3. A b s o r p t io n C u r v e 410 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

T a b l e 1. R e c o v e r y o f A r s e n ic i n W h o l e B l o o d T a b l e II. Recovery of Arsenic from U rine (10-ml. samples) (50-ml. samples) Arsenic R eagent A rsenic Arsenic Reagent N o. A dded B lank Found Arsenic Recovered No. A dded Blank Arsenic Recovered Micrograms Microgram M xerograms Micrograms % M icrograms Microgram Micrograms % 1 0 .0 0 .2 0 .2 0 .0 1 0 .0 0 .0 1.6 2 0 .0 0 .2 0 .2 0 .0 2 0 .0 0 .0 1.4 3 0 .0 0 .2 0 .2 0 .0 3 0 .0 0 .0 1.4 4 1 .0 0 .2 1.2 1 .0 100.0 4 1 .0 0 .0 2 .4 9 0 .0 5 1 .0 0 .3 1.2 0 .9 9 0 .0 5 1 .0 0 .0 2 .6 110.0 6 5 .0 0 .2 5 .3 5.1 102.0 6 1 .0 0 .0 2 .6 110.0 7 10.0 0 .2 10.0 9 .8 9 8 .0 7 5.0 0.0 6.7 104.0 8 15.0 0 .3 15.4 15.1 100.7 8 5 .0 0 .0 6 .6 102.0 9 3 0 .0 0 .3 2 8 .8 2 8 .5 9 5 .0 9 10.0 0 .0 11.2 9 7 .0 10 4 0 .0 0 .3 » 3 9 .8 3 9 .5 9 9 .0 10 15.0 0 .1 1 6.8° 101.3 11 6 0 .0 0 .3 6 3 .3 6 3 .0 105.0 11 2 0 .0 0 .0 2 1 .5 100.0 12 8 0 .0 0 .3 7 8 .8 7 8 .5 98.1 12 4 0 .0 0 .1 4 3 .6 ° 105.0 13 60.0 0.1 63.1° 102.5 14 80.0 0.0 80.5 98.9 ° Reagent blank is included. phosphorus which might be present in the first distillate. Those mice to which no arsenic was added still gave sub­ T a b l e III. Recovery of Arsenic from N orm al Mice stantially the same results. Hence, phosphorus was com­ Arsenic Reagent Arsenic pletely eliminated in the first distillation, although it was No. Weight Added Blank Found Arsenic Recovered accumulated from the bones in a relatively large quantity. Grams Micrograms M icrogram AI icrograms Micrograms % 1 19 0 .0 0 .4 1.4 1 .0 D iscussion 2 20 0 .0 0 .4 1.4 1.0 3 18 10.0 0 .4 11.0 10.6 9 6 .0 4 17 2 0 .0 0.4 20.9 20.5 97.5 The apparatus employed in this method, except the photo­ 5 18 3 0 .0 0.4 29.9 29.5 95.0 electric colorimeter, is of the simplest type. With a good 6 18 4 0 .0 0 .4 40.7 40.3 98.2 hood there is no necessity for ground-glass joints or other expensive equipment. Since the actual distillation requires only 2 to 3 minutes, only one set of distillation apparatus biological materials, it is also applicable to any other type is required. This consists of nothing more than the Kjeldahl of analysis where the microdetermination of arsenic is re­ flask employed in the digestion, a bent connecting tube with quired. rubber stoppers at each end, and the small Fresenius flask used as the trap. The time of evaporation of the distillate C onclusion has been decreased by employing a thermostatically controlled A simple, rapid, and accurate method has been developed hot plate instead of a steam bath. At 120° to 130° C. there for the microdetermination of arsenic in biological materials, is no loss of arsenic during evaporation. If the temperature including blood, urine, bone, and tissue. The sample is of the hot plate is allowed to rise to 130° to 135° C., there digested with concentrated nitric and sulfuric acids, and the is very little if any loss and the time of evaporation is about arsenic is distilled as the trichloride and trapped in dilute 1 . 5 hours. At 140° C. or above the loss of arsenic becomes nitric acid solution in a Fresenius flask. The characteristic appreciable. Because of the total savings in time, six to eight molybdenum blue color is developed from the evaporated res­ samples of blood or urine may be analyzed in an 8 -hour idue and read in a photoelectric colorimeter with a 625- day. millimicron filter. The range is 0 to 100 micrograms of No interference was caused by the presence of antimony, arsenic. Based on the recovery of arsenic, the accuracy is bismuth, selenium, or phosphates either in the blank or in about 0.2 microgram in the lowest range. Very recently the recovery of arsenic. The amount of arsenic in the re­ the maximum absorption of the arsenic color reaction has agents employed appears to be negligible. In the separation been found to be in the infrared at about 840 millimicrons. of arsenic by the generation of arsine as in the Gutzeit or It is expected that the employment of this wave-length band Marsh tests, the use of zinc or tin is required. These metals will more than double the sensitivity. are difficult to obtain entirely arsenic-free in normal times The chief advantages of the method are the simplicity of and now the exigency is greatly increased. If the dilute the apparatus required, the rapidity of the distillation of the nitric acid employed in the trap and the water employed in arsenic trichloride, and the increased accuracy and sensitivity the color-developing solution are sufficiently pure and due to a low reagent blank. This blank may be reduced to protected from laboratory fumes, the blank may be reduced zero by the employment of sufficiently pure distilled water. to zero. No filtration nor adjustment of the pH value of the color- Acknowledgmen ts developing solution is necessary. This solution is placed The author wishes to acknowledge his sincere appreciation directly in the dried Erlenmeyer flasks. The characteristic to Leon A. Sweet and C. Kenneth Banks for their valuable molybdenum blue color produced by heating is stable and assistance. obeys Beer’s lawr, as shown when the color density is plotted He also desires to thank J. M. Vandenbelt for determining against the arsenic concentration. The readings in this the absorption of the molybdenum blue color of the arsenic procedure are made on the transmission through 1 -cm. solution submitted and for drafting the curve shown in cells. The sensitivity may be increased by placing the same Figure 3. volume of the arsenic solution in longer cells. It is expected also that the sensitivity will be more than doubly increased Literature Cited by the employment of the new absorption filter which is (1) Bang, Ivar, Biochem. Z., 161, 195 (1925). being obtained. Although the range of the color test will be (2) Hubbard, D. M., Ind. Eng. Chem., A n a l. Ed., 13, 915 (1941). decreased, that is not serious, since a smaller aliquot can be (3) Morris, H. J., and Calvery, H. O., Ibid., 9, 447 (1937). (4) Rodden, C. J., J . Research Natl. Bur. Standards, 24, 7 (1940). taken.' It is planned to continue the investigation of these (5) Sobotka, H., Mann, W., and Feldau, E., Arch. Derm. Syph., 42, items and to report in a subsequent paper. 270 (1940). Although this method has been employed successfully P r e s e n t e d before the Division of Medicinal Chemistry at the 105th Meeting in the analysis of about a thousand samples of various of th e A m e r i c a n C h e m i c a l S o c i e t y , Detroit, Mich. Mustard Gas in Air

Sensitivity of Qualitative Tests and a Rough Quantitative Determination

WILLIAM RIEMAN III School of Chemistry, Rutgers University, New Brunswick, N. J.

