372 ABSTRACTS OF CHEMICAL PAPERS INORGANIC ANALYSIS. Detection of Small Quantities of Alkali Bicarbonate in Presence of much Normal Carbonate. R. T. Haslam. (J. Amer. Chem. Xoc., 1912, 34, 822-823.)-The carbonate suspected to contain bicarbonate is dissolved in recently boiled water, and excess of calcium chloride is added. If alkali bicarbonate was originally present, some calcium bicarbonate will remain in solution with the calcium chloride.After standing five minutes, a portion of the solution is filtered free from calcium carbonate, and a few drops of ammonia are added to the filtrate. An immed- iate precipitate of calcium carbonate indicates a large amount of bicarbonate in the original sample. With only small amounts of bicarbonate present, the precipitate may not appear for ten minutes.The test will detect 0.1 per cent. of bicarbonate in normal sodium carbonate. The presence of ammonium salts vitiates the results because of the solubility of calcium carbonate in such solutions. G. C. J. Neutral Ammonium Citrate Solution. A. J. Patten and C. S. Robinson. (J. Ind. Eng. Chem., 1912, 4, 443-446.)-According to the official methods of analysis drawn up for the guidance of American agricultural chemists, the ammonium citrate solution for use in determining available phosphoric acid is to be prepared by dis- solving citric acid in water, adding ammonia until the solution is neutral to corallin, and then diluting it with water until the sp.gr. is 1.09. The authors point out that the determination of the end-point with corallin is very difficult, and that purified litmus, though better, is also unsatisfactory as an indicator.Solutions ‘( neutralised ” by four experienced analysts were used to extract a phosphatic manure, with the result that the manure was found to contain 2-83, 3-69, 4.17, and 4-82 per cent. of insoluble phosphoric acid. The authors think that the amount of ammonia necessary for neutralisation should be determined by the conductivity method as suggested byINORGANIC ANALYSIS 373 Hall and Bell (J.Amer. Chem. SOC., 1911, 33, 711 ; J. I d . Eng. Chem., 1911, 3, 559). The solution is nearly neutralised, and then equal measured portions are transferred to measuring-flasks, and a measured and different amount of dilute ammonia is added to the contents of each flask, so that two flasks at least contain an excess of ammonia and two an excess of acid.The contents of the flasks are then diluted to the mark, and their relative conductivities determined. It is unnecessary to determine the absolute conductivities. Provided the same apparatus is used for all the determina- tions, the Wheatstone bridge readings may be plotted against the volumes of dilute ammonia, and the break in the curve will indicate the volume of ammonia necessary to neutralise so much of the solution as was taken for the test.The error of this method, even in the hands of persons with no previous experience of conductivity measurements, is only one-tenth the error of the most careful titration with corallin or litmus. G. C. J. Separation of Beryllium from Aluminium.M. Wunder and P. Wenger. (Zeitsch. anal. Chem., 1912, 51, 470-473.)-From a mixture of beryllia and alumina, the former may be separated as follows : The mixture is fused with 5 grms. of sodium carbonate in a, crucible, which is covered at first, but subsequently uncovered, whilst the mixture is maintained in a state of fusion for two to three hours.The melt is boiled out with water, about a gram of sodium carbonate is added, and boiling con- tinued for a few minutes. Beryllia remains undissolved. It is filtered off, washed, dried, ignited, and weighed, and re-fused with sodium carbonate, and the whole process repeated to insure the solution of the last traces of alumina. In the mixed filtrates, alumina is precipitated by boiling with excess of ammonium nitrate, and is filtered, washed, dried, ignited, and weighed in the usual manner.The worst recorded results are 0-1466 grm. beryllia and 0.1385 grm. alumina found with 0.1472 and 0.1379 grm. respectively present. The results are not stated in such a way as to show what additional error, if any, might be expected to result from the omission of the second fusion.In practice, more or less iron will, as a rule, be present in the material presented for analysis, and the whole of this iron will remain as insoluble