摘要:

J O U R N A L OF THE CHEMICAL SOCIETY. H. E. ARMSTRONG, Ph.D., F.R.S. WYNDHAM R. DUNSTAN. F. R. JAPP, M.A., Ph.D., F.R.S. H. F. MORLEY, M.A., D.Sc. HUGO MULLER, Ph.D., F.R.S. W. H. PEREIN, PbD., F.R.S. W. RAMSAY, Ph.D., F.R.S. W. J. RUSSELL, Ph.D., F.R.S. J. MILLAR THOMSON, P.R.S.E. T. E. THORPE, Ph.D., P.R.S. W. P. WYNNE, B.Sc. &bim : C. E. GROVES, F.R.S. Sub-dlbitar : A. J. GREENAWAY. VOl. LVII. 1890. TRANSACTIONS. LONDON: GURNEY & JACKSON, 1, PATERNOSTER ROW. 1890.LONDON : HARItISON AND SONS, PRINTERS IN OEDINA11Y TO HER MAJESTY, Sl‘. DfARTIN’Y LAKE.C O N T E N T S . PAPERS READ BEFORE THE CHEMICAL SOCIETY. PAQE 1.-Contributions to Cellulose Chemistry. Acetylation of Cel- 11.-Compounds of Phenanthraquinone with Metallic Salts. 111.-Action of Aldehydes and Ammonia on a-Diketones.By G. H. WADSWOIZTH, Associate of the Normal School of Science . . # 1V.-Action of Dehydrating Agents on aw-Diacetylpentanc. Synthesis of Methylethylhexamethylene. By P. STANLEY KIPPING, Ph.D., D.Sc., and W. H. PHZKIN, Jun., Ph.L). V.-aw-Diacetyl-aw-diethylpentane. By F. STANLEY KIPPING, Ph.D., D.Sc., and W. H. PERKIN, Jun., Ph.D. . . 29 V1.-Frangulin. By T. E. THORPE, F.R.S., and H. H. ROBIN- SON,M.A. . . 38 VI1.-The Action of Chloroform and Alcoholic Potash 011 Hydrazines. (Part 111.) By S. RUHEMANN, Ph.D , &LA. . 5 0 VII1.-Note on the Identity of Cerebrose and Galactose. By HORACE !l’. BIZOWN, F.R.S., and G. HAERIS Moitiers, Ph.D., F . I. C. . 57 1X.-Ambinon, the Saccharon of Arabinose. By C. O’SurmVAx, F.R.S. . 59 X.--The Nature of Solutions as elucidated by a Study of the Dmsity, Electric Conductivity, Heat Capacity, Heat of Dissolution, and Expansion by Heat of Sulphuric Acid Solutions. By SPENCER UMFREVILLE PICKEI~INU, MA.. 64 XI.- A New Method of Estimating the Oxygen dissolved in Water. By JOHN C. THRESH, D.Sc., M.B., Medical Officer of Health . . 185 XIL-The Constituents of Flax. By C. F. CROSS and E. J . BEVAN . . 1% XII1.-Milk of Abnormal Quality. By FrlEnEnIcK JAMES LLOYD, F.I.C. . . 201 X1V.-Ethyl aa,-Diacetyladipate. By W. 13. PICRKIN, Jun., PI.1.L). . 2u4 lulose. By C. F. CROSS and E. J. BEVAN . . l By FRANCIS R. JAPP, F.R.S., and ALFRED E. TURNELL . . 4 . 13iv CONTENTS. XV.--ww,-Diacetylbutane. By T. RHYMER MARSHALL, D.Sc., and W. H. PERKIN, Jun., Ph.D. . XV1.--The Action of Chromium Oxychloride on Nitrobenzene.By G. G. HENDERSON, B.Sc., MA., F.I.C., and J. MORROW CAMPBELL, B.Sc., Donaldson Chemical Scholar, University of Glasgow . XVII.-Semithiocarbazides. By AUGUSTUS E. DIXON, M.D., Professor of Chemistry, Queen’s College, Galwny XVIII.-Note on a Phenylic Salt, of Phenylthiocarbamic Acid. By ATJGUSTUS E. DIXON, M.D., Professor of Chemistry, Queen’s College, Galway . XIX.-The Behaviour of the more Stable Oxides at High Tem- peratures. (Part I.) Cupric Oxide. By G . H. BAILEY, I).Sc., Ph.D., and W. 6. HOPKIKS, the Owens College . XX.-The Influence of different Oxides on the Decomposition of Potassium Chlorate. By G. J. FOM7LER, N.Sc., and J. GRANT, the Owens College . In- teraction of Benzyl Chloride and Ally1 Bromide, respec- tively, with Thiocarbamide, Monopheny lthiocarbamide, and Diphenylthiocarbamide.By EMIL A. WERNER, Assistant Lect,urer in Chemistry, Trinity College, University of Dublin . XXK-Derivatives of Phenylhexamethylene. By F. STANLEY & ? P I N G , Ph.l)., D.Sc., and W. H. PERKIN, Jun., Ph.D. . XXII1.-Some Crystallised Substances obtained from tlhe Fruits of various species of Cilms. By WILLIAM A. TILDEN, D.Sc. Lond., F.R.S., and CHARLES R. BECK . XX1V.-Synthesis of Triazine - derivatives. By RAPHAEL MELDOLA, F.R.S.. XXV.--The Nature of Solutions, as Elucidated by the Freezing Points of Sulphuric Acid Solutions. By SPENCER UMFRE- VILLE PICKERING . Part I. Hydromuconic Acid. By S. RUHEMANN, Ph.I)., MA., and F. F. BLACKMAN, B.Sc. SXVI1.-The Molecular Weights of Metals when in Solution. By C.T. HEYCOCK, M.A., and F. H. NEVILLE, M.A. XXVII1.-Tho Formation of Indene-derivatives from Dibrom- a-naphthol. By RAPHAEL MELDOLA, F.R.S., and FRANK HUGHES . XXIX.-The Evidence afforded by Petrographical Research of the occurrence of Chemical Change under great Pressure. By J. W. JUDD, F.R.S., F.G.S., Professor of Geologyin the Normal School of Science and Ro~a.1 School of Mines . XX1.-Contributions to thc Chemistry of Thiocarbamides. XXV1.-Contributions to the Knowledge of Mucic Acid. . PAQ E 241 25:; 25 7 2GS 269 2 7 2 283 304 323 32s 331 370 3 i 6 393 404CONTENTS. V Annual General Meeting . XXX.-Kesearches on the Germination of some of the Gra1nint.z.. Part I. By HORACE T. BROWN, F.R.S., and G. HARRIS MORRIS, Ph.D., F.I.C.. XXX1.-The Terminology of Hydrolysis, especially as affected by " Ferments." By HENRY E. AR&fSTRONG . XXXI1.-Action of Phosphoric Anhydride on Fatty Acids. Part I. By F. STANLEY KIPPING, Ph.D., D.Sc. . XXXII1.-Contributions from the Laboratory of Gonville and Caius College, Cambridge. No. XVIII. On the Sulphates of Antimony. By R. H. ADIE, B.A., Scholar of Trinity College . XXX1V.-Phosphorous Oxide. Part I. By T. E. THORPE, F.R.S., and A. E. TUTTON, Demonstrator of Chemistry in the Normal School of Science, South Kensin,yton XXXV.-An Investigation of the Conditions under which Hydrogen Peroxide is formed from Ether. By Professor WYNDHAM R. DUNSTAN and T. S. DFMOND . XXXV1.-Note 011 the Preparation of Pyrocatechol. By W. H. PERKIN, Jun., Ph.D. . XXXVI1.-Ni trogen Trioxide and Peroxide.By W, RAMSAY, Ph.D., F.R.S. . XXXVII1.-The Action of Light on Phosphorus and some of the Properties of Amorphous Phosphorus. By ALEXAXDER PEDLER, F.I.C. . XXXIX-The Action of Chlorine on Water in the Light, and the Action of Light on Certain Chlorine Acids.--By ALEXANDER PEDLER, F.I.C. . XL.-Notes on the Explosion of Sulphuretted Hydrogen and of the Vapour of Carbon Bisulphide with Air and Oxygen.- By ALEXANDER PEDLER, F.I.C. . XL1.-A Third Naphthnquinone. By RAPHAEL MELDOLA, F.R.S., and FRANK HUGHES. XLI1.--Diethylphosphorous Acid. By T. E. THORPE, F.R.S., and BARKER NoRI~H, Associate of the Normal School of Science, South Kensington . XLII1.-The Relative Antiseptic Powers of Isomeric Organic Compounds. By THOS. CARNELLEY, D.Sc., Aberdeen, aiid W.FREW, F.C.S., Dundee . XL1V.-Contributions from the Chemical Laboratory of the University of Edinburgh. No. I. On Tertiary Butyl Mercaptan. By LEONARD DOBBIN, Ph.D. . XLV.-Contributions from the Chemical Laboratory of the University of Edinburgh. No. 11. On Resylacetophenone. By ALEX. SMITH, B.Sc., P1i.D. . - PACIE 426 458 528 532 540 545 574 58 7 590 599 613 625 t i 3 1 634 636 639 643V i CONTENTS. PAGE XLV1.-Action of Ethyl Oxalate on Camphor. By J. BISHOP TINGLE, Ph.D. . . 652 XLVI1.-On the Molecular Weights of Metals when in Svln- tion. By C. T. HEYCOCK, M.A., and P. H. NEVILLE. Part I1 . . 656 XLVII1.-a@-Dibenzoglcinnamene and the Constitution of Zinin’s Lepiden and its Derivatives. By FRANCIS Itt. JAPP, B’.Et.S., and FELIX KLINGENANN, Ph.D.. . 662 XL1X.-Crystallographical Relations of the Derivatives of Di- benzoylcinnamene. By ALFRED E. TuTToN, Demonstrator in Chemistry at the Normal School of Science, South Ken- sington . . $14 L.-Action of Carbon Monoxide on Nickel. By LUDWIG MOND, L1.-The Milk of the Gamoose. (Preliminary Notice.) Ry A. LI1.-The Interaction of Iodine, Water, and Potassium Chlorate. T~III.-The Action of Heat on the Chlorides and Hydroxides of Mixed Quaternary Ammonium Compounds. By N. COLLIE, Ph.D., and S. B. SCHRYVER, B.Sc., the University College, London . . 767 L1V.-Note on a Compound from Benzoyn a,nd Acetone. By FRANCIS R. JAPP, F.R.S., and JULIUS RASCHEN, Ph.D. . . 78f3 LV.-Reseamhes on Normal and Mixed Diazoamides. By RAPHAEL MELDOLA, F.R.S., and F. W. STREATFEILD, F.I.C..785 LV1.-Note on the Action of Nitric Acid on Dibrom-as-naphthol. By RAPHAEL MET~DOLA, Y.R.S., and FRANK HUGHES . . 808 LVI1.-A New Method for the Estimation of Nitrates and Ni- trites in Water. By R. ORMANDY arid 5. B. COHEN, PhI-)., Owens College, Manchester . . 811 By FRANK PGLLINGER, B.A., B.Sc., late Scholar of Corpus Christi College, Oxford . . 815 Dr. CARL LANGER, and Dr. FRIEDRTCH QUINCKE . . 749 PAPPEL and H. DROOP RICHMOND . . 754 By HENRY BASSETT . . 760 LV1II.-Action of Zinc on Dilute Sulphuric Acid. L1X.-A New Nonobromocamphor. By J. E. MARSfI . . 828 LX.-Invertase: a Contribution to the History of an Uuor- ganised Ferment. By C. O’SULLIPAN, F.R.S., and P. W. TOMPSON . . 834 LY1.-Contributions to the Knowledge of Mucic Acid. Part 11. Action of Phosphorus Pent,achloride on Mucic Acid.By LXI1.-Contributions to the Knowledge of Mucic Acid. B y S. RUHEMANN, Ph.D., 8. RUHEMANN, Ph.D., M.A., and W. J. ELLIOTT, B.A. . Part 111. Hydromnconic Acid. D1.B. . . 937 . 9331CONTENTS. vii PAGE LXIIL-Diphenylfurfuran. By W. H. PERKIN, Jun., Ph.D., LX1V.-Note on the Reduction of Aromatic Amides. By A. HUTCHINSON, B.A., Scholar of Christ's College, Cambridge . 957 LXV.-Some Improved Vacuum Joints and Taps. By W. A. SHENSTONE . . 958 LXTT-The Production of Camphor froin Turpentine. By J. E. MARSH, B.A., and R. STOCKDALE, B.A. . . 961 LXVI1.-pDesylpheno1. By FRANCIS R. JAPP, F.R.S., and G. H. WADSWORTH, Associate of the Normal School of Science . . 965 By GERALD T. MOODY, D.Sc., Demonstrator in the Chemical Department, City and Guilds of London Institute, Central Institution, and T.G. NICHOLSON . 974 LX1X.-Action of Phosphoric Anhydride on Fatty Acids. Parh 11. By F. STANLEY KIPPING, Ph.D., D.Sc. . . 980 LXX.-An Investigation of the Conditions under which Hydrogen Peroxide is formed from Ether. (Second Notice.) By WYNDHAM R. DUNSTAN and T. S. DYMOND . . 988 LXX1.-Contributions from the Laboratories of the Heriot Watt College, Edinburgh. On Berberine. Part 11. By W. H. F.R.S., and A. SCHLOESSER, Ph.D., M.Sc. . . 944 LXVII11.-Parnxylenesulphonic Acids. PERKIN, Jun., Ph.D., F.R.S. . . 991CONTENTS. vii PAGE LXIIL-Diphenylfurfuran. By W. H. PERKIN, Jun., Ph.D., LX1V.-Note on the Reduction of Aromatic Amides. By A. HUTCHINSON, B.A., Scholar of Christ's College, Cambridge . 957 LXV.-Some Improved Vacuum Joints and Taps. By W. A. SHENSTONE . . 958 LXTT-The Production of Camphor froin Turpentine. By J. E. MARSH, B.A., and R. STOCKDALE, B.A. . . 961 LXVI1.-pDesylpheno1. By FRANCIS R. JAPP, F.R.S., and G. H. WADSWORTH, Associate of the Normal School of Science . . 965 By GERALD T. MOODY, D.Sc., Demonstrator in the Chemical Department, City and Guilds of London Institute, Central Institution, and T. G. NICHOLSON . 974 LX1X.-Action of Phosphoric Anhydride on Fatty Acids. Parh 11. By F. STANLEY KIPPING, Ph.D., D.Sc. . . 980 LXX.-An Investigation of the Conditions under which Hydrogen Peroxide is formed from Ether. (Second Notice.) By WYNDHAM R. DUNSTAN and T. S. DYMOND . . 988 LXX1.-Contributions from the Laboratories of the Heriot Watt College, Edinburgh. On Berberine. Part 11. By W. H. F.R.S., and A. SCHLOESSER, Ph.D., M.Sc. . . 944 LXVII11.-Parnxylenesulphonic Acids. PERKIN, Jun., Ph.D., F.R.S. . . 991

ISSN:0368-1645    

DOI:10.1039/CT89057FP001

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

4 JAPP AND TURNER: COMPOUNDS OF 11.-Compounds of Phenanthraguiszone with Metallic Salts. By FRANCIS R. JAPP, F.R.S., and ALFRED E. ‘TURNER. IN t,he course of an investigation of the condensations of phenanthra- quinone with other organic compounds, a solut,ion of zinc chloride in acetic acid was employed as a condensing agent. I t was observed that the phenanthraquinone was precipitated from the accetic acid solution in the form of dark reddish-brown needles, which proved toPHEKANTHRAQUINONE WITH METALLIC SALTS. 5 contain zinc chloride. We therefore examined the behaviour of various quinones and other diketones towards metallic salts and found that, in certain cases, by employing the proper solvent, mole- cular compounds could be obtained. The following have beell analysed :- (1.) An equimolecular compound of phenanthraquinone with zinc chloride : Cl4H8O2,ZnCl2.(2.) A compound of 2 mols. of phenanthraquinone with 1 mol. of mercuric chloride : (C14H80J2,HgC12. ( 3 . ) A corresponding compound with mercuric cyanide : The difference in the combining value of the mercury and zinc salts is worthy of note. No compounds could be obtained with benzoquinone, a-naphtha- quinone, anthraquinone, diacetyl, or benzil ; but /3-naphthaqninone gave a compound (not analysed) with mercuric chloride, closely re- sembling in appearance the corresponding compound from phenan- thraquinone. No compounds with oxy-salts could be obtained. If it is permissible to generalise from so few cases, it would appear that in order that a dicarbonyl-compound may combine with metallic salts, its two carbonyl-groups (1) must be directly united and (2) must form part of a closed chain, that is, must be in the quinone condition. No such compounds are formed with para-quinones or with open- chnin a-diketones.The above molecular compounds are intensely coloured, show- ing that the quinone retains in them its distinctive character. They are in this respect sharply distinguished from the colourless coni- pounds of the orthoquinones with hydrogen sodium sulphite, the formation of which involves a reduction and which are to be re- garded as quinol derivatives : phenanthraquinone hydrogen sodium sulphite being ClaH8( OH) (OS0,Na). Plzenanthrapuinone Zinc Chloride, C14H802,ZnCl,.-To prepare this compound, a hot, strong solution of zinc chloride in glacial acetic acid is added in excess to a hot solution of phenanthraquinone in the same solvent,.The molecular compound separates almost instantly. As i t is decomposed by the moisture of the air it is best filtered, with the aid of the filter-pump, through a funnel consisting of a wide glass tube drawn out to a narrow neck, plugged with cotton ~'001, and connected at the top with it drying tube. The compound is washed on the funnel with boiling glacial acetic acid, in which it is only very sparingly soluble, quickly transferred to a weighing bottle, and dried at 130" until the weight is constant. n;!6 JAPP AND TURNER: COMPOUNDS O F Even when the liquid containing the crystals is allowed t o cool before filtering, there is no danger of the excess of zinc chloride crystallising out.On the other hand, the washing with hot glacial acetic acid is absolutely necessary to remove adhering zinc chloride. The analysis was conducted by decomposing a weighed quantity of the substance by boiling it with water, allowing the liquid with the precipitate to stand overnight, collecting the phenanthraquinone on a weighed filter, washing with cold water to which two or three drops of hydrochloric acid had been added, and drying at 100" before weighing. Substance. Phenanthraquinone. I ........... 1.101 0-663 I1 ........... 1.117 0.676 Found. Calculated for r--- 7 (314 H,02,ZnC12 I. 11. C,,H,O, in 100 parts .... 60.46 60-21 60.51 Different preparations were analysed. The compound forms dark reddish-brown needles which may be heated above 300" without melting.Both water and alcohol de- compose it. When exposed to the air at ordinary temperatures, i t attracts moisture, its colour changing to the orange-yellow of phe- nanthraquinone. A very small quantity of water in the acetic acid used in the preparation of the compound prevents its formation altogether. Phenanthraquinone Mercuric Chloride (C,,H,O,) 2,HgC12. - Some difficulty was experienced in finding a solvent from which this com- pound would crystallise in a pure state. In alcoholic solution it is not formed at all ; glacial acetic deposits mainly phenanthraquinone, with mere traces of the new compound ; chloroform deposits the mole- cular compound as a brilliant scarlet powder, contaminated, how- ever, with free mercuric chloride, which is very sparingly soluble in chloroform.When hot concentrated solutions of mercuric chloride and phe- nanthraquinone in acetone are mixed, the liquid, on cooling, deposits crystals of the double compound, but phenanthraquinone is apt to separate out at the same time. This is avoided, however, by the following method : To a boiling strong solution of mercuric chloride in acetone, finely divided phenanthraquinone is added and the boiling is continued. The double compound separates almost immediately as a heavy scarlet powder. The hot supernatant liquid is poured off from the powder, and the latter is dissolved in boiling acetone. On cooling, this second solution deposits the double compound in red, Acetone was found to give the best results.PHENANTHRAQUINONE WITH METALLIC SALTS.7 obliquely truncated prisms, which, after washing with acetone and drying at loo", are pure for analysis. As it was feared that boiling with water would not extract the whole of the mercuric chloride, a weighed quantity of the crystals was dissolved in boiling alcohol, which decomposes them ; the plze- nanthraquinone was precipitated by the addition of an excess of water; the aIcohol was boiled off, afterwards allowing the liquid with the precipitate to stand overnight : and finally the phenanthra- quinone was collected as before on a weighed filter. The filtrate in this process was always tinged yellow, owing to the presence of traces of unprecipitated phenanthraquinone, which accounts for the slight deficit in the analyses.They melt at 222-223". Substance. Phenanthraquinone. I ........... 1.842 1.101 I1 ........... 1.131 0.682 I11 ........... 1.077 0.647 Found. Calculated for rL- 7 (C14HS02)2,HgC12. I. 11. 111. ClaH80z in 100 parts.. 60.55 59-77 60.30 60.08 In the foregoing analyses, different preparations were employed. Phenanthraquinone Mercuric Cyanide ( C~~H8@2)zHgCyn. - Hot saturated solutions of phenanthraquinone and mercuric cyanide in acetone were mixed and allowed to cool slowly. Well-shaped red crystals were deposited, which showed on certain faces a greenish reflex, visible only when the crystals were under the mother-liquor. The melting-point was 222-223", the same as that observed in the mercuric chloride compound. The method of analysis was identical with that employed in the case of the mercuric chloride compound, and the percentage of phenanthraquinone found was, for the reason already assigned, some what low. Substance. Phenanthraquinone. I.. .......... 1.122 0.691 11.. .......... 1-688 1.0'44 I11 ............ 1.288 0,794 Found. 7 Calculated foi- ---A- (C14Hs02)2,HgCYz- I. 11. 111. 100 parts ......... 62.28 61.58 61.85 61.65 Phenanthraquinone in Different preparations were analysed. Normal School of Xcience, South Kewington.

