ANALYTICAL CHEMISTRY.1. INTRODUCTION.THE demand for increased production of goods and materials-and forincreased rate of production-created by war-time exigencies and intensifiedby post-war plans for economic recovery have done much to accelerate thenatural development of analytical methods that are both rapid and accurate,yet suitable for routine measurements by semi-skilled or hastily trainedoperatives. Nowhere does this appear so clearly as in the field of emissionspectrography where a remarkable degree of inechanisation has alreadybeen achieved. The special interests of soil-scientists and biochemists havelikewise speeded the developments of the Lundegkrdh technique fore-shadowed in the Report for 1941 and now described in the section of flamephotometry.Yet despite the ever-increasing importance of physical methods, morethan half the papers published annually still relate to the procedures ofvolumetric analysis, and although there have been no spectacular develop-ments the past decade has seen steady progress in its techniques andapplications, some of which are reviewed in Section 4 of this Report.Thelast section deals with a complex naturally occurring material of greateconomic importance-sen water-which presents many absorbing anddifficult problems tlo the analyst. H. I.2. ANALYTICAL EMISSION SPECTROGRAPHY.The value of emission spectrography as an analytical technique wasthoroughly established when the subject was last reviewed 1 but its studyreceived a major impetus a t the outbreak of war in 1939, when industrywas faced with rapid expansion.I n the metallurgical industries, andparticularly in those concerned with light a.lloys, the provision of increasedanalytical facilities presented serious difficulties since the additional skilledstaff and buildings needed for chemical methods of analysis were rarelyavailable. The problem was happily solved by the installation of spectro-graphic equipment which provided analyses of a satisfactory degree ofaccuracy in a fraction of the time taken by the chemical methods then inuse, whilst it was economical in both laboratory space, and numbers andquality of personnel.The process employed in quantitative analytical spectrography is treateda t length in the standard textbooks and various specific techniques have1 Ann.Reports, 1937, 34, 454.2 W. R. Brode, “ Chemical Spect,roscopy,” John Wiley & SQns Inc., New York,and Chapman and Hall Ltd., London ; F. Twyman, “ The Spectrochemical Analysisof Metals and Alloys,” Charles Griffin & Co. Ltd., London; D. M. Smith, “ CollectedPapers on Metallurgical Analysis by the Spectrograph,” The British Non-FerrousMetals Research Association, London; “ Analysis of Aluminium and its Alloys :Spectrographic and Polarographic Analysis,” The British Aluminium Co. Ltd.From that time progress has been continuousULAYTON : ANALYTICAL EMISSION SPECTROCRAPHY. 317been described in the literature.3 When spectrography is used as ananalytical tool in industry the details of the method employed dependlargely on the nature of the material to be analysed, the speed and accuracyrequired, and the equipment available, but the following steps a, re commonto most procedures.The material is first sampled, either by taking it intosolution in a suitable solvent, by pelleting, or by casting it into the formof rods or discs which can form one or both of the electrodes of the excitationdischarge. After being sampled, the material is suitably excited, thespectrum is recorded photographically and the densities of the spectrumline images are measured by ineans of il microphotometer. Subsequentcalculations allow the concentrations of the constituents of the sample tobe deduced from these density measurements.Errors in the conventional method of quantitativc analysis may arisca t all stage^.^ Pnefficient sampling is an obvious source of inaccuracy butone that can be readily minimised.The numerous errors associated withthe excitation of the spectrum and the use of the photographic plate areless obvious and are more difficult to overcome.It is generally accepted that attempts to accelerate the spectrographicprocess by shortening the exposure time and by using fast photographicemulsions may have a deleterious effect on the overall accuracy of theanalysis, and it is better to increase the accuracy of the method and togain speed by reducing the iiuniber of replicate analyses carried out oneach sample.The accuracy of analysis has been the subject of considerable attention,Gand after an improvement brought about by a proper understanding ofthe photographic process and the requirements of accurate photometry,attention was directed- a t improving the spark source normally employedfor the routine analysis of metallic materials.Several types of improvedexcitation unit have been described, and their use may, in general, besaid to result in an improvement in analytical accuracy of some 50% whenthey are applied to light alloys ; with ferrous materials the improvementS. Levy, J . AppL Physics, 1940, 11, 480; J. van Calker, Spectrochim. Acta, 1944,2, 333; P. Cohen, J . Opt. SOC. Amer., 1946, 36, 489; A. Walsh, ‘‘ Collected Papers onMetallurgical Analysis by the Spectrograph ” (see ref. Z), p. 65 ; A. Cornu, Compt. rend.,1946,222, 1341 ; J.Wilken, Metallwirts., 1940,19, 121 ; A. von Zeerleder and F. Rohner,Helv. Chim. Acta, 1940, 23, 1287; H. Correll, Aluminium, 1940, 22, 525; R. W. Callonand J. E. Burgener, J . Opt. Soc. Amer., 1944, 34, 543; H. L. Collins and R. T.V. Callon,Canad. Metals, 1945, 8, 20; G. S. Smith, Met. I d . , 1945, 67, 226; 1947, 70, 23;“ Reports of A.S.T.M. Committee E-2 on Spectrographic Analysis,” Proc. Amer. SOC.Test. Mat., 1939 onwards.T. A. Wright, Amer. SOC. Test. Mat., Reprint No. 112,1940; H. Mader and R. Poet-zelberger, Metallwirts., 1940, 19, 381; N. V. Buyanov, Zavod. Lab., 1940, 9, 69;V. K. Prokof’ev, Izvest. Akad. Nauk, S.S.S.R., 1940, (Phys.), 4, 5 ; A. G. Quarrell andG. E. A. Bramley, J . Inst. Metals, 1941, 07, 25; W. Seith and H.Hessling, Z. Elektro-chem., 1943,49, (4/5), 210; S. Levy and 0. W. Christine, J . Opt. SOC. Amer., 1946,36,503.ti H. Miider and R. Poetzelberger, Spectrochim. Acta, 1039, 1, 213.6 A. E. Ruehle, Bull. Amer. SOC. Test. Mat., 1941, 33; H. B. Vincent and R. A.Sawyer, J . Opt. Soc. Anter., 1942, 32, 686318 ANALYTIOAL CHEMISTRY.is less marked but is still appreciable. Microphotometry of the photo-graphic negative, the introduction of which had enabled consistently satis-factory analyses to be carried out by relatively unskilled operators, involvesminor but appreciable errors. Recent developments aim to supplant itby the direct photoelectric measurement of the spectrum line intensities.The speed of analysis has continuously increased, and it is common-place to find methods using photography of the spectrum and subsequentmicrophotometry of the line images which take no more than 10-15minutes for a complete analysis of a sample for several elements, whilst ifa direct photoelectric measurement of line intensity is made the time maybe no more than 2 minutes.The spectrograph has also been applied to an ever-increasing range ofanalyses outside the metallurgical field and is now widely used in suchdiverse work as the analysis of biological materials, ores, and oil additives.Progress in the paat ten years may thus be summed up as an increasein the accuracy, speed, and scopc of the method.The more recent develop-ments in source units coupled with the application of multiplier photocellsto the direct measurement of spectrum line intensities show promise ofproducing still greater accuracy and speed in the near future,Personal ErPors.-The spectrographic process involves numerous steps,and many of them are liable to introduce errors when they are carried outby inexperienced workers. Attention has therefore been directed at simpli-fying and mechanising the process so that it is as independent as possibleof the mistakes and judgment of the operator.In preparing the electrodes for excitation, manual filing has been largelyreplaced by machining or grinding, and the machines designed for thisoperation provide an electrode tip of standardised shape and finish.' Im-proved electrode stands have been described and in the type due toA. von Zeerleder and F.Rohner three pairs of electrodes are accommodated ;the first pair is automatically pre-burned for a given period, the secondprovides the spectrum lines in analysis, and the third can be simultaneouslychanged for a further pair of fresh electrodes. The photographic exposurehas been controlled by the use of automatic timing switches which give anexposure of predetermined duration lo or by photoelectric devices whichexpose the photographic emulsion until a fixed amount of light energy hasbeen emitted by the discharge.ll Automatic photographic processing as7 C. L. Waring, Metals and Alloye, 1945, 21, 1013; H. Moritz, Aluminium, 1940,22, 421 ; H. Kaiser, Spectrochim. Acta, 1942,2, 288; E. J. Esstmond, J. Opt. SOC. Amer.,1944, 34, 621 ; K.R. Mayors and T. H. Hopher, Ind. Eng. Chem. Anal., 1941, 13, 647.8 H. R. Clayljon, J . Sci. Irsek., 1941,18, 65 ; B. F. Scribner and C. M. Carless, J . Res.Nut. Bur. Stand., 1943, 30, 41; J . Opt. SOC. Arner., 1943, 53, 515; W. D. Owsley endR. C. McReynolds, Rev. Sci. Ins&., 1942, 13, 342.Specdrochdm. Acta, 1940, 1, 400.10 G. Belz and G. Reiniger, &id., p. 323 ; F. Walbank, ibid., 1941, R, 160; R. H. Keck,11 J. S. Sedov, Cmpt. rend. (Doklady) Acad. Xci. U.S.S.R., 1943, 41, 329; H. R.ibid., 1944, 2, 389.Clayton, J. Sci. I w t r . , 1940, 23, 233ULAYTON : ANALYTIOAL EMISSION SPECTROBRAPHY. 319used for large-scale roll-film development has not been applied to spectro-graphy, but commercial equipment of a semi-automatic nature for dovelop-ing, fixing, washing, and drying is available.la Many types of microphoto-meter have been devisedJ13 ranging from models designed for very rapidworking l4 to those which are prepared to sacrifice speed to the attainmentof a higher accuracy.l5 In this country the Hilger non-recording micro-photometer is almost universally employed , partly for preference andpartly because it is the only suitable instrument manufactured here.Itis a basically simple apparatus requiring little or no maintenance and isconvenient to use, but it is considered by some users to compare unfavour-ably with representative Continental and American instruments becauseof the slow speed of response of its galvanometer and its critically focussedoptical system.The calculation of the analytical results from the microphotoinetricmeasurements has been accelerated and simplified by the use of calculatorswhich convert the readings of the microphotometer galvanometer intorelative light intensity values and subsequent,ly concentrations of thevarious minor constituents.16 These instruments have become almost anecessity in those laboratories where large numbers of determinations aremade and various patterns have been described. Three interesting detaileddescriptions of accessory equipment of the type mentioned above havebeen published by H.Brackebusch,l7 J. L. Saunderson and V. J. Calde-~ o u r t , ~ * and H. M ~ r i t z . ~ ~Sampling.