hydroxide with the beryllia, any chromium also present passing as chromate into the solution of sodium carbonate and aluminate. Precipitates of beryllia contaminated by ferric hydroxide are ignited, fused with bisulphate, and the aqueous solution of the melt is poured into a solution of 10 grms.of caustic soda in 150 C.C. of water, and the whole boiled. The precipitate consists of ferric hydroxide free from beryllia, and is washed, re-dissolved in acid, re-precipitated by ammonia, and weighed as ferric oxide as usual. The alkaline solution of beryllia is acidified, and the beryllia precipitated by means of ammonia.When chromium is present in the original material, it will exist as chromate in the filtrate from the alumina, and may be recovered by reduction with alcohol and precipitation with ammonia. G. C. J. Volumetric Estimation of Chloric Acid and Chlorates. A. Kolb. (Chem. ,%it., 1912, 36, 635.)-The high results obtained by the action of hydrochloric374 ABSTRACTS OF CHEMICAL PAPERS acid on chlorate in presence of potassium iodide at the ordinary temperature are due to the presence of atmospheric oxygen, and may be avoided by removing the air from the hydrochloric acid solution by means of a current of carbon dioxide, dissolving the chlorate in water which has been boiled and carrying out the reaction in a vessel filled with carbon dioxide.To 10 C.C. of air-free potassium chlorate solution (about &) are added 1 grm. of solid potassium iodide and 50 C.C. air-free hydrochloric acid of sp. gr. 1.125 in the order given, and the mixture allowed to stand five to ten minutes in a stoppered vessel in the dark. After addition of 200 to 300 C.C. water, the separated iodine is titrated. If the quantity of chlorate solution is more than 10 c.c., the hydrochloric acid must be increased proportionately, otherwise the reaction is slower.An increased quantity of chlorate in the 10 C.C. is without effect on the reaction-velocity, but more potassium iodide is required to keep the iodine in solution. 0. E. M. Quantitative Determination of Perchlorates. A. B. Lamb and J. W. Marden. (J. Amer. Chem.SOC., 1912, 34, 812-817.)-The minus errors which attend estimations of perchlorate by fusion with sodium carbonate or other substances, and determination of the resultant chloride as silver chloride, are shown to be due to loss of chloride by volatilisation. This may be wholly avoided as follows : The substance (0.5 grm.) is weighed into a Jena glass test-tube of about 25 to 30 C.C.capacity. An asbestos plug is pushed down into the tube within 2 inches of the bottom, and a second plug 2 inches nearer the top. The tube is clamped in a nearly horizontal position, and the perchlorate is heated gently until effervescence ceases. A somewhat stronger heat is then applied, until, in from ten to fifteen minutes, the chloride is thoroughly fused. When cool, the contents of the tube, including the plugs, are washed on a Gooch crucible with hot water, and the chloride in the filtrate deter- mined in the usual manner.A correction must be made, if necessary, for chloride in the asbestos used, and in the most exact work 0.1 mgrm. may be added to the weight of silver chloride found, as a correction for the loss with the final wash liquors. Without this last correction, the worst result of eight recorded by the authors is only 0.05 per cent.below the theoretical. G. C. J. Estimation of Chromium in Bronzes containing Tin and Antimony. H. Schilling. (Chenz. Zeit., 1910, 36, 697.)-The following rapid modification of von Knorre’s method is recommended : Two grms. of the finely-divided turnings are dissolved in 25 C.C. of hot aqua regia (lHNO, to 4HC1), and the solution mixed with 40 C.C.of sulphuric acid (l:l), and evaporated until white fumes appear. Complete removal of the aqua regia is essential. When cold, 200 C.C. of water are added, and the precipitated sulphates dissolved by boiling, with the exception of the lead sul- phate. The copper and antimony are now precipitated by means of iron wire in the metallic form, the liquid filtered after five minutes, and the precipitate washed.The filtrate and washings are made up to 500 c.c., heated to boiling, and treated first with a few drops of silver nitrate solution (1 : 20) as a catalytic agent, and then with 10 C.C. of a cold saturated solution of ammonium persulphate. By vigorous boiling the chromium is oxidised to chromic acid, whilst the bulk of persulphate is mean-INORGANIC ANALYSIS 375 while decomposed, The remainder, together with any permanganato that may have been formed, is decomposed by adding 5 C.C.of hydrochloric acid (sp. gr. 1.12) and boiling the liquid for a short time. When completely cold, the chromium is estimated by titration in the usual way. The presence -of tin does not affect the accuracy of the results, C.A. M. Iodimetric Estimation of Copper. K. Sugiura and P. A. Kober. (J. Amer. Chem. SOC., 1912, 34, 818-822.)