ISSN:0368-1645    

DOI:10.1039/CT8905700004

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

8 111.-Action of Aldehydes and Ammonia on a-Diketones. By G. H. WADSWOR'I'H, Associate of the Normal School of Science. TIIF: condensations of a-diketones with aldehydes and ammonia have hitherto been studied chiefly in the case of phenanthraquinone and benzil. The reaction has been shown to vary both with the diketone and with the aldehyde employed. The most frequently occurring reaction of this class is that in which glyoxalines are formed : as, for example, by the action of fatty aldehydes, and also of certain alde- hydes of the benzene series: on benzil. Of less frequent occurrence is the formation of oxazoles (i.e., compounds containing the complex -c-0 I I -C-N >C--) : this reaction is exemplified in the condensations of phenanthraquinone with several of the benzenoid aldehydes.0 ther reactions, such as those which yield the complex compounds obtained f porn beiizil with salicylaldehyde and cinnamaldehyde respectively, occur only in isolated cases.* It appeared of interest to ascertain what the action would be in the case of some typical a-diketone of the fatty series, such as diacetyl. With the kind consent of Dr. von Pechmann, by whom diacetyl was first described, and by whose method the material employed in this investigation was prepared, I have, therefore, studied its reactions with aldehydes and ammonia. The aldehydes employed were benzaldehydc, salicylaldehyde, and cinnamaldehyde; but in each case nothing but a glyoxaline was formed, and none of the abnormal reactions which occur in the case of b e n d were observed. I have also studied the action of fatty aldehydes and ammonia on phenanthraquinone, but failed t o obtain condensation-compounds.A benzene-fatty aldehyde, however-cinnamaldehyde-gave an oxazole with phenanthraqninone and ammonia. Benzaldehyde, Diacetyl, and Ammonia.-10 grams of diacetyl, with the molecular proportion of benzaldehyde, were dissolved in a small quantity of alcohol, and heated with an excess of alcoholic ammonia for half an hour on the water-bath. The new compound was precipi- tated by the addition of water, redissolved in dilute hydrochloric a#cid, * Japp and Strcatfeild, Trans., 1882, 155 ; Japp and Hooker, Trans., 1884, 673 ; Jspp and Wynne, Trans., 1886, 464. A fairly complete summary of the results obtained in this field is given in the new Watts's Dictionary of Chemistry, vol.i, p. 465.ACTION OF AL,DEHYDES AND AMMONIA ON a-DIKETONES. 9 the base reprecipitated from the filtered solution by ammonia, and purified by repeated recrystallisation from hot benzene. Analysis of the substance, dried a t looo, gave figures agreeing with the formula C1,H,,N2 :- Substance. CO,. xr,o. I ........ 0.1525 0.4284 0.0977 11.. ...... 0.1456 0.4089 0.0938 Substance. u t. P- Dry N + NO 22.32 C.C. 20" 411 mm. ,, N . . .... 22.32 ,, 20 3 i 8 ,, 111 .... 0.0832 { Dry N + NO 22.32 C.C. 24 416 mm .... 24 373 ,, IV 0.0827 { ,, N . . .... 22-32 ,, Calculated for Pound. C11 HI aN,. r- --&-- -3 I. 11. 111. IV. 7 (--.A- - GI*.. ...... 132 76.74 76.60 76.58 - HI,. ...... 12 6.97 7.12 7.15 - N2.. ...... 28 16.27 - - 16.30 16.16 - -_ 172 99.98 The compound is formed according to the equation- and would be phenyldi~neth~lglyoxaline. Phenyldimethylglyoxaline crystalli ses from benzene in slender needles containing benzene of crystallisation, which is given off on exposure t o air.It is readily soluble in alcohol. It melts at 230-234", and at a higher temperature volatilises, condensing in needles, which showed the same melting point. 0.6720 gram of the substance, crystallised from benzene and dried by brief exposure to air, lost, on heating at looo, 0.1294 gram benzene, equal to a loss of 19.26 per cent. The formula (cllH12N2)2,c6€€6 requires a loss of 18.48 per cent. The platinichloride is precipitated on the addition of platinic chlor- ide to a solution of the base in hydrochloric acid.It forms yellow needles and is anhydrous. 0.1380 gram of the salt, previously dried at looo, gave on ignition 0.0356 gram of platinum. Calculated for (C1,H,,N2,HC1)2PtC11. Fo 11 n d . Pt in 100 parts.. ...... 25-82 25.7910 WADSWORTH : ACTION OF ALDEHYDES Salicy laldehyde, Diacetyl, and Ammonia.-The operation was con- ducted like the foregoing, employing 10 grams of diacetyl and the molecular proportion of salicylaldehyde ; and the resulting compound was purified by recry stnllisation from dilute alcohol. Analysis gave numbers agreeing wit'h the formula of orthohydroxyphenyldimet?~ylgly- oxnlin e. C13:i*g- XH >C*CsH4*OH (1, 2). C H,*C--N Substance. GO,. H,O. I.. ...... 0.2100 0 5395 0.1213 11.. ...... 0.1475 0.3 790 0.0864 Substance. ZI. t.P. Dry N + NO 22.32 C.C. 29' 396 mm. N . . 22.32 29 375 .... I11 0.0856 { .. .... .. .. Calculated for Found. - c11.. ...... 132 70.21 70.06 70.07 H12. ...... 12 6.38 6.41 6.50 N2.. ...... 28 14.89 - 15,02 0 ........ 16 8-51 188 99.99 - - - - - - -- The compound crystallises from dil Ute alcohol in colourless, satiny needles melting a t 218", and from benzene in star-shaped forms. The alcoholic solution shows a blue fluorescence. The hydrochloride is very sparingly soluble in cold water. From its solution the platinichloride is precipitated in yellow needles of the formula ( Cl,H,2Nz0,HC1)2PtCl~,2Hz0 on the addition of platinic chloride, but owing to the sparing solubility of the hydrochloride, is apt to be contaminated with the latter salt, for whicli reason the per- centage of platinum was found too low. 0.33.58 gram of the &-dried platinum salt lost on heating a t 100" 0.0150 gram, and the remaining 0.3208 gram gave on ignition 0.0772 gram platinum. Calculated for (C11H1JY,O,HC1)2PtC1,,2HzO.Found. H,O in 100 parts.. .......... 4.38 4-46 Calculated for (CllH1,N20,HC1)2PtC1,. Fonnd . P t in 100 parts.. ............ 24.74 24*( J6AYD AMMONIA ON a-DIKETONES. 11 Cinnamaldehyde, Diacetyl, awd Ammonia.-10 grams of diacetyl with the molecular proportion of cinnanialdehyde were heated with alcoholic ammonia in a sealed tube at 100" for about an hour. The contents of the tube, after expelling the alcohol by evaporation, were acidified with strong hydrocliloric acid and diluted with 3 to 4 times their bulk of water; this precipitated most of tho resin, which is formed in large quantity in this reaction.The resin was filtered off as quickly as possible, after which the hydrocliloride of the new base began to separate from the filtrate in the form of yellowish needles, the separation being facilitated by the addition of a large excess of fuming hydrochloric acid to the solution. This salt was freed from adhering resin by washing with ether, then dissolving in a small quantity of alcohol and reprecipitating with ether, by which means it was obtained colourless. From a hot aqueous solution of the hydro- chloride, ammonia precipitated the base, which was then purified by recrystallisntion from benzene. It formed colourless, warty crystals melting at 201-202", very soluble in alcohol and in ether, sparingly soluble in petroleum, very slightly soluble in boiling water.A nitrogen determination showed that in this case also a glyoxaline cinnana en y ldime t h y l g 1 y oxaline- C H3.g - NH C -C H:CH*C,H,. C H3*C-N> had been formed. Substance. V. t . P. o,0878 . . { Dry N + NO . . 13.5 C.C. 19" 597 mm. ) ? N ......... 13.5 ,, 19 582 ,? Calculated for C13H14N2. Found. N in 100 parts.. . . . . . . 14.00 14.14 On the addition of platinic chloride to a solution of the hydro- chloride, the platinichloride is precipitated in the form of microscopic yellow needles, which are anhydrous. 0.1624 gram of the salt, which did not lose weight at 120", gave on ignition 0.0394 gram platinum. Calculated for (C1,H,,N,,I€Cl),PtC14. Found. 24.1 5 Pt in 100 parts.. . . . . . . 24.22 Ciiznarnaldehyde, Phena?2.thrapzLino.ae, and Ammonia.-10 grams of phenanthraquinone with the molecular proportion of oinnamaldehyde were heated with an excess of alcoholic ammonia in a sealed tube at12 ACTIOX' OF ALDEHYDES AND AMMONIA ON a-DIKETONES. 100". On cooling, a substance was deposited in a solid state; this was separated by filtration and dissolved in hot glacial acetic acid, from which it crystallised in slender, yellow needles: these, after recrystallising twice from the same solvent, melted constantly at 171-172". The compound is also readily soluble in benzene and in carbon bisulphide. The results of analysis pointed to the formula C,3H,a0, which is that of a cinnnnzeny ldiphenyleneoxazole. It does not form salts with acids. S Ltbstance. cop H20. I ........ 0.1220 0.3844 0 * 05 32 Substancc. V. t . P- N + NO.. 9.02 C.C. 20" 408 mm. I1 .... 0.1259 { ........ 9.02 ,, 20 405 ,, Calculated for Found. - C,, ......... 276 85.98 85.92 H15. ........ 15 4.67 4.84 - N .......... 14 4.36 - 4.48 0 .......... 16 4.98 321 99.99 - - -- -- ATowncrfl Xchool of Science, fiouth A-ensiny t on.