EVERAL methods for the qualitative detection of The two streams of air were mixed in a T-tube and passed through S /3,/3'-dichlorodiethyl sulfide (mustard gas) have been the test solution contained in a Pyrex test tube. 10 cm. long and 12 mm. in diameter, and fitted with a 2-hole rubber stopper. The described in the literature. Some of these tests were de­ delivery tube entered through one hole, extended almost to the signed for the examination of foods (6), and others for the bottom, and terminated in a constriction of 1-mm. internal diam­ analysis of soils (4). Although some papers deal specifically eter. The other hole held the exit tube which led the air through with the detection of mustard gas in air (1, 2), no data are another mixture of charcoal and soda-lime. The connections be­ tween the mustard bubbler and the test tube were glass-to-glass recorded on the minimum concentration of mustard gas in joints covered with tight rubber tubing. The temperature of the air that can be detected by the several methods. In the event laboratory was kept at about 2 0 °. of an air raid in which mustard is used, it would be very desirable to have ready a portable apparatus for the detection of mustard gas, and also to know the sensitivity of the test. T a b l e II. Q ualitative Tests for M ustard Gas A test of sufficient sensitivity would also be useful after the (Velocity of mixed stream was 170 ml. per minute in all cases.) raid to check the effectiveness of the decontamination or the C oncentra­ natural disappearance of the poison. tion of Mus­ Total Mus­ tard in tard in For these reasons several of the more promising qualitative M ixed M ixed tests were investigated with regard to their sensitivity when R eagent Stream Time Stream R esults applied to contaminated air. nr/ml. M in. y G rignard 0.020 10 34 N egative HAuCU 0.040 10 68 Very faint HAuCU 0.040 20 136 F a in t Materials Used /3-Naphthol 0.040 10 68 F a in t HAuCU paper 0.020 10 34 Negative NaiPtl« paper 0.020 10 34 Negative The mustard gas was prepared in this laboratory by William A. Raimohd, and the fraction boiling between 115° and 116° C. at 25 mm. was used. It froze sharply at 14.5° C. Comparison of these figures with the recorded literature (5) indicates that the compound had a high degree of purity. It was colorless and had Procedure only a faint odor of garlic. The Grignard reagent (KCuI2), alcoholic /3-naphthol, and iodo- The velocity of stream A was usually 170 =*= 20 ml. per minute, latinate solution were prepared according to the directions of the maximum velocity that could be used without risk of blowing tainsby and Taylor (6). Filter paper was impregnated with the solution out of the test tube. It also represents a convenient iodoplatinate solution as recommended by Bradley (1). A 1 rate for a field test when the air is passed by means of a rubbcr- per cent solution of chlorauric acid (3) was prepared, and filter bulb aspirator. The velocity of stream B was varied between 1.7 paper was impregnated with this solution. A solution of acetic and 6 . 8 ml. per minute in order to vary the concentration of mus­ acid, made by diluting 5 ml. of reagent grade acetic acid to 100 tard vapor in the mixed air stream. ml., and a 0.5 per cent solution of potato starch were prepared. For the iodoplatinate test, the air was bubbled through 1.0 A small quantity of mustard was accurately weighed and dis­ ml. of 5 per cent acetic acid for the indicated length of time, then solved in sufficient glacial acetic acid to give a standard solution the test tube was removed from the apparatus, and one drop of containing 0 . 1 0 mg. per ml. iodoplatinate solution was added, followed by one drop of starch solution. The pink iodoplatinate turns blue if 2 micrograms or more of mustard are present. The mustard absorbed by the acetic acid can be determined with an accuracy of 1 microgram by comparing the intensity of the blue color with a series of standards. The color standards T a b l e I. Colorim etric Determ ination of M ustard Gas are prepared by putting 0.02, 0.04, 0.06, 0.08, and 0.10 ml. of the Absorbed by A cetic Acid standard mustard solution in Pyrex test tubes (10 cm. by 12 C oncentra­ mm.), diluting to 1 . 0 ml. with water, and adding iodoplatinate tio n of T o tal Velocity of M u stard M u stard and starch solutions. The variation in the concentration of acetic T est Air Streams in M ixed in Mixed Mustard acid does not appreciably affect the intensity of the blue color. No. A B Stream Time Stream Absorbed The color standards must be prepared at the same time that the M l./m in . y / m l M in . 7 y % color is developed in the unknown solution. 1 0 1 .8 1.0 5 .5 10 7 .5 75 The tests with the Grignard reagent and with chlorauric acid 2 170 1.7 0.010 10.0 17 4 24 consisted merely in letting the air bubble through 1 . 0 ml. of the 3 170 1.7 0.010 5 .0 8 .5 2 24 4 170 3 .4 0.020 10.0 34 10 29 respective solutions and observing at intervals whether a precipi­ 5 170 3.4 0.020 5 .0 17 5 29 tate had been formed. The |S-naphthol test was done similarly, 6 170 3 .4 0.020 2.5 8.5 2 24 except that the test tube was surrounded by a water bath at 40°. The tests with the impregnated papers were performed by putting the papers into the Pyrex test tube and passing the con­ taminated air stream through this tube. The iodoplatinate paper was moistened with water before use, but the chlorauric A pparatus acid paper was used dry. The results of the colorimetric determination of mustard The apparatus was essentially like that of Yablick, Perrott, and Furman (7). A stream of air, denoted as A, from a compres­ gas absorbed by the dilute acetic acid under various con­ sor was purified by passage through a mixture of soda-limc and ditions are summarized in Table I. The values in column 4 charcoal. The velocity of this stream was controlled by a pinch- were calculated from the vapor pressure of mustard (5) on the cock and pressure regulator and was measured by a flowmeter. assumption that the air in stream B was saturated with the A second stream of air, B, was similarly purified, controlled, and measured and was then bubbled through a vessel containing mus­ vapor. Most of the entries in column 7 are the mean of tard liquid immersed in a water bath maintained at 20 =*= I o C. duplicate determinations. 412 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

The results of the qualitative tests for mustard gas are is not applicable in the presence of oxidizing agents such as given in Table II. chlorine or nitrous fumes or in the presence of reducing agents, including the arsenical vesicants in large concentration. D iscussion Table II reveals that the other tests are much less sensitive Test 1 reveals that the absorption of the mustard vapor than the test with iodoplatinate solution. by the dilute acetic acid is nearly complete when the air veloc­ Literature Cited ity is small. Tests 2 to 6 indicate that the absorption is much less complete when the velocity is increased. Never­ (1) Bradley, T. F., Chem. Eng. News, 20, 893 (1942). (2) Cox, H. E .t Analyst, 64, 807 (1939). theless, mustard vapor can be detected by the iodoplatinate (3) England, Department of Air Raid Precautions, "Detection and method under the recommended conditions with a 5-minute Identification of War Gases”, Brooklyn, N. Y., Chemical absorption period when its concentration is only 0 . 0 1 micro­ Publishing Co., 1940. gram per ml. (test 3). Since this concentration requires an (4) Hoogcveen, A. P. G., Chemistry and Industry, 59, 550 (1940). (5) Jackson, K. E„ Chem. Ret., 15, 425 (1934). exposure of 2.5 hours for fatal results (S), the iodoplatinate (6) Stainsby, W. J., and Taylor, A. McM., Analyst, 66, 44 (1941). method is fairly satisfactory for detecting dangerous con­ (7) Yablick, M., Perrott, G. St. J., and Furman, N. H., J . Am. Chem. centrations of mustard vapor. It is stated (6) that this test Soc., 42, 266 (1920).

Microdetermination of Magnesium with the Polarograph

CHRISTOPHER CARRUTHERS The Barnard Free Skin and Cancer Hospital and Washington University, St. Louis, Mo.

A polarographic procedure is given for the Solution B, prepared by diluting 2.7 ml. of 1 per cent gelatin solution to 1 0 0 ml. with the phosphate buffer. determination of microquantities of mag­ Solutions A and B should be made just before use or stored in a nesium as the hydroxyquinolate. refrigerator, as the presence of gelatin can favor a rapid growth of microorganisms.