-Though the iodimetric method for the estimation of copper is capable of yielding results of a high order of accuracy, this is only the case when the thiosulphate is standardised under conditions very closely resembling those prevailing in actual assays.I n consequence of this, many papers have been published on the influence of various amounts of acid, salts, water, etc., the most recent being that of Peters (ANALYST, 1912, 277). Instead of trying to secure uniformity in all the conditions by the methods advocated by Peters and others, the authors think it far more easy to precipitate the copper as hydroxide, to dissolve the washed precipitate in acetic acid, and to titrate the resulting solution with thiosulphate after addition of potassium iodide.One advantage of the method is the stable end-point with starch indicator (no return of the blue colour). The only point in the method requiring close attention is the precipitation as hydroxide. To the solution, containing phenolphthaleln, 33 per cent.alkali is added drop by drop until a slight precipitate of copper hydroxide remains undissolved. $ caustic soda free from carbonate (containing baryta) is then added from a burette until a change of colour is observed. Any excess of alkali containing carbonate would hold some copper in solution, whilst a large excess even of caustic alkali would prolong the operation of washing the precipitate. If the precipitation be performed as described, three or four washings suffice. The precipitate with the filter-paper is returned to the flask in which the precipitation was carried out, the precipitate is dissolved in 25 C.C.of 10 per cant. acetic acid, 3 or 4 grms. of potassium iodide are added, and the titration with thiosulphate conducted as usual.The thiosulphate is standardised by precipitating a measured volume of a standard copper solution with subsequent solution of the precipitated hydroxide and titration as described. Obviously, the concentration and nature of the acids and salts in the original solution are without influence ou the results, if the washing of the hydroxide be sufficiently prolonged. The amount of washing which the authors give the precipitate cannot remove the whole of the salts, but the test numbers given in the paper show that no sensible error is introduced from this cause.G. C . J. Estimation of Oxygen and Occluded Gases in Copper. G. L. Heath. (J. Ind. Eng. Chem., 1912, 4, 402-404.)-The loss in weight on heating copper drillings in hydrogen is not due entirely to oxygen from cuprous oxide and sulphurous acid, but also includes gases derived by tho metal from the fuel and refining in the furnace and, in addition, any trace of moisture.I n recognition of this fact it isnow customary to heat the drillings to constant weight in a current of carbon dioxide before proceeding to the determination of oxygen. The loss in weight during the preliminary heating in carbon dioxide is returned as " absorbed gases," and includes376 ABSTRACTS OF CHEMICAL PAPERS hydrogen, and traces of carbon dioxide, carbon monoxide, and nitrogen.The author finds that the subsequent loss of weight by heating in hydrogen does not express the true oxygen of the cuprous oxide present in the cast metal, for the metal may take UP SO much hydrogen during the process that the amount of oxygen may be under- estimated by more than 20 per cent.An accurate determination of oxygen may be made bypassing a current of carbon dioxide through the heated tube for twenty minutes after reduction by hydrogen is complete. In the author’s laboratory the carbon dioxide is purified by treatment success- ively with & saturated solution of permanganate, solution of silver sulphate, concen- trated sulphuric acid, anhydrous calcium chloride, dry chromous chloride, stick