ANALYTICAL CHEMISTRYIN the period under review there is little new material of outstandingimportance. The most interesting work in analytical chemistryhas dealt with the detailed investigation of known processes andwith the perfection of details and apparatus in such modern processesas electrometric determinations. The following discussion will bedevoted mainly to considering some of the more important subjectsinvestigated .Quantitative Separation of Hydroxides.-The separation of metalsby precipitation as hydroxide is usually incomplete owing t o co-precipitation or adsorption of other metal radicals, to the colloidalcharacter of the precipitates in their initial stages, to the fact thatthe hydroxides can only be precipitated when a particular pH isattained, depending upon the metal, and also in part owing to theprecipitation of more or less basic salt which is hydrolysed veryslowly.The range of pH for precipitation of the hydroxides isconsiderable, in some cases being below 4. A full discussion ofthe mechanism of hydroxide precipitation has been given byH. T. S. Britt0n.lIt is a matter of experience that the separation is more completethe denser the precipitate of hydroxide. Of the substances triedby us, hydroxylamine is found t o be helpful in decreasing thegelatinous character of some hydroxides, e.g., chromium hydroxideprecipitated by ammonia. A promising method is described byF. L. Hahn2 in which a mixture of sodium azide and sodiumnitrite in dilute solution is added t o a faintly acid solution of iron,aluminium, or chromium.This reagent is effective in separatingthis group of metals from large admixtures with the nickel group.The precipitate forms quickly and is easily separated by filtration.A double precipitation will, for example, give a complete separationof iron from manganese. This method of precipitation does notintroduce into the substances under test any foreign metallicradical, in contradistinction to the older barium carbonate process.Frequently, however, the addition of another metal is of no im-portance, and precipitation of the iron group of metals, by meansof zinc oxide, for example, has the advantage of rapidity. Theuse of zinc oxide is not new, but the conditions for completenessof separation, particularly as applied to the analysis of alloy steels,“ Hydrogen Ions,” Chap.15. Ber., 1932, 65, [B], 64; A., 244ELLIS AND FOX. 221have now been carefully in~estigated.~ Britton has shown that zinchydroxide begins to precipitate at pH 5.2, iron and aluminium atlower pH, cobalt, manganese, and bivalent metals generally athigher pH. On this basis a separation might be expected, and it isfound that the zinc oxide method can be successfully applied tothe separation of iron, tungsten, vanadium, chromium, uranium,aluminium, and titanium from cobalt, manganese, and nickel inthe analysis of steel. Although the separation of cobalt andmanganese is easily effected and is complete, nickel cannot be soreadily separated despite the fact that the pE for initial precipitationof these metals is nearly the same.On the other hand, tervalentchromium is completely precipitated by zinc oxide, although theinitial pH for precipitation is very close t o that of zinc. While theforegoing method for determination of cobalt in materials containingiron is accurate, it is nowadays common to require determinationsof cobalt in much larger proportions than formerly in high-speedor magnet steel containing vanadium, tungsten, and chromium,amongst other metals of the alloy. The elimination of iron insuch cases by the older methods is lengthy and tedious, but may beeffected more readily by a modification of the process of extractionof ferric chloride from its solution in hydrochloric acid by meansof ether, followed by the removal of chromium, tungsten, andvanadium by t’reatment with sodium peroxide and sodium hydroxide.Traces of copper and iron may be separated by “cupferron,”leaving a cobalt solution ready for precipitation with a-nitroso-P-naphth~l.~ The process here out’lined briefly is more rapid andcertain than methods hithert,o described in which the cupferronwas not utilised to eliminate the last of the iron, and it is renderedpossible because this reagent does not interfere with the subsequentprecipitation of cobalt by the naphthol derivative.Determination of Boron.-The increasing importance of a know-ledge of the minute quantities of boron in water and natural productsis doubtless the reason for a number of papers dealing with themethod of detecting and estimating this element.There is nogreat difficulty in determining boron even in very small quantitieswhen a solution is obtained in a condition suitable for the estimation,but it is just this preliminary preparation which is not easy to carryout, and the purpose of most of the investigations is to obtain asuitable final solution and avoid loss of boron during its preparationand subsequent treatment. Some natural waters contain 0-4 mg.of boron per litre and upwards, and it has been found possible todetermine these small quantities, either in the water directly, orJ. I. Hoffman, Bur. Stand. J . Res., 1931, 7, 883; A., 137.a I-, ibid., 1932, 8, 658; B., 728222 ANALYTICAL CHEMISTRY.after concentration to a small volume with a little added alkali.At pH 7.6 boric acid is only neutralised t o the extent of about 12%but if enough mannitol is present it may be completely neutraliseda t this pH.Since other weak acids or bases are not in generalaffected by mannitol, it becomes possible t o determine boric acidby bringing the solution to pH 7.6 with sodium hydroxide eitherpotentiometrically or by means of phenol-red as indicator. As amatter of fact, if larger amount,s of mannitol are used, the endpoint may be much lower than pH 7.6, and this procedure lowersthe amount of boric acid neutralised before mannitol is added.5The determinations of boron in plant and animal tissues are nota t all so readily made as in the foregoing process, for the destruction oforganic matter, necessarily carried out as the first stage, leads toserious loss if at any point acidity is developed during heating.If the initial preparation is carried out by maintaining analkaline condition during ashing of the material, it