Procedure NVESTIGATION of possible chemical changes produced I in mouse epidermis by methylcholanthrene (1) requires The investigations of Redman and Bright (9) and of special microprocedures, because of the small amounts of Miller and McLennon (6) have shown that the quantitative tissue available for analysis. Magnesium, for example, can­ precipitation of magnesium hydroxyquinolate is very critical not be determined by any ordinary technique, and even and much affected by the conditions under which it is carried polarographic determination is not practical. This metal out. For calibration purposes, the author found that to prevent gives a very poorly defined wave in solutions of tétraméthyl­ high results it was necessary to precipitate the magnesium ammonium salts as the supporting electrolyte (3, 8). Müller hydroxyquinolate by the method of Kolthoff and Furman (

time in seconds (5) were measured in solution T a b l e I. C alibration D ata fo r Magnesium in Magnesium A at a potential of —1.5 volts with respect to Hydroxyquinolate the saturated calomel electrode. Since the (Various concentrations of magnesium hydroxyquinolate in a mixture of 7 parts of 3.33 potentials at which u were determined de­ molar phosphate buffer of pH 7.6 and 3 parts of 0.1 N hydrochloric acid at 25° C. Solutions contained 0.02 per cent gelatin. Air removed with nitrogen, h = 64 cm., t *= 2.77 sec., creased slightly with decreasing concentrations m *=> 2.925 mg. sec."1. Diffusion currents measured at the potentials indicated with respect to the saturated calomel electrode.) of magnesium hydroxyquinolate, the residual Magnesium Hydroxy- P o ten tial currents were measured from —1.56 to —1.40 quinolate a t W hich volts with respect to the saturated calomel id W as Millimolar equivalent C u rre n t“ Observed Corrected M easured id /C electrode. Also id/C decreased, but the ratio, M icro­ M icroam p./ except for the 0.2295 and 0.1148 X 10- 3 M M g./cc. a m p e r e Microamperes Volls m m o le / l. solutions, was found to be fairly constant. 1.14 0.0279 0 .3 9 13.60 13.22 - 1 . 5 6 11.59 0.918 0.0223 0.31 10.67 10.37 -1.52 11.29 For greater certainty in the analysis, a cali­ 0.860 0.0209 0.31 9 .9 5 9 .6 5 - 1 . 5 2 11.22 bration curve was constructed. 0 .5 "0 0.0140 0.29 6.4S 6.19 -1.50 10.85 0.459 0.0112 0.27 4.88 4.61 -1.48 10.04 Since Maas (cited by Kolthoff and Lingane, 0.2295 0.005G 0 .2 8 2 .2 0 1.94 - 1 . 4 6 8.44 0.1148 0.0028 0 .2 5 1.07 0.8 3 - 1 . 4 0 7 .1 8 6) has shown that id/C may not be constant when the drop time is too short, the pro­ ® Residual current of supporting electrolyte alone. cedure described above was repeated with an -...... • electrode having a drop time of 4.38 sec­ onds. The same general ij/C relationship w'as found The diffusion currents were measured (5) and the polaro- for this second electrode as with the electrode used for the grams made after placing the magnesium hydroxyquinolate calibration data in Table I. 'solutions in shell vials and then removing any oxygen with a stream of nitrogen bubbles, m, the rate of flow of mercury Cathodic Waves of Magnesium Hydroxyquinolate from the capillary in milligrams per second, and t, the drop Solutions 2.29 to 1.72 X 10 ~3 M gave reduction waves which were typical of Figure 1, 1. Two waves, the first very small, were present, but accurate measurement of id of the second wave was found to be impractical for analytical pur­ poses because of the small region of constant diffusion cur­ rent. At a concentration 1.377 X 10~ 3 M, a third wave appeared (Figure 1, 2 ), the second wave being well defined for 0.914 to 0.1148 X 10- 3 M solutions and ideal for analytical purposes (Figure 2 , 1). Unfortunately id of the total double wave (second and third) could not be used because the hydrogen wave interfered with the determination of i, of the latter (Figure 2, 2). The first small wave (Figure 1, 1) was not present at a concentration 0.2295 X 10~ 3 M. The half-wave potential of the second wave decreased with decreasing concentrations of the magnesium hydroxyquino­ late, the values being about —1.37 and —1.30 volts with re­ spect to the saturated calomel electrode for 0.918 X 10 ~ 3 M and 0.459 X 10 ~3M solutions, respectively.

Procedure for Solutions Solutions of known magnesium^ content were made by dissolv­ ing Baker’s magnesium ribbon (acid-washed to remove any oxide) in hydrochloric acid. All solutions were diluted when necessary to a volume of approximately 10 ml. in 50-ml. Pyrex beakers. Various amounts were treated as follows for precipitation of the hydroxyquinolate: To each sample were added 1 ml. of 2 N ammonium chloride and 0.5 ml. of 0 N ammonium hydroxide (method of Kolthoff and Furman, 4). The solutions were brought to the boiling point, and 0.7 ml. of a 1 per cent solution of 8 - hydroxyquinoline in 95 per cent alcohol was added dropwise with stirring. The solutions were allowed to cool, then reheated, and another 0.7-ml. portion was added at or near the boiling point in the same way. This procedure was repeated for a third treat­ ment. Another group of standard solutions was treated with 2- ml. portions of the 8 -hydroxyquinoline at room temperature after the addition of ammonium chloride and ammonium hydroxide, and these solutions were also held at or near the boiling point for 1 0 minutes. After one hour’s standing at room temperature, the precipitates were aspirated from the mother liquor with 10-mm. disk Pyrex filter sticks (medium porosity) covered with a layer of asbestos. The precipitates were washed with two 2-ml. portions of wash alcohol which had been previously filtered as described above, F i g u r e 1. P o l a r o g r a m s o f M a g ­ and precipitates, filter sticks, and beakers were dried at 105° C. for n e s i u m H ydroxyquinolate i n P h o s ­ 0.5 hour. After cooling to room temperature, 3 parts of 0.1 N p h a t e B u f f e r -H ydrochloric A c id hydrochloric acid were added, and the difficultly soluble hydroxy­ M i x t u r e a t 2 5 0 C . quinolate was brought into solution by gentle warming and stir­ 2.29 X 10 _a M (sensitivity 1/10) ring. While still warm, the sides of the beaker were rinsed with 1.337 X 10-» (sensitivity 1/20) the acid, using the filter stick to dislodge any particles of hydroxy- 414 INDUSTRIAL AND ENG NEERING CHEMISTRY Vol. 15, No. 6 quinolate adhering to the walls. Finally, the beaker was tilted so that the lower third of the filter stick was rinsed by the warm acid, 7 parts of solution B were added, and, after thorough mixing, the final solution was filtered through paper to remove the asbes­ tos. A sufficient quantity of acid and of solution B was used to keep the magnesium hydroxyquinolate at or below 1 X 10 “ 3 M. Since the half-wave potential decreased with decreasing con­ centration of the magnesium hydroxyquinolate (Table I), the potential at which the diffusion current was measured was deter­ mined as follows: When the diffusion current was reached (shown by the region of constant deflection of the galvanometer oscillations), the voltage applied to the dropping mercury elec­ trode was noted. Then as soon as the galvanometer oscillations indicated that the region of constant diffusion current was passed, the applied potential was again observed. The mean of the initial and final applied potentials was the potential at which the diffusion current was measured, and also the potential from which the residual current was taken from Table I. The mag­ nesium content was read from the calibration curve.

Typical recoveries from various samples of known magnesium concentration are shown in Table II. Quantities as small as 6 8 to 1 0 0 micrograms could be determined with an error of ± 1 8 to ±2.8 per cent. Amounts of magnesium up to 0.563 mg. were determined equally well by adding the 8 - hydroxyquinoline to a hot or cold solution of the known, provided ammonium chloride and ammonium hydroxide had been added previously. Since 2.8 micrograms (0.1148 X 10- 3 M magnesium hydroxyquinolate) per ml. had a diffusion current of 0.825 microampere (corrected), the method might serve for the determination of mere traces of magnesium with adequate methods for handling minute amounts of precipitate. This method should also be of value for the microdetermina­ tion of the other metal ions which are quantitatively precipi­ tated by 8 -hydroxyquinoline (11).