phosphorus, phosphoric anhydride, and anhydrous calcium chloride.The hydrogen is purified by means of a 10 per cent. solution of caustic potash saturated with permanganate, concentrated sulphuric acid, heated palladium asbestos, a 5 to 10 per cent. solution of pyrogallol in 50 per cent.caustic potash, and anhydrous calcium chloride. The clean drillings (50 grms.) are placed in a hard-glass tube, 30 em. long and 6 mm. diameter, with one long bulb or two round ones; and if the copper is very porous or of unknown origin, the sample is dried at 1000 C. while passing a current of carbon dioxide, and the tube with its contents re-weighed after cooling and replacing the carbon dioxide by air.Air is once more expelled by carbon dioxide, and this gas continuously passed while the tube is heated to full redness for twenty minutes. After cooling and replacing carbon dioxide by air, the tube is re-weighed, the loss being due to occluded gases in the copper. Hydrogen is now passed through the tube, which at its forward end is connected to a gas-washing device containing 10 C.C.of an ammoniacal solution of cadmium chloride (2 per cent.). According to the fineness of the drillings, from one to two hours at a red heat is required to reduce the cuprous oxide. Finally, hydrogen is expelled from the tube and from the copper by replacing the stream of hydrogen by one of carbon dioxide while maintaining a red heat. After twenty minutes, the tube is cooled by an air blast, the carbon dioxide replaced by air, and the final weighing of the tube made.The loss since the previous weighing represents oxygen and sulphur, and the latter is determined by iodine titra- tion of the contents of the cadmium chloride bulbs after addition of hydrochloric acid snd appropriate dilution. Proof of the accuracy of the method is given by a careful electrolytic assay of the reduced drillings.The sum of the copper (and silver) thus found, plus the percentage of arsenic and foreign metals found by complete analysis, should amount to 100 per cent. without any oxygen or occluded gases. In three sets of test numbers given, the totals are 100*0007, 99.9998, and 99.9901. Such check-work implies the use of refined electrolytic methods such as those described by the author (ANALYST, 1911, 36,172), which permit of the determination of copper with an error not exceeding 0.001 per cent.(see also ANALYST, 1900, 25, 253; 1905, 30, 385). G. C. J. Precipitation of the Copper - Arsenic Group and Separation of its Divisions. J. I. D. Hinds. (J. Amer. Chem. SOC., 1912, 34, 811-812.)-The method of precipitation described has regard to the fact that a fairly high concen- tration of acid is necessary for the complete precipitation of quinquevalent arsenic,INORGANIC ANALYSIS 377 whereas cadmium, antimony, and tin are not completely precipitated unless the acid concentration is kept low.To 50 C.C. of the solution, if approximately neutral, 5 C.C.of concentrated hydrochloric acid and 10 drops of nitric acid are added, or a less amount of hydrochloric acid if the original solution is strongly acid, The solution is concentrated to half bulk, whereby not more than one-thousandth part of the arsenic is lost. The nitric acid serves to oxidise all the tin to the stannic condition, which is desired, because stannous sulphide is not readily soluble in colourless ammonium sulphide; at the same time, some arsenic is oxidised to the quinquevalent condition. Arsenic is now precipitated by passing hydrogen sulphide for ten minutes through the hot solution, which is heated once or twice until it actually boils. About 80 C.C.of water are added, and the current of hydrogen sulphide continued until the liquid is cold, or for fifteen minutes.The precipitate, which contains the sulphides of all the metals of the group, is filtered, washed, and transferred to a beaker, where it is covered with strong ammonia. Hydrogen sulphide is passed through the mixture for two minutes, after which the other sulphides are separated by filtration from those of arsenic, antimony, and tin, which pass quantitatively into the filtrate with only a trace of copper.G. C. J. Qualitative Analysis without Hydrogen Sulphide. H. Trapp. (Zeitsch. anaZ. Chenz., 1912, 51, 475-180.)