is possible, byimpregnating turmeric-stained flax threads, to determine withsome accuracy as little as 0.01 y of boron, but the test may be tosome extent vitiated by the presence of large quantities of saltsin the ash under examination.For this reason it is usually prefer-able t o distil the acidified ash with methyl alcohol, thus volatilisingthe boron. The subsequent treatment of the methyl borate dependson the quantity present. For large quantities, such as may befound in artificial borosilicates, we have trapped the borate insodium tungstate for gravimetric determination, or in aqueoussodium hydroxide for subsequent volumetric determination in thepresence of mannitol.The choice of acid for catalysing the reactionbetween boric acid and the alcohol needs consideration in each caseexamined. Usually sulphuric acid is used and the distillation isarranged so that the acid-alcohol mixture loses most of its water,but not all, which is usually secured by passing in the vapour of freshmethyl alcohol a t nearly the same rate as that lost by distillation.For silicates, much larger amounts of alcohol and acid are required,and the water present must be regulated in order t o prevent separ-ation of silica, which adsorbs boric acid should the silica separatein a gelatinous state.6 Phosphoric acid can likewise be employedin place of sulphuric acid, but very careful attention is then neces-sary to keep too much water from distilling away, otherwise wehave found phosphoric esters to distil over with the borate, addingconsiderably to the difficulty of the final determination of the boroneven in large amounts.When plant ashes have been suitably treated, the neat andF.J. Foote, Ind. Eng. Chern. (Anal.), 1932, 4, 39; A., 242.E. Schulek and G. Vastagh, 2. anal. Chem., 1932, 87, 165; A,, 354ELLIS AND FOX. 223promising procedure recently developed for determining the boronspectroscopically is to be borne in mind.' The ashes are acidifiedwith citric acid and the boron is distilled out with methyl alcoholcontaining 5% of phosphoric acid.The methyl alcohol-methylborate mixture is burnt in oxygen in a special form of burner underspecified conditions; the flame is observed with a direct-visionspectroscope through a glass cell containing 50 ml. of water intowhich is dropped O.01N-potassium permanganate until the mostintense green boron band in the spectrum of the flame isabsorbed. A calibration curve is drawn showing ml. of 0.01N-potassium permanganate against mg. of boron. The recorded dataindicate that the method is capable of furnishing satisfactorydeterminations.Sometimes it is necessary to determine both fluorine and boron inorganic substances, and the difficulty then arises of eliminating thecarbon and preparing a residue suitable for these determinations.This problem has been examined afresh * with some measure ofsuccess for the comparatively simple substances chosen for testingthe method.Briefly the essential part of the process is the com-bustion of the substance in a Parr bomb with a mixture of sodiumperoxide, potassium chlorate, and sucrose, first enclosing thematerial in a gelatin capsule in order t o keep it away from theoxidising mixture which would otherwise attack the organic borontrifluorides with violence. The alkali is removed by boiling theliquid washed from the bomb with ammonium chloride, and thefluorine is then precipitated by means of calcium nitrate. In thespecial method of treatment of the calcium fluoride by limitedwashing, it is indicated that a solubility correction for this salt isavoided.The boron in the filtrate from the precipitated fluorideis determined in the usual way with mannitol. The results givenfor three compounds are very close to the calculated values.Conductmetric Titrations.--In view of the fact that most chemistshave made measurements of the conductivity of solutions, it issomewhat surprising that the method is not used in practice asfrequently as it deserves. It is at least as accurate as the ordinarytitration relying upon indicators, and much more so if the apparatusis chosen properly. Further, it can be employed where potentio-metric methods fail, for example, in the titration of weak acids orbases when proper account is taken of the varying slopes of theconductivity curves before and after the equivalence or neutralis-ation point is reached.The titration of strong acids and bases7 J. S. McHargue and R. K. Calfee, Id. Eng. Chm. (Anal.), 1932,4, 385 ;A., 1221.* D. J. Pflaum and H. H. Wenzke, ibid., p. 392; A., 1269224 ANALYTICAL CHEMISTRY.when coloured impurities or the natural colour of the liquid precludesthe use of indicators, is effectively carried out conductometrically,for it is not necessary to know the absolute conductivity of theliquid, the reciprocal of the resistance being sufficient for the purpose,if plotted against volume of the titration liquid. A simple form ofapparatus has been described for use with such liquid^.^ Thedifficult titrationof a weak acid by a weak basewhich must some-times be undertaken can be made conductometrically in manycases, for the electrolyte produced as soon as titration is begun is,as a rule, sufficiently strongly ionised to give a large increasein conductivity up to the point of equivalence, after whichfurther addition of titrating liquid produces very little change inconductivity.I n this case the ‘( neutral ” point of intersection isfound by producing the curve (usually a straight line as titrationprogresses) for the initial increase of conductivity and the finalstationary curve. Cases often arise where a titration, whether aneutralisation or a precipitation process, must be conducted in thepresence of alcohol ; the neutral salts produced during a titration inaqueous alcoholic solutions do not alter the “ neutral ” point, butthe curves in the neighbourhood of equivalence gradually merge intoeach other on both sides of the equivalence point.These effectshave been further studied and corrections applied €or the titrations.lOA closely allied matter is the shifting of the end-point of thetitration of a weak acid with sodium hydroxide in alcoholic solu-tions, as the proportion of alcohol increases. This is, of course,of the greatest importance in determining free