ISSN:0368-1645    

DOI:10.1039/CT8905700008

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

13 IV.-Action of Dehydrating Agents on aw-Diacety lpentane. Synthesis of ikIethylethylhexamethylene. By F. STANLEY KIPeIxG, Ph.D., D.Sc., and W. H. PEKKIN, Jun., Ph.D. IN a paper which we had the honour of communicating to the Scciety some time ago, we described the preparation and properties of am-diacetylpantane and of aw-dibenzoylpentane, and stated also that our object in preparing these diketones was to study their behaviour witb dehydrating and reducing agents. It was thought possible that diketones of the constitution R*CO*CH2*[ CH,],*CH,*CO*R might, under the influence of de- hydrating agents, undergo internal condensation, and be converted into compounds having the const'itution a reaction which, to some extent, would be analogous to the formatioil of mesityl oxide from acetone.The study of the action of reducing agents on diacetyl- and dibenzoyl-pentane also possessed great interest, as it was considered probable that in this niaiiner dihydroxy-derivati ves of the con- stitution R*y( 0 H) - 7 (OH) OR CH2*[CH2]n*CH2 would be obtained. Both these reactions would, naturally, be con- trolled by the value of w. Our views regarding the probable course of these reactions have been confirmed experimentally, and in the present paper we beg t o lay before the Society the results which we have obtained in studying the action of dehydrating agents on diace tylpentane. When diacetylpentane is treated with concentrated sulphuric acid, it is converted into a colourless liquid of the composition C,H140. CgHiGOa - HZ0 = CsH140. Considerable difiiculty has been experienced in determining the con- stitution of this interesting substance, as, in accordance with the present views on the structure of organic compounds, a, substance derived from diacetylpentane by the elimination of one molecule of water might be represented by any one of the following constitutional formuls :- TOL.LVLI. c14 KIPPING AXD PERKIN: ACTION OF I. CH,*C i C [ CH,] *-C 0. CH, 11. C H i C . [ C H2] 5-C 0.C H, 111. CH,:C:CH*[CH,!,.CO.CH, IV. CH3*g-- 0 --C.CHs CH*[CH,l,k!H > VII. CH,-C=CH.CO VIII. CH,*f: :C( COGH,) CH,*[CH2],hH2 CHz- [CHz], I IX. CH3*Cz:- C.[CH,],’*CO*CH3 n = 1, 2, or 3) (n‘ = 3, 2, or 1) L[ CHz1n-I The results of experiments which bear directly on the question of the constitution of this compound CgHl40, briefly stated, are the following :- It undergoes no change when treated with 80 per cent.sulphuric acid, or when heated with a coiicentrated solution of mercuric chloride. It is not acted on by phosphoric anhydride in ethereal solution, It combines readily with hydroxglamine, yielding an oxime, C,II,,NO, and with phenylhydrazine, forming a hydrazone, CI5H,,N,. When treated with sodium in moist ethereal solution, it combines directly with 4 atonis of hydrogen, and is converted into an alcohol, C9Hl*O. The iodide from this alcohol is reduced when it, is heat,ed with fuming hydriodic acid and amorphous phosphorus, yielding a hydro- carbon, C9H1+ aw-Diacetyl-aw-diethylpentane, COMe-CHEtf CH,],.CHEt-COMe, is not acted on by cold concentrated sulphuric acid.m-Dibenzoylpentane, COPh*[ C€12]5.COPh, when warmed with phosphoric anhydride, is converted into a cornpound of the composi- tion ClgHl,O. From these facts, the constitution of the compound C,H,,O may be argued with a considerable degree of certainty. Its negative behaviour with 80 per cent. suiphuric acid and with mercuric chloride is strong evidence against the constitution I o r 11, because substances of this nature (derivatives of acetylene), under the same conditions as those employed, almost invariably combine with 1 mol. of water, and are converted into saturated compounds. The fact that the compound C9H140 is not acted on by phosphoric anhydride goes to show that the first three formulae are all improbable, as a compound possessing the constitution I, 11, or I11 would pro- bably be converted into an unsaturated hydrocarbon CgHlz.The formation of the oxime and the hydrazone shows that it is a ketone, and cannot have the constitution represented by formula 1V.DEHPURATIKG AGENTS 0s aw-DIACETTLPENTANE. 15 A conclusive proof of the inadmissibility of the formulae I, 11, 111, and IV is afforded by the behaviour of the compound C,H,,O 011 redaction ; when treated with sodium in moist ethereal solution, and subsequently with fuming hydriodic acid, it is converted into a hydro- carbon C,H,,, whereas a substance represented by any one of these four formule would yield a normal paraffin C,H,,. It is, therefore, an easy task to prove conclusively that the tori- stitution of the compound CgH140 is not represented by any of the first four formulae, but it is much more dificult to distinguish between the remaining five.A compound having the constitution shown in V, TI, VIT, VIII, or TX would be expected to behave in most respects like the sub- stance CgH,,O, that is, to give an oxime and a hydrazone, to yield an alcohol CgH,,O on reduction, and when heated with hydriodic acid to be converted into a hydrocarbon CgEI8. The behaviour of aw-diacetyl-aw-diethylpentane with sulphuric acid affords, however, a means of deciding between these formulae. AE examination of the forrriulze V to I X shows clearly that the presence of a methylene-group in direct combination with the carbonyl-group is necessary to the formation of the compound CgH1,O, only if it possesses the constitution shown in V I I I ; if, on the other hand, it had the constitution T;, VI, VII, or IX, its formation could take place if one of the hydrogen-atoms in both these methglene-groups were replaced by a radicle siich as ethyl, or in other words, diacetyl- diethylpentane, COMe*CHEt*[CH',],*CHEt*COMe, should give a con- densation-product when treated with sulphuric acid just as readily as diacetylpentane.Supposing, for example, that the formula V I I represents the con- stitution of the compound CgHIIO, diacetyldiet hylpentane should yield an aualogous condensation-product of the constitution CH$Y CH-YO CHEt [CH,],*CHEt' If, however, the formula VIII represents its constitution, it is obvious that no analogous reaction could take place in the case of diacetyl- diethylpen tane.The results of two separate experiments have shown that diacetyl- diethylpentane does not undergo internal condensation when subjected to the action of sulphuric acid under exactly the same conditions as those employed in the preparation of the compound C,H,,O, and that not a trace of a compound corresponding with this anhydro-derivative is formed. This behaviour can only be explained by assuming that the con- c 216 RIPPING AND PERKIN: ACTION OF stitut,ion of the compound C,H140 is represented by formula VIII, and CH, that it is A'-orthomcthylacetyltetrahydrobenzene,* ? (A*-orthomethyltetrahydrobenzene methyl ketone). Dibenzoylpentane, COPh*[ CH2I5*COPh, when warmed with phos- phoric anhydride, is readily converted into a compound C,,Hl,O ; i f , as is probable, this compound sho-illd, on further investigation, prove to have a constitution analogous t o that, of the condensation-product of diacetylpentane, this fact would be additional evidence in favonr of our views regarding the mechanism of these internal condensations by proving the inadmissibility of the formula? 11, 111, VI, and VII.A s me are a t present engaged in examining this compound, and iri iiivestigating the behaviour of dibenzoylpentane with reducing arents, we hope to be able to communicate the results to thc Society at an early date. Ort hornet hy 1 tetrahydrobenzene Met h y 1 Ketone, Orthomethyltetrahydrobenzene methyl ketone (orthomethylacetyl- tetrahjdrobenzene) is formed by the action of dehydrating agents on aw-diacetylpentane ; it is best prepared by dissolving the diketone in concentrated sulphuric acid.Dincetylpentane (not more than 30 grams) is added in small portions at ft time to six or eight times its volume of concentrated sulphuric acid ; as considerable evolution of heat occurs, the solution is cooled under the tap after each addition of the diketone, otherwise the yield is decreased, The resulting solution, which varies in colour from brownish-red to dark brown, is kept in a loosely-stoppered flask a t the ordinary temperature from one to two days, during which time it darkens in colour, and a slight evolution of sulphurous anhydride occurs. The solution is then poured into a large volume of cold water, the greenish-yellow oil which is precipitated is extracted by shaking two or three times with ether, and the ethereal extract, after having been dried over potassium carbonate, is evaporated.A yellow, mobile oil, smelling strongly of peppermint, is obtained ; the yield of the crude product is, on the average, 85-90 per cent. of the diketone employed. When this oil is distilled under the ordinary pressure, a small quantity passes over between 120" and 190", but the principal * Compare Bacysr. " Tieber die Constitution des Benzols" (Annalen, 245, 111).DEHYDRATISG AGENTS ON GXU-DIACETYLPEXTANE. 17 portion distils between 190" and 210"; if the distillation is discon- tinued as soon as the thermometer rises above 210--8L5", a small quantity of a thick, brown oil, which will be referred to below, remains in the flask. The fraction 190-210" still contains traces of unchanged diketone ; pure orthomethyltetrahydrobenzene methyl ketone, boiling constantly at 205-206", can only be obtained by again treating this produch with sulphuric acid and fractionating two or three times under the ordinary pressure.The analysis gave the following results :- 0.1855 gram of substance gave 0.5294 gram C02 and 0.1720 gram H,O. Calculated for C,H,,O. Found. C ........ 78.26 per cent. 77.84 per cent. H.. 10.14 ,, 10.30 ,, 0 11.60 ,, 11-86 ,,. ...... ........ OrthomethyltetrahFdrobmzene methyl ketone is a colourless, mobile oil with a strong odour of peppermint, and a sharp, somewhat bitter taste. It boils at 205-206" (about 755 mm.) without decom- position, and is readily volatile with steam. It is insoluble in and specifically lighter than water, but it is miscible with the ordinary organic solverits in all proportions.It does not combine with sodium hydrogen sulphite even when kept for a long- time with a saturated aqueous solution of the salt, but it combines readily with hydroxyl- ainine. When treated with excess of sodium in moist ethereal solution, it is converted into orthomethylhexamethylene methyl carbinol. In order t o determine the constitution of this compound, C9H,A0, i t was necessary, in the first place, to ascertain whether it contained ;t " treble linking ; " the following experiments were made with thi< object :- A small quantity of the pure compound was dissolved in 80 per cent. sulphuric acid, and the solution kept for a few days a t the ordinary temperature ; it was then diluted with water, warmed for about half an hour on the water-bath, and after adding a, large volume of water, extracted with ether.The boiling point and other properties of the oil which was obtained showed that it co:isisted entirely of the original substance, no reaction having taken place. About 4 grams of the pure compound was mixed with a warm, concentrated, aqueous solution of mercuric chloride, and the mixture heated a t 95-98' for three hours with constant shaking. A small quantity of the mercuric salt seemed to be reduced, but otherwise DO reaction was observed. The mixture was then warmed with a little18 KIPi'INQ AND PERKIK: ACTIOI'; OF concentrated hydrochloric acid, the oil extracted with ether, a,nd distilled. The whole passed over within a range of a few degrees, and consisted of the original substance.These experiments having shown that the compound C9H,,0 does not combine with 1 rnol. of water when submitted to the treatment usually employed in the case of substances containing trebly-bound carbon-atoms, it seemed interesting to ascertain, whether, under the influence of dehydrating agents more powerful than concentrated sulphuric acid, it would be converted into a hydrocarbon C,H,,. The formation of such a compound would have shown that the original oil was most probably an open-chain derivative of the constitution given in one of the first three formulae (see p. 14). About 10 grams of the pura ketone was dissolved in about 100 C.C. of pure, dry ether, and a consideyable quantity of phosphoric anhydride added to the solution ; no reaction was observed even after keeping for about 24 hours a t the ordinary temperature.The ethereal solution was filtered and evaporated, arid the residual oil distilled under the ordinary pressure. It commenced t o boil at about 193", and almost the whole passed over between 2U3 and 206O, traces of a thick, yellowish oil, probably a condensation-product, remaining in the flask. The fraction 203-206" consisted of unchanged ketone, ::s was shown by its boiling point and other properties. ACTION OF HYDROXYLAMINE AND OF PHENYLHYDRAZINE ON ORTHO- METHYLTETRAHYDROBENZENE METHYL KETONE. The compound produced by the action of sulphuric acid on a w - diacetylpentane combines readily with hydroxylamine, yielding a colouriess monoxime ; it also formu a hydrazone-derivative when i t is heated with phenylhydrazine.Orthomethyltetrnliydrobenzene niethyl ketone-oxime is best pre- pared by treating the ketone with hydroxylamine by AUWWS' method. About 15 grams of potash are dissolved in dilute alcohol, 5 grams of hydroxylamine hydrochloride added t o the solution, and 10 grams of the ketone poured into the mixtime. After being kept for about 24 hours, the whole is warmed gently 011 the water-bath until most of the alcohol has evaporated ; i t is then cooled, diluted with water, and slightly acidified with dilute hydrochloric acid. The yellowish oil which is precipitated is extracted with ether, the ethereal extract dried over calcium chloride, and evaporated.I.)EHPDRATING AGENTS ON C~W-DIACETYLPENTANE.19 The oxime remains as a yellowish oil which, after being kept over sulphuric acid under reduced pressure, is obtaiiied in an almost pure condition, as is shown by the following analyses made with different samples of the crudo product :- I. 0.1640 gram gave 13.6 C.C. of nitrogen measured at 742 mm. 11. 0.1560 gram gave 12.3 C.C. of nitrogen measured a t 740 mm. and 20". and 20". Found. Calculated for r-h- -3 C,H,,NO. I. 11. N . . . . . . . . 9-15 per cent. 9-23 8.86 per cent. It is a thick, yellowish liquid, with a rather pleasant odour, ant1 shows no signs of crptallising even when kept for a long time ovcr sulphuric acid. It is insoluble or very sparingly soluble in cold, and only very slightly soluble in warm water, but it dissolves freely in potash and concentrated mineral acids, and it is miscible with alcohol, ether, &c., in all proportions.It dissolves in concentrated sulphuric acid with development of heat, yielding a clear yellow solution ; on adding alkali to the diluted solution, no oil is precipitated, but on boiling the alkaline solution ammonia is evolved. It cannot be dis- tilled under the ordinary pressure, as when heated, even i n small quantities, it is completely decomposed. Beckmann (Berichte, 20, 2580) has shown that some oximes, under the influence of acetic chloride, glacial acetic acid, &c., undergo an intramolecular change, the group of at,otns C:N*O H being converted into CO*NH ; the compounds thus produced are readily decomposed on hydrolysis into an amine and an acid.Considerable qnantiiies of the oxime of orthomethyltetrahydro- benzene methyl ketone were prepared in order to study its behaviour wit,h Beckmann's reagents, as it mas thought that in this way the constitution of- the ketone might be ascertained. For this purpose, 7 grams of the oxime was dissolved in 70 grams of glacial acetic acid and 14 grams of acetic anhydride, and the mixture, after having been saturated in the cold with hydrogen chloride, was heated a t 100" for four or five hours in sealed tubes. At the end of this time, the solution had turned dark reddish-brown, but no separation of crystals occurred. On iieutralising with sodium carbonate, a strong basic odour was perceptible, and a brownish- black oil separated a t the surface of the solution.The oil was ex- tracted with ether, and the ethereal extract dried and evaporated, when 5 grams of a dark-coloured oil remained ; this was insoluble in water arid in cold hydrochloric acid, but when heated with hydrochloric20 RIPPING AND PERKIN: ACTION OF acid it was almost completely decomposed, yielding large quantities of resinous products and a small quantity of a yellow neutral oil which was not investigated very closely. The sodium carbonate solution from which the oil had been extracted was rendered strongly alkaline with potash, and boiled for two hours, the distillate being collected in dilute hydrochloric acid. On adding platinic chloride to the dilute acid solution, a precipitate was produced, but this precipitate consisted of pure ammonium platinochloride, as was shown by its characteristic crystalline form, and also by the following analysis :- 0.3865 gram of the platinochloride gave 0 1604 gram of platinum.(NH3)2H2PtClp Found. Calculated for Pt.. . . . . . . 43.9 per cent. 43-83 per cent. The experiment was repeated under similar conditions, but with the same results ; a small quantity of a neutral oil was obtained, and the only basic product which could be isolated was ammonia. When the oxime is treated with acetic chloride at the ordinary temperature, an extremely violent reaction occurs, and a dark-red solution is obtained. I f the excess of acetic chloride is evaporated, and the residue is heated for a long time on the water-bath, it, darkens in colour, and finally becomes black and very thick.A small quantity of a colourless crystalline compound can be isolated from this residue; this substauce is insoluble in water, and only very sparingly soluble in ether; it does not give any basic odour when boiled with potash. It was not investigated, owing to the very small quantity at our disposal. Orthomethyltetrahydrobenxerze Methyl Ketone-hydrazone, The hydrazone-derivative of the ketone is best prepared by heating pure orthomethyltetrahydrobenzene methyl ketone with a slight excess of phenylhydrazine for two to three hours on the water-bath. The product is purified by dissolving it in ether, and washing the et;hereal solution with dilute hydrochloric acid until it is free from phenylhydrazine. The solution is then dried over calcium chloride, the ether evaporated, and the residue kept, for 24 hours over sixlphnric acid under reduced pressure.The analysis gave the following results :- 0.2218 gram of substance gave 24-95 C.C. of nitrogen measured at 18' and 740 mm.DEHPDRATISG AGENTS ON W-DIACETY LPENTASE. 21 Calculated for C15HaoN2. Found . N . . . . . . . . 12.3 per cent. 12.3 per cent. It is a thick, yellowish-brown oil, which shows no signs of crystal- lising ; wben left exposed to the air it gradually darkens in colour, and decomposes with evolution of gas. It is insoluble or only very sparingly soluble in water, but readily soluble in alcohol and ether. REDUCTION OF ORTHOMETHYLTETBAIXYDROBENZENE METHYL KETONE. Methyltetrahydrobenzene met'hyl ketone yields three compounds when it is reduced with sodium in moist ethereal solution.When a large excess of sodium is employed, i t is converted principally into orthomethylhexamethylene methyl carbinol, small quantities of a condensation-product being also formed. If only a slight excess of sodium is used, orthomethyltetrahydrobenzene methyl carbinol seems to be the principal product of the reaction. Methylhexamethylene Methyl Carbinol, Hz'c €IMe CH* C HMe.0 H. C H z < ~ ~ , - CH,> Orthomethylhexamethylene methyl carbinol is prepared by treating pure methyltetrahydrobenzene methyl ketone with sodium in moist ethereal solution. The ketone is dissolved in about twenty times its volume of pure ether, the solution placed in a large bottle containing about 60 C.C. of water and provided with a reflux condenser, and small quantities of sodium are introduced from time to time.As soon as about 20 times the quantity of sodium theoretically necessary for complete reduction has been added, the ethereal solution is separated, and the residue extmcted twice with ether. The ethereal solutions are mixed together, dried over potassium carbonate, filtered, and evaporated. A moderately thick, slightly yellow oil remains, the quantity of which is approximately equal to that of the ketone employed. On distilling it under 40 mm. pressure, the ther- mometer rises rapidly to 120°, and between this temperature and 280" tohe whole passes over. The fraction 120-280" is now repeatedly refractionated under reduced pressure (40 mm.), and is thus separated into a portion boiling at 127-1337', which consists O F slightly impure methylhexamethylene methyl carbinol, and a portioii boiling a t about 250-260".The higher boiling compound is only formed in comparatively small quantities, and will be again re- ferred to. The principal fraction, boiling at 127-137" (40 mm.), is again22 RIPPING AND PERKIN: ACTION OF submitted to fractional distillation under reduced pressure, and finally under the ordinary pressure, the methylhexamethylene methyl carbinol being thus obtained in a pure state. The analyses gave the following results :- I. 0.1160 gram of substance gave 0.3221 gram of CO, and 0.1325 11. 0.1494 gram of substance gave 0.4150 gram of CO, and 0.1685 gram of H,O. gram of H,O. Found. -7 Calculated for r,-L Cg HISO. I. 11. C ........ 76.06 per cent. 75.71 75.72 per cent.0 ........ 11.27 ,, 11.70 11-75 ,, H ........ 12-66 ,, 12.59 12-53 ,, Methjlhexametbylene methyl carbinol is a thick, colourless oil with a smell very like that of menthol; it boils a t 195-200" under t,he ordinary atmospheric pressure, and at 130-133" a t 40 mm. without decomposition. It is only sparingly soluble in water, but miscible with alcohol, ether, &c., in all proportions. It dissolves in concentrated hydriodic acid with development of heat, and in a. very short time the iodide separates from the solution. Methylhexamethylene Meth y 1 Carbinyl Acetate, CH,*COO-CHMe.CH<cH2- CHMe.CH,> CH, C'& Methylhexamethylene methyl carbinyl acetate is formed by treat- ing methylhexamethylene methyl carbinol with acetic anhydride. The pure alcohol is boiled €or 2 to 3 hours with excess of acetic anhy- dride in a, flask provided with a reflux condenser; the excess of anhydride is then distilled off, and the residue is fractionated under the ordinary pressure.The thermometer rises rapidly to about 193', almost the whole distilling between 200 and 210". By fractionating the distillate two or three times, the acetate is obtained in a pure condition, as is shown by the following analysis :- 0.1541 gram of substance gave 0.4058 gram of CO, and 0.1567 gram of water. Calculated for GlH2002. Pound. C ........ 71.74 per cent. 71-75 per cent. H ........ 10.87 ,, 11-29 ,, 0 . . ...... 17.33 ,, 16.96 ,, It is a colourless, mobile oil with a sweet rather pleasant odour,DXHYDRATINQ AGENTS ON aw-DIACETYLPENTAKE.23 and boils at 204-208" under the ordinary pressure without decom- position. It is seemingly insoluble or only very sparingly soluble in wa,ter, but it is miscible with ether, alcohol, &c., in all proportions. When boiled with alcoholic potash, it is readily hydrolysed, being reconverted into the alcohol. a- Iodoet h y lmet h y 1 hex nineth y 1 ene, C H, < g$E5z > C Ha CHMeT . Methylhexamethylene methyl carbinol dissolves in conceni~ated hydriodic acid, a considerable rise of temperature taking place, and after a very short time the iodide separates from the solution i n the form of an oil. This compound is best prepared by heating the alcohol with excess of concentrated hydriodic acid (sp. gr. 1.36) a t 180" in a sealed tube for about 4 hours.After diluting with water, the precipitated iodide is extracted with ether, the solution washed, first with water and then with very dilute sodium carbonate solution, until free from acid, then dried over calcium chloride, and evapo- rated, The residual oil can be easily obtained in a pure condition by submitting i t to fractional distillatioii under reduced pressure. An iodine determination by Carius' method gave the following result :- 0.2420 gram of substance gave 0.2270 gram of silver iodide. Calculated for CSH,;I. Found. 50.29 per cent. I . . , . . . . . 50.68 per cent. Iodoethylmethylhexamethylene is a colourless mobile oil with an d o u r very like that of the iodides of the higher norma1 alcohols. I t boils a t 178-180" (110 mm.), seemingly without decomposition, and is insoluble in water, but readily soluble in the ordinary neutral organic sol vents.It rapidly darkens 011 exposure to light. The fraction boiling at 240-260" (40 mm.), obtained in small quantities in the preparation of methylliexarnethyleiie methyl carbi- nol, was distilled again under reduced pressure (50 mrn.), and the port,ion passing over between 255-265" was coliected separately. The analysis of this product gave results agreeing with the composi- tion C,,H3202. 02366 gram of substance gave 0 6655 gram of CO, and 0.2454 gram of H20. Calculated for C18H3202. Found. C . . . . . . . . H . . . . . . . . 11 -43 ,? 11-52 ,, i7.14 per cent. 76.70 per cent.24 KIPPING AND PERRIN: ACTION OF It is a very thick, yellowish oil with it peculiar aromhtic odour, in- soluble in water but miscible with alcohol, ether, &c., in all propor tions.No experiments have yet been made to ascertain the Constitution of this compound, but it is most probably formed by the condensn- tion of 1 mol. of orthomethyltetrahydrobenzene methyl ketone with 1 mol. of methylhexamethylene methyl ketone in a manner somewhat analogous to that in :vhich pinacone is formed by the reduction of acetone, so that its constitution may be expressed by the formula- n~eth?lltetrahydrobenxene Methyl Carbinol, CH,< ~ ~ ; : , C ~ ~ > C - C H M ~ - O H . OrthomethyltetJrahydrobenzene methyl carbinol is formed, as has been already stated, together with methylhexamethylene methyl carbinol when the ketone is reduced with a slight excess of the theo- retical quantity of sodium in moist ethereal solution.It is obtained as follows :-Methyltetrahydrobenzene methyl ketone is dissolved in pure ether and treated with sodium under the conditions describeil in the preparation of the hexamethylene compound. After adding a little more sodium than is tlieoretically necessary, the ethereal solu- tion is separated, dried over potassium carbonate, and evaporated. When the residual oil is carefully and repeatedly fractionated under reduced pressure (50 mm.), it is separated into two portions, the principal fraction, boiling at, 141-143", consisting of methgltetra- hydrobenzene inethyl carbinol, mixed possibly with small quantities of the hexamethylene-derivative and unchanged ketone. It is a colourless, mobile oil, boils a t 141-143" (50 mm.), and resembles methylhexamethylene methyl carbinol in its behaviour with solvents.The following results were obtained on analysis :- 0.1599 gram of substance gave 0.4509 gram of CO, and 0.1703 gram of H,O. CaIcnlated for C9H,,O. Found. 7i.14 per cent. C . . . . . . . . 0 ........ 11.43 ,, 11.27 ,, 76.90 per cent. H .. . . . . . . 11.43 ,, 11.83 ,,DEHTDKATIXG AGENTS ON ~w-UIACET PLPENTANE. 25 REDUCTIOK OF IfETHFLHEXAMETHYLENE NETHYL CARBINOL. Orthomethylethylhexamethylene is the final product of the reduc- tion of methyltetrahydrobenzene methyl ketone, and is obtained when inethylhexamethylene methyl carbinol is reduced with hjdriodic acid. Its preparation is carried out as follows :-About 12 grams of the pure alcohol is heated with excess of fuming hydriodic acid for 2 - 3 hours in a flask provided with a reflux condenser, and, after dilut- ing with water, the precipitated iodide is extracted with ether.The crude iodide, in quantities of about 3 grams, is heated in a sealed tnbe at 230-240" for 8 hours with a large excess of hydriodic acid of sp. gr. 1.96 and a little amorphous phosphorus. There is great pressure in the tube when cold, and the hydrocarbon forms a layer on the surface of the acid; this is separated from the acid, dried over potassium carboilate, and distilled. The thermometer rises rapidly to 120°, and the whole distils between 120 and 160". The crude product is then repeatedly fractionated over sodium and finally over potassium. The hydrocarbon is thus obtained in a pure state and on analysis gave the following results :- I.0.1211 gram of substance gave 0.3795 gram of CO, and 0.1590 11. 0.1100 gram of substance gave 0.3150 gram of GOz and 0.1434 gram of H,O. gram of HzO. Found. Calculated for r--h--7 C9H,** I. 11. C . . . . . . 85.7 per cent. 85.47 85-54 per cent. H.. . . . . 14-3 ,, 14.58 14-47 ,, A vapour-density determination made by Hof mann's method in 0.0502 gram of substance yielded 57.0 C.C. of vapour. Tempera- ture of vapour 184". Bar. 765 mm. Difference of level 582 mm. Temperature of air 15". the vapour of aniline gave the following results :- Calculated for CgHI8. Found. Molecular weight.. . . 126 125 Orthomethylethylhexamethylene is a colourless, very mobile oil, the odour of which cannot be distinguished from that of the liquid paraffins. It boils a t 150-152" and is insoluble in water and concen- trated hydriodic acid, but miscible with ether, alcohol, &c., in all pro-26 KIPPISG AND PERKIN: ACTION OF portions.When treated with cold concentrated nitric acid, oxidation quickly commences and the mixture becomes very hot. A small quantity of the hydrocarbon was treated with nitric acid in the cold ; as soon as the first energetic reaction was at an end, the mixture was heated on the water-bath in a flask provided with a reflux condenser. The hydrocarbon gradually disappeared, dense, brown fumes being evolved, and finally a homogeneous liquid was obtained. The solution remained clear on adding water, and on evaporation it yielded only a very small quantity of a thick, syrupy residue which was not investi- gated owing to the small amount of material at our disposal. CONDENSATION-PRODUCTS OF WJ-DIACETYL PENTANE.A small quantity of two condensation-products are obtained in the preparation of diacetylpentane by the hydrolysis of et,hyl diacetyl- caproate (Trans., 1889, 55, 333) ; these products remain in the form of a thick brown oil when the crnde diacetylpentane is distilled under reduced pressure. The residues obtained in this way from several operations were mixed together and submitted to fractional distilla- tion under a pressure of 35 mm. The thermometer rose rapidly to about 230", and between about 250 and 280" a considerable proportion of the oil distilled; the temperature then rose quickly to 31Uo, the remainder passing over between 310" and 340". 'I'he portion collected between 250 and 280" was fractionated two or three times under the same pressure, and in this way a yellow oil boiling a t 265-275" was obtained.The analysis of this compound gave the foIlowing results :- I, 0*1418 gram substance gave 0.4030 gram of CO, and 0,1329 11. 0.1118 gram of substance gave 0,3190 gram of CO, and 0.1047 gram of H20. gram of H20. Found. Calculated for r--- --7 CISH280n. I. 11. C ........ 78.26 per cent. 77.82 77.70 per cent. H.. ...... 10.14 ,, 10.40 10.30 ,, 0 ........ 11.60 ,, 11.78 12.00 ,, A molecular weight determination was carried out by Raonlt's Freezing point of the acetic acid 15,740". 1.4882 gram of substance dissolved in 67.25 grams of acetic acid lowers the freezing point 0.30°.method in glacial acetic acid solution and resulted as follows :-DEHYDRATING AGENTS ON CZW-DIACETYLPENTAXE. 27 Calculated for C18H,O,. Found. Molecular weight ........ 276 28 7 These results show that this condensation product is formed by the combination of 2 mols. of diacetylpentane with elimiiiatlion of 2 mols. of water ; as the diketone is readily converted into methyltetrahpdro- benzene methyl ketone with elimination of 1 mol. of water, the con- stitution of the compound Cl8HZ8O2 is most probably expressed by the formula- CH,<cHi CH *CM .cH:>C*CMe:CH*CO.[ CH,],*COMe. It is a yellowish, very thick oil, boiling at 265-235" (35 mm.) seemingly without decomposit'ion, and it shows no signs of crystal- lising even when kept f o r a long time in a desiccator under reduced pressure.It is miscible with alcohol, ether and other neutral solvents in all proportions, but it is insoluble in water. The other fraction, boiling at 310-340" (35 mm.), which is ob- tained from the residues in the preparation of diacetylpentane was likewise submitted to fractional distillation, and in this way a yellow oil boiling a t 320-330" can be isolated without much difficulty. This cornpound gave the following results on analysis :- 0.1988 gram of substance gave 0.5702 gram of CO, and 0.1828 gram of H,O. Calculated for C27H4.203. Found. C ........ 78.26 per cent. 78.22 per cent. H.. ...... 10.14 ,, 10.20 ,, 0 ........ 11.60 ,, 11.58 ,, It is a yellow oil with a slight aromatic odour, and resembles the preceding compound in its behaviour with solvents ; it is of an almost s ~ ~ u p y consistency, but shows no signs of crystallising when kept for a long time in a desiccator under reduced pressure.The higher boiling point of this compound indicates a more Corn- plicated constitution than that of the condensation-product described above, and it is probably formed by the combination of 3 mols. of diacetylpentane, with elimination of 3 mols. of water, according to the equation, 3C9Hls02 = C27H,,O, + 3HZO. ACTION OF DEHYDRATINU AGENTS ON ~WDIBENZOYLPENTANE. The investigation of the behaviour of dibenzoylpentane with de- hydrating and reducing agents has been commenced and has already28 ACTION OF DEHYDRATING AGENTS ON aw-DIACETYLPENTANE. led to interesting results, but the compounds obtained have not yet been examined very fully. A very short account of some of our experiments may, however, be suitably given in this paper on account of their bearing on the question of the constitution of methyltetrs- hydrobenzene methyl ketone. Concentrated sulphuric acid seems to have no dehydrating action on dibenzoylpentane in the cold, as is shown by the following experi- ment. About 0.5 gram of the pure diketone was dissolved in con- centrated suiphuric acid, and the solution kept a t the ordinary temperature for three days ; it gradually became violet in colour and a small quantity of sulphurous anhydride was evolved. The addition of water caused the precipitation of a colourless compound, which, after recry stallisation from dilute methyl alcohol, melted a t 66-67', and was found to consist of unchanged dibenzoylpentane; the quantity obtained was about 0.46 gram. When dibenzoylpentane is warmed with phosphoric anhydride it is converted into a compound having the composition CI9Hl8O, but cou- siderable quantities of resinous products are also formed. The compound C,,H180 crystallises from alcohol in beautiful, colour- less needles, melting a t 110-111". An analysis gave the following results :- 0.1187 gram oE substance gave 0,3787 gram of CO, and 0.0744 gram of H,O. Calculated for C,,Hl@. Found. C ........ 87-02 per cent. 86.84 per cent. 0 ........ 6.11 ,, 6.20 ,, H . . ...... 6.87 ,, 6.96 ,, This substance is at present under investigation, and will form the subject of a paper which we hope soon to be able to communicate to the Society. Heriot W a t t College, Edii.1Lbwgh.