-Higher oxides are first reduced by passing sulphur dioxide through the solution. If insoluble sulphates separate at this stage, they are filtered off and examined separately. After boiling off excess of sulphur dioxide, hydrochloric acid is added, and any precipitate formed is filtered off and examined in the usual manner.To the solution, or filtrate from insoluble chlorides, nitric acid is added to peroxidise the iron, then ammonia until a precipitate begins to form or until the solution is no longer acid, and finally ammonium sulphide in excess. Any insoluble sulphides are filtered off, and the filtrate is acidified to throw down the sulphides of tin, antimony, and arsenic, if present.These metals, as well as the alkali metals, alkaline earth metals, and magnesium, which pass into the filtrate, are separated and identified in the usual manner. The ammonium sulphide precipitate, which may contain mercury, lead, bismuth, chromium, aluminium, zinc, iron, manganese, copper, cadmium, cobalt, and nickel is treated with nitric acid, and any undissolved mercuric sulphide is filtered off.Lead is precipitated as sulphate, and bismuth tested for by diluting a portion of the solution. If bismuth is present, it is next separated, after expelling the excess of acid in the solution. On addition of caustic soda and a little bromine and warming, all the remaining metals are precipitated as hydroxides except chromium, aluminium, and zinc, which pass into the filtrate as chromate, aluminate, and zincate, except when iron is also present, when chromium will be found in the precipitate.The filtrate is divided into three portions, in one of which aluminium is precipitated by ammonium chloride; in another, zinc is detected as sulphide, whilst the third is reduced with sulphur dioxide, and any chromium precipitated as hydroxide. The hydroxides of iron, manganese, copper, cadmium, cobalt, and nickel are digested with ammonium sulphide and washed.From the mixed sulphides iron and manganese are extracted by dilute hydrochloric acid and separated from each other378 ABSTRACTS OF CHEMICAL PAPERS by means of sodium acetate.The sulphides of copper, cadmium, cobalt, and nickel, are dissolved in hot concentrated sulphuric acid, and the solution is neutralised with ammonia. Copper and cadmium are precipitated by shaking with zinc dust, the solution of cobalt and nickel is filtered, and these metals precipitated as hydroxides by boiling with excess of caustic soda. The residue of zinc dust with any copper or cadmium is treated with hydrochloric acid, which dissolves the zinc and cadmium, and the latter may be separated from the zinc by boiling the filtrate from the undissolved copper with excess of caustic soda.An alternative method of analysis, also dispensing with the use of hydrogen sulphide, is described in the paper. G. C . J. Determination of Iodides by Direct Titration.J. W. Turrentine. ( J . Id. Eng. Chem., 1912, 4, 435-436.)-The method, an empirical one, is intended for the rapid estimation of small amounts of iodide in presence of much chloride, such as is required in a laboratory where large numbers of kelp samples have to be reported on. The solution (10 to 100 c.c.) containing the iodide to be estimated is transferred to a 250 C.C.separator, into which 15 C.C. of dilute (1 : 10) sulphuric .acid and 15 C.C. of carbon tetrachloride are also introduced. A solution of permanganate, containing about 1 grm. per litre and standardised in an exactly similar manner on known amounts of iodide, is next added from a burette with shaking after each addition. With the decrease in the concentration of iodide, the solubility of the iodine decreases until, as the end of the titration is approached, the solution appears colourless.The easiest end-point to work to, however, is the persistence for one minute of the pink tint due to excess of permanganate. From the reading, an empirically determined deduction of 0.2 C.C. is made, and the remainder is then the measure of the iodide present.In standardising the permanganate, measured volumes (10 to 100 c.c.) of an exactly 0-1 per cent. solution of potassium iodide are treated in the manner described. By taking varying volumes, not only is the iodine value of the perman- ganate determined, but also the empirical correction already referred to. A series of numbers is given showing that in tbe author's hands this correction is about 0.2 C.C.permanganate, and that with this correction the method enables the analyst