-To avoid heterogeneity in the sample, many workers havebeen attracted by methods of analysis involving the use of a solution ofthe material under test.20 This procedure has several advantages ; non-metallic and non-conducting materials can be satisfactorily dealt with, andl2 See, e.g., trade literature of Associated Research Laboratories, Glendale, Cali-fornia.l3 R.C. Machler, Proc. 7th Summer Conf. on Spec. and its Applications, Mass.Inst. Toch., 1939, 1940, 65; E. M. Thorndike, Ind. Eng. Chem. Anal., 1941, 13, 66;A. Gatterer, Spectrochim. Acta, 1941, 1, 352 ; H. B. Vincent and R. A. Sawyer, J. Opt.SOC. Amer., 1941, 31, 639; W. S. Baird, ibid., p. 179; H. W. Diotert and J. Schuch,ibid., p. 54 ; R. Poetzelberger, Spectrochim. Acta, 1943, 2, 296.l.1 W. A. Kerr, Proc. 7th Summer Conf. on Spec. and its Applications, Mass.Inst.Tech., 1939-1940, 68; R. Fiirth, Nature, 1942, 149, 7 3 ; E. M. Thorndike, I d . Enq.Chem. Anal., 1941, 13, 66-67.l6 G. 0. Langstroth, K. B. Newbound, and W. W. Brown, C a d . J . Res., 1941,A , 19, 103.l6 G. Balz, Aluwziniurn, 1940, 22, 343; C. King, J. Opt. SOC. Amer., 1942, 32, 112;N. S. Bmmmelle and H. R. Clayton, J. Xoc. Chem. Id., 1944, 63, 83; D. A. Sinolair,J. Opt. SOC. Amer., 1944, 34, 689; A. P. Vanselow and G. F. Liebig, ibid., p. 219;J. C. Henderson-Hamilton and A. Lourie, J. SOC. Chem. Ind., 1945, 64, 309.l7 Spectrochim. Acta, 1941, 8, 18.Aluminium, 1942, 24, 394.2o R. Bauer, ibid., 1940, 82, 9 ; W. D. Treadwell and R, Walti, Helv. Chim. Acta,1940, 23, 1446; A. Beerwald and W. Brauer, 2. Metallk,, 1941, 33, 44; R.Walti,Diss. Eidg. Tech. Hoohsch., Zurioh, 1943; R. J. Kiers w d D. T. Englis, I d . Eng.Ghem. And., 1940,12, 275,J . Opt. SOC. Amer., 1944, 34, 116320 ANALYTICAL CHEMISTRY.the preparation of standards of comparison is greatly facilitated since theymay be synthesised from pure salts. Excitation may be by arc, by spark,or by a controlled flame. In use, solution methods are generally moretime-consuming than those employing solid electrodes of the material tobe analysed, and if the solution is used as such, without evaporation, thespray from the discharge may damage other apparatus in the laboratory.Pelleting or briquetting of metallic filings 21 or non-metallic powders 22 hasbeen used to minimise the heterogeneity of the sample but the techniqueis not widespread.The most generally accepted procedure for the analysisof metals is to employ electrodes of the material under test, and in thefield of metallurgical analysis these are usually prepared by casting in oneor two forms. If the samples are cast as rods23 the discharge may bepassed between two of them; if they are cast in disc form the discharge ismade to take place between the surface of the disc and a counter electrodeof another material, usually a pointed rod of graphite.24 In using elec-trodes of these types attention must be paid to the casting technique.25The moulds employed are generally designed to give rapid chilling in orderthat the grain size of the metal comprising the sample shall be as fine aspossible. The temperature of both the mould and the metal before samplingis usually well defined.Such measures ensure a consistently high standardof sampling, but in order to reduce further the effects of heterogeneity inthe sample itself it is customary to make replicate photographic records ofthe spectrum of each sample, a different part of the sample being used foreach exposure.26 Rotating electrodes and discs have been tested tominimise sampling errors but are not generally accepted. as necessary.Calibration and Photographic Procedures.-For many years quantitativeanalysis was carried out almost exclusively by B. A. Lomakin’s method,27in which the photographic plate is calibrated by the inclusion of spectra ofstandard alloys of which the composition has been determined by carefulchemical analysis.This procedure suffers from two disadvantages : it iswasteful, since much of the space on the photographic plate is taken up bythe spectra of the standard alloys, and the casting and analysis of thelarge numbers of these standards which are required for a full and usefulapplication of the method are laborious tasks. Other methods of platecalibration have therefore been developed.In geceral, tzhese methods iiivolve the inclusion on each photographicplate of an intensity pattern consisting of steps of known relative intensities,fcrined, for example, by exposing a portion of the plate to a light source21 H. C. Harrison and C. C. Ralph, Ind. Eng. Chem. Anal., 1943, 15, 466; C. J.Neuhaus, J . Opt. SOC. Amer., 1943, 33, 167; P.A. Leichtle, ibid., 1944, 34, 454; H. W.Dietert, ibid., 1941, 31, 693.2% E. J. Fitz and W. M. Murray, Id. Eng. Chem. Anal., 1945,17, 145; S. H. Wilsonand M. Fieldes, New Zealand J. Sci. Techn., 1941, 23, 47B.H. Moritz, Aluminium, 1940,22,421; 1943,25, 389.24 H. V. Churchill and J. R. Churchill, J . Opt. SOC. Arner., 1941,31, 611.25 “ Analysis of Aluminium, etc.” (see ref. 2), 2nd edn., p. 16.Ibid., p. 19. 27 2. U W Q . ohem., 1930, 187, 76CLAYTON : ANALYTICAL EMISSION SPECTROQRAPHY. 32 1through a, stepped optical wedge or rotating stepped sector. After measure-ment of the density of the photographic images of the steps, the character-istic curve of the photographic plate may be constructed, an artificial originbeing employed because the absolute intensity of the calibrating intensitypattern is unknown.From this curve the intensity ratio of any spectrumline pair may be determined from the difference in densities of the twolines, and since the intensity ratio is a function of the concentration of theminor constituent, the amount of minor constituent present may be deter-mined. Preliminary work is involved in finding the relationship betweenthe intensity ratio of the two spectrum lines and the concentration of theminor constituent, but this is easily and accurately done by using analysedstandards.A method of plate calibration of this type is now commonly employedin laboratories dealing with large numbers of similar samples, but where avariety of materials are analysed the method is not so useful because ofthe large amount of preliminary calibration required.A proper under-standing of the photographic process, as it affects photometric photometry,is essential for the successful application of plate calibration techniquesto spectrography and the subject has been fully discussed by E. H. Amstein.28The photographic emulsion even on a single plate does not necessarily behaveas if it were uniform in its reaction to light, whilst the image after develop-ment often lacks uniformity through an incorrect processing techniquewhich must be carefully established and standardised. The characteristicsof photographic emulsions to ultra-violet radiation have been determined 29and by reference to these data workers have been able to select the mostsuitable type of plate and spectrum lines for their particular needs.Correction for spectrum ‘‘ background ” has been considered from aphotographic aspect and is usually allowed for in trace analysis,30 but itseffect is inappreciable in the analyses of constituents present in higherconcentrations.Excitation Sor~~es.-Probably the most outstanding contribution toimproving the speed and accuracy of the spectrographic process during thepast ten years has been the development of improved excitation units toreplace the D.C.arc and the condensed spark units previously employed.The condensed spark is not an ideal source for spectral excitation, par-ticularly for those metals whose oxides are good insulators, since the amountof energy passing through the analytical gap a t each individual spark isdetermined by the voltage a t which the gap breaks down.This breakdownvoltage may vary between wide limits depending on the condition of theelectrode tips and the degree of ionisation of the vapour between them a tthe instant when the discharge starts. With gap conditions which do not28 J . SOC. Chem. Ind., 1943, 62, 51; A. C. Coa,tes and E. H. Amstein, ibid., 1942,61, 21.E. H. Amstein, ibid., 1944, 03, 172.30 L. W. Strock, J . Opt. SOC. Amer., 1942, 32, 103; R. 0. Scott, J. SOC. Cheni. Ind.,1944,63, 26; J. Cholak and R. V. Story, J . Opt. SOC. Amer., 1941, 51, 730.REP.-VOL. XLV. 322 ANALYTICAL CHEMISTRY.provide a constant breakdown value, therefore, the discharge is not repro-ducible.In order to stabilise the breakdown potential it is common practiceto irradiate the gap with ultra-violet radiation of short wave-length, or toprovide a " leading point " across the spark gap to produce a corona dis-charge before the passage of the spark.31 These methods are not entirelysatisfactory and other methods of controlling the discharge have beeninvestigated.The first controlled condensed spark was described by 0. FeussnerF2who employed a synchronous rotary spark gap to apply the discharge voltageto the gap at predetermined intervals. More recently, J. T. M. Malpicaand T. M. Berry 33 have developed an electronically-controlled condensedspark in which the control is on the primary side of the high-voltage trans-former. In order to define the discharge conditions more accurately, otherworkers have attempted to separate the high-current-density spark phaseassociated with the initial breakdown of the gap from a subsequent low-current-density arc phase which provides most of the light output fromthe discharge, but itself plays no part in gap breakdown.The first circuitof this type was developed by K. Pfeil~ticker,~~ and subsequently, improvedsources working on the same basic principles have been described by otherworkers. Each of these sources has its own particular merits and dis-advantages; for instance, that due to A. Walsh 35 is so designed that it iseasily constructed from readily available components, and several circuitsto this design are in operation in this country with satisfactory results.The source unit described by C.Braudo and H. R. ClaytonY36 and producedin this country by the Metropolitan Vickers Electrical Co. Ltd., is morecomplicated, but dispenses with auxiliary spark gaps or synchronous inter-rupters by adopting electronic methods of synchronising. The " Multi-source " of M. F. Hasler and H. W. Dietert 37 and the circuit due toV. J. Caldecourt and J. L. Saunderson38 are commercially available inAmerica, where their versatility and stability have proved to be of greatvalue in analytical work.In the composite discharge units mentioned above, the analytical gapis first bridged by a very high-voltage discharge of low power and shortduration. The voltage is applied to the analytical gap a t a predeterminedtime and its value is so high that breakdown of the gap occurs practicallyinstantaneously irrespective of the condition of the electrode tips or ofthe state of the vapour between them.Once the gap has been bridged, asecondary discharge at a comparatively low voltage (250-2000 v.) isallowed to cross it, and the main spectral emission is due to this discharge.Conditions in this low-voltage discharge circuit may be modified by alter-31 @. Balz, H. Kaiser, and P. H. Keck, Spectrochim. Acta, 1941, 2, 92; G. Balz,Aluminium, 1944, 26, 60.33 2. techn. Physik, 1932, 15, 673; 2. Metallk., 1933, 26, 73.33 Gen. Elec. Rev., 1940, 43, 333.34 2. Electrochem., 1937, 43, 719; 1938, 30, 211 ; 1941, 33, 267.35 Met.Id., 68, 243, 263, 295,37 J . Opt. SOC. At)26T., 1983, 83, 218.36 J . Xoc. Chem. Id., 1947, $8, 259.a6 Ibid., 1946, 36, 90CLAYTON : ANALYTICAL EMISSION SPECTROGRAPHY. 