Figure 2. Polarograms of Mag­ T a b l e II. Assay of Known M agnesium C hloride Solutions nesium H ydroxyquinolate in Phos­ Magnesium Added Recovered Difference phate Buffer-Hydrochloric Acid Mg. M o. Mg. M ixture at 25 ° C. 0.5625 0.5570 — 0.0055 1. 0.8C x io -> M 0.3369 0.3400 + 0.0031 2. 0.2295 X 10-• M 0.2250 0.2225 - 0 .0 0 2 5 0.1800 0.1750 - 0 .0 0 5 0 0.1350 0.1370 + 0.0020 0.1125 0.1140 + 0 .0 0 1 5 0.1000 0.1020 + 0.0020 0.0800 0.0815 + 0.0015 buffered magnesium hydroxyquinolate without changing 0.0680 0.0700 + 0 .0 0 2 0 the final pH. This addition became necessary at times in order to keep the molarity at or below 1 X 10~3. However, after a few trials, it was not difficult to estimate the propor­ tions of 0.1 N hydrochloric acid and of solution B required Procedure for Tissues to keep the diffusion current within the range of the calibra­ The following technique has been used for the determina­ tion curve. tion of magnesium in 100 to 400 mg. (wet weight) of mouse epidermis: Literature Cited

The mouse epidermis was ignited at 450° C. in a silica crucible (1) Carruthers, C., and SuntzeiT, V., J. Natl. Cancer Inst., 3, 217 in a muffle furnace until all the organic matter was destroyed. (1942). The ash remaining was dissolved with 1 ml. of 0.1 N hydrochloric (2) Heyrovaky, J., and Berezicky, S.. Collection Czechoslov. Chem. acid, and the crucible rinsed into a 50-ml. Pyrex beaker with four Commun., 1, 19 (1929); Chem. News, 138, 180, 195 (1929). 2-ml, portions of distilled water. (Since the calcium content of (3) Kimura, G., Collection Czechoslov. Chem. Commun., 4, 492 (1932). mouse epidermis had been previously found to be low, 10, and the (4) Kolthoff, I. M., and Furman, N. H., "Volumetric Analysis”, iron content very low, 1, removal of these ions was unnecessary. Vol. II, New York, John Wiley & Sons, 1929. Iron and calcium are the only elements found in sufficient (5) Kolthoff, I. M., and Lingane, J. J., "Polarographie Analysis and quantity in limited samples of animal tissue that might interfere Voltammetry Amperometric Titrations", New York, Inter­ with the precipitation of magnesium hydroxyquinolate as de­ science Publishers, 1941. scribed below.) (6) Miller, C. C., and McLcnnon, I. C., J. Chem. Soc., 1940, C56. The magnesium was then precipitated by adding the 1 per cent (7) Müller, O. H., Chem. Rev., 24, 95 (1939). alcoholic 8-hydroxyquinoline reagent to the sample at 96° to (8) Miiller, O. H., Collection Czechoslov. Chem. Commun., 7, 321 100° C. as previously described. The precipitate was washed, (1935). dried, and brought into solution as outlined above. (.9) Redman, J. C., and Bright, H. A., Bur. Standards J . Research, 6, 113 (1931). (10) Suntzeff, V., and Carruthers, C., Cancer Research, in press. From the diffusion current, the magnesium content could (11) Yoe, J. H., and Sarver, L. A., "Organic Analytical Reagents", be read on the calibration curve. When working with un­ New York, John Wiley & Sons, 1941. knowns it was found convenient to have solution A at hand, P u b l i c a t i o n aided by a grant from Tho International Cancer Research so that an aliquot of this mixture could be added to one of the Foundation. Reproducibility of Weighings Made on Micro chemical Balances

CLEMENT J. RODDEN, Chairman, JULIUS A. IOJCK, Secretary, A. BENEDETTI-PICHLER, ALSOPII CORWIN, A N D EDW. W. D. HUFFMAN, Committee on Micro chemical Balances, Division of Analytical and Micro Chemistry,

A m e r i c a n C h e m i c a l S o c i e t y

AT THE Detroit Meeting of the American Chemical S o c i e t y in September, 1940, G. E. F. Lundell, T a b l e II. Condensed Summary of R esults (October, 1942) chairman of the newly organized Division of Analytical and Number of microbalances studied 29 Micro Chemistry, asked the divisional secretary to form a Standard deviation of individual weighings (median value) ±3.4 micrograms Committee on Microchemical Balances, for the purpose of Probable error of individual weighings ±2.3 micrograms formulating specifications for performance and suggestions Probable error of individual weighings, Corner and Hunter ±3.2 micrograms for materials and construction of these instruments for the Largest error to be expected in any one weighing benefit of American balance manufacturers. (twice standard deviation) ±7 micrograms At the meeting of this committee at Atlantic City in September, 1941, it was decided first to find out what performance, as regards reproducibility of weighing, was The test consisted in weighing two 1-gram weights one actually being obtained by microanalysts in the course of against the other, each weighing being followed by a zero their regular work. To obtain this information, the mem­ point determination. The rider was required to be reset bers of the division were circularized with the request that for each weighing and for each zero point. Details of reading they perform a test under definitely stated conditions and the scale, proper conditions of temperature equilibrium, report their results to the committee secretary. These have minimizing vibration effects, etc., were left to the discretion been summarized and the necessary statistical calculations of each analyst, who was asked simply to perform these performed by the divisional secretary, Rev. Francis W. Power, weighings with the usual careful technique he would use S.J., Fordham University, and were presented in sub­ in a regular series of micro weighings such as those involved stantially this form at the Buffalo Meeting in September in a succession of carbon and hydrogen determinations. The 1942. cooperators were asked to make a series of ten such weighings on the 1 -gram weights and another series of ten using two T a b l e I. Summary of Reproducibility Experiments on 1 0 -gram weights; the number ten for the series was selected M icrobalances as a compromise between the requirements of statistical (Requested by the Microbalance Committee of the Division of Analytical stability for the standard deviations and the amount of time and Micro Chemistry, 1942) that a busy analyst could be reasonably expected to spend D ate T ype and R econdi­ Standard Deviation on such a task. Serial No.° Purchase tioning 1-gram load 10-gram load The standard deviations which are given in Table I are M xerograms those of the individual weighings, and were calculated from K 2741 1935 7 .4 7 .2 K 2531 1935 Í.937 2 .8 the formula K 2776 1936 3 .2 B (special) 1941 12.8 b 13ÍÓ& K 2340 1930 Í936 6 .6 6 .3 2 d- K 2943 1938 3 .1 4 .9 2.8 (recheck) 3 .9 n 1.5 K 2731 1936 3 .6 5 .2 K 2413 1931 6 .1 8 .3 K 2447 1933 1.0 1 .6 where d is the deviation in micrograms of each of the n( = 1 0 ) B 27,225 1937 Í939 22.3b 17.0& weighings from the mean. The number in the denominator K 2561 1933 1939 14.0b 3 8 .0 b B none 1935 5 .9 3 .8 is decreased by 1.5 to correct for the underestimation of a K 2769 1937 1.7 2.1 A 14,772 1942 1.2 * 2.1 standard deviation derived from such a short series. Further K 2881 1937 2 .2 2 .8 calculation indicates that there is no significant difference A (special) 1936 0 .9 1.1 B 31,636 1942 9.9b 2 8 .3 b between the values for 1 -gram load and those for 1 0 -gram 8.4* (recheck) 22.2b 5 . 1 c 5.2 load. K (not given) 1937 2 .5 6 .4 In the summary, Table II, the median values for the B 29,734 1940 1.2 0 .7 K 2863 1937 Í94Ó 1 .8 2 .3 standard deviations are also expressed as “probable errors” K 2790 1937 1940 4.7 6.7 K 2423 1933 1939 2 .2 1 .7 in order to make the data more easily comparable with a B 26,730 1939 3 .3 2 .5 somewhat similar study made by Corner and Hunter (4 ) B 30,833 1941 4 .0 5 .2 K 2877 1938 5 .9 5 .6 which came to our notice about the time that our requests K 2545 1934 2 .2 3 .6 B 21,406 1930 i 935 8.0 6.5 for data were being sent out. They give figures for the S-R 6164 1938 3 .5 2 .0 probable error of weighings of two 1 -gram weights made on A 12,745 1940 ----- 1.3 1.3 8 different microbalances, one of which (an Oertling) was Results on Other B alances A sem im icro 1941 4 8 studied extensively. The standard deviations in their paper B X 11,661 vary from 2.5 to 12.6 micrograms, the poorer values being B macro 1931 21 (2-gram load) 27 (50-gram 22,984 load) obtained on rather old balances used by students. On the S m acro 1915 52 (20-gram load) 20,450 whole their study leads to a probable error for the individual B semimicro 1933 6 (5-gram load, weighings of about ±3 micrograms, which is in very good 24,675 Sept., 1938) B semimicro 1933 16 (5-gram load, agreement with the results of the present survey. They also 24.675 Oct., 1942) conclude that most of this error is due to the setting of the ° Designations of balances: K, Kuhlmann; B, Becker; A, Ainsworth; S-R, Sartorius (Ramberg). rider. ¿N ot used in establishing median. Our figures, as well as those of Corner and Hunter, ob­ c After elimination of vibration. viously upset any chemist’s pious belief that because his 415 416 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 microchemical balance will respond to a difference in weight balances (including some domestic ones) and a recommenda­ of 1 microgram it will also reproduce weight differences as tion to the analyst to ascertain for himself just how well his closely as this on repeated trials. If we assume that the own balance will actually perform. outside tolerance or maximum error to be expected is twice The work of the committee along lines of performance the standard deviation, we find that by and large our analysts specifications and suggestions for materials and construction should work on about 7-mg. samples if they wish to be reason­ is being continued, and the chairman would appreciate ably sure of their sample weights to 1 part per 1 0 0 0 , the suggestions from those interested. We wish also to express value to be used by any given analyst depending on the re­ our gratitude to all those who have cooperated in making producibility figure which he gets. The actual sample weight these tests: for the over-all process is also controlled by its chemical factor and of course by the error that one is willing to tolerate * CooperatorB Cooperators Ricman (Rutgers) R odden (N. Bur. Standards) for the whole analysis. This matter has been fully dis­ Kuck (C. C. N. Y.) Chapman (Cyanamid) cussed by Benedetti-Pichler (1, S), who gives figures for the Nichola (Cornell) Haagen-Smit (Cal. Tech.) Tuemmler (Shell Development Co.) Bushey (Aluminum Co.) reproducibility of weight by some analytical balances. Power (Fordham) Hallett (Eastman Kodak) Alicino (Fordham) Duncan (W. Va. University) The discussion in his book should also be read in this con­ Royer (Calco) Galitzinc (G. E., Pittsfield) nection (2). Seeing that only about one third of the re­ Shrader (Dow) Clarke (Wesleyan) Van Etten (Northern Regional Lab.) Signeur (Canisius) ported figures are such as would permit 5-mg. samples to be Huffman (Denver) weighed to within 1 part in 1 0 0 0 , it is recommended that the analyst check his balance by some such reproducibility test Literature Cited as described here, and be guided by what he may find. (1) Benedetti-Pichler, A. A., Ind. E no. Chem., A nai.. E d., 11, 226 (1939). C onclusion (2) Benedetti-Pichler, A. A., "Micro-technique of Inorganic Analy­ sis”, pp. 173-84, New York, John Wiley & Sons, 1942. While we do not as yet wish to go on record as recom­ (3) Benedetti-Pichler, A. A., Mikrochemie, 27, 339 (1939). mending a certain maximum reproducibility as a specifica­ (4) Corner, Mary, and Hunter, Harold, Analyst, 66, 149 (1941). tion, we feel that the foregoing figures warrant an expression P r e s e n t e d before the Division of Analytical and Micro Chemistry at tho of surprise at the poor performance of so many microchemical 104th Meeting of the A m e r i c a n C h e m i c a l S o c i e t y , Buffalo, N. Y.

