antimony in 1,500,000 parts of solution may be thus detected. of small quantities is complete in one hour. The precipitation Mercury is separated from a nitric acid solution as metal, on a small closelywound platinum spiral. A current of 4-5 c.c. per minute can be used. As a confirmatory test the spiral is washed, dropped into a test-tube, heated to sublime the mercury, and then converted into the iodide by the addition of a small crystal of iodine and warming gently: 0·0001 grm. of metal can be detected thus in 150 c.c. of solution. Lead is precipitated either as peroxide at the anode from a nitric-acid solution or as metal from an ammonium-oxalate solution; the latter method is more delicate, but the former has the advantage that it can be made approximately quantitative. In this case also 0.0001 grm. is readily detected, the confirmation being effected by converting the metal or oxide into the sulphide or iodide. Copper is electrolysed as usual from an acidified solution, and 0.00005 grm. can be readily detected, the confirmation being effected by dissolving the precipitated metal in acid and testing with potassium ferrocyanide. Quantitative results with 1 mgrm. of metal are obtained thus. Cadmium is best deposited from a potassium cyanide solution, with a current of 0.2 c.c. per minute. The yellow sulphide serves as the confirmatory test: 0.0001 grm. of metal can be thus detected. The detection of the above metallic poisons in urine can be effected directly by these methods as described, but owing to the presence of the organic matter it is necessary to pass the current twice as long as when aqueous solutions are employed. In twenty-four hours a current of 1-2 c.c. per minute completely decomposes urine, leaving a clear solution. In the case of lead the electrolysis of an ammoniumoxalate solution gives a more delicate reaction than the separation as peroxide from nitric acid solution. To detect these poisons in other cases the destruction of the organic material with which they are associated, by the ordinary means, is necessary. These electrolytic tests are one and a half time more delicate than the colorimetric tests for antimony and copper by means of sulphuretted hydrogen, and ten times more delicate than the tests for mercury and lead by the same reagent. It is further to be noted that the methods are in many cases methods of separation as well as of detection, e.g., the separation of lead from iron by electrolysis in nitric acid solution. Also where it is desirable to obtain approximately quantitative results, electrolysis possesses a marked advantage over the usual colorimetric processes, because the erroneous results due to the influence of the varying constituents in the solutions tested on the reliability of the reaction are entirely obviated. 3. Interim Report on the Proximate Constituents of Coal. The Committee desired reappointment, as their investigations are not yet completed. 4. Apparatus for Extraction for Analysis of Gases Dissolved in Water. By EDGAR B. TRUMAN, M.D., F.C.S., Borough Analyst, Nottingham. The A glass flask of 500 c.c. capacity is joined by means of its tubular termination to a second lower flask of 200 c.c. capacity by means of a water-joint. In the lower flask is suspended from the upper one a thermometer, reading up to 150° C. From the neck of the upper flask proceed two millimetre tubes. right-hand one, after receiving a stopcock, expands into a cup having a capacity of 30 c.c. The tube on the left rises to the level of the bottom of the cup. This tube has two tubes supplied with stopcocks joined on to it at right angles-one above and the other below. To the one above is attached, by a water or glycerine joint, a mercury tube doubled on itself above and below, and having a length when so doubled of 880 mm. This tube is graduated in mm. from 0 to 400 in two directions, downwards in the open limb and upwards in the long limb, starting in each case from the level of the horizontal tube. This tube is filled with mercury up to the zero points, and indicates the rate of exhaustion of the apparatus, and is also a test of leakage. The second tube a little further on points downwards for attachment to a Geissler's water-pump. Still further on a stopcock is let into the main horizontal tube, which then bends downwards for communication with a Sprengel pump. The apparatus is put into connection with both mercurial and water-pumps, and the stopcock at the base of the cup is closed. By means of the water-pump the apparatus is exhausted in a great measure of air; five minutes' pumping with highpressure water produces a vacuum of 730 mm., when the barometer stands at 753. The water-pump stopcock is then closed, and exhaustion is completed by the Sprengel in about thirty minutes more. The liquid to be examined for gases is then, after measurement, introduced by the cup into the upper flask, whence it flows into the lower one. The liquid is allowed to stand for an hour, so that gases disengaged at ordinary temperatures may come off. These are collected by the Sprengel and analysed in the usual way. The vacuum having been restored, heat is cautiously applied to the lower flask by means of a Bunsen burner. If carefully done there is no bumping. The effect of heat is, by disengaging gas, to increase tension, and to enable the water to become hotter. The mercury in the mercurial tube and that in the thermometer rise. When the mercury in both places remains constant the Bunsen burner is removed. The gas given off by boiling is then collected and analysed. 