323ation of the electrical parameters to produce any type of excitation rangingfrom a high-current, short duration spark to an arc of long duration. Theexcitation conditions can thus be adjusted to suit any particular problem.Minor constituents present in concentrations up to some 5% can beeffectively dealt with by these source units and an accuracy of some 24%of the amount of minor impurity present is generally attainable on singleanalyses. This performance is a marked advance in light alloy analysis,but does not show a great improvement over the performance of the con-densed spark for steel analysis.Coupled with the versatility of the units,however, the increase in accuracy is a very valuable attribute.Direct Measurement of Spectrum Line Intensities.-The conditions inthe electrical discharge having been stabilised, attention was directed toeliminating the photographic plate and its subsequent microphotometry.During recent years the measurement of the low light intensities encoun-tered in emission spectrography has been made practicable by the develop-ment of stable photo-electron multipliers.39 Interest in this field has beenlargely centred on the electrostatically focussed multiplier photocells origin-ally developed in America and now manufactured in this country. Thecharacteristics of these cells are described in the makers’ literature 4o andhave also been studied in detail by K.G. Kessler and R. A. Wolfe41 andR. W. Engstr0m.4~ In America, where the cells were available some yearsago, the direct measurement of the intensity of lines in the emission spectrumhas been carried out in numerous lab~ratories,~~ and with the developmentof ultra-violet transparent envelopes this method has now become estab-lished and commercial direct-reading spectrographs are marketed by a tleast two scientific instrument manufacturers.&These instruments allow direct measurement of the intensities of up to16 predetermined spectrum lines, and hence, as one line is employed as aninternal standard, the determination of the concentration of 15 minor con.stituents can be carried There seems little doubt that this methodof spectrographic analysis is still in its infancy, but even so it shows manyvaluable advantages over the more orthodox procedures employing photo-3B J.A. Rajchman and R. L. Snyder, EEectronics, 1940, 13, 20; K. Zworykin and40 R.C.A. Manufacturing Co. Inc., Harrison, N. J.; Cosmos Manufacturing Co.41 J . Opt. SOC. Amer., 1947, 37, 33.43 D. H. Rank, R. J. Pfister, and P. D. Coleman, ibid., 1942, 32, 390; D. H. Rank,R. J. Pfister, and H. H. Grimm, ibid., 1943, 33, 31 ; E. A. Boettner and G. P. Brewing-ton, ibid., 1944, 34, 6; G. A. Nahstoll and F. R. Bryan, ibid., 1945, %, 646; M. F.Hasler and H. W. Dietert, ibid., 1944, 34, 751; M. F. Hasler, J. W. Kemp, andH. W. Dietert, A.S.T.M. Bull., 1946, No.139, 22; J. L. Saunderson, V. J. Caldecourt,and E. W. Peterson, J . Opt. Soc. Amer., 1945, 35, 681; G. H. Dieke and H. M. Cross-white, ibid., p. 471 ; J. L. Saunderson and T. M. Hess, Metal Progress, 1946, 49, 947.44 Applied Research Laboratories, Glendale, California ; Baird Associates, Cam -bridge, Mass.dB Applied Research Laboratories, Glendale, California. Trade literature on the“ Spectrograph Quantometer Adaptor.”J. A. Rajchman, Proc. I.R.E., 1939, 2’7, 558.Ltd., Brimsdown, Middlesex.42 Ibid., p. 420324 ANALYTIOAL UHEMISTRY.graphic recording of spectral line intensities. The main advantage of themethod is its speed, and it is generally claimed that a sample can be analysedfor ten elements in under 3 minutes. The choice of spectrum line pairs ismade easier since a much wider range of intensities can be measured thanis practicable using photographic means; G.H. Dieke and H. M. Cross-white,"6 for example, report that they have successfully used a pair of Lineswhose relative intensity ratio was 1 : 40,000. On the other hand, the com-paratively large size of the photomultipliers, or of the mirrors used to directthe images of the spectrum lines on to them, makes it difficult to use lineswhich are close together.It is apparent that this type of instrument is most valuable in thoseapplications where large numbers of similar samples have to be analysed,and it is therefore eminently suitable for use in the control laboratories oflarge metallurgical works. When a complete survey of a material is requiredor when its impurities or constituents are not known the photographicmethod is of greater value, although direct-reading instruments in whichthe whole of the spectrum can be scanned by zt single photocell can beused for this purpose.The main disadvantage of direct-reading instruments is their high cost ;a high-dispersion spectrograph is essential for a full application of themethod, and the commercial instruments incorporate a specially designedgrating spectrograph of the required characteristics ; the recording apparatusis far from simple and for the best results a high-power controlled sourceunit is used for excitation.It is generally claimed that the analytical accuracy of such an equip-ment is a t least as high as the orthodox photographic method, and that,since the intensities of the spectrum lines used may vary between widelimits, a greater range of minor constituent concentrations may be sntis-factorily covered.Rage of Appkation.-Although the most marked advances in emissionspectrography in the past ten years have occurred in the analysis of metals,particularly light alloys, progress in other fields has been considerable.Many reviews of the scope of spectrographic methods have appeared in theliterature, and its general application to specific branches of science hasalso been dealt with.For example, B. L. Clarke and A. E. Ruehle4' havereviewed its applications in communications research, V. It. Ells 48 givesdetails of the spectrographic analysis of plant derivatives, and its use inagricultural investigations is reviewed by L. H.Rogers.49 For thesemiscellaneous analyses it is customary to use an arc between carbon orgraphite electrodes, the sample being held in a hole drilled in one or bothof the electrodes. fertili~ers,~~ and plants 53 The analyses of water,5046 J . Opt. SOC. Amer., 1946, 36, 192.48 J . Opt. SOC. Amer., 1941, 31, 634.6o L. W. Strock and S. Dexter, ibid., p. 167.61 R. 0. Scott and R. L. Mitchell, J . SOC. Chem. Id., 1943, 62, 4 ; G. W. Fox andR. A. Goodwin, Iowa State Coll. J . Sci., 1941, 15, 119; R. Q. Parks, J . Opt. SOC. Amer.,1912, 32, 233.47 Bdl System Tech. J., 1938, 17, 381.4s Ibid., p. 260CLAYTON : ANALYTICAL EMISSION SPECTEOGRAPHY . 325are of direct application in agricultural work, whilst in the biological fieldwe find methods ranging from those dealing with food 54 to those whichdeal with traces of metallic and non-metallic derivatives in blood.55 Arcexcitation has also found wide application in the analysis of minerals, ores,and slags,BG and a specific application to cement analysis is described byA.W. H e l ~ . ~ ’Many of these materials, particularly those which are easily taken intosolution, may be analysed by a flame technique.58 An oxy-acetylene flamehas been used for the analysis of fruit and plants,59 and the Lundeghrdhapparatus 6O has been adapted for the analysis of biological material ingenera1.6l H. LundegGrdh and H. Bergstrand 62 has used it for the examin-ation of liver, the material being ashed and dissolved in acid before beingatomised in the flame.Interesting modifications of the flame methodproduced by H. Ramage,63 in which a spill of filter paper is impregnatedwith the solution under test, have been described by M. N. Thruston 64and by F. C. Steward and J. A. Harrison.6” The former method uses acopper arc in which to burn the filter paper, whilst the latter involvesfeeding the spill into the flame a t a controlled rate to ensure regular spectralemission. Further developments in this field are described in Section 3of this Report (2. 326).Por insulating mat’erials a spark technique is usually employed.J. R. Churchill and R. G. Russell 66 pellet the material with sodium fluorideand graphite powder before sparking, and J.van Calker 66 has used 2 solidsample painted with a conducting material to enable the discharge to strike.The halogens and certain non-metals may be detected by using a glowdischarge in the vapour of the material, or by the use of spark excitationa t reduced pressure. Using the spark technique, K. Pfeilsticker 67 hasbeen able to detect the presence of gases in metallic electrodes.62 E. H. Melvin and R. T. O’Connor, Ind. Eng. Chem. Anal., 1041, 13, 520; R. T.O’Connor, ibid., p. 597.63 W. R. Brode and I. W. Wander, J . Opt. SOC. Amer., 1941, 31, 402; B. C. Brun-stetter and A. J. Myers, ibid., p. 163; M. L. Nichols and L. H. Rogers, Ind. Eng. Chent.Anal., 1944, 16, 137.64 J. K. Brody and D. T. Ewing, ibid., 1945,17,627 ; D.A. Warper and N. Strafford,J . SOC. Chern. Id., 1942,61, 74.6 5 A. Tracey and J. McPheat, Biochem. J., 1943, 37, 683.56 J. M. Bray, Arner. Min., 1942, 27, 769; W. W. A. Johnson and D. P. Norman,Astrophys. J., 1943, 97, 46; J. R. Churchill and R. G. Russell, I d . Eng. Chem. Anal.,1945, 1’7, 66; C. G. Carlsson, Jernkont. Ann., 1943, 127, 572; P. D. Korzh, Izvest. Akad.Nauk S.S.S.R., 1945, (Fiz), 9, 665.67 J . Res. Nat. Bur. Stand., 1945, 34, 129.69 M. A. Griggs, R. Johnstin, and B. E. Elledge, Ind. Eng. Chern. Anal., 1941, 13, 99.6o H. LundegBrdh, ‘’ Die quantitative Spectralanalyse der Elemente,” Jena, Gustav61 J. Cholak and D. M. Hubbard, Id. Eng. Chem. Anal., 1944, 16, 728.62 Regiae SOC. Sci. Upmliensis, 1940, 12, 1.63 Nature, 1936, 137, 67.66 Ann.Bot., 1939, 3, 427.6 7 Ibid., p. 424.5 8 Ann. Reports, 1941, 38, 274.Fischer, Vol. 1, 1929; Vol. 2, 1934.64 J . SOC. Chem. Id., 1942, 61, 144.6 6 Spectrochim. Acta, 1940, 1, 403326 AN&YTICAL CHEMISTRY.In the metallurgical field methods have been described for the investig-ation of inclusions and segregates in steel samples,68 and a particularlyinteresting paper on this subject has been written by J. Convey andJ. H. Oldfield.69 In their apparatus the photographic plate moves insynchronism with a traversing spark, and so the image on the plate showsthe point to point variation of the composition of the sample. The require-ments for the application of spectrography to rapid foundry control werediscussed in 1945 by H.W. Dietert and J. A. S~huch,~O and, apart fromthe omission of recent developments in direct-reading equipment, the con-siderations put forward by these authors still apply. The application ofnormal spectrographic equipment to foundry control in England hasrecently been described by H. R. Clayton,71 the analysis having beenaccelerated by shortening the time spent in photographic processing. Inthis method an alloy may be analysed for 5 elements in less than 10 minutes.H. R. C.3. FLAME PHOTOMETRY.The Lundeghrdh method of exciting the emission spectrum of an elementby atomisation of its solution and spraying into an air-acetylene flamehas become firmly established as a standard spectrographic technique (seep. 325). Plame photometry represents a logical development of this pro-cedure whereby the same means of spectrum excitation are employed, butthe subsequent measurement of emission intensity is greatly simplified bythe introduction of relatively cheap filters and direct-reading photocell-galvanometer combinations in place of the expensive spectrograph andassociated equipment required for the photographic recording and deter-mination of line intensities.In the typical flame photometer ordinarylight filters are introduced to select those regions of the spectrum whichcontain euitable spectral lines of the particular element concerned and tocut out any radiation emitted by other elements present in the solution atthe same time. This selection, of course, may be achieved far more pre-cisely by means of monochromators as in the case of the Beckmanspectrophotometer attachment,2 but the introduction of such devicesdetracts considerably from the simplicity and cheapness of the apparatus.The low emission energies available in such methods of direct photometrynecessitate the adoption of extra sensitive means of detection (see p.324).In simple flame photometry the use of relatively wide-band filters and the6 8 F. G. Barker, J. Convey, and J. H. Oldfield, J . Iron Xteet Inst., 1941,144, 143 P ;R. Weihrich and W. Schwarz, Arch. Eisenhuttenw., 1941, 15, 83; G. Thanheiser andJ. Heyes, ibid., 1940, 14, 543.69 J . Iron Steel Inst., 1945, 152, 473 P.70 Trans. Amer. Pound. Assoc., 1945, 52, 889.71 J . SOC. Chem. I d . , 1948, 67, 270.1 G.Thanheiser and J. Heyes, Mitt. Kaiser Withelm Inst. Eisertfmsch., 1937, 19,113; 1939, 21, 327; W. H. Jansen, J. Heyes, and C. Richter, 2. physikal. Chem., 1935,A , 174, 291; J. Heyes, Angew. Chem., 1937, 50, 871.2 R. H. Muller, Anal. Chenz., 1947, 19, No. 8, 21aLEYTON : FLAME PHOTOMETRY. 