Detection of Gold in Plating Procedure The anode clamp is attached to the article to be tested after MELVIN LERNER the removal of any lacquer present thereon. A piece of spot-test paper or absorbent filter paper moistened with a drop of the test New York Customs Laboratory, Treasury Department, solution is then placed against the spot on the metal surface to be New York, N. Y. tested. The platinum cathode is used to press the moist paper against the metal surface for a period of about a second. If any doubt arises as to whether or not a closed circuit has been ob­ tained, a milliammeter may be inserted in the circuit. An intense ESTS to detect gold in platings must frequently be made purplish-brown stain on the side of the paper in contact with the T specimen indicates the presence of gold. in U. S. Customs Laboratories, as the Tariff Act (3) If silver is present in the object under examination, the test prescribes different rates of duty for plated and unplated should not be performed in direct sunlight, nor be exposed too articles. Various methods have been proposed for the elec­ long to diffused light, as the purple color of the partially reduced trographic detection of gold without material injury to the silver chloride will develop in a short time. surface. Each is limited in either failing to reveal gold in Interferences very thin “flash” plates or in not providing a sharp distinction between gold and other metals. Approximately forty common metals and alloys have been tested by the above method and negative results obtained in Arnold (1) uses a solution of benzidine in acetic acid for the each case. « electrolysis, but differentiation between gold and copper in flash Nickel, cadmium, zinc, tungsten-nickel-tin alloy, brass, plates on brass is extremely difficult, as both produce similar copper-vanadium alloy, titanium, silver, phosphor-tin alloy, greenish-blue colorations. Calamari, Hubata, and Roth (2) use a peroxide solution containing nitrate ion, but their method, as copper-molybdenum alloy, sterling silver, tungsten-nickel admitted by the authors, is inapplicable to thin gold plates on alloy, copper, molybdenum, tungsten, iron, carbon steels, copper or brass because the characteristic gold coloration is stainless steels, tungsten-tin-copper alloy, chrome-titanium masked by a strong greenish brown from the copper. Yagoda alloy, nickel-silicon alloy, platinum, rhodium, tin-nickel '4 ) electrolyzcs with a 1 per cent sodium chloride solution and follows with a spot test using stannous chloride, but results are alloy, lead, pewter, tin solder, Monel metal, bronze, Ger­ negative with very thin gold plates. man silver, antimonial lead, type metal, tantalum, Babbitt metal, aluminum, bismuth-lead alloy, magnesium, tin, and For the past 6 months, this laboratory has used the follow­ indium gave no perceptible colorations. ing method, which yields a distinctive gold test even for thin Chromium gave a light yellow and cobalt a very light pink plates without interference from other metals or alloys yet color. encountered. N o te . While extensive tests have not been made on gold alloys, this method gives excellent results with the usual 1 0 -, Apparatus and Test Solution 14-, and 18-carat jewelry. Three No. 6 dry cells, connected in series, are used (approxi­ Literature Cited mately 4.5 volts). A small clamp is connected to the anode, a piece of platinum wire (B. & S. gage 18) to the cathode. The test (1) Arnold, E., Chem. Lisly, 27, 73-8 (1933). solution is prepared by adding approximately 1 0 grams of stan­ (2) Calamari, J. A., Hubata, R., and Roth, P. B., Ind. Eno. Chem., nous chloride to 100 ml. of dilute sulfuric acid (1 to 9). Any pre­ Anal. Ed., 14, 535 (1942). cipitate formed on mixing can be ignored. A piece of metallic (3) Tariff Act of 1930, Paras. 339, 367, and 397. tin should be added to the solution to prevent its oxidation. (4) Yagoda, H., private communication, 1941. lune 15, 1943 ______A N A L Y T I C A L E D I T I O N _____ 15 HEÏROVSKÏ POLAROGRAPH

The application of the polarographic method of an­ manent photographic recording of every analysis are alyses expands steadily. Some of the analyses being some of the reasons why the Heyrovsky Polarograph is made with the Heyrovsky Polarographs now in use becoming so widely accepted. include the analysis of brass; of steel and iron; of lead, The procedures established thus far by no means magnesium, nickel, and zinc alloys; of metallic impuri­ define the field of polarography—the perfected in­ ties in aluminum; of lead and zinc in paints; of major strumental system of the Heyrovsky Polarograph constituents in plating solutions; and the differentia­ creates unlimited possibilities for analytical and re­ tion of waters. search applications. In the industrial health field, the Polarograph is be­ ing successfully used for such important analyses as A bibliography of more than 700 papers dealing with the detection of incipient lead poisoning. The sensi­ the polarographic method of analysis and a booklet tivity, simplicity and speed of the polarographic method discussing the Polarograph and polarographic analysis for determination of lead in urine permits complete are available without charge on request. and periodic examination of plant personnel at low S-29301 POLAROGRAPH—Heyrovsky, American Model XI, cost without the need of highly trained operators. Indicating-Recording. Complete with three vessels, 29305, 30 Accuracy; rapidity; the possibility of detecting and MM, 1 electrode, 29340; 21 CM Capillary tubing, 29350; 4 packages bromide paper, 29347; three feet Neoprene tubing identifying minute quantities and of making simul­ Pi inch; three rubber stoppers for vessels; mercury reservoir taneous determinations of several components; small and connecting wires. For 115 volts, A.C. 60 cycle cir­ sample requirement; preservation of sample; and per­ cuits ...... 5550.00