ISSN:0368-1645    

DOI:10.1039/CT8905700013

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

29 V .-aw -Diacet y 1-a w -dieth y lpen tame. By F. STANLEY KIPPING, Ph.D., DSc., and W. H. PERKIN, jun., Ph.D. IN the preceding paper we have described the preparation of a com- pound of the composition C9H140, and have given an account of various experiments, from the results of which it seems to be proved that this substance is methyltletrahydrobenzene methyl ketone, In order to distinguish between several of the theoretically possible formulse for this compound (compare p. 14), it became necessary to prepare a diketone of the formula R*CO*CHR*[ CH,]3-CHR*CO*R, and to study its behaviour under the same conditions as those em- ployed in the preparation of the compound C9HI4O from aw-diacetyl- pentane, COMe*[CH2],*COMe. A short account of these experiments will, we hope, not be with- out some interest.When ethyl ethylacetoacetate (2 mols.) is treated with sodium ethoxide (2 mols.) and trimethylene bromide (1 mol.) in alcoholic solution, a moderately energetic reaction occurs on warming gently, and, after proceeding in the usual manner, a yellow oil is obtained. The crude product, which consists of several compounds, contains about 40-45 per cent. of ethyl aw-diacetyl .am-diethylpimelate, COOEtCEt( COMe)*[CH2]3.CEt(COMe)*COOEt, produced by the combination of 1 mol. of trimethylene bromide with 2 mols. of ethyl sodioethylacetoacetate in accordance with the equation- 2CH,*CO*CNaEt*COOEt + C3H,Br2 = COOEt*CEt(COMe)*[CH2]3*CEt(COMe)*COOEt + 2NaBr. This ethereal salt is readily hydrolysed when boiled with alcoholic potash, the principal products being aw-diacetyl-aw-diethylpentaue and w-acetyl-aw-diethylcaproic acid. The equations representing the decomposition are as follows :- C 0 OE t *CEt (C OMe).[ CH2] ,*CE t (C OMe)*C 00 E t + 4ROH = C0Me-C HEtf CH,),*CHEt*COMe + 2K2c03 + 2C2H6*OH, COOEt. CEt (COMe) . [ CH2I3* CE t (COMe)*CO OEt + 4KOH = COMe*CHEt*[CH2],*CHEt*CO0K + 2CzH5*OH + K,CO, + rtud CHs* GO OK. VOL. LVII. P30 RIPPING AND PERKIN : CW-DIACETYL-CW-DIETHYLPENTANE. It seems probable that, in addition to the two compounds referred fo above, small quantities of aw-diethylpimelic acid are also produced by the following reaction :- COOEt*CEt(COMe)*[CH2]3*CEt~(COMe).COOEt + 4KOH = COOK*CHEt*[CH~]~*CHEt.COOK + BC,H,*OH + BCH,*COOK. Action of Trimethylene Bromide on Ethyl Sodioethylacetoacetate. Formatiom of Ethyl aw-Diacet?iE-aw-dietk ylp irnelate, C 0 OE t C E t ( C OMe) [ CH,] s*CE t ( C 0 Me) C 0 0 E t .Ethyl diacetyldiethylpimelate is obtained, together with considerable quantities of compounds of lower boiling point, when ethyl sodio- ethylacetoacetate is treated with an alcoholic solution of trimethylene bromide. Ethyl ethylacetoacetate (2 mols.) is added, in small quantities at a time, to a solution of sodium (1 mol.) in about 12 times its weight of alcohol, the temperature being kept below 25-30" by cooling under the tap from time to time; aft.er adding the trimethylene bromide (1. mol.), the mixture is heated on the water-bath in a flask provided with a reflux condenser. As soon as the temperature rises to about 70-80", an energetic reaction commences, sodium bromide separates from the solution, and the heat developed is sufficient to keep the alcohol in ebullition for several minntes.After heating for 2-3 hours to complete the reaction, the alcohol is evaporated, the residue zixed with water, and the precipitated oil extracted with ether. When the ethereal extract is evaporated, there remains a yellow oil, the quantity of which is about equal to that of the ethyl ethylaceto- acetate employed. The crude product is distilled with steam as quickly as possible, the distillation being continued as long as oily drops collect i n the receiver. I n this way the ethyl diacetyldiethylpimelate, which forms from 40 to 45 per cent. of the whole, is separated from the more volatile compounds. The dark yellow oil which remains in the flask is extracted with ether, the ethereal solution dried over calcium chloride and evaporated.That this process of separation is very efficient is shown by the following analysis, which was niade with a portion of the residual oil after i t had been kept over sulphuric acid under reduced pressure for about 24 hours. 0.1474 gram of substance gave 0.3430 gram of CO, and 0.1220 gram of H,O.RIPPING AND PERKIN : DXU-DIACETYL-go-DIETHYLPENTANE. 31 Calculated for Cl9 H,?O, * Found. C ........ 64.05 per cent. 63.46 per cent. H.. ...... 8.99 ,, 9.22 ,, 0 ........ 26-96 ,, 27.32 ,, When the crude ethyl diacetyldiethylpimelate is distilled under reduced pressure (45-50 mm.), the thermometer rises rapidly to about 205", then a little more slowly to 235"; almost the whole distils between 235" and 255", a small quantity only passing over between 255' and 280".The oil boiling at 235 -255" was fractionated again under the same pressure, and the portion boiling constantly at 249-252" collected separately. This fraction quickly solidifies to it mass of colourless crystals, and after having been spread on a, porous plate to free it from traces of oily impurities, it melted at 42-43" with previous softening. The analyses of the crystalline compound purified in this way gave the following results :- I. 0.2109 gram of substance gave 0,4937 gram of CO, and 0.1689 11. 0-2500 gram of substance gave 0.5840 gram of CO, and 0.2016 gram of H,O. gram of H,O. Found. Calculated for r------7 C19H3205- I. 11.C ...... 64.05 per cent. 63.80 63.71 per cent. H . . .... 8-99 ,, 8-90 8.96 ,, 0 ...... 26.96 ,, 27.30 27.33 ,, Ethyl diacetyldiethylpimelate separates from ether and alcohol in long. very slender needles melting at 44-45" with slight previous softening ; it is very readily soluble both in ether and in alcohol, so that it cannot well be purified by crystallisation from these solvents. On adding water to a cold alcoholic solution, the ethereal salt is pre- cipitated in an oily condition, but it almost immediately solidifies to a mass of ill-defined crystals. It is very readiig soluble in benzene, light petroleum, xylene, chloroform, and most ordinary solvents, but it seems to be quite insoluble in cold water. Ferric chloride produces no coloration in alcoholic solutions.It is readily hydrolysed by warm alcoholic potash, the principal product being acetyldiethylcaproic acid. When treated with concentrated alcoholic potash in the cold, it is slowly decomposed ; the solution contains small quantities of diace tJyldiethyI- pentane, and the potassium salt of a thick, colourless acid, probably diacetyldiethylpimelic acid, which is mixed with some other acid richer in carbon. D 232 -PING AND PERKIN : #CU-DIACETYEC~UJ-DIETHYLPENTANE. Bye-products obtained in the preparation of Ethyl Diacetyldiethyl- pimelate. The crude product of the action of trimethylene bromide on ethyI sodioethylacetoacetate contains, as has been stated above, a consider- able proportion (55-60 per cent.) of an oil which can be separated from the ethyl diacetyldiethylpimelate by distilling with steam ; when the distillate is extracted with ether, and the extract dried over calcium chloride and evaporated, a yellowish, mobile oil with a pleasant ethereal odour is obtained.This oil is probably a mixture of unchanged ethyl ethylacetoacetate and ethyl ethylallylacetoacetate ; the last-named compound would probably be produced by the inter- action of the trimethylene bromide and ethyl sodioethylacetoacetate according to the equations- CH,*CO*CEtNa*COOEt + C3H6Brz = C H 3* C 0- C E t ( C 0 OE t ) C H2* CH2*C H2B r and C H3G 0 CE t (CO 0 E t ) C Hz*C H2*C HzBr + C H3* C 0. C E tNa. C 0 0 E t = CH3*CO*C14t(COOEt)*CH2*CH:CHz + CH,*CO*CHEt*COOEt + NaBr. It was repeatedly submitted to fractional distillation under the ordinary pressure in order to try and isolate the ethyl ethylallylaceto- acetate, but the attempt was unsuccessful.Various fractions were analysed, but in all cases the results agreed with those required by a mixture of ethyl ethylacetoacetate and ethyl ethylallylacetoacetate, the analyses giving from 64.4 to 65.3 per cent. of carbon and about 9.1 per cent. of hydrogen instead of 66.7 per cent. of carbon and 9.0 per cent. of hydrogen as required by ethyl ethylallylaceto- acetate. aw-DiacetyZ-aw-cliethylpentane, CEL3*CO*CHEt*[CHz],*CHEt*C0*CH, Diacetyldiethylpentane is best prepared by treating ethyl diacetyl- diethylpimelate with alcoholic potash as described in the preparation of diacetylpentane. The ethereal salt is dissolved in a small quantity of alcohol, and the solution heated to boiling in a flask provided with a reflux condenser ; about one-third of a hot, moderately concentrated, alco- holic solution of potash (4 mols.) is then added, and the mixture is boiled again.As soon as no further separation of potassium carbonate occurs, another third of the potash solution is poured in and the mixture is heated again for about 10 minutes ; the remainder of the alcoholic potash is then added and the alcohol is immediately evaporated. The alkaline residue is mixed with water, the precipitated diketoneKIPPING AND PERKIN : #CU-DIACETYL-~CU-DIETHYLPENTANE. 33 extracted with ether, the ethereal solution dried over potassium car- bonate and evaporated. The yellowish-brown, mobile oil which is thus obtained consists of impure diacetyldiethylpentane ; the quantity of the crude product is only 11-12 grams from 50 grams of the ethereal salt, the small yield being due to the formation of large quantities of acid decomposition-products.As the crude diketone showed no signs of crystallisation even when cooled below O", it was submitted to fractional distillation under a pressure of 110 mm. The thermometer rose rapidly to about 160": and about 10 per cent. of the whole distilled below 206" ; the thermo- meter then remained very constant at 207-208", about 50 per cent. passing over between 200" and 210°, and the remainder at a slightly higher temperature. The fraction 200-210" was distilled again under the same pressure and the portion boiling at 207-208" collected separately. This liquid on analysis gave the following results :- 0.1690 gram of substance gave 0.4533 gram of CO, and 0.1751 gram of H20.Calculated for C,,H,,O,* Found. C . . . . . . . . 0 . . . . . + . . 15.09 ,, 15.58 ,, 73.59 per cent. 73-10 per cent. H.. .... .. 11.32 ,, 11-32 ,, Diacetyldiethylpentane is a colourless, moderately mobile oil with a slight, rather pleasant, aromatic odour similar to that of diacetyl- pentane. It boils at 20'7-208" (110 mm.), and shows no signs of crystallising. It is insoluble or only very sparingly soluble in water, but miscible with alcohol, ether, and other ordinary solvents in all proportions. It does not combine with sodium hydrogen sulphite even whenleft f o r a long time in contact with a concentrated aqueous solution of the salt. It dissolves in cold concentrated sulphuric acid with a yellowish-brown coloration, but it undergoes no change even after two days' time, as is proved by the experiments described below.aw-Diacety I-aw-diethy lpentanedioxinae, NOH:CMe.CHEt* [CH,],*CHEt.CMe:NOH. Diacetyldiethylpentane combines readily with hydroxylamine in alkaline solution, yielding the dioxime. The diketone (about 1 gram) is dissolved in alcohol, treated with an alcoholic solution of hydroxylamine hydrochloride (1.5 gram) and potash (5 grams), and the mixture kept at the ordinary temperature for 24 hours ; the alcohol is then partially evaporated on the water-34 KIPPINO AND PEREIN : ~~-DIACETYL-~W-DIETHYLPENTANE. bath, the residue mixed with water and treated with a slight execss of dilute hydrochloric acid.The oil which is precipitated is extracted with ether, and after drying the extract over calcium chloride, the ether is evaporated. The dioxime remains in the form of a thick, yellowish oil which soon solidifies to a mass of crystals. After removing oily impurities by spreading the crystals on a porous plate, a nitrogen determination was made with the fol- lowing result :- 0.1184 gram of substance gave 22 C.C. of nitrogen measured at 12" and 755 mm. pressure. Calculated for C,,H?,N202. Found. N . . . . . . . . 11.6 per cent. 11.9 per cent. Diacetyldiethylpentanedioxime separates from a mixture of ben- zene and light petroleum in colourless, microscopic crystals melting a t 110-111". It is readily Roluble in alcohol, ether, acetic acid, benzene, and other ordinary solvents, but only very sparingly in cold light petroleum ; on adding light petroleum to the benzene solution, tlie dioxime is precipitated in a crystalline condition. It dissolves freely in alkalis and in moderately concentrated mineral acids. Behaviour of Diacetyldiethy~entane with Concentrated Sulphuric Acid.The chief object in view in preparing diacetyldiethylpentane was to study its behaviour with sulphuric acid. I t has been shown in the previous paper that diacetylpentane dissolves in cold concentrated sulphuric acid, and is thereby converted into a compound C,H,,O ; tlie various theoretically possible formula for this compound have been already discussed, and it has been shown that its constitution is probably that of an orthomethyltetrahydrobenzene methyl ketone.One of the chief arguments in support of this view is the fact that diacetyldiethylpentane does not yield an analogous product under the same conditions, and is, in fact, unchanged by concentrated sulphuric acid. About 8 grams of crude diacetyldiethylpentane was dissolved in about 60 grams of concentrated sulphuric acid ; a slight development of heat was observed and the solution gradually turned brown. After keeping for about 24 hours, the solution was poured into a large volume of cold water and the precipitated oil extracted with ether. A yellowish-brown oil remained when the ether was evaporated, and the weight of this product was about 6.5 grams. The crude oil was submitted to fractional distillation under the ordinary pressure ; the thermometer rose rapidly to 230", and only a small quantity distilled between 230" and 240", the greater oortion passing over at 245-255".mPPINQ AND PERKIN : ~CU-DIACETYL-~UJ-DIETHYLPENTANE. 35 This fraction was collected separately, and analysed without any further purification.0.2228 gram of substance gave 0.5863 gram of GO, and 0,2239 The following results were obtained :- gram of H,O. Calculated for r--- C13H220. CMH242. Found. C ...... 804 73.6 per cent. 71.7 per cent. H . . . . . . 11.3 11.3 ,) 11.2 ,, 0 ...... 11.3 15.1 ), 17.1 ,, Thia analysis shows that the fraction boiling at 245-255" consisted o F impure diacetyldiethylpentane, and proves conclusively the absence of any compound richer in carbon. Since this fraction formed about 60 per cent. of the crude product, and most of the remainder boiled at a temperature above 255") the experiment would show that t h e formation of a compound C13Hz20 does not take place under these conditions.In order to prove beyond doubt that such is really the case, the experiment was repeated in the following manner :-lo grams of almost pure diacetyldiethylpentane was dissolved in 300 grams of cold concentrated sulphuric acid and the solution kept for two days at the ordinary temperature; there was only a slight evolution of sulphurous anhydride, but the solution gradually darkened in colour. I t was poured into a large volume of cold water, the oil extracted with ether, the extract dried over potassium carbonate and eva- poratted ; the residual oil weighed about 8.8 grams.When the oil was distilled under the ordinary pressure, the thermometer rose at once to 215", and the whole passed over between 215-2;0", the greater portion boiling at about 250". The fraction 215-270", that is to say, the whole of the crude product, was distilled again under the ordinary pressure, and the portions passing over at 215-225" and 22.5-235" were collected separately. These two fractioiis were analysed with the following results :- I. Fraction boiling at 215-225" : 0.1568 gram of substance gave 11. Fraction boiling at 225-235" : 0.1654 gram of substance gave 0.4141 gram of CO, and 0.1617 gram of H,O. 0.4371 gram of CO, and 0.1733 gram of H,O. Calculated for Found. r--- 7 CnH220. C13H42. I. 11. C .... 80.4 p. c. 73.5 p. c. 72.06 76.5 p. c. H .... 11.3 ), 11.3 ..11.40 11.6 ,, 0 .... 11.3 ,, 15.2 ,, 16.64 15.9 ,,$6 RIPPING). AND PERRIN : LXW-DIACETYL-C~ W-DIETHYLPENTANE. These analyses show that both fractions consisted of impure di- acetgldiethylpentane, the purer of the two being naturally that of higher boiling point, as it approaches more closely that of the pure diketone. NOW, since, if any dehydrating action had occurred, the resulting compound must have been contained in these two lowest boiling portions of the crude oil, itl is clear that no compound of the composition C,,H,,O had been formed. This experiment, 'therefore, confirms the previous results and proves conclusively that diacetyldiethylpentane is not converted into a compound of the composition C13H230 under conditions which, in the case of diacetylpentane, result in the formation of a compound of the composition C9Hla0 ; this difference in behaviour can be explained only by assuming that the latter has the constitution of an ortho- methyltetrahydrobenzene methyl ketone, a view which is fully borne out by other experiments.w-AcetyZ-aw-diethzJZcu~roic Acid, CH&O*CHEt*[CH,],~CHEt*COOH. The alkaline mother-liquors from which the diacetyldiethylpentane has been extracted with ether contain the potassium salts of acetyl- diethylcaproic acid, acetic acid, and probably also diethylpimelic acid ; these compounds having been produced by the hydrolysis of ethyl diacetyldiethylpimelate in accordance with the equations given on pp. 29 and 30. On acidifying wikh dilute sulphuric acid, a thick, yellowish oil is precipitated ; the solution is shaken three or four times with ether, the ethereal extract dried over calcium chloride, and the etherevaporated. AS the thick, sour-smellingoil which remained showed no signs of crystallising even when cooled below 0", it was submitted t o fractional distillation under reduced pressure (110 mm.).The acetic acid which passed over at the cornmencement of the distillation was readily identified by its odour and by converting it into ethyl acetate ; the thermometer then rose rapidly to 200", a small quantity only distilled below 240", and the whole of the remainder passed over between 240" and 280". The portion boiling at4 240-280" was fractionated again under a pressure of 90 mm., and a fairly large quantity, boiling at 253-255", collected separately.The analysis of this fraction (b. p. 253-255") showed that it con- sisted of acetyldiethylcaproic acid : 0.1420 gram of substance gare 0.3490 gram of CO, and 0.1305 gram of HzO. Calculated for C12H2203. Found. C . . . . . . . . 0 .. .. .. .. 28.43 ,, 22.76 ,, 67.28 per cent. 67.03 per cent. H . . .. .. .. 10.28 ,, 10.21 ,,KIPPING AND PERKIN : 2~-DIACETYL-aw-DIETHYLPENTANE. 37 Acetyldiethylcaproic acid is a thick, colourless liquid having an odour very like that of pyruvic acid; it becomes semi-solid when cooled strongly, but it has not yet been obtained in a crystalline condition. It is insoluble, or only very sparingly soluble, in water, but miscible with alcohol, ether, benzene, and other ordinary solvents in all proportions. The silver salt, CH,*CO*CHEt*(CH,),-CHEtGOOAg, was pre- pared by dissolving a portion of the fraction boiling at 253-255" in ammonia, and fractionally precipitating the neutral solution with silver nitrate.It is a colourless, seemingly amorphous compound, and rather sparingly soluble in hot water ; it seems to be stable in the light. A silver determination, made with a porbion of the second fraction of the salt, gave the following result :- 0.4714 gram of substance gave 0.1605 gram of' silver. Calculated for C1,H,,O,Ag. Found. Ag . . . . . . . 33.64 per cent. 34.03 per cent,. The first fraction gave a somewhat larger percentage of silver probably owing to the presence of small quantities of diethylpimelic acid. I n neutral aqueous solutions of the ammonium salt, lead acetate and mercuric chloride produce a colourless precipitate ; ferric chloride, a buff-coloured precipitate ; and copper sulphate, a dark-green pre- cipi tate. The calcium salt and the barium salt are readily soluble in water. w- Acety I-aw-diethy lcaproic Acid Oxinze, NOH:CMe.CHEt.[ CH2I3*CHEt*COOH. Acetyldiethylcaproic acid oxime can be obtained by treating the acid with hydroxylamine hydrochloride and excess of potash in dilute alcoholic solution ; after keeping for about 24 hours, the alcohol is evaporated on the water-bath, the residue dissolved in water, the solution slightly acidified with hydrochloric acid, and extracted with ether. The crude product is a thick, yellowish oil, but it slowly solidifies to a mass of crystals ; the crystalline substance, after having been spread on a porous plate and washed with a little light petr- oleum, melted at 102-103". A nitrogen determination gave the following result :- 0.0961 gram of substance gave 5.05 C.C. of nitrogen measured at 14' and 745 mm. pressure.38 THORPE AND ROBINSON : FRANGULIN. Calculated for 6.11 per cent. CI2HaNO3. Found. N.. . . . . . . 6.12 per cent. Acetyldiethylcaproic acid oxime crystallises from a mixture of benzene and light petroleum in colourless, microscopic plates ; it is readily soluble in alcohol and benzene, but only sparingly in light petroieum, and, seemingly, insoluble in water ; it dissolves freely in alkalis and in concentrated hydrochloric acid. Heriott- Watt College, Edinburgh, Chemical Laboratory.