327consequent increase in available radiant energy enables the experimenterto employ less elaborate detectors such as barrier layer cells or photo-electric cells in conjunction with a sensitive galvanometer. Visual methodsemploying filters have also been described but are usually not to be recom-mended for routine work.* The photometer is calibrated against solutionscontaining known amounts of the element under test, the range of con-centration being determined by the sensitivity of the detector galvano-meter combination.For low concentrations, e.g., up to 10 p.p.m. of sodium,a linear relationship has been found between concentration and galvano-meter ‘reading, but for higher concentrations a calibration curve is usuallyrequired.5The key factor in flame photometry is, of course, the efficiency of thefilters in cutting out unwanted radiation. Sodium light is particularlydifficult to eliminate and early attempts to determine potassium in thepresence of sodium by substituting filters for monochromators were onlypartly successfuL6 Combinations of filters such as Jena (Schott) typesRG9 and BG17 proved to be more efficient than single filters. W. Xchuh-knecht obtained a satisfactory separation using a combination of threefilters, BG19, RG8, and BG3, and by means of a gas-filled photocell andsensitive galvanometer, was able to determine potassium in solution aschloride with an accuracy of f5y0 over the range 0.04-0-008% of K20.He claimed that interference due to sodium, calcium, and magnesium, alsopresent in the solution, was negligible.In 1938 a flame photometer based on this design was produced by thefirm of Zeiss for the routine determination of potassium.It employed agravity feed for the test solution to the atomiser, and a special filter : thefiltered radiation was measured with a cmium photocell and mirror galvano-meter. At the same time Messrs. Siemens introduced a model with asuction feed, an all-glass atomiser, and a single filter (RGS) : by means of astabilised amplifier, the phot’ocell currents could be read on an ordinarymilliammeter.Both these instruments underwent exhaustive tests in theanalysis of plant, food, and fertiliser extract^,^^ and in general the resultsagreed well with the figures for potassium as determined by chemicalanalysis, down to a concentration of 20 mg. of K20 per 100 C.C. The out-standing feature of this new technique was the ease and rapidity withwhich the analyses could be made, the average time for each determinationbeing about 2 minutes. The majority of the investigators found the photo-cell-galvanometer combination to be more reliable as a means of detectionbecause of instability in the photocell amplifier. Using a similar techniqueand the combination of filters RC.19, RGS, and BG3, H.LundegBrdh andS. Coy, Angew. Chem., 1937, 50, 301.Ind. Eng. Chem. Anal., 1945, 1’7, 605.F. Kertscher, Bodenk. Pjlanz., 1938, 10, 758; W. Lehrnann, ibid., p. 766; F. Gei-secke and W. Rathje, ibid., p. 776; L. Rohmlehrer, Mezogs. Kutat,, 1944, 17, 51 ;Chem. Abs., 1947, 41, 7165.* L. Schmitt and W. Breitweiser, Bodenk. PJEanz., 2938, 10, 750.ti Jansen, Heyes, and Richter, loc. cit., ref. 1. Anyew. Chem., 1937, 50, 299328 ANALYTICAL CHEMISTRY.K. Boratynski concluded that the method might be used successfullyfor the routine determination of potassium over the range 0.00025-0.004~.with an accuracy of &lo%.Further work revealed that interference due to the presence of otherelements, particularly calcium in soil extracts, could seriously affect theaccuracy of the potassium determinations. Substitution of the filter RG8 byRG9 improved the performance of the instrument, but it was still unsatis-factory and various other suggestions were made to overcome this inter-ference such as precipitation of the calcium and the use of illuminatingga,s instead of acetylene, whereby the intensity of the calcium radiationwas reduced owing to the lower flame temperature produced.lO Theaddition of ammonium phosphate and calibration of the instrument withpotassium in ammonium nitrate solution was also recommended.l1Various modifications were proposed, particularly in the design of theatomiser and burner, to increase the accuracy and adaptability of theapparatus and to allow for the substitution of ordinary illuminating gas.12S.D. Boon l3 discusses in some detail the development of the flame photo-metric technique and gives much valuable information as to the relativemerits of different types of filters and photocells together with a descriptionof the apparatus used in his investigations. A somewhat similar design hasbeen described for the determination of sodium and potassium in biologicalfluids and for the determination of serum potassium using an Ilfordfilter 609 and Cintel GS1S photocell.15 Results obtained with this photo-meter agreed with chemical determinations of potassium within 2 nig.per 100 C.C. Filters based on compounds of the rare earths praseodymium,neodymium, and dysprosium (e.g., Wratten No.77, Chance ON16) havestrong absorption bands in the region of the intense sodium lines (5890,5896 A.) and have been used successfully for the elimination of this element.describe amuch more compact apparatus designed for routine laboratory use, incor-porating a metal atomiser with gravity feed, an ordinary Meker-type Fisherburner, and a barrier-layer cell with galvanometer. Domestic coal gas isused, and the monitoring of the gas and air supplies is achieved by pressuregauges instead of the more cumbersome manometers of earlier designs.Various Corning glass filters are described for the determination of sodium,potassium, calcium, and lithium. A number of experimental models basedR. B. Barnes, D.Richardson, J. W. Berry, and R. L. HoodSvemk Kern. Tidskr., 1938, 50, 136.lo H. Riehm, Bodenk. P&nz., 1945, 36, 109.l1 K. Nehring, Chim. et Ind., 1942, 30, 36; G. Varrallyay, Mezogs. Kutat., 1944,17, 95; Chirn. et Id., 1946,56, 413.l2 H. Riehm, Bodenk. PJEanx., 1940,21/22, 277 ; E. Rauteberg and E. Knippenberg,ibid., 1940, 20, 364; Erniihr. P$anz., 1941, 37, 73; H. Riehm, Bodenk. PjZanz., 1942,28, 246; R. Hermmn and P. Lderle, ibid., 1942, 30, 189; R. Hermann, Forsch.Dienst, 1943, 16, 239.l3 “ Vlam-fotometrie,” D. B. Centen’s, Amsterdam, 1945.l4 W. R. Domingo, W. Klyne, and W. Weedon, Biochem. J., 1948,42, xxxvi.l5 W. Klyne, ibid., 1948, 43, xxvLEYTON FLAME PHOTOMETRY. 329on this design were produced by the American Cyanamid Co., but a similarmodel made by the Perkin Elmer Corporation is commercially available.Using these instruments, the applicability of the flame photometric tech-nique has been thoroughly investigated.16 The design of the atomiserwould appear to be most critical, and in particular cases metal typeswith suction feed were found preferable to all glass types.Further im-provements in design have also been suggested.ls Satisfactory results, itis claimed,l’ could only be obtained when a more uniform gas supply(cylinder gas) was substituted for the mains supply.The technique involves a number of possible sources of error : (a) Vari-ations in the gas and air supplies which affect the temperature of the flameand therefore the intensity of emission; ( b ) non-uniformity of the spraywhich is dependent upon the air pressure and atomiser ; (c) surface-tensionand viscosity differences between standard and test solutions affecting therate of atomisation; (d) mutual interference between elements in theflame; and ( e ) filter limitations.Most of these factors have been discussedin relation to the flame excitation technique as used in spectrographicwork and are admirably reviewed by R. L. Mitchell and others.20 Forpure solutions, results consistent within A3 yo have been obtained withoutundue diffi~ulty,~ but for complex solutions, like biological extracts, extraprecautions are necessary. The presence of acids, bases, or salts affectsthe accuracy of the analyses to a varying degree according to their con-centration in the solutions.Most biological extracts, for example, areobtained by acid digestion, but because of their interference in the extracts,acids cannot be used indiscriminately. Certain acids are particularlytroublesome; the presence of O - O l ~ o of phosphoric acid is sufficient tocause a decrease of 14% in the estimation of sodium and 9% in that ofpotassium compared with the readings obtained for these two metals inpure solutions.21 Even in pure solutions, the nature of the anion affectsthe calibration of the instrument and must therefore be taken into con-sideration. Errors due to the presence of other cations are generally smallunless they occur at a concentration equivalent to or greater than thatof the test element, in which case they may interfere quite seriously andmust be allowed for.Alcohol and acetic acid give rise to positive errorsin the determination of potassium and sodium in all cases where they occurat concentrations greater than l”/.21 It is therefore evident that greatl6 0. J. Attoe and R. Truog, Soil Sci. SOC. Amer. Proc., 1946, 11, 221; R. It. Over-man and A. K. Davis, J . Biol. Chem., 1947, 168, 641; P. 31. Hald, ibid., 1947, 167,499; T. D. Parkes, H. 0. Johnson, and L. Lyklren, AmZ. Chem., 194S, 20, 827; S. J.Toth, A. L. Prince, A. Wallace, and D. S. Mikkelson, SoiZSci., 1948, 66, 459.l7 Hald, loc. cit., ref. 16.A. T. Myers, I d . Eng. Chem. Anal., 1946, 18, 555; V. Toscani, ibid., 1947, 19,“ The Spectrographic Analysis of Soils, Plants and Related Materials,” Tech.820.Comm.Bur. Soil Sci., 1948, No. 44.2o H. C. T. Stace, J . Proc. Austral. Chem. Inst., 1947, 14, 144.21 Parkes, Johnson, and Lykken, loc. cit., ref. 16330 ANALYTIUAL CJHEMISTRY.care has to be taken before the simple flame photometer may be used fara particular investigation.The usual procedure adopted has been to calibrate the photometer withsolutions approximating in composition to that under test, and in thisway fairly reliable results can be obtained. This method, however, is notalways practicable, and for this reason an internal-standard technique hasbeen suggested22 whereby any change affecting the light intensity