fatty acids inglycerides, where alcohol is commonly used. For accurate work,corrections must be applied t o the values obtained on titration;naphtholphthalein is a better indicator for the purpose than eitherphenol- or thymol-phthalein.Of the practical applications of conductometric titrations, animportant development is the titration of small quantities of fattyacids by means of a solution of sodium hydroxide in methyl alcohol.An alcoholic solution of the fatty acid is obtained in a specialextraction apparatus, and to the solution is added a small quantityof hydrochloric acid, whereby a conducting solution results.Ontitration with the sodium hydroxide, the conductivity curve, asmeasured by galvanometer deffexions with a “ thermo-cross ”apparatus, is found to consist of three straight lines. The firstis the diminution of conductivity due to neutralisation of the smalladded quantity of hydrochloric acid, the second is the slowly9 T. Callan and S. Horrobin, J . 800. Chern. Ind., 1928,47, 3 2 9 ~ ; B., 1929,10 W.Poethke, 2. anal. Chern., 1931, 86, 399; A., 135.154ELLIS AND FOX. 225increasing conductivity due to the neutralisation of the fatty acid,and the third arises from the greatly increased rate of rise of con-ductivity due to the excess of sodium hydroxide. The projectionof the second line on the abscissa (representing volume of sodiumhydroxide added) gives the quantity of reagent required forneutralisation of the fatty acid.11 Owing to the small dissociationof the fatty acid and the use of an aqueous-alcoholic medium, itseems very likely that errors arising from shifts of the equivalencepoints are largely eliminated in the conditions of test.An unusual application of the conductometric method is thetitration, in an atmosphere of nitrogen, of tetraisoamylammoniumiodide dissolved in diethylzinc with ethylsodium in the same solvent .12It was observed, by following the course of titration, that the con-ductivity reached a minimum when equimolecular quantities hadreacted, the suggestion being that ethyltetraisoamylammonium,N(C2H5)(C5Hll)4, had been formed.Colorimetric Measurements and " Boundary Layer " PhotoelectricCells.-In using colorimeters which depend upon comparison ofdifferent thicknesses of coloured liquids, as, for example, themodern forms of the Duboscq colorimeter, the errors arising fromdifferent intensities of the coloured solutions, peculiarities of theright and left eyes, and fatigue are often ignored.Even when thetwo solutions to be compared are interchanged, divergencies ex-ceeding 1 mm.in height may be obtained. I n any general con-siderations of the best method of using colorimeters, certain factorsmust be recognised, especially, on the one hand, that the observer'seyes must be " trained," and on the other, that they become fatiguedvery rapidly. Even in an observation lasting less than ten secondsthe eye passes through the three stages, of increasing, maximum, anddiminishing delicacy of power of matching two coloured solutions.Before repeating the observation, at least five seconds are necessaryfor the eye to recover its initial power. With the object of diminish-ing errors due t o this rapid fatigue of the eye, a detailed investig-ation of the best conditions has been made and rules for the greatestaccuracy are laid down.13 The ordinary method of keeping the eyea t the eyepiece while adjusting the height of the standard colouredliquid is given up, since it involves a time of observation which bringsthe eye into the period of fatigue.The procedure recommended isto make a match of colour by varying the height of liquid in theordinary way, as an approximate measure of the height. A slightvariation in height of the standard liquid is then made and anl1 G. Jander and K. F. Weitendorf, Angew. Chem., 1932, 45, 705.la F. Hein and H. Pauling, 2. E'lektrochem., 1932, 38, 25; A,, 243.l3 N. E. Pestov, 2. anal. Chern., 1932, 89, 9;- A., 920.REP.-VOL. XXIX. 226 ANALYTICAL CHEMISTRY.observation is carried out rapidly (about 3 seconds) before the eyehas time to tire.These rapid observations are repeated withslightly different heights above and below the match points andtabulated. I n this way it is possible to ascertain the position ofmatching to 0.2 mm. Certain general considerations are worthy ofnote for accurate work. For each coloured substance there arecertain concentrations and depths of liquid which give the bestresults. For example, in DenigBs's method l4 for determiningphosphoric oxide by formation. of molybdenum-blue, the closestresults are obtained with a concentration of 1-5-26 mg. of P,O,per litre. It is also to be observed that better results are found byusing one eye, in preference to both eyes alternately.Improvements in photoelectric cells, and especially the develop-ment of cells of maximum sensitivity for different parts of thespectrum, point the way to developments in colorimetric determin-ations which should replace the " normal " eye.In this field therecent developments of the boundary-type cells such as the cuprousoxide or the selenium-chromium type offer advantages in facility ofuse, and rapidity and robustness of apparatus. For this kind of cellno external batteries are required. The lay-out is simple : aconstant source of light, a good condensing lens, and a galvanometeror micro-ammeter according to the method of use decided upon.15It is desirable to select a coloured screen, e.g., special green glass,so that a fairly narrow band is cut out of the white light for pro-jection through the test liquid on to the photo-cell.Mr. L. G.Groves and one of the authors l6 have for some time used a cuprousoxide cell covered with a thin film of gold for which the maximumsensitivity to light of colour temperature about 3800" Abs. is inthe green region of the spectrum. Such a cell shows a comparativelylarge change in current for a small change in colour in the regiongreen-yellow, towards either red or blue, and is excellent fortitrations with bromocresol-green or methyl-red indicators. Sincethe response of the photo-cell is very rapid-much less than lo3second-the use of the cell for measuring the depth of colour producedby vitamin A with antimony trichloride is indicated. Usuallythis is determined by means of Lovibond tintometer glass unitsand an endeavour is made t o read maximum intensity. Withthe cuprous oxide cell we find that readings can be takenimmediately after mixing and the rate of variation of the trans-1 4 Compt.rend., 1927, 185, 777; A., 1927, 1156; 1928, 186, 1052; B.,1928, 420.R. H. Muller, Mikrochem., 1932, 11, 353; A . , 1016; H. M. Partridge,I n d , Eng. Chem. (Anal.), 1932, 4, 315; A., 934.16 Unpublished workELLIS AND FOX. 227mitted light can be followed readily wit'h a fairly delicate micro-ammeter. The curve of growth and decay of colour can then beobserved continuously .A further extension for continuous recording purposes is obvious,since the current produced by illurninatlion is ample for applicationto a recording galvanometer of the type made by the CambridgeInstrument Company.Fluorescence Methods of Analysis.