5. A Discussion on Explosions in Coal Mines, with special reference to the Dust Theory, was opened by Professor H. B. DIXON, F.R.S. 6. The Application of the Hydrogen Flame in an Ordinary Miner's Safety Lamp to Accurate and Delicate Gas Testing. By Professor FRANK CLOWES, D.Sc. Lond. The 'flame cap' or halo seen in the dark above a pale flame in air containing combustible gas serves as the most rapid and practical means of detecting the presence of inflammable gas or vapour in the air. The method has been in common use by the miner, but the oil flame which he uses for the purpose is wanting, not only in delicacy, but also in accuracy. It will not readily detect the presence of less than 3 per cent. of fire-damp in the air, whereas for modern purposes it should detect less than 0.5 per cent.; and, owing to the variation in the size and adjustment of this flame when applied to testing, its indications are very variable, and are not of a standard character. Many objections exist to the employment of a separate alcohol lamp carried for testing purposes. None of these applies to the use of the hydrogen flame, especially when it is applied in an ordinary safety lamp burning oil from a wick in the usual way. The hydrogen flame is the most delicate indicator known, and it is applied of uniform size, giving standard and invariable indications. The author's early work consisted in measuring with accuracy the height and noting the appearance of the flame cap appearing over the standard 10 mm. (=0·4 inch) hydrogen flame. The flame was exposed for this purpose to air containing known percentages of gas in the 'test chamber' specially devised for the purpose. The statement, previously made, that the hydrogen flame is the most delicate gastesting flame known was fully confirmed by comparing its indications with those yielded by a small alcohol flame and by a reduced oil flame. The small alcohol flame could not detect less than 1 per cent. of fire-damp, even under the most favourable conditions; the reduced oil flame could not detect with certainty less than 3 per cent. The author then directed his attention to applying the hydrogen flame in a practical way to the detection and measurement of minute quantities of fire-damp in the air. This was ultimately effected by supplying the hydrogen from a small steel cylinder containing the gas in a compressed condition. The cylinder can be readily carried in the pocket, and, when necessary, it can be immediately attached to the ordinary safety lamp, and made to furnish the standard hydrogen flame burning at a jet in the lamp. The gas is kindled at the jet by the oil flame, which is then extinguished. The accurate estimation of proportions of fire-damp in air varying from 0-2 to 3 per cent. is rapidly and easily effected by the standard hydrogen flame. Higher percentages are estimated either by reducing the size of the hydrogen flame, or by employing the oil flame diminished in size until it becomes non-luminous. The small pocket cylinder is under a pound in weight, and when freshly charged, by being connected with a store cylinder at 120 atmospheres' pressure, it carries a store of gas sufficing for over 200 tests. This combined lighting and testing safety lamp has been found to be thoroughly practical in its nature after lengthened use in several collieries, and it surpasses in convenience all the delicate and accurate mine gas-testing apparatus yet described. The lamp, in a modified form, has been adapted to detecting and measuring petroleum vapour in the air. 7. On the Gases enclosed in Coal Dust. By Professor P. P. BEDSON. 8. A Note on the Temperature and Luminosity of Gases. 9. On Ethyl Butanetetracarboxylic Acid, and its Derivatives. By BEVAN LEAN, B.A., B.Sc., Bishop Berkeley Fellow of Owens College. When sod-malonic ether is treated with ethylene bromide, the chief product is ethyl trimethylene dicarboxylate (1.1), thus: CH2Br CH2Br CH +2CHNA: (COO C,H,), = c = (COO CH ̧), + CH2(COO C2H ̧)2 CH2 But at the same time a small quantity of an oil of high boiling-point is formed, constituting only about 3 per cent. of the whole, which is ethyl butanetetracarboxylate,' thus:: CH,Br CH2Br CH-CH: (COO C2H)2 +2CH Na: (COO C2H;)2= CH2-CH:(COO C ̧H.)2 +2NaBr. The fact that this interesting substance is produced in such small quantities made its further investigation a matter almost of impossibility. More recently, however, Professor Perkin has found that the substitution of ethylene chloride for the bromide is effectual in greatly increasing the yield of ethyl butanetetracarboxylate. As soon as the new method for the preparation of this substance had been thoroughly worked out, I investigated, at the suggestion of Professor Perkin, some of its derivatives, and I desire to give a brief notice of some of the results at which we have arrived. When treated with sodium ethylate, ethyl butanetetracarboxylate forms a di-sodium compound, which reacts readily with the iodides or chlorides of the alcohol radicals. For example, when acted on by methyl iodide the reaction takes place as follows:— In the course of this investigation I have already made a detailed study of the di-methyl, di-ethyl, di-cetyl, and di-benzyl derivatives of ethyl butanetetracarboxylate, formed by the action of alcohol radicals on its di-sodium compound. 