due toone element affects the internal standard in the same way. Using a speciallymodified photometer to enable them to measure independently the radiationdue to both test element and internal standard, the authors have appreciablyincreased the accuracy of the technique. By adopting the standard methodof analysis with this modified instrument it becomes quite feasible todetermine two elements in the same solution at the same time.The flame photometric technique has been successfully used for thedetermination of boron as methyl borate with a sensitivity down to 5 pg.of the element per ml.l3 Other rapid methods for the determination of thiselement based on similar principles have also been described.2*Hence, although flame photometry is obviously a rather crude techniqueand certainly limited in its applications, it nevertheless possesses certainadvantages over spectrographic methods, particularly when factors like costand simplicity in operation have to be considered.Constructional andmaintenance costs are negligible compared with those of the more elaboratespectrograph, and with a certain amount of care and preparation, consistentresults can be obtained even by relatively unskilled operators. The methodhas already proved to be particularly valuable for the routine determinationof sodium, potassium, and calcium in biological fluids and extracts, butmight well be adapted for the analysis of other elements which are excitedat flame temperatures provided that satisfactory filters can be found.L. L.4. VOLUMETRIC ANALYSIS.much work has been carried out in this important but unrewarding field,and many valuable collaborative studies have been sponsored by theAssociation of Official Agricultural Chemists.Dipotassium paraperiodate,K2H,10,,3H,0,2 salicylic acid,3 and o-chlorobenzoic acid 4 have been pro-posed as acidimetric standards, but although sulphamic acid, NH2*S03H,Solutions and Standards for Volumetric Analysis.-Since the last Reporta2 J. W. Berry, D. G. Chappell, and R. B. Barnes, Id. Eng. O h . AmZ., 1946,2* J. S . McHargue and R. K. Calfee, ibid., 1932, 4, 385; 1937, 9, 288; H. C. Weber18, 19.and R. D. Jacobson, ibid;., 1938,10, 273.Ann. Reports, 1937, 34, 480.L. Mdaprade, Cong. Chim. id., Compt. rend. 18me Cong., Nancy, 1938, 91;Chern. Abs., 1939, 33, 6192.3 E. Latiu, 2. anal. Chem., 1943, 126, 184.I.G. Murgulescu and V. Alexa, &id., 1943,1!%, 260IRVING : VOLUMETRIC ANALYSIS. 331has the advantage of being a strong acid and relatively soluble in water,6hydrochloric acid prepared by the constant boiling-point method @tillappears to be more exact and convenient to prepare.6It is now generally agreed that a temperature of 300" should not beexceeded when heating sodium hydrogen carbonate for the preparation ofanhydrous sodium carbonate,' but W. R. Carmody states that up to0.1 % of water is held tenaciously but can be partly eliminated by powderingand re-igniting. The standardisation of acids against sodium carbonate andborax has been studied c~llaboratively,~ and W. Young lo proposes 8-di-phenylguanidine as a primary standard, and A. J.Berry l1 advocatesthallous carbonate, which has a high equivalent weight and serves as alink with iodornetric standards since it is quantitatively oxidised by iodatein acid solution to the thallic state. Silver hydroxide prepared from puresilver has been used to standardise acids, halides, thiocyanate, and silvernitrate.12The stabilisation of solutions of sodium thiosulphate continues to attractattention,13 though it appears that sterile solutions of pH not greater than6.2 maintain their titre for long periods.l* C. W. Jordan l5 deprecates theuse of borax as a preservative, but to safeguard against adventitious inocul-ation by sulphur bacteria, sodium benzoate,16 chloroform,17 mercuriciodide,l* and amyl l 9 and octyl20 alcohol have been suggested.Chloroformand mercuric iodide effectively stabilised solutions stored a t 40" for twomonths, but alkalis promoted decomposition.18 Although pure crystallinesodium thiosulphate pentahydrate slowly decomposes in the solid state,21the anhydrous salt is thermally stable for 79 days a t 120" and has beenproposed as a primary standard.22 Changes in the composition of standardti S. M. J. Butler, G. F. Smith, and L, F. Audrieth, Ind. Eng. Chem. Anal., 1938,10, 690.W. H. King, J . Assoc. 08. Agric. Chem., 1942, 25, 653.Ind. Eng. Chem. Anal., 1945, 17, 577.L. Vandaveer, J . Assoc. OBc. Agric. Chena., 1939, 22, 563; H. W. Conroy, ibid.,' L. Ramberg, Svemk Kem. Tidskr., 1940, 52, 137.1941, 24, 636.lo Canadian J . Res., 1939, 17, B, 192.l1 Analyst, 1939, 64, 27; cf.E. Jensen and B. Nilssen, Ind. Eng. Chem. Anal.,l2 L. G . Escolar, Anal. Pis, Quim., 1945, 41, 1071, 1086; 1946, 42, 203, 211.l3 Ann. Reports, 1935, 32, 454.l4 J. L. Kassner and E. E. Kassner, fnd. Eng. Chem. Anntl., 1940, 12, 655; G. M.lb Arner. J . Pharm., 1938, 110, 316.l6 J. Ehrlich, Id. Eng. Chem. Anal., 1942, 14, 406.1939, 11, 508.Johnson, J . Assoc. Off. Agric. Chem., 1942, 25, 659; 1945, 28, 594.Krassner and Kmsner, loc. cit., ref. 14; S. 0. Rue, Id. Eng. Chem. Anal., 1942,Johnson, bc. cit., ref. 14; A. Baudouin and (Mlle.) P. Hillion, Bull. Soc. Chim.14, 802. 18 Idem, ibid.biol., 1944, 26, 490.2o Idem, ibid.22 13. M. Tomlinson and F. GI. Ciapetta, iba., 1941, 13, 639.V. K. LaMer and H. M. Todinson, I d .Eng. Chem. A w l . , 1937, 9, 688332 ANALYTICAL CHEMISTRY.iodine solutions have been discussed by C. K. and for standardis-ation of thiosulphate it is invariably obtained from potassium iodide byoxidation with potassium dichromate 24 (a reaction effectively catalysed bycupric ions 25), or by potassium iodate or better cupric perchlorate ; 26addition of potassium thiocyanate improves the e n d - p ~ i n t . ~ ~Pure potassium iodide has been prepared 28 and examined as a primarystandard in permanganatrometry29 and arsenious oxide can be used inpreference to oxalate if a suitable catalyst (e.g., iodine chloride) is pre~ent.~OOther oxidants such as potassium di~hrornate,~~ iodate,32 br~mide-bromate:~and cerate solutions 34 have been carefully studied, and E.C. Deal 35 reportson the standardisation and stability of thiocyanate solutions. Solutions ofsodium hypochlorite retain their titre in the and when stronglybasified37 and are preferred t,o bromate in the determination of antimonyand other ~ u b s t a n c e s . ~ ~ , ~ ~ For some purposes they can be replaced bysolutions of chloraniine-~.~~ More work has been carried out on the reactionsof sodium chlorite,40 and its potentialities as a volumetric reagent.41 Whereacidified bromate-bromide mixtures are inappropriate a solution of brominein potassium bromide can be stored and dispensed from apparatus describedby A. J. Henry.42With regard to reducing agents, stable solutions of mercurous per-chlorate have been used for the determination of ferric iron though thereaction is not strictly stoi~heiometric.~~ Agreement has not yet beenreached on the best means of standardisiiig titanous chlorideu but theuse of buffers to increase the pH and raise its reduction potential is well23 J .Arner. Pharm. ASSOC., 1948, 37, 6.24 Johnson, Zoc. cit., ref. 14.26 J. J. Kolb, Ind. Eng. Chem. Anal., 1944, 16, 38.27 G. C. Oglethorpe and C. G. Smith, AnaZyst, 1943, 68, 325.28 J. J. Lingane and I. 31. Kolthoff, “Inorganic Syntheses,” New York, 1939,29 I. M. Kolthoff, H. A. Laitinen, and J. J. Lingane, J . Amer. Chem. SOC., 1937,ao D. E. Metzler, R. J. Myers, and E. H. Swift, Ind. Eng. Chem. Anal., 1944,16, 625.3 1 J. R. Pound, Chew&. Eng. Min. Rev., 1945, 38, 87.32 S. M. Berman, J .Assoc. Off. Agric. Chem., 1937, 20, 590.33 H. C. Van Dame, ibid., 1947, 30, 502.34 G. F. Smith and C. A. Getz, I d . Eng. Chem. Anal., 1940, 12, 339.35 J . Assoc. Off. Agric. Chem., 1942, 25, 661; 1945, 28, 595; 1947, 30, 496.36 J. Bitskei, Magyar Chem. Pol., 1944, 50, 97.N. I. Goldstone and M. B. Jacobs, Id. Eng. Chem. Anal., 1944,16, 206.38 J. Bitskei and K. Petrich, Magyav Chem. Lapja, 1947, 2, 230.39 D’Costaa C,. Macris, Ann. Chim. analyt., 1946, 28, 165; B. Samek, C?:asopsiS4o M. C. Taylor, J. F. White, G. P. Vincent, and G. L. Cunningham, Id. Eng.41 D. T. Jackson and J. L. Parsons, Ind. Eng. Chem. Anal., 1937,Q, 14; L. F. Yntema42 Analyst, 1945, 70, 259.43 W. Pugh, J., 1945, 588.44 J. E. Breit, J . Assoc. 08. Agric. Chern., 1947, 30, 504.25 B.D. Sully, J . , 1942, 366.p. 163.59, 429; 1039, 61, 1690.Czechoslov. Le‘k., 1941, 21, 77.Chem., 1940, 32, 899.and T. Fleming, ibid., 1939, 11, 375IRVING : VOLUMETRIC ANALYSIS. 333established.45 J. E. Lindsay 46 has examined electrolytic iron as a standard,and ferrous ethylenediamine sulphate, [C2H4(NH,)2]2S04,FeS04,4H20, isfound to be much more stable than Mohr's salt.47 P. R. Duke 4s ensurescomplete reduction of standard ferrous solutions by running them down acolumn of lead amalgam just before use. The simple and direct preparationof chromous chloride or sulphate solutions of determinate concentrationdescribed by J. J. Lingane and R. L. Pecsok 49 should facilitate the extendeduse of this powerful reducing agent whose storage, standardisation, andreactions have recently been reviewed.@, 5ORapid methods have been described for preparing standard solutions ofalmost all the reagents in conimon use in volumetric analysis.51 Whenthese are dispensed from a large storage reservoir and replaced by dry airthe evaporation of water to restore saturation conditions must cause anincrease in the concentration of the residual solution, but H.A. Liebhavsky 52has shown that; the error is quite negligible.In view of the great importance of accurate pH measurements to theanalyst it will not be inappropriate to recall that the pH of O.O5~-borax isnow stated 53 to be 9-18 a t 20". Acid salts of benzoic, phenylacetic, andother organic acids give highly buffered solutions suitable as pH standards,S4and saturated potassium hydrogen tartrate solution is said to be betterthan aqueous potassium hydrogen phthalate.55Apparatus.-Drastic modification in the design of apparatus for macro-volumetric analysis is scarcely to be expected, though minor improvementscontinue to be made. For instance, J. T. Stock and M. A. Fil156 proposetwo methods of modifying burette taps to effect finer control of delivery,and F. C. Guthrie 57 describes a simple reading device. Copiously illustratedand referenced reviews of microvolumetric apparatus have been given byG. H. Wyatt,58 and by R. Belcher and C. L. Wils0n.~9 On this scale taplessburettes are increasingly used.60 An entirely new type of microburettedesigned by J. A. Saunders,61 an electrically operated burette,62 and devices46 0.L. Evenson, ,T. Assoc. Off. Agric. Chein., 1945, 28, 633; P. G. Butts, W. J.46 Chemist Anulyst, 1942, 31, 8.47 K. P. Caraway and R. E. Oesper, J. Chem. Educ., 1947, 24, 235.48 Ind. Eng. Chem. Anal., 1945, 17, 530.49 Anal. Chem., 1948, 20, 425.50 H. W. Stone, ibicE., p. 747; R. Flatt and F. Sommer, Helv. Chim. Acta, 1942, 25,51 E. Shulek and F. Szegho, 2. anal. Chem., 1942, 123, 252.52 Ind. Eng. Chem. Anal., 1944, 16, 349.ti3 A. D. E. Lauchlan, Nature, 1944, 154, 577.64 J. C. Speakman and N. Smith, ibid., 1944, 165, 698.6 6 J. J. Lingane, Anal. Chem., 1947, 19, 810.5G Analyst, 1946, 71, 142. Bi Chem. and Ind., 1947, 240.5 8 Analyst, 1944, 69, 81, 180. 