-The appearance of a fullsummary of the methods of application of fluorescencef7 to thedetection or characterisation of various substances is opportune inview of the far-reaching claims sometimes made for this method oftest.Perusal of this paper shows that the correct interpretation ofthe fluorescence observed is by no means simple, unless the materialunder test is known and the operator is experienced. The apparatusrequired is simple : a source of ultra-violet light, e.g., a mercury-vapour lamp or metallic arc, an " ultra-violet glass " filter such asWood's ultra-violet glass, which transmits a large proportion ofradiation from about 2800 to 4000 B., and a box to hold the sourceof radiation behind the glass screen.The quantitative utilisationof fluorescence has not yet advanced very far, one difficulty arisingfrom the circumstance that measurement of the int,ensity of fluor-escence is a matter of individual experience. Despite this, someprogress has been made and a few of the results are interesting.The use of the fluorescent " stick '' of J. Eisenbrand l8 is capableof extension. Briefly, this consists in dipping part of a closed tubeof some fluorescent material, say quinine sulphate solution, into theliquid under examination. If the liquid absorbs ultra-violet light,then the part of t'he stick inside the liquid becomes darker than thatoutside and may even become invisible, so that a promising methodof evaluating the concentration of fluorescent liquids is indicated.Many attempts have been made to measure the intensity of thefluorescent radiation, some met hods depending upon examinationof the scattered light as in nephelometric observations, some uponphotometric wedges, and one utilising the standard Guild Colorimeterwhereby the content of red, blue, and green light of the fluorescentsubstance may be reconstituted on a purely physical basis.19Discussion has arisen on the subject of the most suitable concen-tration for quantitative work in fluorescent liquids, and it has beensuggested that a maximum fluorescence occurs at a particular con-centration.The fact seems to be that there is an optimum con-l7 M. Haitinger, Mikrochem., 1932, 11, 429; cf.P. W. Danclcwortt, " Die18 2. angew. Chem., 1929, 42, 445; A., 1929, 666.ID G. E. Troase, Pharm. J . , 1930, 124, 264.Lumineszenz-Analyse im filtrierten ultravioletten Licht," Leipzig, 1929228 ANALYTICAL CHEMISTRY.dition depending upon the mutual effect of absorption of theincident ultra-violet light by the liquid and the resulting fluorescence,and that if a thin enough layer is observed the interference is sodiminished that the fluorescence can be examined t o best advantage.This further indicates that quantitative results will be obtainedbest with the more dilute solutions. Apart from this physical con-dition, the pH of the fluorescent liquid has a large effect on theintensity of fluorescence. Within as little as 0.2 unit of pH as muchas 70% variation in fluorescence has been observed with naphthol-sulphonic acids.20 Prom this fact a necessary precaution is toemploy water buffered to the same pH as the fluorescent solution fordiluting test liquids.I n all attempts t o use the fluorescence arising from the incidentultra-violet light as a quantitative method, it has to be rememberedthat the fluorescence may change owing to photochemical effectspurely, or to the combined effect of light and oxygen.The fadingof the red fluorescence of porphyrin through oxidation is an exampleof this ; 21 the general oxidation occurring during fluorescence withdyestuffs such as eosin and rhodamine, and natural subst,ances likevegetable oils, iesculin, and phloxin, is a subject which has attractedattention from its bearing on the physical condition of thesematerials.22The change of fluorescence with pH is now an established meansof determining by titration very minute quantities of acids or baseswith the aid of ultra-violet light, in many cases with coloured orturbid liquids.While Boyle knew nothing of pH values-and mightconceivably have declined to apply the terminology if he had known-he used an infusion of lignum nephriticum to test for alkalinityand acidity. However, it is satisfactory to note that the method ofusing change, or suppression, of fluorescence for indicator purposesis extending. The important observation of J. Eisenbrand 23 thatquinine has two ranges of inflexion, one at pH 6 and the other atpH 9-5-10, has been utilised by him quantitatively.Strong acidsare indicated very sharply, even in very dilute solutions a t pH 6,whereas weak acids which at first show blue-violet fluorescence, giveno colour a t pH 9.5. A full range of this class of indicator dependingupon fluorescence in ultra-violet light is now available. For ex-ample, eosin can be used t o pH 3, by its green fluorescence ; salicylicacid changes from colourless to dark-blue a t pH 3 ; umbelliferone is2o L. J. Desha, R. E. Sherill, and L. M. Harrison, J. Amer. Chem. SOC.,1926, 48, 1493 ; A., 1926, 996.21 S. Rafalowski, 2. Physik, 1931, 71, 798; A., 1931, 1212.