20 20 These derivatives on hydrolysis yield tetracarboxylic acids, which possess some very remarkable properties, which have not been, so far as I know, observed in the case of any other organic acids. These acids, although they contain four carboxyl groups, do not in all cases behave as tetrabasic acids. On determining their basicity by titration with standard solution of potassium hydrate, some of them react as di-basic acids. Notably is this the case with di-benzyl butanetetracarboxylic acid, the result being the same whether phenol phthalein or litmus be used as the indicator. In this connection it is to be noted that on forming the silver or calcium salts of di-benzyl butanetetracarboxylic acid, they were found to have the formulæ CHO.Ag, and C2H2O,Ca + 2H,0 respectively. On the other hand, di-methyl and di-ethyl butanetetracarboxylic acid on titration with potassium hydrate give different results according as phenol phthalein or litmus is used as an indicator. They behave as tetrabasic acids when phenol phthalein is employed. If, however, one or two drops of litmus solution be added to the solution of these acids in potassium hydrate, which, as shown by phenol phthalein, had been neutralised by hydrochloric acid, a distinctly blue colouration is produced. On adding more hydrochloric acid the blue colouration changes gradually to a red tint, and the solution appears to become neutral to litmus, only when sufficient hydrochloric acid is added to neutralise one half of the potassium hydrate, which was equivalent, as shown by phenol phthalein, to the tetracarboxylic acid present. The silver salts of di-methyl and di-ethyl butanetetracarboxylic acid, unlike that of di-benzyl butanetetracarboxylic acid, are tetrabasic. The di-substituted butanetetracarboxylic acids we have obtained, when heated to 200°, all lose two molecules of carbonic anhydride, yielding di-substituted adipic acids. The study of these acids appeared to be especially interesting in view of the recent work on the di-substituted succinic, and glutaric, and pimelic acids. In accordance with Van 't Hoff's theory, the di-substituted succinic and glutaric acids are found in two modifications. The substituted pimelic acids, on the other hand, have only been found in one modification. Considerable interest is therefore attached to the question of isomerism in the substituted adipic acids. We have found that the di-substituted adipic acids, obtained from substituted butanetetracarboxylic acids, invariably exist in two modifications, which are readily capable of separation by crystallisation from benzene or toluene. The difference between the melting points of the two modifications is usually 60-80°. For example, two modifications of di-benzyl adipic acid were isolated, one crystallising in diamond-shaped crystals, which melted at 211-3°, the other crystallising in six-sided prisms melting at 152°. Of these derivatives of adipic acid the di-methyl alone have been previously obtained. They were prepared by Zelinsky, by the hydrolysis of ethyl dicyandimethyl adipate.' Experiments on succinic acid have shown that the more alkyl groups there are introduced, the more readily can an anhydride formation take place, and it was thought that this would also be the case in the adipic series. Now, the anhydride of adipic acid has been formed, yet on attempting to form anhydrides by heating the substituted adipic acids in sealed tubes with acetyl chloride, in no case could any evidence of an anhydride formation be obtained. On the other hand, whether the higher melting or lower melting modification was employed, a partial conversion into the other modification was effected. This result is remarkable, and cannot at present be understood. The author has also formed ethyl dibromobutanetetracarboxylate, by the action of bromine on a solution of ethyl butanetetracarboxylate in chloroform. It crystallises in magnificent prisms, which melt at 82-3°. The author is engaged in the investigation of this substance, and expects interesting results from the study of its derivatives and its use in synthetical chemistry. 1 Ber., 24, ii. 997 10. On the Salts of a new Platinum-sulphurea Base. By W. J. SELL, M.A., F.C.S., F.I.C., and T. H. EASTERFIELD, M.A. The authors have obtained the salts of a base Pt(CSN2H,),(OH), by the action of platinic chloride upon a hot solution of thiocarbamide in dilute hydrochloric acid, and subsequent addition of the acid the salts of which are required. The chloride Pt(CSN,H) Cl2, sulphate Pt(CSN,H,),SO,, and the picrate have been prepared and analysed." The free base corresponding apparently to Reiset's first base Pt(NH2),(OH), has not yet been obtained pure, for its solutions undergo partial decomposition upon evaporation even in a vacuum. That the crystalline residue thus formed contains the base is evident from the fact that the above-mentioned salts can be regenerated from it. 11. On Citrazinic Acid. By W. J. SELL, M.A., F.C.S., F.I.C., T. H. EASTERFIELD, M.A. and A. W. v. Hofmann has shown that citrazinic acid is in all probability to be regarded as a, a' dioxyiso-nicotinic acid. In this paper it is shown that in a number of cases the tautomeric keto formula more readily represents the reactions of the substance. The chlorine, bromine, and isonitroso, and derivatives prepared with a view to testing the constitution are quite in accordance with the keto formula, whilst the phenylhydrazo derivative has not been sufficiently studied for the authors to decide its constitution. Isonitroso-citrazinic acid is a somewhat unstable substance, yielding a beautiful silver salt; when boiled with dilute sulphuric acid it yields quinhydroketopyridin which on oxidation with dilute nitric acid yields the corresponding quinone, and by reduction appears to yield the hydroquinone which has not yet been obtained in the analytically pure condition. The above-mentioned quinhydroketopyridin dissolves in alkaline solutions with the production of a deep-blue solution, and it appears to be to this cause that the characteristic nitrite' test for citrazinic acid is due. By the oxidation of isonitrosocitrazinic acid with nitric or nitrous acids a bright yellow acid results which gives very characteristic salts. The acid potassium and ammonium salts are precipitated in the crystalline condition by adding the chlorides of these radicles to an aqueous solution of the acid. Although the acid contains only two hydrogen atoms these are both replaceable by metals. Reduction of the yellow acid leads to the formation of the quinhydroketopyridin. By the action of cold nitric acid upon citrazinic acid a substance is produced which appears to be represented by the formula |