59 Metcsllurgia, 1946, 34, 337; 35 47.6o I.Liitgert and E, Schroer, 2, physikal. Chem., 1941, 49, B , 257.61 Analyst, 1946, 71, 528.62 F. C. Nrtchod, I d Eng. Chem. Anal., 1945, 17, 602.Meikle, J. Shovers, D. L. Kouba, and W. W. Becker, Anal. Chem., 1948, 20, 947.684334 ANALYTICAL CHEMISTRY.for varying the rate of efflux 63 and for obtaining drops as small as 0.1 mm.in diameter 64 may also be noted. By fabricating a glass electrode in theform of a re-entrant bulb of capacity -1.5 ml. W. Ingold 65 is able totitrate 300-900 pg. of acid with an accuracy of &Is%. The principle ofthe hypodermic syringe from which the displacement of liquid is controlledby a micrometer screw underlies many precision micro-pipettes and burettesdescribed recently.66 When such apparatus is motor-driven, the rate andextent of the delivery of titrating fluid can be controlled by potentialchanges of indicator electrodes in solution so that potentiometric titrationscan be carried out and recorded automatically.6~ An alternative system inwhich the titrant is added a t a constant rate has been described by GonzalezBarredo and Taylor.66 Benedetti-Pichler has considered the possible errorsarising from the evaporation of standard solutions from the tips of niicro-burettes68 and describes the technique of titration with pg.samples wherea low-power microscope is needed to observe operations conducted with theaid of mechanical manipulators.69Titrations in Non-aqueous Solvents.-Where the material to be titratedis insoluble in water, solvents such as ethyl, n-butyl, and amyl alcohol oracetone have often been substituted.A. E. Ruehle 70 extends the range todioxan and ethylene glycol monoalkyl ethers (Cellosolves) with anisolerecommended as a solvent for pitches and asphalts in titrations againstpotassium hydroxide and sodium butoxide.use alcohol or naphtha as a solvent for thiols in titrations with copper alkylphthalates.That non-aqueous solvents may provide a solution to the problemsinvolved in determining certain salts, or acids and bases too weak fortitration by conventional methods in aqueous solution, follows from aconsideration of Bronsted’s equationE. Turk and E. E. ReidHA + S HS+ + A-For if the solvent S used is more basic than the conjugate base A- of theacid HA which is to be determined, equilibrium will be displaced appreciablyto the right.Acids whioh are weak in water will thus appear stronger in amore basic solvent. Conversely, weak bases will appear stronger whenwater is replaced by a more acidic solvent, as was first demonstrated experi-63 J. T. Stock and M. A. Fill, Metdlurgia, 1944, 31, 103; F. P. W. Winteringham,Analyst, 1945, 70, 173.e4 W. R. Lane, J. Sci. Imtr., 1947, 24, 98.66 Helv. Chim. Acta, 1946, 29, 1929.86 P. A. Shaffer, P. S. Farrington, and C. Niemann, Anal. Chem., 1947, 19, 492;cf. G. H. Wyatt, Metallurgia, 1945, 32, 240; V. Stott and (Miss) I. H. Hadfield, B.P.584,841 ; J . J. Lingane, Anal. Chem., 1948, 20, 285; H. A. Robinson, Trans. Electro-chern. SOC., 1947,92, Preprint 38, 503 ; J.M. Gonzalez Barredo and J. K. Taylor, ibicE.,Preprint 26, 303.6’ Lingane; Robinson, Zocc. cit.88 A. A. Benedetti-Pichler and S. Siggia, I d . Eng. Chem. Anal., 1942,14, 662.1 3 ~ A. G. Loscalzo and A. A. Benedetti-Pichler, ibid., 1945, 17, 187.70 I d . Eng. Chem. Anal., 1938, 10, 130. 71 Ibid., 1946, 17, 713IRVING : VOLUMETRIC ANALYSIS. 335mentally by N. F. Hall and J. B. Conant.72 For reactions with glacialacetic acid as solvent the titrant is prepared by adding acetic anhydrideto aqueous perchloric acid in proportion to its water content, diluting withacetic acid to the desired strength, and standardising against anhydroussodium carbonate. '3 Blumrich and Bandel 73 found t h a t primary, secondary,and tertiary amines (but not amides of carboxylic acids or acetylatedamines) could be titrated potentiometrically : the titre after acetylationof a mixture thus gave the amount of tertiary amine alone.73 Up to 50%of water in a sample is admissible but a special procedure must be adoptedwhen sterically hindered secondary arnines are present, 74 Since salicyl-aldehyde condenses with ammonia and primary (but not secondary ortertiary) amines to form azomethines of decreased basicity, a method becomesavailable for determing all the components of an ammonia-amine mixture. 75When less than 0.274 of water is present, many organic bases and amino-acids and alkali, alkaline-earth, and ammonium salts of carboxylic acidsbehave as strong bases in glacial acetic acid and can be titrated with 0 4 N -perchloric acid, crystal-violet, thymol-blue, and neutral-red being used asvisual indicators.76 a-Naphtholbenzein is preferred for dimethylaniline 77and quinine, the latter titrating as st di-acid base.78 In chloroform thecinchona bases and nicotine are accurately titratable as di-acid bases withtoluene-p- sulphonic acid and picric acid, respectively , dimeth yl- yell0 w beingthe indicator.However, in aqueous alcohol they both behave as mono-acid bases towards mineral acids (methyl-red) thus permitting an assay ofnicotine in tobacco.79 Salts of weak monobasic organic acids (notably the" soaps ") dissolve quite readily in 1 : %glycols and better still in admixtureswith higher aliphatic alcohols or chlorinated solvents and can be titrateddirectly with solutions of hydrochloric, perchloric, or other strong acids inthe same solvent, the end-point being determined potentiometrically orvisually.80 Phenolphthalein and methyl-red can be used in a double-indicator method to determine free alkali and soap, and salts of inorganicacids such as metaborates, aluminates, etc., mixtures of weak and strongacids, and weak bases can be determined in the same solvent.8lWeak acids can be titrated if the solvent is more basic than water, butto minimise solvolysis it should have a small autoprotolysis constant and thetitrant must naturally be even more basic than the solvent.Using an-hydrous ethylenediamine as solvent and sodium 2-aminoethoxide as titrant,72 J . Amer. Chem. SOC., 1927, 49, 3047, 3062; 1930, 52, 5115.73 K.Blumrich and G. Bandel, Angew. Chem., 1941, 54, 374; H. Haslam and74 C. D. Wagner, R. H. Brown, andE. D. Peters, J . Amer. Chem. SOC., 1947, 69, 2609.75 Idem, ibid., p. 2611.713 J. C. Oppenheim, J . Soc. Chem. Ind. Victoria, 1945, 45, 647.?7 Haslam and Hearn, loc. cit., ref. 4.7 8 R. L. Herd, J . Amr. Pharm. Assoc., 1942, 31, 9.7 O E. M. Trautner and 0. E. Neufeld, Australian Chem. Inst., 1946, 15, 70.8o S. R. Palit, 14. Eng. Chem. Anal., 1946, 18, 246.P. F. Hearn, Analyst, 1944, 69, 144.Idem, Oil and Soap, 1946, 23, 58336 ANALYTICAL CHEMISTRY.M. L. Moss, J. H. Elliot, and R. T. Hall82 find that aromatic carboxylicacids and phenols behave as strong acids and give very satisfactory inflec-tions in potentiometric titration curves.Amino-acids titrate as simplecarboxylic acids arid salicylic acid behavcs as a dibasic acid. Even resorcinolgives two inflections, the second being that of a very weak acid, and allthree stages of dissociation of boric acid are detectable. In acetic anhydrideas solvent, sodium acetate acts as a strong base, changing. indicators such asmethyl-orange to their alkaline colour and reacting instantaneously withtrichloroacetic acid and more slowly with acid ~hlorides.8~In addition to work in aqueous alcoh01,~4 acefone,B5 and glacial acetica~id,~2-789 s6 there are many scattered observations relating to titrations innon-aqueous solvents and there can be little doubt that this subject willsteadily gain importance as its potentialities come to be more generallyrealised.Coulometric Analysis.-The previous sections will have exemplifiedevolutionary trends in classical volumetric analysis, and the search forgreater speed and accuracy with ever smaller samples of material.Theinconvenience of having to prepare standard solutions, the difficulties inherentin their maintenance, and problems of burette design and instrumentationcould be circumvented if the titrant could be generated in situ by an electro-lytic method. This was first realised experimentally by L. Szebellkdy andZ. Somogyi ,S7 who standardised hydrochloric acid by adding potassiumchloride and passing an approximately constant current between a platinumcathode and a silver anode until the change in colour of bromocresol-greenshowed that neutralisation was complete.The quantity of electricityemployed was measured by a silver weight coulometer in series and theextent of chemical action was calculated from this, and the known valueof the Paraday, 100% current efficiency being assumed. This procedure,described appropriately as coulometry, was extended to the standardisationof sulphuric acid, and coulometric determinations of thiocyanate, hydrazine,hydroxylamine, and even caustic alkali could be effected by generatingbromine electrolytically.Though capable of very precise results, applications were limited (sincethe electrode potentials were not controlled) to cases where a single cellreaction could take place and where a specific indicator was available,whilst the use of a weight coulometer was an obvious disadvantage.Nowall oxidation-reduction processes involve electron transfer ; and whether thisis effected electrolytically a t suitable electrodes or by means of an appro-priate ovidising or reducing standard solution is dictated sometimes bychoice, sometimes by necessity. Though Szebelledy and Somogyi appliedAnal. Chem., 1948, 20, 784.83 M. Usanovitsch and K. Jazimirski, J. Qen. Chem. Russk, 1941, ll, 957.84 H. Baggesgaard-Rmmussen, 2. anal. Chem., 1936, 105, 269.85 G. M. Richardson, Proc. Roy. SOC., 1934, B, 115, 121, 142, 170, 180.86 I. M. Kolthoff and A. Willan, J . Amer. Chem. SOC., 1934,56, 1014; G . F. Nadesuand L. E. Branchen, ibid., 1935, 57, 1363.2. anal. Chem., 1038, 112, 313, 323, 332, 385, 391, 395, 400IRVlNa : VOLUMETBIC ANALYSIS.337their coulometric technique only to familiar volumetric determinations, nosuch arbitrary limitation is necessary, for all electrochemical determinationscan legitimately be included within its scope. Provided the electrodereaction is reproducible and exactly defined in a stoicheiometric sense, itneed be neither chemically nor thermodynamically reversible. Withmixtures of reducible ions, control of potential becomes imperative, andJ. J. Lingane 88 points out that a mercury-pool cathode with a silvsr-silverchloride anode possesses many advantages since it is relatively easy toobtain current efficiency in the electrolytic reduction of certainorganic compounds and various metal ions,g0 whilst conventional polaro-graphic methods serve to establish optimum conditions of cathode potentialand electrolyte composition and concentration for any specific determin-ation.Lingane developed a hydrogen-oxygen coulometer as a direct-reading instrument to indicate continuously the progress of an electrolysis,88and the potential control can be effected manually or automati~ally.