ISSN:0368-1645    

DOI:10.1039/CT8905700029

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

38 THORPE AND ROBINSON : FRANGULIN. V I.-Frangu lin. By T. E. THORPE, F.R.S., and H. H. ROBINSON, M.A. ‘rHE constituents of the bark of the Alder Buckthorn (Rhamnus /i-angula), and more especially the yellow colouring matters that can be got from it, have occupied the attention of several chemists since the beginning of the century. Prior t o 1857, no ultimate analyses of these colouring matters were made, although the substances obtainable from the bark had been frequently described. Gerber, in 1828 (Brande’s Archiv. f. Pharrn., 26, l ) , published a proximate ana- lysis of the bark, and among the substances enumerated was a yellow resinous colouring matter, which with alkalies gave a dark-red colour. Binswanger, in 1850, published a work on the constituents of the bark (Pharm.Studien uber R. frangula und cathartica. Miinchen; also in Buchner’s Repertorium f. Pharm. [3], 4, 47) ; among these he mentions a yellow colouring matter which he states Buchner had examined and named Rlzarnno-Xanthin. Buchner’s work was not published until some years later, namely, in 1853 (Annulen, 87,218) ; he had noticed that the paper in which some of the root-bark had been wrapped was coloured yellow, and that the inner surface of the bark was covered with golden-yellow crystals. He prepared some of the substance, examined its properties, and described its points of resemblance and dissimilarity to chrysophanic acid. In a later paper (Zeitschrifi &fur Chem., 1865, 699), he described some yellowish-red, needle-shaped crystals he got from Rhamno-Xanthin by sublimation, but gave no analyses.Winckler (Archiv. der Pharm., 1856, 335), Phipson (Compt. rend., 47, 153), Salm Horstmar (Pogg. Ann., 109, 539), and Enz (Viertel-THORPE AND ROBINSON : FRANGULIN. 39 jahrschrift fiir Pharrn., 16, 106) have also treated of these yellow colouring matters. Casselmann, in 1857 (Annulen, 104,77), was the first to publish an ultimate analysis of a yellow colouring matter got from 3. frangula. This substance he named Fr angulin, considering that Buchner's name Rhamno-Xanthinwas liable to be confused with Xantho-rhamnin, a substance Kane had obtained from R. cathartica and tinctoriu. He used ammonia-water to extract it from the bark, and describes it as forming lemon-yellow, crystalline masses with a slight silky lustre and as appearing under the microscope in bright-yellow, transparent, square tables.He made three analyses of frangulin prepared by different methods and dried at 100"; the mean of these analyses was, carbon 57.19 per cent. and hydrogen 4.98 per cent., from which he deduced the formula C6H603 (new notation). He examined the action of fuming nitric acid on frangnlin, and concluded that it yielded oxalic acid and a nitro- derivative ; this he analysed, together with its silver and copper salts, and assigned to it the formula C20H1004N02)5 + $H,O. Kubly, in 1866 (Phurm. Zeitschrift fiir Rusdand, 5, lSO), obtained a substance from the bark which he named A v o r n i n , and which he found to be a glucoside, yielding on hydrolysis, sugar, an acid which he called avornic acid, and an amorphous resin.Faust repeated his work and found that avornin was impure frangulin, but to Kubly is due the discovery of the glucosidal nature of frangnlin. Faust, in 1872, published a paper (Annulen, 165, 229) on frangulin and frangulic acid : he prepared a small quantity of frangulin, using alcohol to extract it from the bark ; he did not see the square tables mentioned by Casselmann, and found that its melting point was 226", also that when boiled with hydrochloric acid it split up into frangulic acid and sugar ; but he had not enough frangulin either to make a quantitative hydrolysis o r an analysis. He accepted Casselmann's analysis, and multiplying C6H603 by 3*, gave C20H20010 = Cl&O4 + C6HI2O6 as the equation representing its hydrolysis.He prepared frangulic acid directly from the bark, using boiling caustic soda solution; he found that on drying this at 120" it lost 6.07 per cent. in one case and 6.4 per cent. in another, and when thus dried gave carbon 67.26 per cent. and hydrogen 4 per cent. as a mean of five analyses : but in these the carbon varied from 66.78 to 67.91 per cent. He assigned the formula 2C14H804,3H20 to the undried substance and 2C14H804,H20 to it when dried at 120'. He next dried a sample at 180°, when i t lost 10.97 per cent. (2C14H804,3H20 in losing 3H20 would give up 10.11 per cent.), and on analysing the product thus dried he got carbon 70.33 per cent. and hydrogen 3.67 per cent, He gives the melting point as about 249".40 THORPE AND ROBINSON : FRANGULIN. (C,H,Oa would give carbon '70 per cent. and hydrogen 3.3 per cent.).He further prepared and analysed a bromine-derivative and an acetyvl- derivative of frangulic acid. By distilling frangulic acid with zinc-dust he obtained anthracene, and concluded that the acid was dihydrozyanthraquinone and therefore an isomer of alizarin. Liebermann and Waldsteiii were the next authors t o treat of the yellow colouring matter got from B. f r a n y u l a (Berichte, 1876, 17'75) ; they did not prepare frangulin, but examined a colouring matter got from the bark by means of caustic soda solution. They made five analyses of this; in two cases drying the substance at 190", and in one case subliming it, crystallisiiig the sublimate from alcohol and drying at 150". The mean oE their analyses was, carbon 67.1 per cent.and hydrogen 4 per cent., the carbon ranging from 66.7 to 67.2 per cent. ; they therefore assigned tlhe formula C,,H,,O, to the sub- stance. They prepared and analysed an acetyl-derivative and made a distillation of the colourinp matter with zinc-dust, obtaining a mixture of anthracene and methylanthracene, resembling that obtained from emodin and from chrysophanic acid, and giving on oxidation ccntliraquinonecal.bozzJ lie acid. They compared the properties of the colouring matter and its behaviour with various reagents with those of emodin, and they found the melting point of both to be 257". Finally, they concluded that the colouring matter they examined was identical with emodin. As regards Faust's frangulic acid, they con- sidered it to be probably a different substance, and that possibly it and emodin may occur jointly in the bark and mutually replace one another as alizarin and purpurin do in madder.They suggest C~~HzoOlo + H,O = C15H,o0, + C6H,,06 as the equation representing the hydrolysis of frangulin. Keussler, who published his paper in 1878 (Pharnz. Zeitschrift fiir Bassland, 17, 257), did not prepare frangulin, but used caustic soda and sodium carbonate in his operations on the bark, and obtained a substance having the same external appearances as the frangulic acid of Faust and the emodin of Liebermann and Waldstein. He came t,o the conclusion, however, that this was not emodin, but con- tained an additional CH, group, and was trihydroxyethyl- or t r i h y d r o q - dimethylanthraquinone.He made five analyses of it dried at 180", and obtained carbon 67-54 per cent. and hydrogen 4.38 per cent. as a, mean result, the carbon varying from 67.41 to 67.69 per cent. (CI6H1,O5 corresponds t o carbon 67.6 per cent. and hydrogen 4.2 per cent.). On distilling it with zinc-dust, he obtained and analysed a mixture of anthracene and methylanthracene, a considerable evolution of gas taking place during the distillation, possibly due to the decom- position of dimethylanthracene. He also prepared and analysed a nitro- derivakive and its silver salt.THORPE AND ROBINSON : FRANQULIN. 41 The only analysis of frangulin since that of Casselmann was made by Schwabe (Arch. der Pharm., 1888, 26, 569), who assigned to it the formula C2,Hzo09 ; he hydrolysed it quantitatively, and obtained from it a substance insoluble in water, which he analysed and found to be emodin, and a substance soluble in water which reduced Fehling’s solution.He suggests that the la.tter is the rhamnodulcite described by Liebermann, and that the equation for the hydrolysis is CzlH2,09 + H20 = C,,H,,05 + C6H1206. In previous communications one of us, in conjunction with Mr. T. H. Greenall (Trans., 1887, 51, 52), and subsequently in conjunction with Dr. W. T. Smith (Trans., 1888, 53, 171), has shown that morindin, the characteristic colouring matter of A’], the root-bark of Morinda citrifolia, also yields on hydro- lysis a trihydrosyniethylanthraquinone, CI5H1,,O5, which, however, seems not t o be identical with emodin. I n view, therefore, of the dis- crepancies in the statements of Faust, of Liebermann and Waldstein, and of Keussler, it became of interest to determine (1) what was the true composition of the frangulin, and what was the real nature of the products of its hydrolysis, and (2), if a trihydroxymethyl- anthraquinonc should turn out to be oneproduct of the hydrolysis, as Liebermann and Waldstein, and also Schwabe assert, whether this body was identical with morindon.Preparation of Frangzdim. The preparation of frangulin from the bark is a somewhat troublesome and tedious operation, for not only is it present in very small quantities, but it is accompanied by plant fat and resinous substances, from which it is not easily separated. The method we adopted was first to remove the plait fat by repeatedly extracting the crushed bark with low-boiling petroleum, which dissolves very little except the fat and chlorophyll, and then to extract the residue with methylated spirit, the extraction being continued until the liquid running off contained but little colouring matter.These extractions were performed in an apparatus, Fig. 1, designed on Soxhlet’s plan, but modified so as to suit the large quantities of material dealt with. It consisted of a cylindrical tin can, E, 10 inches in diameter and J 7 inches in height, provided with a removable conical top, C, which could be secured to the can by means of a stout brass flange which was pressed by means of a number of clamps to a similar flange attached to the top of the cylinder. A washer cut out of sheet indiarubber and placed between the two flanges kept the apparatus air-tight when the clamps were screwed up.The charge of bark, weighing 14 lbs., was contained in a muslin bag and rested on a false bottom made of thick tin-plate, perforated with holes and sup-42 THORPE AND ROBlNSON : FRANGULIN. ported by legs a t a height of about 1$ inches from the bottom of the can. In the side of the can and close to the bottom was a tin tubulure u, giving access to the space below the false bottom. Into this, by means of a cork, the end of a tube, s, was fitted which rose vertically outside the can to a point a little below the flanges and then bent over and acted as a syphon. To avoid too great rigidity, there was a junction in the tube a t b. At the top of the conical lid was a tubulure, into which a glass T-piece was fitted.The syphon led to a bottle-shaped copper vessel, R, of about four gallons capacity, in which the petroleum or methylated spirit was boiled. The vapour of the solvent was conducted by a wide-bent glass tube, t, up to the sido branch of the T-piece and through it into a reflux condenser which was joined to the upper extremity c of the T-picce. The glass tube t was wrapped round with strips of woollen cloth to prevent con- FIG. 1. densation occurring in it. By employing this apparatus, the same quantity of liquid was made to serve for numerous extractions, as it was syphoned off into the copper can as soon as the tin was full, and w a ~ then boiled off again, passed to the condenser., and returned as fresh solvent to the tin.Both the ext,racting tin and the copper can were surrounded by water-baths and could be heated to suitable temper- atures.THORPE AND ROBINSON : FRANQULIN. 43 The methylated-spirit extract on standing for two days deposited a considerable quantity of dark-brown resinous matter ; it was filtered, and to the filtrate a solution of lead acetate in methylated spirit was added in order to remove tannin, the brown precipitate thus pro- duced was filtered off, and through the filtrate a current of sulphu- retted hydrogen was passed to throw down fhe lead present. After separating the lead sulphide by filtration, the solution was put aside for five weeks, and a t the end of that time it was found to have de- posited some frangulin, which separates from such impure liquids in characteristic little orange-red spheres of some 4 mm.or so in diameter. This deposit was purified by heating with spirit, which dissolved some brown impurities and was poured off; the residue was hhen recrystallised from methylated spirit, of which a large quantity was required to effect solution. This preparation, which will be denoted by A for reference, amounted to 2.0 grams and was of an orange- yellow colour. FIQ. 2 . The liquid which had deposited the fraiigulin spheres was next evaporated to dryness with atidition of calcium sulphate, which was added in order to get a residue that could be submitted to extraction with ether ; it would otherwise have given a brown, pitch-like mass. The residue obtained was further mixed with fragments of pumice44 THORPE AND ROBINSON : FRANGULIN.to facilitate the ext~action, which was performed in a modified form of King's extractor, Fig. 2 (Chem. News, 1888, 57,235), constructed out of an inverted bell-glass, $7. This was closed at the bottom by a cork through which a vertical tube, s, passed. This tube was ground off at an angle at its upper extremity, and over it was inverted a somewhat wider tube closed at its upper end and reaching nearly but not quite down to the cork at the bottom. A plug of glass wool was placed on the cork, and then the residue mixed with fragments of pumice was packed in, taking care to have a column of pumice fragments round the centre tubes. The wide end of the bell-glass was closed with 8 large cork through which a glass T-tube passed.This cork was coated with plaster of Paris to prevent escape of ether. The lower end of the tube s was bent and passed for a short distance into the end of a slightly wider tube, p, and the joint was made good with a piece of indiarnbber tubing. The tube p led through the cork of a flask, R, in which the ether was heated. The vapours of the ether were conducted up a wide tube, t, to the side arm of the T-tube, and thence to a reflux condenser connected to the upper end of the T-tube. The condensed ether flowed into the bell-glass, and when it rose above the top of the tube s, it was syphoned back into the flask R again; thus the operation went on automatically, as in the case of the large extractor. Some of the extracts obtained deposited fairly pure frangulin ; others were so impure that but little frangulin could be got, from them.The best extracts were obtained when the amount of calcium sulphate added was sufficient to produce a residue that could be crumbled between the fingers. The frangulin thus obtained, together with about half a gram obtained in preparing A, but which wanted another purification, was recrystallised from alcohol and gave 2.7 grams (preparation B) . Besides the above two preparations, a little more frangulin was got from the calcium sulphate residue and from various mother-liquors, making the total quantity obtained from 14 lbs. of the bark about 5+ grams, a yield equivalent to 0.09 per cent. Frangulin is an orange-yellow powder which under the microscope has a crystalline appearance, best seen by reflected light; when in suspension in a liquid, or when a mass has dried in a cake, it often exhibits a silky lustre.It is not very soluble in hot alcohol and is still less soluble in cold, separating as the solution cools ; the cold alcoholic solution, however, has a distinctly yellow colour. The presence of a, trace of alkali imparts a red tinge to its solutions. It melts at about 225". Composition.-The preparations A and B were now analysed ; they had been placed over sulphuric acid in a vacuum to dry, and aTEORPE AND ROBINSON : FRANUULIN. 45 portion of each was weighed out i n a platinum boat and then dried at 120" until the weight became constant. I. 0,2099 gram of frangulin (preparation A), after drying at 120" until constant, became 0.2061 gram, a loss equivalent to 1.8 per cent., and then on combustion gave 0-4643 gram of CO, and 0.0997 gram of H,O. 11.0.1884 gram of frangulin (preparation B), after drying a t 120" until constant, became 0.1836 gram, a loss equivalent to 2.5 per cent., and then on combustion gave 0.4115 gram of GO, and 0.0862 gram of HzO. Percentage Composition of Fyangulin. I. Preparation A. C ......... 61.44 H ......... 5-37 11. B. 61.13 5.22 The agreement of the analyses of two independent preparations pointed t o the substance being homogeneous, and not a mixture, and it will be seen that further analyses confirmed this. Hydrolysis. A preliminary experiment showed that when frangulin in solu- tion in methylated spirit was boiled with the addition of aqueons hydrochloric acid, it was decomposed into a substance (which sub- sequently proved to be emodin) precipitated on diluting with water, and a substance which remained in solution and had the power of reducing Fehling's solution.Quantitative hydrolyses of preparations A and B were next made. Weighed quantities of A and B were dried a t 120" until constant in weight, and then dissolved in alcohol, 10 C.C. of strong aqueous hydrochloric acid added, and the solution boiled for about 3 or 4 hours, using R reflux condenser. A€ter the alcohol had been boiled off until the solution was reduced to a half or a third of its original bulk, it was largely diluted with water, which caused an orange-yellow precipitate to separate ; evaporation was continued for some time i n order to drive off: all alcohol, aud the precipitate was collected on a dried and weighed filter paper ; it was then washed with water and dried a t 120" until constant.The filtrate was almost colourless, and considering the strong tinctorial power of the preci- pitated substance, practically all of it must have been separated by the treatment adopted. I. 0.8457 gram of frangulin (preparation A), on drying a t 120" VOL. LYIJ. E:46 THORPE AND ROBINSON : FRANGULIN. until constant, lost 0.0131 gram (equivalent to 1.55 per cent.). The resulting 0.8326 gram of dry frangulin on hydrolysis gave 0.5710 of product (emodin) insolnble in water. IT. Some frangulin (preparation B), after drying at 120" until constant, weighed 0.7722 gram, and on hydrolysis gave 0.5146 gram of product (emodin) insoluble in water.I. IT. Preparation A. B. Mean. Yield of product (emodin) 68.6 p. c . 66-6 p. c. 67.6 Product of Hydrolysis insoluble in W a t e r (Emodin). The product of the hydrolysis thrown down on diluting with water was of an orange colour, and under the microscope proved to be a mass of interlacing needles ; it was much more soluble in alcohol than fraagulin ; with caustic soda solution, it gave a characteristic cherry-red colour . The products from the hydrolyses of preparations A and B were separately crystallised from alcohol, and combustions were made of them after drying a t 120" until constant in weight. I. 0*2200 gram of the product from the hydrolysis of frangulin (preparation A) after drying a t 120" until constant, became 0.2086, a loss equivalent to 5.18 per cent., and then on com- bustion gave 0.5076 gram of C02 and 0.0723 gram of H,O.11. 0.1508 gram of the product from the hydrolysis of frangulin (preparation B) after drying a t 120" until constant, became 0.1401 gram, a loss equivalent to 7.10 per cent., and then on combustion gave 0.3430 gram of GO, and 0.0500 gram of H,O. Some more frangulin was hydrolysed in the manner described above, but using a 10 per cent. solut8ion of sulphuric acid in larger volume instead of the hydrochloric acid. The insoluble product obt,ained was crystallised from benzene, and an attempt was made to sublime it in a vacuum a t a temperature of 240°, using a bath of diphenyl- amine. The sublimate, however, formed too slowly, so the substance was again crystallised from benzene, and used for making two analyses.It was first dried a t 180", but lost nothing by that treat- ment. 111. 0.2073 gram of the insoluble product, dried a t 180", gave on combustion 0.5000 gram of CO, and 0.0701 gram of H,O. IV. 0.2003 gram of the insoluble product,, dried a t 180", gave on combustion 0.4898 gram of COz and 0.0691 gram of H20.THORPE AND ROBINSON : FRANGULIN. 47 Percentage Composition of the Insoluble Product. Calculat,ed for I. 11. 111. LV. Emodin, Product from Preparation A. B. Mean. Cl5HI0O5. Carbon.. ........ 66.36 66.77 66.91 66.69 66-68 66-67 Hydrogen.. ...... 3-85 3',97 3-82 3.83 3.87 3.70 - - 29.63 - - - Oxygen. ......... 100~00 Thus the product of the hydrolysis has tbe same eomposition as emodin, as stated by Liebermann and Waldstein, and by Schwabe.The product obtained from frangulin was compared side by side with some emodin obtained from rhubarb, which Dr. Hugo Muller had kindly furnished, in respect to its colour reactions with strong sulphuric acid and with strong potash solution, and the same cherry- red colours were obtained from both specimens. Purther Analyses of Frangulin. As the anaIyses published by Schwabe differed notably from those given on p. 45, we subjected our preparations to further treatment to remove any possible impuritiesj a d then made other combustions, the results of which agreed with those first obtained. Preparation B was twice recrystallised from alcohol, and then gave the following results :- 111. 0.2133 gram of frangulin, after drying at 120" till constant, lost 0-0045 gram, equivalent to 2.11 per cent.; 0.2081 gram of the frangulin thus dried gave on combustion 0.4669 gram COz and 0.0994 gram H,O.IV. 0.2041 gram of frangulin, after drying at 120" nntil constant, became 0.1997 gram, a loss equivalent to 2.16 per cent., and then on combustion gave 0.4485 gram CO, and 0.0950 gram V. 0.2540 gram of frangulin, after drying at 120" until constant, became 0.2485 gram, a loss equivalent to 2.17 per cent., and then on Combustion gave 0.5567 gram of CO, and 0.1176 gram of water. HZO. Percentage Composition of Frangulin. (Preparation B twice recrystallised from alcohol.) C ........ 61.19 61.25 61-10 H ........ 5.31 5.29 5.26 111. IT. V. E 248 THORPE AND ROBINSON : FRANGULIN Another method of purifying it for analysis was next tried, and still the results proved to be the same as before.Portions of preparations A and B, and of B recrystallised, were mixed and warmed with ether, but not dissolved ; 'filtered and washed with ether; then crystallised from alcohol and washed, first with ether and then with alcohol. Two analyses were made of the frangulin thus prepared. VI. 0.1543 gram of frangulin, after drying at 120"'until constant, lost, 0.0033 gram, equivalent t o 2.14 per cent. 0.1501 of the frangulin thus dried gave on combustion 0.3366 gram of CO, and 0.0721 gram of E,O. VII. 0.1998 gram of frangulin, after drying at 120" until constant, lost 0.0055 gram, ,equivalent to 2.75 per cent. 0.1939 of the frangulin thus dried gave on combustion 0.4348 gram of CO, and 0.092G gram of H,O. (Preparations A and B mixed, treated with ether, and recrystallised.) VI.VII. C ................. 61-16 61.16 H . .............. 5-34 5.27 Collecting these analyses together and taking the mean, it will be seen that they are quite concordant, and agme with the formula C23H2209- Peroentage Cowyosition of Frangulin. Calculated for I. 11. 111. IT. V. VI. VII. Mean. C22H2209. 6C .. 61.44 61.13 61.19 61.25 61.10 61.16 61-16 61-20 61-40 H.. 5.37 5.22 5.31 5.29 5.26 5-34! 5-27 3.29 5.11 33-51 33.49 0 .. 100~00 100~00 _ - - - - - - -- Schwabe found, as a mean of five analyses, the figures given below and assigns to frangulin the formula C21H2009- Calculated for Mean. c,, 1 3 ~ ~ 0 ~ . C ........... 60.38 60.57 H ........... 5.32 4.80 0 ...........34.30 34.63 100~00 100~00 T t will be noticed that the mean of Schwabe's analyses differs con- The disparity may perhaps be siderably from that of our analyses.THORPE AND ROBINSON : FRAKGULIN. 49 duo to a difference in the way of drying ; our frnngulin was in every case dried a t 120" till constant, losing about 2 per cent. in the process. Schwabe remarks that, on drying, frangulin lost no water of crystallisation, but does not state a t whet temperahre he dried it. If our frangulin had contained 1$ per cent. of water when snalysed, the figures would have been carbon 6@28, and hydrogen 5.38, which are very close to those obtained by Schwabe, T h e Product of the Hydrolpds Xoluble irt W n t e r . In order to determine the nature of the product soruble in water obtained by the hydrolysis of frangulin, the filtrate from the emodin obtained in hydrolysis I was concentrated t o a small bulk, and i t was heated with the addition of sodium acetate, phenylliydrairine hydrochlo- ride, and water, and evaporated to dryness.The residue was treated with water, when a brown substance remained undissolved, and was collected; this proved to be insoluble in hot water, but dis- solved in alcohol; on diluting the alcoholic solution, an orange- yellow precipitate with a brownish tinge separated, this was collected and dried, and its melting point found to be about 158". Some pbenglglucosazone was prepared for comparison, and was found to be quite different from the nsazone of the produck from frangulin. The product i n question is undoubtedly not glucose. Owing to Bhe small quantities of frangulin hydrolysed, our attempts to isolate the product from the filtrates obtained in the other hydrolyses were unauccessf ul. Co~cZusioms.-The results arrived at are :- 1. That frangulin is a glucoside of the formula C,?H,,O,, the term glucoside being used in its wider sense as including substances which on hydrolysis yield products capable of reducing Fehling's solution. 2. That the products of its hydrolysis are emodin, identical with the emodin of rhubarb, and a compound which reduces Fehling's solution, but which is not glucose. As the yield (67-6 per cent.) of emodin on hydPolysis was higher than the theoretical amount (62.8 per cent.) which CnHB20, should yield, i t was imaginable that the formula was really C21H2009, and that the higher percentage of carbon obtained was due to the presence of free emodin as an impurity; hut this view was discarded, firstly, because the presence of 10 per cent. of free emodin would be required to raise the carbon from 60.58 to 61-20, and this could scarcely be present in each of the four samples that had received different treatment yet gave concordant figures ; and, secondly, because the presence of 10 per cent. of emodin would have made the50 RUHEMANN : THE ACTION OF CHLOROFORM hydrogen 470 per cent., an amount differing widely from the 5.29 per cent. obtained. We are engaged in preparing larger quantities of fmngnlin with a view of redetermining the yield on hydrolysis and of identifying the soluble product' ; but as we are unable to continae the investigation together we are now communicating this secticbn to the Society, and the aubjeict will be continued by .one of us.