*2 R. W. Wood, Phil. Mug., 1922, [vi], 43, 757; J. C. McLennan, Proc.Roy. SOC., 1923, [ A ] , 102, 256.23 Ph~rm.Z., 1929, 74, 249; A., 1969, 525ELLIS AND FOX. 229especially delicate, changing from colourless to blue in the rangeA few applications of fluorescence or ultra-violet colorimetry t othe detection of elements will serve to illustrate the extraordinaryutility of this field of investigation. Traces of uranium can befound by the yellow fluorescence produced on fusion of the substanceunder examination with sodium fluoride, thus giving a good methodfor use with animal tissues and plant structures. The detection ofarsenic with mercury bromide papers in quantities less than 0.001 yby ultra-violet light is now so well established that it is nearlyin ipossible t o obtain reagents which will not give positive indicationsin these conditions.More recently determinations of cadmium inconcentrations of the order of 1 in 2,500,000 have been carried out,and the details for accuracy defined.25 I n very dilute solution it isdifficult to see cadmium sulphide, and in any case the colour of thissulphide is not suitable for accurate colorimetric use in the ordinaryway; but in ultra-violet light the colour comes out with greatdistinctness. To apply the method t o organic substances, thematerial is oxidised by means of a mixture of sulphuric acid andnitric acid, as in the Kjeldahl process, and the sulphide is precipitatedfrom the resulting solution after suitable treatment in a prescribedmanner.Inorganic Analysis.Organic reagents have been applied to inorganic analyticaloperations for a very long period and several now rank among theclassical methods ; of these may be cited oxalates, tartrates, anddimethylglyoxime.One of the most interesting developments inthe analytical field during the past 10 years or so has been theextension of the application of such substances, and while muchof the work has been carried out on new additions to this class ofreagent, the older ones have not been neglected. Broadly, thesecompounds may be divided into three sections : indicators, reagentsfor colorimetric tests, and those forming insoluble complex com-pounds with metals. Of these, the last are often particularlyadapted to microanalysis on account of their high molecular weightand corresponding low content of the metal.The following account, by no means exhaustive, is intended tosummarise briefly the possibilities of some of these compounds.Xalicylaldoxime, which is readily prepared from the aldehyde,was proposed by P.Ephraim 26 as a precipitant for copper from a24 Y . Volmar and E. Widder, Bull. SOC. chim., 1929, 45, 130.25 L. F. Fairhall and L. Prodan, J . Amer. Chem. Soc., 1931, 53, 1321 ; A . ,36: Re?., 1930, 63, [B], 1928; A., 1930, 1393.PH 6*5-7*6.241931, 701230 ANALYTICAL CHEMISTRY.slightly acetic solution, and separations from various metals wereeffected though 0. L. BradyZ7 found that under the prescribedconditions of acidity, some nickel was also precipitated. I n thepresence of ferric iron, precipitation is best effected in very dilutehydrochloric solution,28 or in the presence of tartrate.29 D.G. Ivesand H. L. Riley 30 used this method to determine the copper in itssalts with certain alkylmalonic acids. The oximes of various otherphenolic aldehydes can also serve to precipitate but possessno advantage over salicylaldoxime. F. Feigl and A. Bondi 32 haveinvestigated the atomic grouping which seems to be more or lessspecific for copper.Benxoinoxirne is another precipitant for copper, t'hough E.Azzalin33 found that it was less satisfactory in the presence ofother metals than was first indicated.34 For copper, it is certainlyinferior to salicylaldoxime, but H. B. Knowles 35 has shown thatmolybdates are completely precipitated by it from quite stronglyacid solutions ; provided vanadates and chromates are first reduced,tungs-hates are practically the only interfering salts and this canreadily be allowed €or by precipitation with ciiichonine from asolution of the mixed ignited and weighed oxides.Dimethylglyoxime gives a pink coloration with ferrous salts onaddition of ammonia 36 which has been applied to the detection ofvanadium.37 The colour is transient owing to the readiness withwhich alkaline ferrous solutions tend to oxidise, but can be pre-served by covering the solution with ligroin.I n connexion withthe better-known use of this reagent for nickel, it may be recordedthat the precipitate can be prepared for weighing by washing withalcohol and ether,3* which, though it has been disputed, wouldappear to be satisfactory provided an adequate current of air is27 J., 1931, 105.28 F.Ephraim, Ber., 1931, 64, [B], 1216; A . , 1931, 813.29 W. Reif, Mikrociienz., 1031, 9, 42.1; A., 1931, 927; 2. anal. Chem., 1932.30 J., 1931, 2003.31 F. Ephraim, Ber., 1931, 64, [B], 1210; A., 1931, 813.32 Ibid., p. 2819; A., 160.33 Ann. Chim. anal., 1925, 15, 373; A., 1926, 140.3 4 F. Feigl, Ber., 1923, 56, [B], 2083; A . , 1923, ii, 880.35 Bur. Stand. J . Res., 1932, 9, 1 ; A , , 1104.36 P. Slawik, Chem.-Ztg., 1912, 36, 54; A., 1912, ii, 299; L. A. Tschugaevand B. P. Orelkin, 2. anorg. Chem., 1914, 89, 401; A., 1915, ii, 489; W.Vauhel, 2. oflentl. Chem., 1921, 27, 163; A., 1921, ii, 596; P. M. Koenig,Chin?. et Ind., 1922, 7, 55; E. J. Kraus, 2. anal. Chem., 1927, 71, 189; A .,1927, 746; 'R. Nakaseko, Mem. CoZZ. Sci. Kyoto, 1928, [ A ] , 11, 113; A , ,1928, 727.88, 38; A., 589.37 F. Ephraini, Helw. Chirn. Acta, 1931, 14, 1266; A., 137." J. Dick, 2. anal. Chem., 1929, 7'7, 354; A., 1929, 901ELLIS AND FOX. 231finally drawn through the crucible ; 39 precipitation in the presenceof much cobalt can be performed by converting that metal intothe form of cobalticyanide, which is not affected by the subsequentaddit'ion of formaldehyde t o reconvert the nickel into the ionisedcondition.40 Among the less common applications may be men-tioned tests for bismuth,41 cobalt,43 and for the platinummetals, in particular palladium.448-Hydroxypuinoline (" oxine ") forms insoluble complex com-pounds with most metals, some of which were known for manyyears before their value in analytical operations was realised byR.Berg and F. L. Hahn independently about six years ago. Sincethat time the publication of well over Mty papers serves to indicatethe interest which this reagent has aroused. The subject has beenreferred t o in recent Reports, but it may be useful to make arecapitulation here.I n his first paperY45 R. Berg presented in tabular form the re-actions of the commoner metals in acetic acid, in ammoniacal, andin caustic alkaline