~~During electrolysis a t constant potential the current decreases expon-entially with time, and each determination would theoretically require aninfinite time for its completion. In practice 99% reduction is achieved(irrespective of the initial concentration) by the time the current has droppedto 1% of its original value and little is gained by pursuing the electrolysisfurther.If a constant electrolysing current is employed some means ofdetecting the end-point must be provided. J. Epstein, H. A. Sober, andS. D. Silvers2 determine acid gases in the air (or materials which can bepyrolysed to acids) by absorbing them in the cathode chamber of a U-shapedelectrolysis vessel and titrating with hydroxyl ions generated by the elec-trolysis of sodium bromide. The end-point is determined potentiometricallyby a Pinkhof system, and since a constant current is employed the timetaken for complete neutralisation is a measure of the acid present.Unstable bromine solutions are an inconvenient feature in the deter-mination of di- (2-chloroethyl) sulphide by oxidation of thiodiglycol preparedtherefrom by hydrolysis) to its sulphoxide, and J.W. Sease, C. Niemann,and E. H. Swift eliminate them by generating the bromine electrolyticallyin an apparatus suitable for the determination of pg. quantities of thiodi-glycol 93 or 30-1000 pg. of arsenic.94 The constant current of less than10 ma. is derived from a dry storage battery, and the coulometer isreplaced by a stop-watch. Polarised electrodes are used to detect the end-point in what is effectively a combination of the dead-stop end-point 95 andan amperometric titration, for since the concentration of bromine in excessdetermines the rate of diffusion of this element to the indicator cathodeand thus the amount of depolarisation, the magnitude of the indicatorcurrent affords a reliable measure of the end-point correction. H.I.88 J . Amer. Chem. SOC., 1945, 67, 1916.so Ibid., 1943, 65, 1348; cf. R. Pasternak, Hel~u. CJ&n. Act& 1948, 31, 753.I d . Eng. Chem. Anal., 1944,16,147. 91 Ibid., 1945, 17, 332.98 Anal. Chem., 1947, 19, 675. 93 Ibid., p. 197.s4 J . arner. Chem. SOC., 1948, 76, 1047. 95 D. P. Evans, Analyst, 1947, 72, 99338 ANALYTICAL UHEMISTRY.5. ANALYSIS OF SEA WATER.There are present in the sea widely different concentrations of a largenumber of inorganic ions as well as colloidal and particulate inorganic andorganic matter. Some fifty elements have already been detected andthe presence of others may be inferred from their occurrence in marineorganisms.2 Of the major elements present some (e.g., sodium and potas-sium) are amongst the most difficult to determine, others (e.g., calcium,strontium, and magnesium) are not easy to separate, and constituents ofgreat biological importance (e.g. , phosphates and nitrates) are present inconcentrations far below those normally dealt with by the microchemist,a few pg./l.being of great importance. The high chloride-ion concentrationand the salt content are constant complicating factors which often rendermodification of conventional methods essential.The major constituents of the oceanic water bear a virtually constantratio to the total salts, being unaffected by land drainage, so that exceptfor special purposes the determination of more than one element is rarelymade. It is usual to determine the silver-precipitated halides (chlorinity).An international standard for chlorinity independent of atomic weights hasbeen maintained by referring all determinations to the so-called Copenhagen‘‘ Normal Water,” which has been established as a permanent standard 3in terms of the mass of silver required to precipitate completely the halogenin one kg. of that water.Using recent values, J. Lyman and R. H. Flem-ing4 give values for the major constituents in terms of chlorinity andsalinity.The concentrations of other elements are affected by biological agencies,and for reason of space attention will be confined to a selection of the mostimportant of these ; the concentrations are expressed in units recommendedby the Association D’Ochanographie Phy~ique,~ and approximate rangesare given for the elements considered.By drawing attention to the marineliterature it is hoped to minimise the regrettable duplication of effort sonoticeable in the analysis of nutrient materials such as phosphate andnitrate which are important in many biological and biochemical studies,and by concentrating on those elements present in minute concentration,the attention of physical chemists may be drawn to a field in which reactionkinetics a t high dilution are of great importance.Phosphorus and Arsenic.-Phosphates (0-4.003 mg.-atom of PO,-P/1.).The earlier work 6 involved either evaporation or precipitation with ferric salts.H. U. Sverdrup, M. W. Johnson, and R. H. Fleming, “ The Oceans,” New York,1942.D. A. Webb, Sci. Proc. Roy. Dublin SOC., 1937, 21, 505.a J.P. Jacobsen and M. Knudsen, Assoc. Oceanog. Physique, Publ. Sci. No. 7, 1940.J . Marine Res., 1940, 3, 134.B. Hellmd-Hwen, J. P. Jacobsen, and T. G. Thompson, Assoc. Oceanog. Physique,D. J. Matthews, J . Marine Biol. ASSOC., 1916, 11, 122; 1917, 11, 251; E. Raben,Publ. Sci. No. 9, 1948.Wksensch. Meermnters., 1916-1920, 18, 1BARNES: ANALYSIS OF SEA WATER. 339W. R. G. Atkins and E. G. Wilson were the first to apply the Denigbsreaction and all subsequent work has been done using this method, a recentsummary of which is given by R. J. Robinson and T. G. Thompson.8Much work has been done on this method of determining phosphates, sincethe element is of great importance to many branches of biochemical work.However, a great deal of the work is repetitive and a fundamental study ofthe reactions and their kinetics is still required.A blue colour can beproduced under the appropriate conditions by the action of many reducingagents upon the heteropoly-acids of molybdic and phosphoric acids andthe intensity of the colour is dependent upon many variables, but for thequantities of phosphate occurring in sea water, only stannous chloridereaches the required sensitivity. K. Kalle's important s t ~ d i e s , ~ much ofwhich were repeated by J. Tischer,lo show that the visual blue is affectedby halides and the absorption in the violet is appreciably higher in seawater owing to the production of yellow tints in the formation of whichmolybdate, chloride, and bivalent tin ions are considered to be involved.L.H. N. Cooper l1 suggested that these yellow colours, particularly notice-able with excess of stannous chloride, are due to the formation of complexmolybdenyl halides and their subsequent hydrolysis. When comparingthe colour developed in sea water with standards made in distilled waterit is necessary to apply a correction for the amount of salt, and since tem-perature affects the colour development, the sensitivity, and the salt error,both the standards and unknown should be a t the same temperature. Aspecial acid molybdate reagent being used, conditions were found under whichthe extinction was proportional to phosphate content and the salt error wasminimal. It was found best to add the stannous chloride in two portionsat an interval of 10 minutes, the intensity of colour being measured5 minutes after the second addition, H.W. Harvey l2 has recently con-firmed and extended these results.Dissolved and particulate phosphorus (0-0.6 mg.-atom of Pll.). I naddition to inorganic phosphate, dissolved organic phosphorus compoundsare present as a result of organic decomposition. Complete oxidation oftraces of organic matter in the presence of large quantities of salts, andthe reduction of the arsenate formed from arsenite during this oxidationwhich is necessary in order to prevent its interference in the subsequentphosphorus determination, give rise to technical difficulties when onlytraces of organic phosphorus compounds are present. K. Kalle andalso others l3 found that the method developed for fresh waters l4 were' Biochem.J., 1926, 20, 1223; J . Marine Biol. ASSOC., 1923, 13, 119; 1925, 13, 700.J . Marine Res., 1948, 7, 33.Ann. Hydrog. Marit. Meteor., 1933, 61, 124; Ber. Deutsch. Wise. Komm. Meere-forschung, 1933, 6, 273; Ann. Hydrog. Marit. Meteor., 1932, 60, 6 ; 1935, 63, 58, 195.lo Z. PJanz. Dung., 1934, 33, 192.l1 J . Marine Biol. ASGOC., 1938, 23, 171.l3 Intern. Rev. ges. Hydrobiol., 1933, 29, 221.L. Titus and V. W. Meloche, ibid., 1931, 26, 441.l2 Ibid., 1948, 27, 337.R. I . Robinson and G. Kemmerer, Tram. Wiscon. Aead. Sci. Arts, 1930, 85, 117340 ANALYTIUAL CHEMISTRY.unsatisfactory. E. Kreps and M. Osadchih l 3 used hydrogen peroxide inthe Oxidation, but Kalle considered that the danger due t o production oforganic acids (e.g., oxalic acid) which would iiiterfere with the phosphorusdetermination renders this reagent unreliable.He therefore fumed thesolid with sulphuric acid with the addition of a little copper salt. F. Berger,15working on marine sediments, showed that Kalle’s method did not give acomplete oxidation, nor were persulphate and perhydrol completely eEec-tive; he recommends fuming nitric acid. Arsenic was not reduced by theearlier workers and Kalle used thiourea for this purpose, but L. H. N. Cooper l6had no success with this method of reduction.A. C. Redfield, H. P. Smith, and B. H. Ketchum l7 used a digestionsimilar to that of Kalle, but effected reduction of arsenate by prolongedheating with sulphite in stoppered bottles.H.W. Harvey l2 has used an alternative method for determining thedissolved organic phosphorus in which considerable modifications areeffected. The organic phosphorus compounds are hydrolysed with acid byautoclaving a t 3 0 4 0 lb./sq. in. for 5-6 hpurs, sulphite being added toprevent oxidation of arsenite.For the phosphorus determination in plankton, L. H. N. Cooper l1found the methods developed by Robinson and Kemmerer l4 and Titusand Meloche l4 to be unsatisfactory owing to the difficulty of removing thelast traces of the oxidising agent. Cooper used perhydrol and sulphuricacid, followed by the molybdenuni- blue estimation, adding stannous chloridebefore the molybdate.W. R. G. Atkins and E. G. Wil-son ls suggested that the discrepancy between their results and those ofD.J. Matthews,6 when compared with the values obtained by E. Rabenon the phosphate content of sea water, were due to the fact that they usedDenigbs’s colorimetric method and Matthews used the method of L. Pougetand D. Chouchak,lg whilst Raben’s method involved evaporation withnitric acid. It is suggested that, as arsenic in sea water exists largely asarsenite, after oxidation (Raben) this would be included in the phosphate.Atkins and Wilson found that Pouget and Chouchak’s method gives animmediate opalescence in the cold (if dilute, on standing) with phosphates,opalescence only on warming with an arsenate, and with arsenite only itfaint opalescence on warming owing probably to oxidation to arsenate.They also showed that the Deniges reaction could be used for the deter-mination of arsenate, but only a faint colour was produced with arsenites,again probably owing to oxidation.N.W. Rakestraw- and F. B. Lutz 2O ‘used the Gutzeit method, and amodified Gutzeit method has been described 2 l which is essentially a modi-Arsenic (0.1-0.5 pg.-atom of As/l).Intern. Rev. ges. Hydrobiol., 1938, 37, 420.l6 J . Marine Biol. ASSOC., 1937, 21, 673.l8 J. Marine Biol. ASSOC., 1927, 14, 609.l@ Bull. SOC. chim., 1909, 5, 104; 1911, 9, 649.21 S. Gorgy, N. W. Rakestrrtw, and D. L. Fox, J . Marine Rea., 1948, 7, 22.1 7 Biol. Bull., 1937, ‘53, 421.2O Biol. Bull., 1933, 65, 397BARNES: ANALYSIS OF SEA WATER. 341fication of that due to M.B. Jacobs and J. Nayler.22 After reduction ofAs” by acid bisulphite, arsiiie was absorbed in sodium hypobromite andthen reduced by hydrazine sulphate in the presence of acid molybdate.Arsenic (0.5 pg.-atom/l.) present in the sea was fractionated into arsenite(50-60 yo), arsenate, dissolved organic arsenic, and particulate arsenic(each 8--16%).Silicates (0.0007-4-14 mg.-atom of Si/l.).