E. H. SARGENT & CO., 155-165 E. Superior St., Chicago, Illinois Michigan Division: 7959 East Jefferson, Detroit, Michigan

SARGENT SCIENTIFIC LABORATORY SUPPLIES INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6

They are Dependable

WHATMAN Filter Papers are dependable. When you standard­ Braun Model PC Electrolytic Apparatus ize on a grade for a particular determination, you know that FOR PRECISE ANALYSIS papers from box after box, month OF STRATEGIC METALS after month will have the same The work of sample analysis in the "Strategic Metals” industries requires dependable, highly accurate, yet filtering speed and retentiveness easy to operate electrolytic apparatus. Since Braun Model PC Electrolytic Apparatus was designed to meet as those in the box you originally the requirements for the most exacting technical analysis it satisfies every need in either routine or tested. That is why, in labora­ research work. tories all over the country, Braun Electrolytic Apparatus is built to save you every possible moment of time in making routine analyses, Chemists specify WHATMAN yet it gives sufficiently accurate results for the most exacting research work. It is the ideal equipment for Filter Papers for all their work. the busy laboratory working under wartime pressure.

They know, from long experi­ Uniform and accurate determinations of copper, lead, antimony, cadmium, nickel, tin, zinc and other metals ence, that WHATMAN means are rapidly made with this apparatus, permitting routine analyses to be made with a minimum expenditure of dependability in quality and de­ labor. All units operate independently of each other so that as many as six individual analyses can be made livery. simultaneously on one 6-unit outfit.

These outfits are available in 2, 4 and 6-unit assemblies. Samples cheerfully sup-plied. Motor generators can be supplied for operating the apparatus.

For complete information address Dept. 1-6 H. REEVE ANGEL & CO., INC. 7-11 Spruce St., New York, N. Y.

BRAUN CORPORATION 2260 East Fifteenth Street^os Angeles, California Son Francisco, Calif. : Senil,'e, Washington BRAUN-KNECHT-HEIMANH-CO. SCIENTIFIC SUPFLIES CO. lune 15, 1943 ANALYTICAL EDITION 17

K l e t t ....

Photometers A p t R Ï ° R

^ During the past 32 years we have solved numerous measurement problems by supplying PUotoelectrUc standard or special instruments to accurately Q l a l i G e ll measure pressures, vacuums and flows of liquids and gases. We have constantly improved, devel­ G oiosüsnetesi oped and diversified the line so as to meet fully all needs of the field.

U-TYPE MANOMETERS N o. 900-3 The Clean-Out Manometer for line pressures up to 100 lbs. per sq. in. A wing nut permits the Manometer body and tube to be disconnected For the varied needs of industrial, clinical from the head—the tube cleaned with a brush furnished or tube re­ and agricultural laboratories. Complete placed, and the instrument reas­ and self contained — no accessory equip­ sembled without disconnecting the ment necessary. Fused cells for readings head section from the piping. Send on solution depths of 2.5, 10, 20, and 40 mm. for Bulletin No. 1. Selective light filters available. The Model A-204 Manometer for line pressures up to 400 lbs. per sq. in. These instruments are built with heavy walled annealed straight ^JUe fC lett Pyrex tubes, gland packed top and bottom in steel end blocks and fybuvU m etesi clamped to the body at fixed inter­ vals to prevent distortion. Bulletin N o. 10 gives complete details. WELL-TYPE MANOMETERS Direct reading Manometers for line pressures up to 2 50 lbs. per sq. in.—of standard construction, available in both wall and flush N o. 3071 front panel mounting designs in Designed for the ranges from 6" to 100". Ask for B ulletin B. rapid and accurate determ ination of SIGHT FEED BUBBLER thiamin, riboflavin, and other substances which This Sight Feed Bubbler with bowl fluoresce in solution. The sensitivity and sta­ of Pyrex glass is good for line bility are such that it has been found particu­ pressures up to 50 lbs. per sq. in.; larly useful in determining very small amounts plastic, for pressures up to 100 of these substances. lbs. Sturdily built in brass or semi­ To further increase its adaptability a separate steel construction. Needle valve circuit and scale is provided for colorimetric controls bubbler rate. Ask for Bulletin 21. determinations. LITERATURE SENT UPON REQUEST THE MERIAM CO. 1969 West 112th St., • Cleveland, Ohio

S IN C E 1911 K l e t t Manufacturing Co. 179 EAST 87TH STREET NEW YORK, N. Y. MANOMETERS, METERS AND GAUGES FOR THE ACCURATE MEASUREMENT OF PRESSURES, VACUUMS AND FIOWS OF LIQUIDS AND GASES INDUSTRIAL AND EN GINEERING CHEMISTRY Vol. 15, No. 6

COORS • CHEMICAL AND SCIENTIFIC PORCELAIN WARE H u U ,qu!Pm.nt O f A l l MODERN TESTING LABORATORIES

OIL —earth's black m agi«"—turns the mighty wheels of industry on land, on sea, and in the air.

Truly COORS ware is an essential prod­ uct in the manufacture of all necessary ad­ juncts to living.

M ills, washers and calenders in a range of sizes for laboratory work in natural and synthetic rubber and plastics. Heavily con­ structed and precision built. For full details write NATIONAL RUBBER MACHINERY CO. General Offices: Akron, Ohio C oors P o r c e l a in C o m p a n y GOLDEN, COLORADO

LUMETRON Photoelectric COLORIMETER and FLUORESCENCE METER A high precision instrument of unusual flexibility for all tests involving . . . TRANSMISSION 211 G R A M FLUORESCENCE CAPACITY TURBIDITY Without the Use of U. V. ABSORPTION Separate Weights 10 M G. SENSITIVITY Hereisa low-priced baianee which will be found most useful wher­ ev er fast# accu rate w eig h in g Is r e ­ quired. With its sensitivity of 10 milligrams and its capacity of 211 grams—without the use of any loose weights— it has innumerable applications in the analytical or in the educational laboratory. W eigh­ ing is greatly speeded up by the ingenious provision for damping: the beam can be stopped by sim­ ply pressing the finger against the beam-end, which swings in a slot­ Mod. 402EF for determination o£ V it Bi and Bj ted recess. A lso No. 577 Schaar Triple Beam Reflection Meters Electronic Vit. A Meters Balance, 211 Gram Capacity, Clinical Colorimeters Electronic Photometers Finished in 8Iaclc Continuous Flow Colorimeters Electronic Interval Timers E n a m e l...... Barrier-Layer Photocells Write lor Literature. Prompt Delivery. SCHAAR & COMPANY PHOTOVOLT CORP. Complete Laboratory Equipment 95 Madiion Ave. New York City 754 WEST LEXINGTON ST., CHICAGO June 15, 1943 ANALYTICAL EDITION

Evaporated silver film, magnification 100,000 x. The RCA Electron micrograph of zinc oxide smoke film formed by burn­ Electron Microscope reveals the film is not continuous but ing pure zinc in oxygen. For the first time the very fine parts of consists of extremely fine particles, whose average size is less zinc crystals can be seen. Below is diffraction pattern of same than one two-hundred-fifty-thousandths of an inch. specimen taken with Electron Microscope diffraction adaptor.

A transmission electron diffraction pattern of the thin silver The reflection method is here used to obtain the diffraction film shown above. Diffraction patterns indicate the structure pattern of zinc oxide. When the material is too thick for trans­ and physical arrangement of the atoms in the material. mission, the reflection method achieves equally good results.