to determine small quantities of potassium iodide with an error not exceeding 6 per cent. with as little as 10 mgrms. present, nor exceeding 3 per cent. with quantities of 20 to 100 mgrms. The application of the method may be extended by effecting the standardisation in presence of large quantities of any salt, in presence of which the analyst is called upon to determine traces of iodide; but where this other salt is sodium chloride, such procedure is unnecessary.The author frees his carbon tetrachloride from reducing substances by prolonged treatment with iodine, subse- quently removing the excem of iodine with thiosulphate and washing the carbon tetrachloride with water.G. C. J. Potassium Iodide and Mercurous Nitrate as Sensitive Reagents for Tungsten and Molybdenum. E. Kafka. (Zeitsch. anaZ. Chem., 1912, 51,482.)- To the neutral solution of tungstate or molybdate, 1 drop of a saturated solution of mercurous nitrate is added, and this is followed by 1 C.C. or more of concentratedINORGANIC ANALYSIS 379 hydrochloric acid and excess of potassium iodide.The mixture is shaken vigorously until the green mercurous iodide first formed redissolves. In presence of much tung- state or molybdate, the solution assumes a blue colour at once. With little tungstate or molybdate present, the colour develops more slowly. As little as 0.2 mgrm. of sodium tungstate gives a distinct reaction. To distinguish between tungsten and molybdenum, potassium thiocyanate is added to the blue solution, which will change to blood-red in presence of molybdenum.Solutions containing so little molybdenum as to develop no blue colour in the first reaction may give a distinct red with thiocyanate. The function of the mercurous nitrate in the reaction first described is to increase the reducing power of the hydrogen iodide by forming a stable complex with the liberated iodine.G. C. J. Separation of Nickel and Palladium by Means of Dimethylglyoxime. W. Wunder and V. Thuringer. (Ann. Chim. anak, 1912, 17, 201.)-The solution of the two metals is slightly acidified, preferably with hydrochloric acid, and treated with an excess of a, 1 per cent. solution of dimethylglyoxime in 2 per cent. hydro- chloric acid.After standing for thirty minutes on the hot-water bath the precipi- tated palladium is collected, washed with boiling water, ignited, and weighed in the metallic form. The filtrate is boiled and rendered slightly alkaline with ammonia, and the precipitated nickel dimethylglyoxime is collected in a Gooch crucible, washed with boiling water and then with 20 per cent.alcohol, and dried at 100' C. until constant in weight. The weight multiplied by the factor 0.2031 gives the amount of nickel. In solutions containing 0.025 grrn. of palladium and 0.0846 grm. of nickel, the amounts thus found in six estimations were 0.0248 to 0.0252, and 0-0840 to 0.0843 grm., respectively. C. A. M. Use of Oxygen under Pressure for Estimating Carbon in Ferro-Alloys.P. Mahler and E. Goutal. (Compt. rend., 1912,154, 1702-1705.)-The use of the calorimetric bomb in the estimation of carbon in steel (ANALYST, 1911, 36, 562) has been extended to the most refractory ferro-alloys. With these alloys, not only must one use litharge as a flux as in the case of very hard steels, but an auxiliary com- bustible is required. As auxiliary combustible the authors use iron of known carbon content.From 0.5 to 3 grms. of the alloy, according to its supposed carbon content, is taken for combustion, together with two or three times its weight of iron. With alloys very high in carbon, of which only 0-5 grm. can conveniently be taken, the proportion of iron must be increased, as there is some difficulty in effecting complete combustion of charges with a total weight of less than about 3 grms.Of litharge a weight equal to balf that of the combustible charge is taken. A second experiment is made under identical conditions, save for the omission of the ferro-alloy. This second experiment provides the correction for the carbon in the soft iron and for any constant errors of the method. The authors now employ caustic soda for absorbing the carbon dioxide, and estimate the amount of the latter by titration before and after combustion, using phenolphthalein as an indicator.The test numbers given in the paper show that, when the method does not give results in close agree- ment with