ISSN:0368-1645    

DOI:10.1039/CT8905700038

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

50 RUHEMANN : THE ACTION OF CHLOROFORM VII.-!l%e dcti'on of Chloroform and Alcoholic Potash on Hydraaines. (Part 111. ) By S. RUHEMANN, Ph.D.,MA. IT has been previously shown )that the action on phenylhydrazine of the agents mentioned in khe title gives rise to the formation of a basic RubEitance which is {to .be regarded as a diphenyl-derivative of tetrazine, N<NH,c,>N. The application of the yeaction to homo- logues of phenylhydrazine has resulted in the formation of homo- logues of d>iphenyltetrazine, and it is to the description of several of these compounds that this paper will be devoted. CH*NH Paraditoly ltetrazine. In the previous paper (Trans., 1889, 242), although diphenyl- fetrazine was the chief subject o€ study, it was mentioned that a ditolyltetrazine could be prepared from paratolyl hydrazine.I have subjected this substance to a closer study, and have found that it is a weaker base than diphenyltetrazine, but that, like the latter, it dissolves, although with difficnl ty, i n boiling hydrochloric acid, from which, on cooling, slender, white needles of the hydrochloride are deposited. These are, however, very unstable, and, on drying over potash and sulphuric acid, continually lose hydrogen chloride. This circumstance explains the discrepancy which will be observed in the following numbers :- Found. *Calculated far +-7 46H16N4,HC1- I. 11. C1. .. .... .... 71.81 10-76 11.0 Pnraditolyltetrnzine combines with 1 mol. of methyl iodide, when it is heated with a solution of the iodide in methyl alcohol at aAND ALCOHOLIC POTASH ON HYDRAZISES.51 temperature of IOO", and yields a methiodide of the formula C,6&N4,CH,I, as shown by the analysis !- Calculated. Fonnd. I ........... 31-28 31.16 N . .......... 13.793 14.00 This compound is sparingly soluble in boiling watw, and crystal- lises from the aqueous solution in colourless needles which melt a t 242". Nitroparaditoly ltetrazine is formed by adding fuming nitric acid to the solution of the tetrazine in acetic acid, and is thrown down by water as a yellow precipitate which is easily soluble in glacial acetic acid and in alcohol, and cr.ystallises from methylated spirit in orange needles which melt at 144". The formula: Cl,H1,(N02)N4, requires- Found. 7-- 7 Theory. I. 11. - C ......... 62-14 62.5 H.. ........ 4.85 5.0 N.......... 22.65 - 22.75 - The formation of a nitro-compound, so readily soluble i n acetic acid and with such a low melting point as compared with that of nitro- diphenyltetrazine (above 300°), led me to repeat this experiment ; but I arrived at the same result. There is only one nitro-derivative formed. I, therefore, reinvestigated the action of nitric acid on diphenyl- ketrazine, and found that besides the nitro-derivative already described, there is formed another which is easily soluble in acetic acid or alcohol, and crgstallises from the latter in yellowish-red needles which melt at 145-146". A nitrogen determination proved this substance to be an isomeric ~itrodi~hen?lZtetrazine. The formula, ClrHll (NO,) Na, requires :- Theory. Found. N . ..........24.91 24.98 The action of bromine on paraditolyltetrazine results in the forma- tion of two bromine-derivatives. On adding bromine to the solution of the tetrazine in glacial acetic acid, a white, crystalline precipitate separates which, when recrystallised from the same solvent or from boiling alcohol, forms needlea melting a t 245" with decomposition. A nitrogen determination of this substance proved it to be a dibromo- paraditoly ltetrazine. Its formula, C16H14Br,N4, requires-52 RUHEMANN : THE ACTION OF CHLOROFORM Theory. Found. N . .......... 13-27 13.67 Besides this bromine-derivative, another is also formed, which remains dissolved in the acetic acid after the first has separated out, but can be precipitated from it by water. It is extremely soluble in alcohol, and crystallises from dilute spirit in colourless, slender needles which melt a t 81".Lack of material prevented me from analysing this substance ; but from analogy to the action of bromine on diphenyltetrazine, it is probable that the compound melting at 81" is an isomeric dibrominated substitution-product. 0rthoditotyItetrazin.e. This is prepared by a method similar to that which yields the other tetrazines ; but in this case it is necessary to take the precau- tion of adding the alcoholic potash to the solution of the hydrazine and chloroform in spirit very gradually ; warming only slightly until the reaction begins, or else the hydrazine suffers complete decomposi- tion, and a brown, viscid mass is formed. The product of the reaction was treated as in the other cases, namely, after dilution with water it was shaken with ether, the unaltered hydrazine removed from the ethereal solution by dilute sulphuric acid, and the ether evaporated. A brown oil was then left, and from this, after some time, crystals separated ; these are readily soluble in alcohol, and crystallise from it in slightly coloured plates.This substance is orthoditolyltetrazine. For complete purification, it was found advisable to transform it into the hydrochloride and to decompose this by boiling with an aqueous solution of ammonia. The tetrazine is thus obtained in colourless crystals which melt at 141". The following numbers correspond with the formula- Found, 7 Theory for r-d- C1GHl6N4. I. 11. C ........ 72-73 72.704 - H.. ......6.06 6-09 - N ........ 21-21 - 21.32 This formula was, moreover, verified by the determination of the molecular weight, and by the study of its behaviour towards nitric and sulphuric acids.AND ALCOHOLIC POTASH ON HYDRAZINES. 53 A molecular-weight determination by Raoult's method gave the following reanlt :- Weight of substance. .......... 0.4104 grams. ,, acetic acid .......... 23.9212 ,, 14.60" C. 7 9 mixture 14.8 6 , , 0.26 ,, Freezing point of acetic acid.. .. Depression of freezing point .... ...... Molecular weight derived from the above data .............. 258 Theory for C16HIGNd.. .......... 264 Orthoditolyltetrazine is dissolved by boiling hydrochloric acid, and t h e solution on cooling deposits white needles of the hydrochloride which, when dried a t 100", lose hydrogen chloride, but may be dried over sulphuric mid and potash without suffering decomposition-as shown by the analysis, which gave :- Calculated for C16H,,N,,HC1- Found.C1 ............ 11.81 11.61 N . ............ 18.64 18.i7 Orthoditolyltetrazine also combines with 1 mol. of methyl iodide when the two are heated together in methyl alcohol solution a t 100" for a few hours. If the solution is evaporated on the water-bath and the residue dissolved in boiling water, white nodules crystallise out on cooling: these melt a t 198" and t u r n slightly yellow on drying at 100". The following numbers correspond to the formula CI6Hl6N4,CH3I :- Theory for C1;H,9NJ. Found. N . .......... 13-79 13.93 I ........... 31.28 30.74 The action of sulphuric acid on orthoditolyltetrazine gives rise to a sulphonic a c i d ; this is obtained on warming the base with con- centrated sulphuric acid, when it dissolves.On adding water, a slightly bluish precipitate is thrown down, which is sparingly soluble in boiling water, and on cooling crystallises in colourless prisms which are free from water of crystallisation, and gave on analysis values corresponding to the formula c L6H15N,.HS03 :- Theory. Found. s ........... 9.3 9-15 N . .......... 16.28 16.5654 RUHEMANN : THE ACTION OF CHLOROFORM On adding fuming nitric acid to the solution of the tetrazine in glacial acetic acid, crystals are deposited after a short time which are sparingly soluble in alcohol, and crystallise from it in yellow needles melting a t 206-207". Analysis shows this compound to be a mononitro-derivative of orthoditolyltetrazine. The formula, C,sH,5(N0,)N4, requires- Found.Calculated for T-- 7 C16H15N502. I. 11. - C ........ 62.13 61.873 H.. ...... 4.85 4.91 N.. ...... 22.65 - 22.80 - Whilst the formation of tetrazines from phenylhydrazine and para- tolylhydrazine is accompanied by that of the formyl-derivatives of the corresponding hydrazines, formylorthotolylhydrazine could not be obtained by the isonitrile reaction; it is, however, easily pre- pared by heating orthotolylhydrazine with formamide in an oil-bath a t about 120" until ammonia ceases to be evolved. The product of the reaction solidifies on cooling and crystallises from water in colourless plates melting at 120" ; these on analy.3is gave numbers corresponding to the formula CH3*CsH,*NH*NH*OHO [ CHs : NH = 1 : 2J:- Found.Calculs ted for 7-- 7 C8HlON20. I. 11. 111. - - C ........ 64-0 64.2 H . . ...... 6-67 6.87 - N ........ 18.67 - 18.53 18.80 Although the preceding researches are in themselves sufficient to show that in general the action of chloroform and alcoholic potash on phenylhydrazine and its homologues gives rise to substitut'ed tetr- azines, yet I thought it desirable to apply this reaction t o a still higher member of the primary hydrazines. I have chosen pseudo- cumylh ydrazine. Pseudocumyl hy clrazine. This compound has already been obtained by Ealler (Berickte, 18, 89) by the method described by E. Fischer (Annnlerz, 190, 67) for the preparation of phenylhydrazine. I can confirm Haller's state- ment that pseudocumylhydraxine canriot be prepared according to V.Meyer arid Lecco's method (,Berichte, 16, 2976), as 9n mixing the stannous chloride and diazo-solution evolution of nitrogen takes place, and a tarry product is formed.AND ALCOHOLIC POTASH ON HY DRAZINES. 55 Pseudocumylhydrazine is sparingly soluble in ether and decom- poses readily. Before applying the isonitrile reaction to this base I have subjected it to a closer study with the object of characterising it by some derivatives. Acetylpseudocum~ylhjdrazine.-Acetic anhydride acts on the hydr- azine with development of heat. The reaction was completed on the water-bath and the product recrystallised from boiling water. It foyms colonrless plates which are readily soluble in alcohol and melt at 156-1-57".When submitted to analysis, it gave numbers corre- sponding to the formula C6H2(CH,),*;IJH*~H.Co.cH, :- Found. Theory for 7- 7 C11HU3N 2O. I. 11. C ............ 68.75 68.57 - H 8.41 - ............ 8.33 N ............ 14.57 - 1460 Pseudocumylsenaicarbazide is formed by adding an aqueous solution of potassium cyanate to a solution of the hydrazine hydrochloride in water. The precipitate thrown down is sparingly soluble in water, but readily in alcohol, and crystallises from the latter solvent in colourless needles which melt a t 195". The formula, C6Hz( CH3)3*NH*NH.CO*NH2, requires :- Found. Theory for r--- 7 c10E16N30. I. 11. 61-93 - C ............ 62.17 H 7.78 - ............ 7.77 N ............ 21-76 - 21.97 BenzyliBeize-~seudocum.ylhydrazine, C6H2( CH,),*NH*N:CH* C6H5, pre- pared from the hydrochloride of the hydrasine in the usual manner, cryst,allises from alcohol in coloured needles which decompose a t 100" and also, after some time, at the ordinary temperature.A nitrogen determination of this substance dried in a vacuum over sulphuric acid gave- Calculated for ~',6H18N!2* Found. N ............ 11.76 1 2 0 Pseudocumythydrazine-pyruvic acid, C6H2( CH,),NH*N:C( CH3) *COOH. On adding pyruvic acid to at1 aqueous solution of the hydraxine hydrochloride, a lemon-yellow, flocculent precipitate is at once throwti down which is soluble in alcohol and crystallises from it in yellow needles melting a t 148" with decomposition.56 THE -4CTION OF CHLOROFORM, ETC., ON HTDRAZINES. The. formula, Cl2Hl6N2o2, requires- Found. 7-- -.Theory. I. 11. 111. - - C ........ 65-45 65.20 H ........ 7.27 7.36 N ........ 12-73 - 12-69 12.97 - - Pseudocumylhydrazine, when treated with chloroform and alcoholic potash in the manner previously described, yields a compound which ia very sparingly soluble in alcohol but readily in boiling glacial acetic acid, and ci-ystallises from the latter in yellow needles which melt at 254". This substance is, without doubt, dipseudocumy Z- tetrazine. I hope to communicate to the Society the results of the further stndy of this tetrazine in another paper. I may 'here mention that I have already begun to stndy the act>ion of chloroform and alcoholic potash on other derivatives of phenyl- hydrazine. I n a note published recently in conjunction with F. F. Blackman (Trans., 1889, 612) it was stated that the above-ment8ioned reagents do not yield the corresponding tetrazine with benzophenyl- hydmzine. I have since then applied the isonitrile reaction to para- bromphenylhgctraeine. When this hydrazine, which has been described by Neufeld (Annalen, 248, 94), is treated with alcoholic potash ;tncl chloroform it yields, after removal of the unaltered hydrazine, a crys- talline residue. Hot water extracts from it a substance which, when repeatedly recrystallised from boiling water, can be obtained in colourless, irridescent needles melting at 198". Analysis of this compound proved it to be formyl-parabromophenylhydrazine. I t s formula, C6H,Br*NH*NH*CH0 [Br : NH = 1 : 41, requires :- Found. Theory for r--- 7 C? H ;N Br . r. IT. 111. - N ........ 13.0'2 13-44 13-19 Br ....... 37.2 I - 37.12 The formyl-compound is the chief product of the isonitrile reac- tion on parabromophenylhydrazine, but along with it a substance is formed which decomposes so readily tha't'hitherto I have been pre- vented from analysing it. I shall, however, repeat this experiment, and hope to lay the result before the Society in my next comruuni- cat ion. Uga iversit y Laboratory, Cambridge.