solutions, and showed that in the last case, withtartrate also present, copper, magnesium, zinc, cadmium, andferrous iron alone are precipitated-with the proviso that mercury,bismuth, manganese, cobalt, and nickel are partially precipitatedwhen present in very large proportions.Of this " oxine " group ofmetals, the iron can be eliminated by prior oxidation and all exceptmagnesium are precipitated from acetic solution.The precipitates are crystalline, or can be made so by suitablywarming the solutions, and can readily be filtered and washed.They can be weighed after drying a t an appropriate temperature,usually loo", or often after careful ignition t o oxide, although thisprocedure involves the loss of the advantage of weighing a com-pound of high molecular weight ; for example, the aluminium com-pound contains only 5.87% of the metal. Alternatively, a volu-metric method may be applied whereby the oxine residue is titratedA. A. Wassiljew and A. K. Sinkowskaja, 2.anal. Chem., 1932, 89,262.*O F. Feigl and H. J. Kapulitzas, ibid., 1930, 82, 417; A., 1931, 455.41 H. Kubina and J. Plichta, ibid., 1927, 72, 11; A , , 1927, 1048.43 S. G. Clarke and B. Jones, Analyst, 1929, 54, 333; A., 1929, 900.43 F. Feigl and L. von Tustanowska, Ber., 1924, 57, [B], 762; A., 1924,ii, 504.p4 M. Wunder and V. Thuringer, 2. anal. Chem., 1913, 52, 101, 660; A.,1913, ii, 252, 884; A. Gutbier and C. Fellner, ibid., 1915, 54, 205; A., 1915,ii, 493; A.M. Smoot, Eng. and Min. J., 1915, 99, 700; B., 1915, 554; C. M'.Davis, U.S. Bur. Mines, Rep. Investigations 2351 (1922); A., 1922, ii, 662;H. E . Zschiegner, J . Ind. Eng. Chem., 1925,17, 291; A., 1926, ii, 443; R. A.Cooper, J . Chem. Met. Min. S. Africa, 1925, 25, 296; A., 1925, ii, 827.45 J.pr. Chern., 1927, 115, 178; A., 1927, 674232 ANALYTICAL CHEMISTRY.with standard bromate-bromide mixture ; H. T. Bucherer andF. W. Meier 46 have applied their filtration method in the case ofnickel and cobalt. For small quantities, a colorimetric process hasalso been de~cribed.~'The most useful applications of this reagent are for magnesium 48and aluminium.49 For the determination of the former metal inparticular it has already reached the technical text-books. Theexcess of reagent can readily be removed for subsequent deter-mination of alkali metals, while separation from the alkaline-earthmetals can also be effected, though double precipitation may benecessary. In the case of aluminium, the metal can be precipitateddirectly from solutions which contain organic compounds such astartrates or glycerol which would have to be destroyed before theusual treatment with ammonia could be applied; other usefulseparations are from beryllium,50 and from phosphate, borate,fluoride, uranium, titanium, e t ~ .~ 1Many useful separations of metals can be effected by means ofthis reagent; in addition, it provides a useful means of bringingmetals, which have been separated by other means, into a formsuitable for weighing. Amongst the practical applications whichhave been described are the evaluation of some pharmacopceialsalts and preparations,52 the determination of the magnesium hard-ness of water,53 of magnesium in Portland cement,= in blood 55and in organic liquids, 56 the analysis of cadmium-red pigments46 2.anal. Chem., 1932, 89, 161; A., 1012.4 7 R. Berg, with W. Wolker and E. Skopp, Mikrochem., Emich Festschr.,1930, 18; A,, 1930, 1546; W. A. Hough and J. B. Ficklen, J. Amer. Chern.SOC., 1930, 52, 4752; A , , 1931, 327.R. Berg, 2. anal. Chem., 1927, 71, 23; A., 1927, 639; F. L. Hahn andK. Viewig, ibid., p. 122; A., 1927, 639; H. Fredholm, Svensk Kem. Tidskr.,1932, 44, 79; A., 489.49 R. Berg, 2. anal. Chem., 1927, 71, 369; A., 1927, 848; F. L. Hahn andK. Viewig, loc. cit. (ref. 48).5O I. M. Kolthoff and E. B. Sandell, J. Amer. Chem. SOC., 1928, 50,1900; A., 1928, 981; M. Niessner, 2. anal. Chem., 1929, 76, 135; A., 1929,285.51 G. E. F. Lundell and H. B. Knowles, Bur. Stand. J. Res., 1929, 3, 89;A., 1929, 1260.s2 H.Matthes and P. Schutz, Pharm. Ztg., 1928, 73, 353; B., 1928, 501 ;I. M. Kolthoff, ibid., 1927, 72, 1173.53 K. V. Luck and H. J. Meyer, 2. angew. Chem., 1928, 41, 1281; B., 1929,114; M. E. Stas, Pharm. Weekblad, 1930, 67, 1245; B., 1931, 92.54 J. C. Redmond and H. A. Bright, Bur. Stand. J. Res., 1931, 6, 113; B.,1931, 396.65 S. Yoshimatsu, Mikrochem., 1931, 9, 628; D. M. Greenberg and M. A.Mackay, J. Biol. Chem., 1932, 96, 416; A . , 764.56 C. Bomskov, 2. physiol. Chem., 1931, 202, 32; A., 35ELLIS AND FOX. 233involving separation of cadmium from ele en ate,^^ and its use insilicate analysis.58Following the observation 59 that 5; : 7-dibromo-8-hydroxyquinol-ine gives a precipitate with copper salts even in mineral acid solu-tion, the influence of various substituents on the solubility andstability of the complex metal compounds has been investigated.60’ Several of the 5 : 7-dihalogeno-derivatives form very insolublecompounds with copper, iron, and titanium, for which metals ascheme of separations has been described, based on the dibromo-derivative .61Aliphatic diamines, e.g., ethylenediamine, form certain insolublecomplexes such as (HgI,)(Cuen,), which has been applied to thedetermination of both mercury 62 and copper.63 Cadmium formsan analogous but in the case of bismuth, the cobaltcomplex has the composition (BiI,),(Co en,)I G5 (compare the com-pound CgH70N,HBi14 given by bismuth with 8-hydroxyquinolinein the presence of iodide).66 Propylenediamine can also be used inthe mercury-copper complex.67A compound of somewhat different character is (Ni en,)S,O,,whose insoluble character permits the detection of thiosulphate inthe presence of most sulphur oxy-acids, thiocyanate, and sodiumsulphide ; 68 this recalls the compound copper pyridine persulphatewhich serves to detect pers~lphate.~~DiaZEyZdiMiocccrbamtes, which were proposed many years ago 70for the detection of copper and iron, have been applied quantitat-ively to the determination of copper, 71 sodium diethyldithiocarb-G 7 C. G. Daubney, Analyst, 1932, 57, 22; B., 234.5 8 J. Robitschek, J. Arner. Ceram. SOC., 1928, 11, 587; B., 1928, 895;A. Benedetti-Pichler and F. Schneider, Mikrochem., Emich Festschr., 1930,1 ; A., 1930, 1544; A.Granger, Ceram. et Verrerie, 1932, 137.50 R. Berg, Z . anal. Chem., 1927, 70, 341; A., 1927, 436.6O Idem, Z . anorg. Chem., 1932, 204, 208; A . , 490.61 R. Berg and H. Kustenmacher, ibid., p. 215; A., 490; see also idem,Mikrochem., Emich Festschr., 1930, 26; A., 1930, 1546; V. Marsson andL. W. Haase, Chem.