those of the ordinary direct dry combustion method, it gives higher380 ABSTRACTS OF CHEMICAL PAPERS results. In these cases-high-grade ferro-chromes and ferro-silicons-the authors think their higher numbers must be preferred.The following numbers serve to illustrate the difference in results by the two methods. A 69 per cent. ferro-chrome showed 10.6 per cent. carbon by the bomb, and 9.7 per cent. by ordinary direct dry combustion. A 75 per cent. ferro-silicon showed 0.20 per cent. carbon by the bomb, and 0-11 per cent.by the ordinary method. Results by the two methods with ferro- vanadium and ferro-molybdenum agree very closely. G. C. J. Separation of the Rare Earths. C. James. (J. Amer. Chern. SOL, 1912, 34, 757-771.)-The paper states in a compendious form the best methods available for the separation of La, Ce, Pr, Nd, Sm, Eu, Gd, Tr, Dy, Ho, E, Tm, Yb, Lu, Ct, Yt, and Sc, in mixtures where all may occur.The methods are suited to large scale operations rather than to analytical practice, one method requiring a few thousand fractional crystallisations. But, though several of these elements can be determined with fair accuracy in minerals or mixtures of oxides in which they are the chief constituents, exact methods for the separation of others are not yet known, Where spectroscopic examination shows one or more of the rarer earths to be present in notable proportion, reference to this paper will show the analyst whether an approximate estimation of these elements is reasonably practical, and, if so, will indicate the necessary steps to be taken.G. C. J. Improvements in the Ludwig-Sipocz Method for the Estimation of Water in Silicates. M.Dittrich and W. Eitel. (Zeitsch. anorg. Chem., 1912, 75, 373-381.)-The method referred to in the title of the paper is that in which the powdered mineral is mixed with dry fusion mixture in a platinum boat, heated in a current of dry air in a combustion tube, and the water absorbed in appropriate drying tubes and weighed. The authors propose to replace fusion mixture by sodium car- bonate, chiefly because it is easily obtained anhydrous by three hours' heating below 300' C., and can be mixed with the powdered mineral without risk of the absorption of more than a few tenths of a mgrm.of moisture. They also agree with Hillebrand ( ' 6 Analysis of Silicate arid Carbonate Rocks," p. 87) that the higher melting-point of sodium carbonate makes its use preferable to that of a mixture of sodium and potassium carbonates.The other innovations of the authors relate to the apparatus employed. They use a quartz combustion tube 45 cm. long by 22 mm. diameter, and 0.5 mm. in the wall. The platin-iridium boat, which is 12 cm. long, is provided with a cover and enclosed in a cylinder of thin platin-iridium foil, just fitting the corn- bustion-tube, to protect the latter in case of frothing.The forward end of the combustion-tube is drawn out and connected by a ground joint to the first tared drying-tube. The other end of the combustion-tube is also ground to receive a quartz plug carrying a tube of small diameter, which is connected by rubber tubing to the apparatus for drying the current of air.Sulphuric acid is used for drying the air and for absorbing the moisture given up by the mineral. Heating is effected by a multiple Bunsen burner, supplemented as a rule by five to tea minutes heating with a multiple blowpipe, or an electric furnace may be used with advantage. In the latter case the temperature is raised to 850" to 900° C. in about twenty minutes,INORGANIC ANALYSIS 381 maintained at this point for fifteen minutes, and then raised to 1000" to 1050Ofor ten to fifteen minutes.One grm. of the substance with 6 grms. of sodium carbonate is a suitable charge, and the residue in the boat is immediately available for the determination of silica, alumina, etc. Duplicate determinations of moisture seldom differ by more than a few tenths of a, milligram.G. C. J. Volumetric Estimation of Sulphuric Acid, Nitric Acid, and Nitrous Acid, in Nitrating and Waste Acids. G. Finch. (Zeitsch. Schiess- a. Sprelzgstofwesen, 1912, 7, 113.)