TAKING THE MYSTERY OUT OF METALS

A case-study of the practical application of the RCA Electron Microscope

Once the pattern of particles in a metal is attachment, it permits study of atomic lattice of known, once their size and structure, their specimens, leading to new, significant analyses. position and relationship to each other can be The RCA Electron Microscope is a com­ observed and studied, the way is open for re­ pact unit, drawing its power from an ordinary search that will yield rich rewards. 110-volt, 60-cycle, single-phase A. C. outlet The RCA Electron Microscope is playing a Its operation can be mastered by any compe­ new and important part in solving metallurgi­ tent laboratory technician. cal mysteries for science. By its inherent high In your plant, the RCA Electron Micro­ resolving power making possible useful magni­ scope can be an instrument for important prog­ fications up to 100,000 diameters, it enables ress in research about your products man to see structural details never before ... Complete data and literature are within his ken. And through the principle available on request. RCA Vic­ of electron diffraction, which the instru­ tor Division, Radio Corpora­ ment makes use of by means of a simple tion of America, Camden, N. J.

RCA ELECTRON MICROSCOPE INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. IS, No. 6

THE FRONT OF INDUSTRY

We bow to you chemists, for your con­ tributions toward beating the Axis. W ithout your investigative and creative minds, how weak would our Mar effort be! In your work, we hope you find Hoskins Laboratory Furnaces are help­ ful and reliable. We build them the best w'e know how, and their long life is shown by many of them still being in use after more than twenty-five years of service. If you need more laboratory Shoicn above are the FH-303-A for carbon combustions; the FA-120 for furnaces, MT-ite to your dealer or to us. measuring the volatile content of coal; and the FR-234 that handles four . . . H oskins M an u facturin g Co., carbon combustions at one time. All Hoskins Furnaces have Chromel elem ents, noted for their durability and ease of renetcal. ( Want our free Detroit, Michigan. Heating Unit Calculator?) HOSKINS PRODUCTS

ElECTRIC HEAT TREATING FURNACES • • HEATING ELEMENT ALLOYS • • THERMOCOUPLE AND LEAD WIRE • . PYROMETERS • • WELDING WIRE • • HEAT RESISTANT CASTINGS • • ENAMELING FIXTURES • • SPARK PLUG ELECTRODE WIRE • • SPECIAL ALLOYS OF NICKEL • • PROTECTION TUBES June 15, 1943 ANALYTICAL EDITION 21

DAIGGER Acid and Alkali Resistant LABORATORY ENAMEL

Daigger NUKEM Laboratory Enamel can be termed a "liquid plastic" and may be easily applied to any surface. W ith H and W ith M ortar and Peatle This product was originally developed for use Homogenizcr in industrial plants where steel and wood parts are exposed to acid and alkali fumes. Daigger Laboratory Enamels are applied with No Emulsion Failures little effort either by brush or spray methods. They will set out-of-dust in about fifteen minutes and will dry hard in one hour. On properly With the IAB0RAT0RY HOMOGENIZE!! prepared surfaces, these enamels will cover in • Test samples, experimental and keep clean. Save time and one coat without showing brush marks; they will batches perfectly homogenized materials. Capacity, 1 to 10 produce rich, full gloss finishes which are quickly, conveniently. Perma­ ounces; sturdily made of molded resistant to heat, alcohol, acids, alkalies, fumes nent suspension with no failures aluminum; stainless steel piston. of all kinds and constant washing or scrubbing. — if ingredient-ratio is sound. Height, 101/2 inches. Still avail­ The microphotos above show able from pre-war stock. Only higher degree of dispersion. S6.50 complete, direct or from PRICED AT P 0 r g a ^ ° n Hundreds in daily laboratory your laboratory supply house. *1.75 per quart use. Easy and simple to operate Satisfaction Guaranteed /

Write for color card and descriptive literature HAND A. DAIGGER & CO. Laboratory Apparatus and Reagents HOMOGENIZED 159 West Kinzie Street INTERNATIONAL EMULSIFIERS, INC. 2401 Surroy Court, Chicago, III.

THE CHEMISTRY o f th e ALIPHATIC ORTHOESTERS B y HOWARD W. POST Department of Chemistry, University of Buffalo AMERICA! It is the purpose of this ivork to co-ordinate and present the material now to be found in the literature on the ortho- esters of aliphatic carbon, namely those of orthoformic acid ANALYTICAL and its homologs and of orthocarbonic acid. Space will also be given to silicon analogs of these compounds. The subject matter will be grouped under headings indica­ FILTER PAPERS tive of reactions rather than of compounds and under sub­ headings arranged in historical sequence. For instance, the “Ash-Low” Grades for Qualitative and preparation of orthoesters will be treated according to Semi-Quantitative Analyses method, subdivided chronologically. “Ash-Free” Grades for Gravimetric Analyses of Greatest Accuracy CONTENTS: — Introduction; Preparation and General Properties; Reactions with or Catalyzed by Inorganic Complete Line of Hand-Folded Filters Acidic Substances; Reactions with Organic Acids, Anhy­ High Wet-Strength Filter Papers drides and Halides; Reactions with Nitrogen Compounds; for Use in Quantitative and Reactions with Organo-Metallic Compounds; Carbohy­ Qualitative Analyses drate Orthoesters and Orthoacids; Miscellaneous Reactions; Analytical Filter Pulp in Moist, Silicoorthoesters: Preparation and Physical Properties; Dispersed Condition Chemical Properties of Silicoorthoesters; Polyalkoxides of Write for Samples Other Elements of the Fourth Column; Physical Properties of Orthoesters; Author Index; Subject Index. CARL SCHLEICHER & SCHUELL CO. A. C. S. Monograph No. 92, 188 Pages $4.00 Plant and Laboratories: SOUTH LEE, MASS. Executive and Sales Offices: 116-118 WEST 14th ST., REINHOLD PUBLISHING CORP. 330 W. 42nd St., New York, N. Y. NEW YORK 22 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 DETERMINATION OF THIAMINE

REAGENT— Ethyl p-Aminobenzoate METHOD — Colorimetric REFERENCE — Kirch and Bergeim, J. Biol. Chem., 143, 575 (1942)

D iazotized ethyl p-aminobenzoate reacts with thiamine to form a colored compound that can be quantitatively extracted from the solution by the use of iso-amyl alcohol. The intensity of the color is pro­ portional to the thiamine present, and is read on a Pulfrich photometer. The method is specific for free thiamine, and can be applied to solu­ tions containing 3-6 gamma per 5 cubic centimeters. Absorption and other methods of purification are unnecessary, thus making the determi­ nation rapid. The reagent is available in a highly purified grade, as Eastman 1177 Ethyl p-Aminobenzoate, M P 88-90°. Write for an abstract of the article in which the colorimetric deter­ mination of thiamine, with ethyl ■p-aminobenzoate, is described. . .. Eastman Kodak Company, Chemical Sales Division, Rochester, N. Y. |4 > There are more than 3400 ER EASTMAN ORGANIC CHEMICALS KODAK

EXPLOSION PROOF MOTOR FOR HAZARDOUS LOCATIONS To meet the demand of the O il Industry for a large centrifuge with an ex­ plosion proof motor, the International M odel A E Centrifuge was developed. The Model A E , equal in size and capacity to the Size 2 Centrifuge, has • maximum speed of 1,700 r.p.m. which is sufficient for the testing of oils. However, the Chemical and Paint Industries require the higher speeds of the new explosion proof motor which is now available.

International Model Be Centrifuge with explosion proof motor of 3,450 r.p.m.

accommodates the regular centrifuge equipment and has the safety features of the M odel A E with the added advantage of higher speed for more gen­ eral industrial uses. The special adjustable high speed motor is approved for Class 1 — Group D locations.

A speed regulator and disconnect switch are mounted permanently in a vaporproof box on the side of the centrifuge and connected through con­ duit and explosion proof fittings to the motor. The sealed ball bearings of the motor require no attention. A conveniently located hand brake permits rapid stopping.

Further information upon request INTERNATIONAL EQUIPMENT CO. 3 5 2 Western Avenue Boston, M a s s .