-W. R. G. Atkins and E. G .Wilson introduced the method of F. Dienert and F. Wandenb~lcke,~~using picric acid standards for the comparison, and further study of thereaction has been made by T. G. Thompson and H. G. Houlton 24 (q.v. forearlier references). The original picric acid standards were shown to bein error and corrections have been made.25 The advantages of borax-buffered standards of potassium chromate 26 to replace picric acid havebeen stressed by R.J. Robinson and H. J. Spoor,27 who found that thefull colour development took place within 3 minutes and there was nofading within 2 hours. Temperature was found to be without effect.S. W. Brujewicz and L. K. Blinov’s results 28 on the effect of salt concen-tration were not confirmed ; they found a correction factor of 1-16 in con-trast to the Russian workers’ value of 1.66. Dienert and Wandenbulckefound that silica in the colloidal form is not determined by these reagents,but further investigations on this and upon the effect of salinity upon thecolour development cppear desirable.Nitrogen.-Ammonia (0.35-3.5 mg.-atom of NH,-N/l.). H. E. Wirthand R.J. Robinson 29 have compared the earlier methods 3O and found aJl thereagents except Treadwell’s to have a non-sensitive region. The sensitivityof Treadwell’s reagent increases with increasing chlorinity, but Beer’s lawdoes not apply a t low concentrations. A. Krogh31 used a vacuum-dis-tillation method after making the water alkaline, the liberated ammoniabeing absorbed in hydrobromic acid and determined by T. Teorell’s naphthyl-red t i t r a t i ~ n . ~ ~ Air a t reduced pressure is used to drive off ammonia, andattention must be paid to blank determinations, The accuracy of a singledetermination is of the order of 0.04 pg. of nitrogen (20-ml. sample). Themethod can be adapted for the determination of ammonia in air.22 I d . Eng.Chem. Anal., 1942, 14, 442 ; see also Ann. Reports, 1944, 41, 282.23 Compt. rend., 1923, 176, 1478.24 Ind. Eng. Chem. Anal., 1933, 5, 417.25 E. J. King, Conlr. Canal. Biol. and Pi.&., 1931,8,119 ; E. J. King and C. C. Lucas,J . Anzer. Chenz. SOC., 1928, 50, 2395; Robinson and Kemmerer, ref. 14, p. 129.26 H. W. Swank and M. G. Mellon, Ind. Eng. Chem. Anal., 1934, 6, 348.27 Ibid., 1936, 8, 455.28 Bull. State Oceanog. Inst. U.S.S.R., 1933, No. 14, 44.29 Ind. Eng. Chena. Anal., 1933, 5, 293.30 R. Witting, Oefv. Finska Vet.-Soc. Forh., 1915, 57, No. 21; H. Wattenberg,Cons. Perm. Intern. Rapp., 1929, 53, 90; Ann. Hydrog., 1931, 59, 95; T. Bramud andA. Klem, Hvalradets Skr., 1931, No. 1 ; L. H. N. Cooper, J . Marine Biol. ASSOC., 1933,18, 677 ; K.Buch, Merentutkimuslaitoksen Julkaisri. Havs. Skrift., 1920, 2.31 Bwl. Bull., 1934, 81, 126.92 Biochem. Z., 1932, 248, 246342 ANALYTIUAL OHEMISTRY.Nitrate (0.1-43.0 mg.-atom of N03/l.). The phenoldisulphonic acidmethod is not applicable in the presence of chloride. Reduction to ammoniahas been but the method is tedious and the results are open toquestion, since ammonia may be formed during reduction from nitrogenoussubstances. Two methods have been employed, both depending uponoxidation of a reagent in strong sulphuric acid with the production ofcoloured compounds. W. R. G. Atkinsa proposed the use of diphenyl-benzidine, but H. W. Harvey’s reduced strychnine reagent 85 is more com-monly employed; directions are given for preparing the reagent, theproperties of which are somewhat capricious.The presence of much dis-solved or particulate organic matter vitiates the results. Difficulties werereported by T. G. Thompson and M. W. J0hnson,3~ but L. H. N. Cooper:’using safranine as an artificial standard, found the method satisfactory,although erratic results were obtained in the presence of nitrites. B. M. G.Zwicker and R. J. Robinson38 confirmed by analysis that strychnidine andtetrahydrostrychnine were the main products in the reduction employedby Harvey and they suggest the use of strychnidine (which gives a moreintense absorption maximum) prepared by electrolytic reduction of strych-nine in sulphuric acid solution using a mercury cathode.3B Data are givenconcerning the effect of reagent concentration on the colour produced withthis strychnidine reagent.Distrychnidyl, although difficult to prepare anduse, is twice as sensitive as the strychnidine reagent. Purther details ofthe preparation of a satisfactory reagent and of its behaviour are given byW. A. Ridde140 and D. Rochford,4l who recommend addition of hydro-chloric acid to increase the sensitivity.(4.v. for earlier work) has determined the necessaryconditions for the use of E. A. Letts and P. W. Rea’s diphenylbenzidinereagent.42 Nitrites give erratic results. In view of the fact that it hasbeen reported that the Harvey reagent and diphenylbenzidine reagentdiffer little in sensitivity, the neglect of the latter would hardly seemjustified .The dissolvednitrogen in lake waters has been determined and fractionated into a numberof components, samples obtained by evaporation of large quantities ofwater being used.& Amino- and non-amino-nitrogen were determined, butW.R. G. AtkinsDissoEved organic nitrogen (0.1-10.0 mg.-atom of N/l.).33 E. Raben, Wiss. Meeresunters., 1905, 8, 81; 1910, 11, 303; 1914, 16, 207.34 J . Marine Biol. Assoc., 1932, 18, 167.35 Ibid., 1926, 14, 71; 1928, 15, 183.36 Publ. Puget Sound Biol. Station, 1929, 7, 345.37 J . Marine Biol. ASSOC., 1932, 18, 161. 38 J . Marine Res., 1944, 5, 214.B. M. G. Zwicker and R. J. Robinson, J . Amer. Chem. SOC., 1942,64, 790.J . Biol. Bd. C a d . , 1936, 2, 1.*l Commonwealth of Australia, C.B.I.R., Bull. No. 220, 1947.4a J., 1914, 105, 1157.43 B.P. Domogalla, C. Juday, a d W. H. Peterson, J . Biol. Chem., 1925, 63, 269;W. H. Peterson, E. B. Fred, and B. P. Domogalla, ibid., p. 287 ; E. A. Birge and C. Juday,Bull. Bur. Pkheries, 1926, 42, 185; E. A. Birge and C. Juday, Proc. Nat. Acad. Sci.,1926, 12, 515BARNES: ANALYSIS OF SEA WATER. 348it is not certain that no decomposition had taken place during evaporation.The albuminoid and total nitrogen of water of the Puget Sound have beeninvestigated by R. J. Robinson and H. E. Wirth,44 using standard methodsof analysis and large volumes of water. in develop-ing a method for smaller quantities of water, abandoned Kjeldahl methodssince ammonia was always obtained in amounts from 0.5 to 2 vg. of nitrogenper ml. on heating sulphuric acid, and all efforts to remove this " organic N "failed.Their method (sensitivity approximately 0.3 pg. of N) involves diges-tion of the sample at 500" with sodium hydroxide in a silver tube in anatmosphere of hydrogen. The ammonia formed is then taken up in ~ / 1 0 0 -hydrobromic acid, combined with sodium hypobromite and the excesshypobromite titrated according to Teorell's method (see p. 341). Detailsare given for purification of the hydrogen, setting up the apparatus, andpreparation of distilled water free from organic nitrogen. T. von Brandand N. W. Rake~traw,~~ using the method, showed that the error is usuallybelow 10% in samples containing approximately 200 pg. /litre of dissolvedorganic nitrogen.Cwbon,--DissoZced and particulate carbon ( 0 .1 4 - 4 mg.-atom of C/l.).A number of methods involving alkaline permanganate have been used todetermine the dissolved organic matter, but none can be considered verysati~factory.~~ A. Krogh and A. Keys45 developed a wet combustiontechnique similar to that of H. Lieb and H. G. Krainich,4* the methodinvolving the removal of salts. After expulsion of carbon dioxide by boil-ing, most of the chloride is precipitated by thallium sulphate, and afterevaporation, the dry residue is oxidised with a mixture of chromic andsulphuric acids in a current of carefully washed air. The carbon dioxideand carbon monoxide are carried through an oxidising combustion tubeinto baryta, the excess of which is titrated with hydrochloric acid. 25 MI.of water are used and careful attention to blanks is emphasised. Theaccuracy approaches 0.1 mg. of carbon/l. but the blank is rather high.By using filters, colloidal, soluble, and particulate carbon were differentiated.Trace Metals.-Zinc can be determined with d i t h i ~ o n e , ~ ~ but beforedetermination af manganese (as perrnanganate) 50 or iron (as the thiocyanatecomplex) 51 halides and organic matter must be removed completely.N. W. Rakestraw, H. E. Mahncke, and E. F. Beach 5z first concentrate theA. Krogh and A.44 J . du Cons., 1934, 9, 15, 187.46 J . Marine Res., 1941, 4, 76.47 W. R. G. Atkins, J . Marine Biol. ASSOC., 1922, 12, 772; 1923, 13, 160; G. J.Pereira, Bol. de Pescas, 1924, 9, 149; W. E. Adeney and B. B. Dawson, Sci. Proc.Roy. Dublin Soc., 1926, 18, No. 17; 0. G. Ibanez, hTotas y Resurnenes, 1928, 11, No. 26;P. Chauchard, Compt. rend., 1932, 194, 1256; Ann. Inst. Oceanog., 1935, 15, 329.48 Mikrochem., 1931, 3, 367.49 K. Buch, Finska Kern. Medd., 1944, Nos. 1-2, 25.61 T. G. Thompson and R. W. Bremner, J . du Cons., 1935,10,39; T. G. Thompson,62 Ibid., 1936, 8, 136.45 Biol. Bull., 1934, 67, 133.T. G. Thompson and T. L. Wilson, J . Amer. Chem. Soc., 1935, 57, 233.R. W. Bremner, and I. M. Jamieson, Ind. Eng. Chem. Anal., 1932, 4, 288344 ANGPTIOAL CHEMISTRY.iron in sea water by precipitation and reduction with ammonium sulphide,the co-precipitated basic magnesium salts acting as excellent carriers.Fluoride did not interfere with the recovery of 0.01-0-04 mg.11. Theauthors noted that ethylene glycol monobutyl ether had a stabilisinginfluence on the red thiocyanate complex which they extracted with amylalcohol: since it was stable to light but not to heat it was necessary tocontrol the temperature before extraction.2 : 2’-Dipyridyl and 2 : 2’ : 2”-tripyridyl53 were introduced into seawater analysis by L. H. N. Cooper,M and a careful study of the reactionbetween dipyridyl and iron a t high dilutions of the latter by K. Buch 55emphasises some of the dficulties likely to be encountered. Thus accord-ing to Cooper and J. H. Boxendale and P. George 56 only undissociateddipyridyl ( K , = 1-2 x molecules enter the complex, sothat both pH, neutral salts, and concentration of reagents affect the reaction.Theoretical calculations using data obtained by L. H. N. Cooper 57 andby P. A. Kriukov and G. P. Awesjewitsch 58 are given by Buch.Copper has been determined by sodium diethyldithiocarbamate. 59 H.Barnes 6O has given methods for the determination of copper and mercuryin sea water solutions in connection with anti-fouling investigations. Otherworkers on this problem have discussed the copper content of marineorganisms and sea water and the relationship to Since the workof H. 3’. Prytherch62 attention has been paid to the copper content ofoysters and sea water under natural and artificial conditions. G. A. Riley 63has used the carbamate reagent with estuarine waters, and K. Buch49gives a dithizone method in a paper which includes useful data on theextraction of copper and zinc dithizonates by different solvents at varyingpH values of the solution.K, -H. B.H. BARNES.H. R. CLAYTON.H. IRVING.L. LEYTON.53 Ann. Reports, 1945, 42, 258.66 Fin-ska Kem. Me&., 1942, 51, 22.s7 Proc. Roy. SOC., 1937, B, 124, 299.6s W. R. G. Atkins, J . Marine BioZ. ASSOC., 1932, 18, 193; 1933, 19, 63.6o Ibid., 1946, 26, 303.61 G. L. Clarke, Bwl. Bull., 1947, 92, 73; C. M. Weiss, ibid., 1947, 93, 56.62 Ecol. Monographs, 1934, 4, 47.03 J . Marine Res., 1937, 1, 60.54 Proc. Roy. SOC., 1935, B, 118, 419.56 Nature, 1948, 162, 777.2. Electrochem., 1933, 39, 884