-Ztg., 1928, 52, 993; A., 1929, 164; L. W. E a s e , Z .anal. Chem., 1929,78, 113; A., 1929, 1159.G2 G. Spacu and G. Suciu, ibid., 1929, 77, 334; 78, 244; A., 1929, 901,1259.63 Idem, ibid., p. 329; A., 1929, 1413.64 Idem, ibid., p. 340; A., 1929, 900.65 Idem, ibid., 79, 196; A., 1930, 184.M R. Berg and 0. Wum, Ber., 1927, 60, [B], 1664; A., 1927, 847.87 G. Spacu and P. Spacu, Z. anal. Chem., 1932, 89, 187; A., 1011.69 G. Spacu, But.SOC. StGnte Cluj, 1923, 1, 583.70 M. DelBpine, Bull. SOC. chim., 1908, [iv], 3, 652; A., 1908, ii, 633.71 T. Callan and J. A. R. Henderson, Analyst, 1929, 54, 650; A.,Idem, ibid., p. 192; A., 1010.1930, 53.H 234 ANALYTICAL CHEMISTRY.amate being the particular salt used. Since iron also gives abrown colour, it must be removed, and this has been effected byelectrolytic deposition of the copper 72 or by means of citric acid; 73the coloured copper salt can then be extracted by carbon tetra-chloride. The method has been applied to ascertain the coppercontent of sea-water 74 and of certain pharmaceutical preparationsand chemicals. 75 Piperidine pentamethylenedithiocarbamate, pre-pared commercially as an accelerator for vulcanisation, can be usedfor the same purp0se.7~Tetramethyldiaminodiphenylmethane (" tetramethyl-base ") haslong been known as a reagent, qualitatively, for ozone 77 and forlead and manganese in their higher states of oxidation.78 Althought'he blue coloration is not too stable, endeavours have been made toutilise the reaction quantitatively for lead 79 and for manganese ;permanent standards are prepared of mixtures of crystal-violet andmethylene-blue.New ground is cut in reactions with vanadateand tungstate, the latter being applied quantitatively.81Cupferron, which derived its name its a precipitant for copperand iron, is also known in the same capacity for titanium andzirconium ; during t,he last ten years, furt<her investigations haverevealed other possibilities, in particular in the case of tin.82 Con-ditions for the precipitation of mercurous mercury 83 and bismuth **are also described.It must not be overlooked that, in neutralsolution, all metals except the alkalis are precipitated by cupferron72 F . Grendel, Pharm. Weekblad, 1930, 67, 913, 1050, 1345; B., 1930,1089; W. R. G. Atkins, J . Marine Biol. ASSOC., 1932, 18, 193; A., 714.73 L. A. Haddock and N. Evers, Analyst, 1932, 57, 495; A., 1011.7 4 W. R. G. Atkins, loc. cit. (72).7 5 N. Evera and L. A. Haddock, Quart. J . Pharm., 1932, 5, 458.7 6 R. G. Harry, AnaZyst, 1931, 56, 736; A., 35.7 7 C . Arnold and C. Mentzol, Ber., 1902, 35, 1324; A., 1902, ii, 352; F.Fischer and F . BrBhmer, Ber., 1906, 39, 940; A., 1906, ii, 224; F.Fischerand H. Marx, ibid., p. 2555; A , , 1906, ii, 627.78 J. A. Trillat, Compt. rend., 1003, 136, 1205; A., 1903, ii, 512; J. B .Ficklen, Chem. and Ind., 1931, 50, 869; A., 1931, 1383.79 A. D. Petrov, J . Buss. Phys. Chem. SOC., 1928, 60, 311; A . , 1928,726.R. G. Harry, Chem. and Ind., 1931, 50, 796; A , , 1931, 1385.M. Papafil and R. Cernatesco, Ann. sci. Univ. Jassy, 1931, 16, 526;A., 1931, 1386.82 A. Kling and A. Lassieur, Compt. rend., 1920, 170, 1112; A , , 1920, ii,452; N. H. Furman, J . I n d . Eng. Chem., 1923, 15, 1071; A., 1923, ii, 881;A. Pinkus and (Mlle.) J. Claessens, Bull. SOC. chim. Belg., 1927, 36, 413; A . ,1927, 848.83 A. Pinkus and (Mlle.) M. Katzenstein, ibid., 1930, 39, 179; A., 1930,1011.a. Pinkua and J. Dernies, ibid., 1928, 37, 267; A., 1928, 1109ELLIS AND FOX. 235and that such separations depend upon the varying solubilities ofthe salts in acid solutions.a-Nitroso-~-mphthoZ is best known as a precipitant for cobalt,s5although it has also been applied to the determination of palladium 86and to the separation of iron from beryllium,87 indium,s8 andgalli~rn.~S Since cupferron does not interfere with the subsequentprecipitation of cobalt by nitrosonaphthol, the former reagentcan be used t o remove traces of iron left after ether extractionin the determination of cobalt in magnetic and high-speed tools teeL90One of the drawbacks to the use of this reagent for cobalt is thatthe precipitate obtained is of indefinite composition and musttherefore be subjected to further treatment to convert it into aform suitable for weighing.It has now been shown,g1 by firstoxidising the cobalt and then working in an acetic acid solution ofthe cobaltic hydroxide obt,ained on addition of alkali, that a definitecompound is obtained which can be dried a t 130" without losingits two molecules of water. The process is particularly applicableto the analysis of cobaltic colours such as smalt.An interesting test for cyanates,92 applicable in the presence ofmost inorganic salts, consists in t,reating the dried silver salt,suspended in ether, with cyclohexene and iodine; the pungent-smelling 2-iodocycZohexylcarbimide is formed. As cyanides alsocause the formation of a somewhat similar-smelling compound,treatment of the test solution with ammonia to give a white, finelycrystalline precipitat,e of the corresponding carbamide serves t odiagnose cyanates.The readiness with which sodium sulphite in solution becomesoxidised is not generally appreciatcd: Experiments a t concen-trations such as are used in volumetric work have showns3 thatsuch oxidation may amount to 30% in 30 minutes, but may beretarded for some hours by the addition of erythritol, glycol, orethyl alcohol.In the case of sulphurous acid there is the additional85 M. Ilinski and G. v. Knorre, Ber., 1885, 18, 699; A., 1885, 840.8 6 M. Wunder and V. Thiiringer, 2. anal. Chem., 1913, 52, 737; A., 1913,ii, 1080; W. Schmidt, 2. anorg. Chem., 1913, 80, 335; A., 1913, ii, 440.87 M. Schleier, Chem.