Makers of Fine Centrifuges for More than Forty Years. June 15, 1943 ANALYTICAL EDITION 23 GLASS ABSORPTION CELLS

jjitie Q ualify u » m m ade k t f , 1 ^ I 6 t t

Optical Flat Fused in an elec­ Walls. Many tric furnace with stock sizes are cem en t that is available. Special acid, alkali and sizes made to solvent resistant. order.

Sole manufacturer in the United States of fused Electrophoresis Cells ■ Makers of complete Electrophoresis Apparatus

K lett Manufacturing Co. 1 7 7 e a s t 8 7 T H street, new york, n. y.

W elch J iix fli V acuA im 5 b u a - S e a l P u w u p A , Designed for quantity output GUARANTEED VACUUM — .05 Micron (.00005 mm Hg) OPTIMUM OPERATING SPEED 300 Revolutions Per Minute FREE AIR CAPACITY 33.4 Liters Per Minute OIL REQUIRED — 650 ml. Duo-Seal Oil Quiet and Sturdy Especially adapted for ELECTRONICS • FREEZING DISTILLATIONS • DRYING

DUO-SEAL VACUUM PUMP, Motor Driven. Vacuum .05 micron — free air capacity of 33.4 liters per m in u te ...... $140.00

Also supplied with a larger motor giving 57 liters free-air capacity per minute and vacuum of 0.1 m icron...... $155.00

LIMITED STOCKS AVAILABLE PROMPT SHIPMENT. ORDER NOW. AN INTERESTING BOOKLET — FREE Tells all about vacuum W . M. Welch Scientific Company techniques and is full of Established 1880 valuable facts. Send for your copy today. 1518 Sedgwick Street Chicago, Illinois, U. S. A. INDUSTRIAL AND ENG INEERING CHEMISTRY Vol. 15, No. 6

Although the Coleman pH Electrometer is designed for the laboratory, it is also used extensively for plant control. ELIMINATES GLASS ELECTRODE BREAKAGE — a special feature of the Coleman. The electrodes are sealed and fitted at the factory and safely mounted to avoid all contact with the sample cup. The KC1 junction is easily and thoroughly flushed . . . no ground joints to carry over from test to test! SEALED CHASSIS prevents moisture from causing leakage error. The only attention required is replacement of batteries, tubes and glass electrode . . . the latter is guaranteed for six months and usually lasts over a year! No. 3537-A Coleman pH Electrometer, for pH only. Double scale, 0-13 pH by 0.1 pH divisions and also in millivolts. Easily read to 0.02 pH and amply sensitive to same limits. Electrical accuracy 0.1%. Elec­ trochemical accuracy 0.05 pH. Complete CHEMISTRY’S with Sealed Ag-AgCl Glass and Reference Electrodes, buffer solution, tubes, batteries. With AUTOMATIC temperature compen­ PART IN VICTORY sator. In ten thousand and one ways, the chemist is bringing the day of Victory closer and closer. Through multiplied production of military explo­ sives . . . through discoveries of new processes, new syn­ thetics, new compounds, new alloys . . . chemis­ try is speeding the complete mobilization of Price $235.00 our war machine and the nation's needs to keep Reasonable priority requirements. it running. Write for Cat. IE 615 Spencer’s contribution in this achievement lies in the field of optical science and in supply­ ing instruments of many types indispensable to the world of chemistry.

opencer LENS COMPANY BUFFALO, NEW YORK SCIENTIFIC INSTRUMENT DIVISION OF AMERICAN OPTICAL COMPANY June 15; 1943 ANALYTICAL EDITION 25

A N e i m ) Catalog and M anu al

Over 800 illustrations, diagrams and plan­ ning ideas. 224 pages. Abounding with inter­ est and information.

WallTable-Fume Hood, groups essential pieces in small space, in easy reach. Hundreds more items of practical de­ sign priced within your budget. Cat. No. 4970 The Foundation of Your Ltiboratory Planning

NEW A. C. S. MONOGRAPHS • e PROTEINS, AMINO ACIDS AND PEP­ acid-base equilibria in amino acid and protein solutions. All these factors are considered with respect to their influence on the solu­ TIDES AS IONS AND DIPOLAR IONS bility of proteins in water, salt solutions, acids and bases, and organic solvents. 686 P ag es. Illu s tra te d .$13.50 by Edwin J. Colin and John T. Edsall Harvard Medical School THE CHEMISTRY OF NATURAL including Chapters by John G. Kirkwood, Hans Mueller, J. L. Oncley, and George Scatchard. COLORING MATTERS The study of proteins has recently become of major importance to The Constitutions, Properties, and Biological Relations chemists, biochemists, immunologists, clinicians, as well as to many industries. Enzymes and most viruses are proteins, as are of the Important Natural Pigments many bacterial toxins and all of the antibodies which neutralize by Fritz Mayer these toxins. Proteins are the most important structural elements of living cells. Silk, wool, and certain synthetic textile fibers are Formerly Professor of Chemistry proteins. The processing of gelatin, casein and soy bean protein University of Frankfort-on-M ain is of great commercial significance. Especially the processing of Translated and Revised by A. II. Cook blood plasma for the use in transfusion has become a vital factor in medical treatment of war injuries. This unique and masterly treatise presents in easily accessible This Monograph is concerned with the composition and the size, form a vast amount of essential data on the constitution, physical shape and electrical properties of protein molecules. The earlier constants, structures, and functions of all the known natural portion of the book, which deals with amino acids and peptides, pigments and related substances. Many of these, such as vita­ lays the foundation for the detailed discussion of proteins which mins A and K, the carotinoids, anthracenes, chlorophyll, ribo­ follows. Many comprehensive tables of data are included. flavin, hemin, and the flavones are in the forefront of modern The molecular weights of proteins, as determined from osmotic biological research. The basic information, extensively docu­ pressure, ultraccntrifugal and X-ray studies, are presented in mented, which is provided in this volume will therefore be of the detail. These data, combined with studies of viscosity, double greatest interest to biochemists, dye technologists, and students refraction of flow, and dielectric dispersion, permit calculations of organic structure. Fully discussed also are such well-known of protein shape, which is of major importance for the biological coloring matters as indigo, alizarin, and the anthocyanins, and functions of proteins. The very remarkable dielectric properties also the insect dyes, and the pigments of fungi, molds, rare woods, of amino acids, peptides, and proteins are fully treated, as are and other materials. 351 Pages. 810.00

Sent to you forjree examination period oj ten days anywhere in continental United States REINHOLD PUBLISHING CORPORATION, 330 West 42nd St., New York, N. Y. 26 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 15, No. 6 NEW E. G. CO. TIMER

QUANTITY Need Noi SacM^ice Actual Size ACCURACY OPTICAL HIS domestic, high quality, 7 jewel, i n T 60 second Timer is the product of MANUFACTURING one of the most famous makers of high quality watches and time-pieces in America. Before the war, our chief aim was the production of optical material and equip­ This fine time-piece is guaranteed for accuracy and operates in the most ment possessing unequalled precision. convenient manner preferred in labora­ tory tests, with start, stop, and fly-back Today volum e of output is all-important from the crown. too, and we are glad to report that al­ It is made to conform to U. S. Govern­ though this output has increased many- ment Specifications GG-W -i 1 1 A, Type B, Class 7. fold, Perkin-Elmer standards of accuracy remain unchanged. Sixty second sweep of large hand indi­ cates split seconds in 1 /5 seconds. The small integrating hand indicates 30 In the postwar period, we shall again be sweeps of the large hand or 1S00 pleased to place Perkin-Elmer facilities at seconds before repetition. your disposal for optical engineering, de­ The case is chrome-plated nickel with sign and manufacturing of the highest dust-proof double backed case and the calibre. face is fitted with a clear unbreakable crystal.

GR976a stopwatch 60 second $29.50 THE PERKIN-ELMER CORPORATION CIEN 8ROOX. CONNECTICUT THE EMIL GREINER CO.

161 Sixth Ave. New York, N. Y. MANUFACTURERS OF PRECISION LENSES - PRISMS and MIRRORS OPTICAL tltKI AND COMIVLTATIOM w