-A wad of cellulose fibre is placed in 50 C.C. of the diluted mixture of acids, excess of barium carbonate added, and stirring continued until the evolution of carbon dioxide nearly ceases, when the liquid is boiled for five minutes.The precipitate of barium carbonate and sulphate is well washed, the filtrate boiled for several minutes with a wad of cellulose and 50 C.C. of sodium carbonate, filtered, the precipitate washed until no reaction is obtained with phenolphthalein, and the excess of sodium carbonate in the filtrate titrated back with hydrochloric acid and methyl orange.The cellulose, prepared by steaming ash-free filter-paper cut into small pieces for two days, and kept for use in alcohol, ensures the precipitation of the barium sulphate in an easily filtered condition. The barium carbonate is prepared by precipitating barium chloride solution with sodium carbonate in the cold, and washing thoroughly by decantation. If & barium hydroxide solution is used instead for the neutralisation, the acids may be estimated simultaneously ; the total acidity is given by the amount of barium hydroxide used, and the nitric acid, present in the filtrate as barium nitrate, i s estimated as before.Nitrous acid, if present, is titrated with permanganate, and: deducted from the total acidity. The nitrometer may show a higher total nitrogen, than the volumetric method; this is attributed to the presence of lower nitric-acid esters of glycerol, not removed in the after-separation.The error is usually negligible. 0. E. M. Rapid Method of Purifying Sulphupic Acid. G. Bressanin. (Gum. Chim. Ital., 1912, 42, 456-458.)-The principle of the method of separating arsenic, etc. (ANALYST, 1912, 206), may be applied to the purification of sulphuric acid for toxicological work.The acid is first concentrated to sp. gr. 1.52, then cooled, and treated with 10 C.C. of a 30 per cent. solution of hydriodic acid. After standing for twelve hours it is filtered through glass-wool and purified asbestos, and the iodine eliminated from the filtrate by boiling. Acid containing as little as 0-0005 mgrm.of arsenic per C.C. gives a precipitate with the reagent after some time. The purific. ation must be carried out in vessels of Jena glass, since ordinary glass or porcelain may yield further traces of arsenic to the acid. Tin, copper, cadmium, lead, anti- mony, and selenium, are also precipitated-the last in elementary form ; whilst nitric and nitrous acids are reduced to nitrogen peroxide.Mercury and bismuth are not precipitated, and no precipitate is given by the members of the third and following groups of metals. C. A. M.382 ABSTRACTS OF CHEMICAL PAPERS New Reagent for Thorium. M. Koss. (Chem. Zeit., 1912, 36, 686.)-Tho solution to be tested is rendered strongly acid with hydrochloric acid and a few drops of a solution of sodium hydrogen hypophosphate (NaHPO,, 2H,O) added. With considerable quantities of thorium a white flocculent precipitate forms at once ; with small quantities heating or long standing is necessary. Less than 0.0001 grm. of thorium oxide in 1 C.C. of solution may thus be detected. Titanium and zirconium are simultaneously precipitated. Previous treatment with hydrogen peroxide converts the titanium into pertitanic acid, which is not precipitated by the reagent. To remove zirconium, the mixed precipitate, after washing, is oxidised with sulphuric and nitric acids, and the solution precipitated with oxalic acid ; zirconium remains in solution. 0. E. M. Quantitative Determination of Yttrium. C. F. Whitternore and C. James. (J. Amer. Chem. Soc., 1912, 34, 772-774.)-Precipitates of yttria are apt to be contaminated with salts of sodium, potassium, lithium, and magnesium, and with ferric oxide and alumina, if the corresponding elements are present in the solutions from which the yttria is precipitated. Thus yttrium may be overestimated by 5 per cent. if precipitated by caustic soda, ammonia, or oxalic acid, in presence of salts of sodium or potassium. The ammonium salts of several organic acids were tried as precipitant, with unsatisfactory results; but it is found that ammonium sebtt- cate precipitates yttria quantitatively and free from sodium even when large quan- tities of sodium are present. In presence of potassium, a, second precipitation as sebacate is necessary to obtain exact results. In presence of iron, aluminium, mag- nesium, or lithium, the whole of the yttria may be precipitated free from any trace of these elements by means of oxalic acid, provided the precipitation be made in the cold, and that excess of ammonium chloride is present. G. C. J.