-Ztg., 1892, 16, 420; B., 1892, 713; E.A. Atkinsonand E. F. Smith, J . Amer. Chrn. SOC., 1895, 1'7, 688; A., 1896, ii, 220.8 8 F. C. Mathers, ibid., 1908, 30, 209; A., 1908, ii, 434.89 J. Papish and L. E. Hoag, ibid., 1928, 50, 2118; A., 1928, 981.01 C. Mayr and F. Feigl, 2. anal. Chem., 1932, 90, 15; A , , 1224.92 M. Linhard and M. Stephan, 2. anal. Chem., 1932, 88, 16; A., 588.g3 J. Lukas, Chem. Listy, 1932, 26, 26; A., 242.See p. 221236 ANALYTICAL CHEMlSTRY.danger of loss of sulphur dioxide although these errors may bereduced by the use in the analysis of a stronger oxidising agentthan the usual iodine.94 I n general terms, acid sulphite solutionsare more stable than neutral solutions.95 These considerationsalso enter into determinations of sulphur dioxide in gases; thusM.D. Thomas and J. N. Abersoldg6 found that substantial pro-portions of the dioxide absorbed in water from very dilute mixtureswith air were oxidised. It follows that accurate determinationsof the dioxide in gaseous mixtures can only be made if thescrubbing reagent can serve as the oxidising agent, e.g., hydrogenperoxide or iodine-starch, or, if alkaline media are used, somestabilising substance such as has been mentioned above is alsopresent ; for tjhis purpose H. F. Johnstone 97 found benzyl alcoholsatisfactory.The iodometric determination of persulphate has attracted muchattention in recent years, and, although reports have, on the whole,been favo~rable,~~ yet erratic results have been recorded ; 99 thesehave now been traced to the presence in the potassium iodide of asmall quantity of organic impurity having a strong reducing actionin alkaline solution.Persulphates, including the ammonium salt,are decomposed by boiling for a short time with excess of pureneutral hydrogen peroxide, and the sulphuric acid formed can thenbe titrated with alkali.2Organic Analysis.Compounds having an enolic or allied structure have a reducingaction on mercurous nitrate; the reaction is more general thanthat with ferric chloride but is also given by certain substanceshaving active unsaturated grouping^.^During recent years, many derivatives have been described forthe purposes of identification of numbers of various classes ofcompound. In the case of acids, particular attention has beenQ4 J.Bicskei, 2. anorg. Chem., 1927, 160, 64; A., 1937, 330.O 5 H. M. Mason and G. Walsh, Analyst, 1928, 53, 142; A., 1928, 497.9 G Ind. Eng. Chem. (Anal.), 1929, 1, 14; B., 1929, 282.Q7 Univ. Illinois Bull., 1931, 28, No. 41, 100.Q8 L. von Zombory, 2. anal. Chem., 1928, 73, 217; A,, 1928, 497; A.Schwicker, ibid., '74, 433; A., 1928, 1107; C. V. King and E. Jette, J. Amer.Chem. SOC., 1930, 52, 608; A., 1930, 441; J. H. van dor Meulen, 2. anal.C'hem., 1932, 88, 173; A., 710.Q9 A. Kurtenacker and H. Kubina, ibid., 1931, 83, 14; A., 1931, 451.1 A. Kurtenacker, ibid., 88, 171 ; A., 710.3 E. V. Zappi, Bull. SOC. chim., 1932, [iv], 51, 54; A., 362.J. H. van der Meulen, Rec. trav. chim., 1932, 51, 445; A., 587ELLIS AND FOX. 237paid to the phenacyl4 and substituted phenacyl esters.Duringthe present year the lists have been extended in the case of theunsubstituted,6 the p-halogeno-,' and p-phenyl-phenacyl esters.I n general, aliphatic acids when warmed with thionylaniline affordanilides .92 : 4-Dinitrochlorobenzene readily reacts with alkyl and arylsodium mercaptides to form sulphides which in turn may be con-verted into the corresponding sulphones by means of permangan-ate.lO The p-toluenesulphonates of aromatic amines can, in general,bc readily prepared in a pure state ; l1 the p-nitrobenzyl derivativesof a number of amines have been described,12 and the carbamatesand carbamides derived from p-nitrophenylcarbimide with alcoholsand amino-compounds.13Pentabromoacetone, formed by the oxidation of citric acidby acid permanganate in the presence of bromide, reacts inalcoholic solution with sodium iodide; one molecule of citricacid in this way liberates six equivalents of iodine.This methodis more rapid than those in which the bromo-compound isweighed.14The claim by L. Nyns l5 for a copper-bicarbonate solution asspecific for fructose has been disproved both in this country and inAmerica. Maltose, dextrose, and arabinose reduce the reagent to aslight degree, the effect increasing in the order given. Nevertheless,the method is of value in sugar analysis, particularly when used incombination with other methods; by working a t 55" instead of at48-549", the period of reduction may conveniently be reduced by$. B. Rather and E. E. Reid, J . Amer. Chem. SOC., 1919, 41, 75; A.,1919, i, 157.W. L. Judehd and E. E. Reid, ibid., 1920, 42, 1043; A., 1920, i, 480;R. M. Ham, E. E. Reid, and G. S. Jarnieson, ibg., 1930, 52, 818; A., 1930,474; S. G. Powell, ibid., 1931, 53, 1172; A., 1931, 621; N. L. Drake andJ. Bronitsky, ibid., 1930, 52, 3715; A., 1930, 1436.13 K. Chen, Trans. Science SOC. China, 1931, 7, 73; A., 529; W. Kimura,J. SOC. Chem. Ind. Japan, 1932, 35, 221; A., 946.7 K. Chen and C. Shih, ibid., p. 81 ; A., 529; C. G. Moses and E. E. Reid,J. Amer. Chern. SOC., 1932, 54, 2101; A., 744; H. Lund and T. Langvad,ibid., p. 4107; A., 1249.* N. L. Drake and J. P. Sweeney, ibid., p. 2059; A., 745.10 R. W. Bost, J. 0. Turner, and R. D. Norton, J. Amw. Chem. SOC., 1932,l1 C. R. Noller and P. Liang, ibid., p. 670; A., 375.l2 E. Lyons, J. Amer. Pharm. ASSOC., 1932, 21, 224; A., 502.l3 C. W. van Hoogstraten, Rec. trav. chim., 1932, 51, 414; A., 597.l4 P. A. Kometiani, 2. anal. Chem., 1931, 86, 359; A., 43.P. Cam6 and D. Libermann, Compt. r e d . , 1932,194, 2218; A., 867.54, 1985; A., 719.Sucr. Belge, 1924, 44, 210; B., 1925, 21238 ANALYTICAL CHEMISTRY.one-half, i.e., to 75 minutes.16 Both phenolphthalein and alcoholreact with iodine in the presence of alkali; in neutralising solutionsprior to the oxidation of aldose sugars by alkaline iodine, the useof aqueous methyl-orange avoids the introduction of errors fromthis cause.17Conditions are described for the quantitative determination ofvarious carbonyl compounds by formation of 2 : 4-dinitrophenyl-hydrazones; the method has been applied to camphor, menth-one, pulegone, citral, furfuraldehyde, methylfurfuraldehyde, andsant onin.B. A. ELLIS.J. J. Fox.16 E. F. Jackson and J. A. Matthews, Bur. Stand. J. RCS., 1932, 8, 403;17 (Miss )C. A. Mallen, AnaZyst, 1932, 57, 244; A . , 603.Is 0. Fernjndez, L. Sociiis, and C. Torres, AnaZ. Pis. Quim., 1932, 30,37 ; A., 411 ; 0. Fernandez and L. SociAs, ibid., p. 477 ; A . , 948 ; E. Simon,Biochem. Z., 1932, 247, 171; A., 763.A . , 836