ISSN:0368-1645    

DOI:10.1039/CT8905700050

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

5'7 VIIL-Note on the Identity of Cercbrose and Galactose. By HORACE T. BROWN, F.R.S., and G. HARBIS MORRIS, PhD., F.T.C. IN a recent number of the Zeitschrift fur physiologische Cheinie there is a paper by H. Thierfelder, showing the identity of the brain-sugar, cerebrose, with the well-known carbohydrate, galactose ( Z e d . physiol. Chern., 14, 209). Cerebrose was first described by Thudichum (see Annals of Chern. Medicine, 2, 209 ; also Treatise on the Chemical Con- s f i f u t i o n of the Brain, p. 143), who found it amongst the products of the long continued action of dilute sulphuric acid at 130" upon certain ni trogeni se d const it uen ts of the brain. It is a crystallisable sugar, resembling milk-sugar in its feebly sweet taste and in the great bardness of its crystals, and it is described by its discoverer as having a cupric reducing power equal to about five-sixths of that of dextrose, and as giving in the polariser a "limited rotation '' of +70° 4.0'.On analysis, numbers were obtained consistent. with a formula of C6Hl& The cerebrose which was employed by Thierfelder (Zoc. cit.) was obtained from cerebrin by the action of dilute sulphuric acid a t high temperatures, It was found, on oxidation with nitric acid, t o yield muck acid, just as does galactose ; and in its melting point, specific rotatory power, and cupric reducing power, gave numbers so closely in accord witb those obhined with galactose, that there can be no doubt as to the identity of the two sugars. In the early part of last year (1888), whilst we were engaged on the determination of the molecular weights of the carbohydrates by Raoult's freezing method, Dr.Thudichum was good enough to place in our hands for examination a specimen of cerebrose, which he had prepared some years previously from pure phrenosin. We made a careful examination of its properties, and proved to our complete satisfaction that it was gulactose. Although we have but few new facts to add to the very complete work of Thierfelder, it seems worth while to place on record the independent proof of the identity of the two sugars, especially as the specimen we have examined is the one described by Thudichum in his first notice of cerebrose. Specijfc Rotatory Power. (5.6198 grams cerebrose per 100 c.c.) (1) [a]j 87.46" = [a]= 78-98" a t 16°C.(2) [dJj 86.78 = [a]= 78.32 at 21" ,,58 ON THE IDEXTITY OF CEREBROSE AND GALACTOSE. SpeciJfic Rotatory Power of galactose. [~x]n 80.56' at 18°C. [ a ] D 79.93 at 21 ,, Cupric Reducing Power. (Strength of solution, 5.6198 grams cerebrose per 200 c.c.) (1) 0.1124 gram cerebrose reduced . . 0.24145 gram CuO. (2) 0.1686 2, ,, . . 0,36273 ,, (1) K = 97.42 Galactose K = 93.01 (a) K = 97.57 The cerebrose used in the above determinations was not recrystal- lised by us, and the slight divergence of the above numbers from those of galactose i8 probably accounted for by the presence of a little dextrose, which, from an examination of the uncrystallisable mother- liquor (also handed to us by Dr. Thudichum), we have reason to believe is formed in the hydrolysis of phrenosin together with the cerebrose.We have prepared, the phenylhydrazine-compounds of cerebrose and of galactose under identical conditions, and find their melting points as follows :- Phenylcerebrosazon, m. p. 142" Phenylgalactosazon, m. p. 146 Thierfelder found a much higher melting point than the above, but this is probably accounted for by thg fact that galactose yields, just as some of the other sugars do, several compounds of this nature. Our results agree in this important point, that when treated under similar conditions cerebrose and galactose yield phenylhydrazine- compounds having identical melting points. Molecular Weight of Cerebrose determined Ey Raoult's Freezing Method. Freezing point of water used, 0.085". Streiigth of solution, 5,6198 grams cerebrose in 96.58 grams water. E. C. A. M. - 0.5 70 0.655 0,112 170 - 0.5 70 0.655 0.112 170 Calculated for C,H,20,. Found. A .. . . . . . . . . . . 0.106 0.112 M .. .. .. .. .. .. 180.0 170.0O'SULLIVAN : AHABINON, THE SACCHARON OF ARABINOSE. 5'3 A comparison of the microscopical appearance of the crystals of cerebrose and of galactose indicates beyond doubt the identity of crystalline habit.

ISSN:0368-1645    

DOI:10.1039/CT8905700057

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

O'SULLIVAN : AHABINON, THE SACCHARON OF ARABINOSE. 5'3 IX.-Arabinon, the Saccharon of Arabinose. By C. O'SULLIVAN, F.R.S. ANONGST the products of the action of sulphuric acid on arabic acid (Trans., 2884, 45, 5 5 ) , I indicated the presence of a compound which I named a-arabinose, and of which I said " the optical activity must be well above [a]j = 140"." Whilst working with the dextro-rotatory gums (Gedda gum), I frequently noticed that the sugar syrups had a higher optical activity than arabinose, the p-arabinose of the above paper, thus indicating the presence of a substance of high rotatory power. It often happened, however, that there was little or no evidence of the presence of this compound, and this frequently undei. conditions such as I thought warranted me to expect it in greatest quantity.There were here, then, evidently some circumstances, a knowledge of which could only, if at all, be arrived at by a study of the substance itself. A 25 per cent. solution of gedtiic acid (an acid obtained from Gedda gum, which I shall describe in a future communication) was heated to boiling, and 2 per cent. of sulphuric acid added, this having pre- viously been diluted with 4 or 5 volumes of water. A t the end of 15 minutes' digestion the solution was cooled, and alcohol, sp. gr. 0.830, added as long as a precipitate was thrown down. When the super- natant liquid became clear, it was decanted off and the acid neutralised with strong baryta-water. The sulphate and other salts of barium were filtered off and the alcoholic liquid evaporated to a syrup in a vacuum ; from this, arabinose crystallised out, and a further crop was obtained from the mother-liquor by the addition ox" a little dry niethyl alcohol.Ethyl alcohol, sp. gr. 0.820, was added to the methylic mother liquor until a slight precipitate formed ; this contained a trace of barium, and if sufficient alcohol had been added and the original syrup was sufficiently free from water, the alcoholic solution would be fairly free from that metal. The clear alcoholic liquid was decanted off and ether added as long as a precipitate formed. An examination of this syrupy precipitate showed that it did not contain barium, and that it was not acid, demonstrating the absence of a gum acid ; and that the solid matter it contained had an optical activity of [a]= = 163".This syrup was dissolved in a little methyl alcohol and allowed toGO O'SULLIVAN : ARABINON, stand for some months; the solution showed no signs of crystallisation ; hence I det'ermined t o make such an examination of the substance as was possible under t.he circumstance. Ethyl alcohol, sp. gr. 0.820, was added to the rnethylic solution ; a large, syrupy precipitate was first formed, but furt)her addition of the solvent partly redissolved i t ; the precipitate was allowed to settle and the clear supernatant liquid poured off. This precipitate is fraction a . Ether was added to the clear liquid until a quantity of precipitate was obtained, estimated as one- third of the whole matter in solution; this was allowed to settle, it is fraction b.A further addition of ether to the clear supernatant liquid yielded another pre- cipitate, fraction c. The optical activity of these three fractions was determined with the following results :- Fraction a gave [a]D = 187.2 ,, b ,, [a]D = 188.9 ,) c ,, [a]= = 187.6 Hence, a large quantity of a, substance of lower rotating power was eliminated, and we have in these three fractions the compound in a state of moderate purity. A determination of the cupric reducing power was made. 2.261 grams of solution of a sp. gr. 1.01426 gave 0.113 gram CuO ; from this K = 62.1. The quantity of substance in solution in this case and in the determination of the specific rotatory power was estimated on the supposition that 10 grams of the substance in 100 C.C. of solution gives a sp.gr. of 1.0385. The true factor will have to be determined later. The three fractions were again mixed, and the solution evaporated in a vacuum to a syrup. This was taken up with methyl alcohol, a small fraction taken out by adding strong ethyl alcohol, and then ether added as long as a precipitate formed. The optical act.ivit(y of the solid matter in this syrup, determined as above, was found to be [a]D = 192.5. A further quantity of low rotating substance was therefore eliminated by this fractional precipitation. No trace of crystallisation appeared in this syrup or in its solution in dry methyl alcohol after standing for some weeks. I had, therefore, to try to further purify the substance by fractional precipitation after the manner already described.Proceeding in this way, I a t last obtained a fraction, of which the specific rotatory power was [PI= = 196.7, with a K = 57.3. Considering the number of fractionations and the slow increase of optical activity and slight diminution in copper oxide re- ducing power, this fract.ion map be looked upon as a pure compound, as pure as, indeed, it is possible to get it in the absence of crystallisa- tion. We may therefore safely proceed to the closer examination of its characters and properties.THE SACCHARON OF ARABIXOSE. 6 1 The syrup was dried in a vacuum over sulphuric acid, and then at a temperature gradually increasing to 75-80" until the weight became constant. In this way a fused, glassy mass was obtained, which, on cooling, solidified to a brittle glass.On pulverisation, this yielded a white, highly hygroscopic powder. A portion of the powder was taken and again gradually dried a t 55" in a current of dry air under a pressure of 165-200 mm. of mercury. When the temperature was raised to 70-75", the substance fused. The weignt became constant at 55" at the pressure and under the circumstances indicated. 2.594 grams of dry substance gave 100 C.C. of solution of sp. gr. 1.01022, and this solution, in a 200 mm. tube, had an optical activity a D = + 10.3". From these observations we find that a solution containing 1 gram dry substance i n 100 C.C. has a sp. gr. of 1*00394-reasoning from analogy, the amount of this substance in any solution, up to a sp. gr. of 1.04500 a t least, can be calculated with practical accuracy by employing this factor-and that the specific rotatory power of the compound is [aID = + 198.5".3.203 grams of a solution sp. gr. 1.02554 yielded on usual treatment 0.26'2 gram CuO ; hence the cupric rediicing power, or K is .58*8, that is, 100 parts of it reduce as much copper as 58.8 parts of dextrose. 4.938 grams of dry substance were digested a t 100" for 30 minutes with 20 C.C. of 2 per cent. sulphuric acid. The acid was separated with baryta-water, and the filtrate from the barium sulphatc made up to 100 C.C. with the washings ; in testing for neutralisation, there was, of course, fiome slight loss. The solution had a sp. gr. of 1.01995, and in a 200 mm. tube gave aD = + 11.25". From this the solid matter has a specific rotatory power [a]D = -l- 108*6", a number closely agreeing with t.hat of arabinose.It was evaporated to a syrup ; on standing, this solidified to a mass of sphenoidal crystals, highly characteristic of that body, crystallking under like conditions. These were washed with as little methyl alcohol as possible, the washings yielded a further crop of crystals, and the mother liquor from these again crystallised on evaporation to a syrup. Over 70 per cent. of the material taken was obtained in the two crops of crystals. Both crops were mixed, as the crystals were identical, and a solut#ion was made of them containing in 100 C.C. 4.97 grams of dry substance; the optical activity of this was found to be aD = + 10.52" in a 200 mm. tube, its sp. gr. 1.01914, and 1.897 grams of it reduced 0.221 gram CuO.From these numbers we have 1.00385 as the sp. gr. of a solution containing 1 gram in 100 c.c., [oc]D = 105.8 as its specific rotatory power, and its K = 108.3. These are practically the factors of pure arabinose ; hence the com- pound acted on by sulphuric acid holds the same relation to arabinose VOL. LVII. F62 0 SULLIVAN : ARABINC N, as does either dextrin (amylin) or maltose (amylon) to dextrose (amylose). Is it an “ i n ” compound o r an “ on ” one ? Arabinose has recently been satisfactorily proved to be C,HloO, ; if the substance with which we are dealing is an “ o n ” compound, its molecule is CloHls0g(2C5HloO~-H,0), if an “in” one it is nC5HSO4, n being considerably greater than 2. Raoult’s method of determining molecular weights seemed to be a simple and an expeditious way of determining the question.The freezing point of a solution of the substance, sp. gr. 1.02554, and consequently containing 6.466 grams of the dry substance in 100 c.c., was 0.535” below that of water ; thence, by Raoult’s law, 19 +!6’”88 x15!!5 = 239.2, the molecular weight of the substance. The molecular weight of Cl,H,,O, is 282 ; hence, although the number obtained is low, when we consider that the molecular weight of some of the ‘‘ on ” compounds comes out low by Raoult’s method (Brown and Morris, Trans., 1888, 53, 615 ; Tollens and Mayer, Ber., 21, 1566), and that the preparation may not be absolutely free from ash, which we know materially affects the results, we have sufficient evidence, that the compound belongs to the “ on ” and not to the “ in ’’ class.The new substance is then C,,H,,O,, the saccharon of arabinose. I, there- fore, propose to call it arabinon : according to the nomenclature pro- posed by Scheibler (Uer., 18, 646), the compound would be arabinbiose. I prefer the vowel nomenclature, and, therefore, the term I propose. 6.466 x 100 A C12H,,011 sugar requires- C .......... 42.11 per cent. H.. ........ 6.43 ,, C .......... 42.58 per cent. And a C,oHl,Og compound- H . . ........ 6.38 ,, The difference in percentage of carbon is suficiently great to he apparent on determination. A combustion of arabinon was made, in a current of oxygen, side by side with one of saccharon (cane- sugar), in a like time and under similar conditions, with the following results :- Dry saccharon employed, 0.3329 gram ; found CO, = 0.514 gram, Dry arahinon employed, 0.3495 gram ; found CO, = 0.541 gram, HEO = 0.198 gram, H,O = 0.205 gram and ash = 0.002 gram.These numbers correspond to- Saccharon. Arabinon. C 42.11 p. c. 42.46 p. c. ........ H . , ...... 6.61 ,, 6.55 ,,THE SACCHARON OF ARABINOSE. 63 There can, therefore, be no doubt that the formula of arabinon is ClOH1809. If we now turn to the results obtained above by acting on arabinon with sulphuric acid, we find that the observed increase due to the hydrolysis is 4.9 per cent., or 100 parts arabinon yielded 104.9 parts arabinose. The equation CioHw.09 + HZO = 2C5HioOs Arabinon. Arubinose. requires 106.3 parts arabinose, a number sufficiently close to that observed, when we consider that the operations could not be per- formed without some loss, to confirm the conclusion as to the nature of the compound.We have, therefore, in the new compound, arabinon, a sugar holding the same relation to arabinose that maltose dose to dextrose. I have not, yet succeeded in crystallising it. In the syrupy state, it dissolves easily in dry methyl alcohol; from this, strong ethyl alcohol, of sp. gr. 0.820, precipitates i t partially, and an excess dissolves it. The highest specific rotatory power observed is [a]D = + 202" (C = 6.466). There is apparently an increase with the concentration. Its K is not greater than 58. If it be considered that 2 mols. of arabinon reduce 9 mols. CuO, the K would be 57.5, and the sp. gr. of a solution con- taining 10 grams dry substance in 100 C.C. is 1.0395. To the taste it is distinctly sweet, and in that respect it is sharp and clean like sac- charon. It is easily diffusible through parchment paper ; sulphuric acid readily hydrolyses it to arabinose. This inversion clearly explains the irregular absence and presence of the compound of high rotatory power amongst the sugar syrups obtained by the action of sulphuric acid on the gums, and points out the direction to be followed in order to prepare it in quantity. I have demonstrated the presence of arabinon in the syrups ob- tained by the modified action of sulphuric acid on arabic acid, the chief gum acid of all the lzevorotatory gums I have hitherto examined, such as gum arabic, Turkey, Senari, Levantine, and East India gums, and on the acids obtained from two varieties of dextrorotatory gum (Gedda gum). In the immediate future I hope t o further add to our knowledge of arabinon, and I shall be disappointed if I do not mcceed in crystal- lising it. Much of the analytical work in this paper, and of the general work that led to the results, was done by my assistant, Mr. A. L. Stern, B.Sc. Ether throws it out of this solution almost completely. My cordial thanks are due to him.

ISSN:0368-1645    

DOI:10.1039/CT8905700059

出版商:RSC    

年代:1890

数据来源: RSC

摘要:

64 X.-The Nature of Solutions as elucidated by a S t u d y of the Density. Electric Conductivity. Heat Capacity. Heat of Dissolution. and .Expansion by He0.t of Szclphwric A c i d Solutions . By SPENCER UMFREVILLE PICKERING. M.A. TABLE O F CONTENTS . PAGE PART I . General and Introductoq . Origin and nat. ure of the work Methods of differentiation employed . 65 65 67 PART IT . Densities 69 Method employed . 69 Experimental error 70 Temperature of the determinations 71 Preparation of the solutions 72 Results obtained . ’73 Second differentiation ’76 Position of the changes of curvature ’77 Conclusions drawn from the densities . 78 Mendeleeff R views 79 Contraction on mixing .79 Construction of B density table 84 PART 111 . Electric Conductivity . 86 PART IT- Heat Capacity 88 Method employed 89 Experimental error 90 Results obtained . 90 PART V . Heat of Dissolution . 94 Discussion of Berthelot’s and Thomsen’s results . 94 Graphic representation of results 97 Methods employed. 98 Calculation of results 98 Experimental error 1 0 0 Results obtained . 101 “ Dissolntion” experiments 101 Effects of excessive smoothing of the curve. . 104 Direct differentiation . 104 “Mixing” experiments .105 Position of the changes of curTature . 108 Comparison with Pfaundler’s and Thornsen’s results . 109 PART VI . Expansion by Heat A . As a function of percentage composition . 114 114 Results 11 A STUDY OF THE NATURE OF SOLUTIONS . 6 5 PAGE Differentiation of the results . 116 Position of the changes of curvature 117 Position of maxima . 119 B . As a function of temperature 119 120 121 Regularity shown by the cases specially investigated . Irregularity shown in other cases . PART VII . Discussion of the Resalts . The method of analysing the results and curves Experimental error Multiplicity of the changes of curvature Agreement of the results from different sources Definite composition of the hydrates indicated .Discovery of the extreme hydrates . Possible relationship between some of the hydrates . Complexity of the highest hydrates lndication of the monohydrate in solution The first differential is probably never rectilincal . The second differential probably is so . Dif€erent nature of the curves representing different properties Influence of temperature on the hydrates in solution . Specific changes of curvature as distinct from general changes of direction . Apparent magnitude of the changes a t the extreme ends of the figures Nature of the curves . Limits of differentiation Summary .Conclusion 122 122 123 124 125 125 128 128 129 129 130 130 132 132 132 133 134 1d4 135 137 TABLES I to XXIV 139 PLATES 1 to 3 184 PART I.-GEKERAL AND INTRODUCTORY . IN the early part of 1887. I began an investigation on the nature of solutions. by making a number of determinations of the heat of di8-solution of solutions of calcium chloride and nitrate. with the view to settling whether the curves representing these results exhibited any sudden changes indicative of the existence of liquid hydrates. or not . The subsequent publication of Meudelkeff’s examination of the densi-ties of alcohol (Trans. 1887. 778) and sulphuric acid (Zeit . physikal . Chem. 1.275). and of Crompton’s examination of F . and W . Kohl-rausch’s determinations of the electric conductivity of solutions of the latter substance (Trans 1888. 116). showed that sudden alterations in curvature might be established more satisfactorily by plotting out the rate of change-that is. the first differential coefficient-. than by a mere inspection of the original curve. or by attempting to appl (i 6 PICKERINGI A STUDY OF equations to it. MendelBeff stated that the rate of change of the densities is represented by a series of straight lines meeting a t points corresponding with hydrates of definite molecular composition whilst Crompton found that in the case of the electrical conductivities two differentiations had to be applied before a rectilineal figurewas obtained.These results are tantamount to the densities s of p per cent. 8 0 1 ~ -tions being represented by a series of 'different parabolas of the form s = A + Bp + Cp2 while the conductivities k are represented by a series of the form of 7c = A + Bp + Cp2 + Dp3 any one equation holding good between two definite hydrates. Crompton's results added confirmation to the particular hydrates indicated by Mende-l6eff. On examining by differentiation my results with the heat of dissolu-tion of the calcium salts I found that they yielded rectilineal figures only after a double application of the piwcess and the hydrates thereby indicated were very numerous ; further confirmation of these hydrates was obtained by determining the densities of the solutions, which were found in accordance with Mendel8eff"s statement to yield a series of straight lines after the first differentiation (see Proc.1888 35). These cases however were so complicated that I proceeded to investigate the presumably simpler case of sulphuric acid. The nature of the heat of dissolution curve was found to be the same as in bhe case of the calcium salts but the hydrates thereby indicated were far more numerous than those which MendelCeff had mentioned, and on examining his density results more closely I found with surprise that the values which he quoted* by no means warranted the conclusion that the first differential consisted of straight lines. These values appeared to me to be too meagre to settle the question a t issue, so I made a series of fresh determinations repeating them at four different temperatures.The results confirmed the existence of' those hydrates which the heat of dissolution had indicated and showed that the curves required a second differentiation before they yielded a rec t ilineal figure. From these densities the expansion by heat was deduced and further confirmation of the hydrates thus obtained. The heat capaci-ties of the weak solutions had also been determined in order to calcu-late out the heat of dissolution results and these also afforded similar evidence of an independent nature. The pyesent communication is confined to the results with aulphuric acid. * Not determined by Mendelkeff himself but collected by him from various 8oiirce8 THE NATURE OF SOLUTIONS. 67 Methods of Diferentiation employed.The results obtained and generally also the curves represeiiting them were examined by the same process of differentiation as that adopted by Mendelkeff. When the difference between the densities &c. of two solutions is divided by the difference in the percentage composition of these solutions the value obtained gives the mean rate of change in the densities bettween the two percentages taken or as it has to be less perfectly expressed the mean rate at a point intermediate between these percentages. This constitutes direct differentiation from the experimental values themselves. Differentiation however is a very dangerous and at the best a very imperfect tool. The differential is strictly the tangent of a curve and the method here adopted assumes that the curve which would represent the results is a straight line between every two experimental points.The method would yield absolutely true results only if these differences were infinitely small ; whereas if they be reduced too much-generally below 1 per cent.-the experimental errors attain such relatively large proportions that the results are useless. The actual differences between the experiments will also affect the form of the differential to a certain extent and irregularities will be caused if the differences are not the same throughout. When the experiments themselves form a curved figure consisting of a series of different curves they will when differentiated yield a series of other ciirves or straight lines between each of which there will generally be one differential point conforming with neither of the adjacent curves it being derived from two experiments belonging to two different primary curves.The existence of an undoubtedly non-conformable point is strong proof of a change in the vicinity,* but when doubtful it only adds to the difficulty of interpreting the results. When a second differentiation has to be performed it is hardly eyer possible to apply it to the experimental values themselves as Cromp-ton did in the case of the conductivities for the quantities dealt with would be of about the same magnitude as the experimental ewors themselves. I n such cases we must diminish the error by drawing a smoothed curve through the first differential points taking readings from this curve and differentiating these.Except in the case of the densities,? I have performed the first differentiation on the smoothed * Any error in an experiment will affect the two differential points derived from it in opposite directions. t To plot the densities out so as to allow 0 *OOO1 gram to 0 -01 inch the whole curve would have to measure about 2000 inches a wholly impracticable length 66 PICKERINQ A STUDY OF curve representifig the experiments as well as on the experimental values themselves. I n differentiating the smoothed curve we have the advantage of being able to take as many points as we like and a t equal distances apart; but on the other hand the drawing being subject to the “ taste ” of the draughtsman it is never safe to dispense entirely with direct differentiation especially as changes in curvature will be less marked in the original curve than in the first differential figure and may be overlooked if we trust entirely to the former.It requires no little practice and the most careful attention to the magnitude of the experimental errors to draw the smoothed curve on which to perform the differentiation. The curves representing the present results have been drawn with the help of a long thin steel or wooden lath. Such a lath when bent under the four points of pressure exerted by one’s two hands does not form a curve of any particular nature and does not necessarily give a curve which will differentiate eventually into a straight line ; but I find that if the experiments form a figure on t o which the bent steel cawnot be fitted that figure certainly does not consist of a single parabolic curve.* The scale which I have used wherever practicable is such that the experimental error is represented by one-tenth to one-twentieth of the distance between the points the maximum error corresponding to about one-tenth of an inch.Readings from these curves were taken with an accuracy of one-hundredth of an inch or less. Drawings on several different scales and with several different points as origins, were made in all cases and the labour entailed in the treatment of the results has by far exceeded that of the determinations them-selves. I n cases where the experimental error was comparatively large the results have been divided into two independent series and each of these differentiated separately ; the results being then plotted out together.I n such a case it must be remembered that we may get two non-conformable points in the vicinity of a change. Sometimes the somewhat less exact method of taking the mean of every two consecutive differential points has been adopted. Considerable difficulty has been experienced in finding any satis-factory method of depicting the curves in print. The size of the original working drawings is such that their reproduction here would * Though differentiation will alone tell us the precise nature of any curve, I would with the experience I have gained generally trust as much to the drawing of the original curve as to differentiation for showiiig whether that curve is con-tinuous or not and where the changes occur in it ; but such a method would carry conviction to no one 9-ho had not made curves their special study to the same extent as I have done THE NATURE OF SOLUTIONS.69 be impossible without considerable and often injurious reduction ; this and the inevitable inaccuracy of printed plates made it evident that any one wishing to examine the results in detail would plot the experiments out for himself and draw his own curves while to the more casual reader small illustrations in the form of woodcuts would be more instructive and convenient than plates. Such woodcuts, therefore have been used almost exclusively ; but their size generally precludes the insertion of the numerous experimental points and also necessitates the use of relative scdes for the abscissE and ordinates different from those which are suitable to the experimentai error they may sometimes therefore give an exaggerated impres-sion of the magnitude of the changes sometimes the reverse 1 have supplemented them however with a statement of the scales which I have found most suitable in the various cases-Table XXIV a t the end of the paper.I must waru my readers that many of the individual changes in curvature which I mention are admittedly of a very doubtful nature, especially when considered by themselves. Their existence can only be established by the concordant indications of independent proper-ties and even with such accumulation of evidence some of them must still remain doubtful. This is inevitable if these changes are really due to the existence of hydrates of diflerent degrees of com-plexity and stability.A study of any one portion of this paper with-out reading Part VI1 where the results from various sources are collected and discussed would probably lead to a very incorrect esti-mate of the value to be attached to the work in general and to the conclusions which T draw from it. Before proceeding to the details of the work I must express my best thanks to Mr. H. Crompton for the help which he has so ungrudgingly given me in it. Not only has he made many impor-tant suggestions which will be specially noticed in their proper places, but throughout the course of the work he has aided me with his superior mathematical knowledge and followed every step in the work with an interest which has almost made it his own.PART II.-DENSITIES, Met hod employed. The piknometer is nearly always an inaccurate instrument owing to the imperfect fitting of the stopper into the neck of the botkle, b u t the very nature of this inaccuracy implies that it may by chance be non-existent in some cases. Such an instrament was used in the following work. The concordance of the results obtained with i t were found to be as nearly as possible the same as in the case o 70 PICKERING A STUDY OF Sprengel tubes,* and the errors noticed were less than might be attri-buted to the balance and thermometer errors these latter being 0*0001 gram and 0.004° respectively. Consecutive determinations of the water contents (23 c.c.) of the bottle at the same temperature only once showed a difference of as much as 0.00015 gram and a more trying test of its trustworthiness was found in the comparison of my results with water at every 2" between 6" and 38" (to be given in a subsequent communication) with the smoothed curves represent-ing those of Kopp or Pierre; the average difference was in each case 0.000012 of the volume or 0*0003 gram and not only does this error include that due to differences in the determination of the actual temperature but it was evident that the greater part of it was due to some constantly increasing source of error which would either be absent in a series of determinations done at the same temperature or, if present would have no appreciable effect on tlhe differences between the consecutive experiments the only important consideration in the present case.To obtain this accuracy it is necessary to place a small glass plate over the open capillary in the stopper while the bottle is in the bath, and bcing weighed,? otherwise an appreciable loss occurs through evaporation even when the liquid is below the bottom of the stopper. The increased accuracy obtained in this way niay be judged by the following values obtained when the bottle was filled with water :-A. Without the plate-Extreme diff. of 0.0006 gram in 4 determinations at 8" Y > 0.0005 , 5 9 1s B. With the plate-Extreme diff. of 0-0001 gram in 3 determinations at 8" , 0 1 2 79 18 9 0 9 3 $1 28 7 9 0.00015 , 5 ?l 38 The calorimeter with its stirrer worked by an electromotor was used as the bath a very small flame being placed under it when temperatures above 18" were required ; at and below 18" the tempera-ture of the laboratory was reduced to that of the determinations.The thermometer was tapped continuously as in calorimetric work. Experirneixtnl Error. The water contents of the bottle were generally determined at the * Dr. Perkin's and Dr. Nicol's results were taken for this comparison. t A cap cannot be used while the bottle is in the bath a8 it increases the evapo-ration or rather distillatiun from the capillary THE NATURE OF SOLUTIOSS. 71 beginning and end (sometimes more frequently) of a series of deter-minations and any slight variation noticed was assnnied to have taken place regularly. The tare of the bottle was taken before and after each day's work and rarely showed any alteration.Thus each determination is practically dependent on one weighing only and the error of the first differentiatial deduced from two determinations is generally of the same magnitude as that of a single determination since there is generally a difference of 2 per cent. between the strength of every two consecutive solutions and the actual differences and their errors will be divided by 2. An error of one estimation figure in the reading of the thermometer (0.1 mm. or 0*004°) will in an extreme case (with acids of high strength and high coefficient of ex-pansion) cause an error of 0.0001 gram ; the balance error may be placed at a like amount ; so that 0,0009 gram or 0400008 of the den-sity may be taken as a safe limit of error in the majority of cases, both for the determinations themselves and for the differential points deduced from them.With solutions above 78 and below 10 per cent., the differences were less than 2 per cent. and the error in the differen-tials here will be proportionately larger. The glass plate on the stopper of the bottle was not adopted till after the series at 18" had been finished so that the error with these results will be somewhat greater," and another source of inaccuracy in this case was that the contents of the bottle were determined once for all it not haviug been ascertained at that time that a change of a few degrees will effect an alteration in the capacity of a well-seasoned glass bottle (see Phil.Nag. March 1890). It would however be ex-cessive I think to place the error of the determinations at 18" at twice the value of those at the other temperatures. Temperature of the Determinations. The exact temperatures at which the densities were determined were 7*978" 17*925" 28*064" and 38.203" & 0.002. All these were ascertained by a direct comparison of the experimental thermometers with a " natural standard," the zero point of which had been recently determined ; the " constant difference " of this instrument was taken to be the distance between the boiling point and freezing point taken immediately after boiling and Poggendorff's correction was applied to the readings. To reduce my determinations of the water contents of the bottle to true densities Kopp's determinations of the latter were taken ; these are 0.999890 0.998712 0.996353 and 0.993070 at the four tempera-* It would not on an avemge be nearly so much greater as the above quoted determinations with the more Tolatile water would imply 72 PICKERING A STUDY OF tures respectively.By mistake the density at 17.965" was taken as 0.998695 in the calculations but this error will scarcely affect the densities themselves and will not affect the differentials at all.* Preparation of the Solutions. I am indebted to the kindness of Dr. Messel for a liberal silpply of the pure acid prepared by repeated cry stallisation. Mr. Crompton was good enough to estimate its strength volumetrically starting with pure silver. His results deduced from the esamiuatim of two weaker solutions prepared from the strong acid were-A.(fI&1]= 99.8325 per cent.H2SO4. B.( }= 99.8665 per cent. H,S04. c 99.897 J Mean = 99.8495 per cent. &Sod. The value was taken as 99.85 per cent. The second half of this lot of acid (which had been supplied in two bottles) appeared from a comparison of about 20 density deter-minations at 18" to be 0.007 per cent. stronger than the first half (the one analysed) and as this small difference could not have been dis-tinguished with certainty by further analysis it was determined by these density results themselves. Subsequent work has led me t o conclude that the strength of the acid used in the present work is somewhat greater than I have taken it to be. I n plotting out the freezing points of a series of acids containing somewhat more water and more anhydride thau H,SO4 itself we get a figure consisting of two nearly straight lines rising up so as to meet at a very sharply marked angle and the point at which they meet must correspond for reasons which 1 need not enter into a t present t o the definite compound H,SO itself ; the point of intersection of these lines may be ascertained to within - + 0.01 per cent.or less. In this way 1 have determined by means more reliable than any analysis the strength of a large quantity of' acid used in subsequent work arid I have compared this acid with that used in the present work the contents of the second bottle above mentioned being taken for this purpose. The comparison, made by taking the densities of 52 and 66 per cent.acids a t 28" gave * It hap been corrected in Table VII THE NATURE OF SOLUTIONS. 73 99.92700 and 99,92914 mean = 99.92812 per cent. as its strength, instead of 99.857 given by the analyses. This result was further con-firmed by determining its freezing point which gave 99.932 & 0.01 per cent;. at8 the strength." As all the present results had been calculated out long before this later work had afEorded this more certain means of ascertaining the strength of the acid and as the error would not appreciably affect these results when differentiated I did not think it worth while to alter them except in a few cases where the introduction of the correction will be specially mentioned. In all the other cases the percentages quoted are 712~1000000 too low.The accuyate dilution of very strong sulphnric acid is not an easy matter. Series A. By exposure to moist air for a month with subsequent, addition of water. Used for about every other determination of the heat of dissolution. Series R. Used in the majority of the density determinations at 18". Series C. Used in the other density and heat of dissolution determinations. The solutions were made up in quantities of 100 to 200grams usiug a balance reading to 0*0001 gram so that their relative composition is accurate to about one unit in the fourth decimal place of the per-centage provided no loss of anhydride occurred in making them. In the case of all three sets the solutions from 58 to 16 per cent. were made from a 58 per cent. solution ; those from 16 to 4 per cent., from a 16 perlcent.; those from 4 to 1 from a 4 per cent. ; and the weaker ones from a 1 per cent. solution. To ensure uniformity in the whole series including the last determination-that is the one with water itself-this water was boiled in platinum a t the same time as that used for diluting the acid and kept for the same time in bottles of the same glass as those containing the acid. Although this water evidently contained air not the slightest difference (that is less than 0*000004) could be detected between its density and that of freshly-boiled water. Three main series of solutions were prepared as follows :-By addition of ice. By the very cautious addition of water. Results obtuined. The densities (not specific gravities) a t 18" are given in Table I : the full series with differences not less than 1 per cent, are first entered and then the more numerous ones with strong and weak * Some acid prepared by W.Kohlrausch (Ann. Pirys. Chim. 17 69) by crystal-lisation gave almost exactly tbe same results a8 my stock acid 1-8342 against my 1.834185 at 18". Perkin found the density of some supplied to him by Messel to be 1.83'148 a t 15" my acid being 1.83726 a t this temperature (Trans. 1886 783) 74 PICKERLXQ A STUDY OF solutions the differentiation of alternate experiments (A and B) are given in the last four columns." Table I1 contains the result's a t 8", 28" and 38" the differentiation of alternate experiments being here given (in the FIQ. few instances where it is adopted) separately.1.-Densities of S o l u t i o n s of S u l p h u r i c Acid. Percentage Strength. 0 20 41) 60 80 100 Fig. 1 is a sketch of the densities a t 18" and 38" plotted against percentage composition. The general change in curvature a t about 80 per cent. and the existence of a maximum a t 97.5 per cent. are noticeable being ignored by most text-books. Fig. 2 A (p. SO) gives a small size illustration of the first differen-tial obtained directly from the results at 8". Plates 1 and 2 (p. 184) show this as well as the results at the other temperatures on a scale . more appropriate to the experimental error and inclination of the various parts of the figures. At 28' and 38" it was found impossible to extend the determina-tions below a 20 per cent.acid owing to the liberation of air-bubbles, and the densities at these temperatures even from 30 per cent down-wards may be affected (though apparently with regularity) from this cause. * Where two solutions of nearly the same strength have been examined the results with each have been differentiated separately with those of the next stronger arid next weaker solutions ; we thus get two pairs of partially independent differential points at nearly the same percentages. This is a more accurate method than taking the mean of the two experiments themselves. Those solutions marked by an asterisk belonged to different sets of preparrttkms from the others THE NATURE OF SOLUTIONS. 75 I n each of the plates (1 2 and 3) the experimental error of 0*0002 gram or 0*000008 of the density would alter the position of the poiiits by about one hundredth of a big division of the paper.Greater errors evidently exist in some cases and are probably due to errors in the composition of the solutions By the omission of the doubtful de-terminations we get the points marked by crosses (the " alternate determinations " of the table) each of which it must be remembered, is a substitute for the two points on either side of it and considerable assistance in determining where such mean points should be taken, is afforded by the fact that the solutions used a t 18" were in nearly all cases,* different from those used at the other temperatures. Starting a t the 100 per cent. end we have in each case a rapid rise represented by a Iine of a strongly-curved nature as far as about 94 per cent.; this is then followed by a long and comparatively flat stretch as far as 84 per cent. it being so flat in the case of the result8 at 18" that i t is within experimental error a straight line. Then follows a more or less sharp curve which gives place to a gentler curve the differential reaches a maximum point a t about 73 per cent,. and then proceeds to fall but not regularly for it forms a most unmistakably wavy line exhibiting two troughs a t about 42 and 3 per cent. respectively after which it rises again in a very abrupt manner. Now even the most general characteristics of this first differential are absolutely conclusive as to its not being rectilineal and not being a single or simple curve and consequently that the densities them-selves cannot be represented by any single curve.No amount of smoothing could mask the wavy character of three quarters of the figure or could reduce it to uniformity with the flat portion from 84 to 94 per cent. and with the abrupt curvature exhibited a t both extremities. Although the figure may be continuous it can only be drawn in separate sections or curves but to determine the exact nature or even number of these constituent curves is a difficult problem. A good deal but as will be shown below not so much as might be expected, depends on the amount of smoothing allowed. In the present drawing this smoothing is sufficient to render the whole figure con-tinuous and to allow Considerable latitude beyond the known ex-perimental error no greater amount of smoothing is legitimate in the first instance and could be justified only in the event of the present drawing leading to contradictory or to no conclusions.The constituent curves of the figures have been extended beyond the points u p to which they are applicable simply in order to show f Except at 5 7 9 and the odd percentages from 79 upwards 76 PICKERING A STUDY OF more clearly which portions have had to be drawn separately. These constituent curves in some instances cut each other on being pro-duced but in the majority of cases they appear to meet tangentially. Curve VIII Plate 2 between 50 and 60 per cent. is of a very doubtful charact,er chiefly ow7ing to the error in the 60 per cent. solution it does not however appear possible to dram curves VII and IX so as to meet without increasing the error of the points very considerably ; they apparently cease to be applicable a t about 50 and 60 per cent.respectively and hence curve VIII has been drawn to represent the points between these two percentages. With Fig. D especially this curve is very doubtful Aince there may be another error a t 58 per cent. ; there is also some doubt as t o the drawing of curve VII. The two series of determinations a t 18" with totally different sets of solutions (A and B Table I) when plotted out separately show a concordance which affords strong additional proof of the correctness of the results. The more numerous results with weak solutions (F and G, P1. 2; show some remarkably abrupt changes the exact position and nature of which must however be doubtful owing to the magni-tude of the error consequent on the proximity of the determinations.The results with very strong solutions at 18" and 8" (F and G, P1. 1) show the existence of a change of curvature at about 97e.5, which did not appear in the 1 per cent. series owing to lack of SUE-cient points.* The alternate determinations only have been differen-tiated in this case ; but even then the error is three times as great aR with the 1 per cent. series. The direct first differentiation having failed t o reduce the densities to a rectilineal figure the effect of a second differentiation was ex-amined the operation being performed on the first differential curves, Table 111 contains the numerical results.? A general review of them is given in Fig.4 B (p. 87) while A B and E P1. 3 represent them in greater detail and on more appropriate scales. The results at the other temperatures closely resemble those at 18" which alone are represented in the plate. The constituent lines are prolonged in the drawings to render the points of junction clearer. The second differential diagram is certainly composed of lines * If however a sufficient number of readings be taken from the continuous and apparently regular first differential curve for the 1 per cent. series between 94 and 100 per cent. n second differential is obtained which consists of two straight lines meeting a t about 97% per cent. showing a change here in a similar though not SO precise a manner as when we start with a curve containing all the experimental points.t In the values in this and some of the other tables thc decimal point has been moved six places t o the riglit ,to economise space THE NATURE O F SOLUTIONS. 77 18". I 8". which are straight within experimental error,* though it is not possible to affirm that they are absolutely straight or that they may not be united with each other by short curved reaches. Each portion of tlhe first differential figure which had to be drawn separately yields an independent straight line for the second differential the only ex-ceptions being (l) in the case of curves VII and XI1 at 8" (not shown here) which on the second differentiation yielded (though doubtfully) two straight lines meeting at 40 and 88 per cent. respectively ; (2) in hhe case of both extremities of the figures where the second differen-tials are undoubtedly curved but by being so they really afford st'rong confirmation of their rectilineal character ; for the ends of the 1 per cent first differential series are as a matter of fact made up of one or two points taken from different contiguous curves and, when we take for the second differentiation points sufficiently close t o each other we find that these end curves split up into straight lines.The results are shown by Fig. E P1.3 (p. 184) and by the dis-continuous lines in Fig. F P1. 3 ; the irregularity in the line 1 of the Jatter is probably due to error. The following table gives the percentages at which the changes of curvature occur. The determination of their exact position is a matter of considerable difficulty ; if the curves constituting the first differential are tangential to each other the second differential lines mill meet at the point of change but if they are not tangential but TABLE A,-Position of the Changes in the Density Curves for Szclphuric Acid.Mean. At 38". 93.5 (94 -5j - (89) 84.4 (84) 78-2 (78) 72 .8 (71) 59 (58 '5) 49 (none) 32 (31) - (42) 28". 94.0 (94) 77 -5 (79 -8) - (88 '5) 84 -5 (85) 73.3 (69-5) 57 (62?) 50 (49.5) 29 (28 '5) -97 '2 (97 '2) - (89) 93.6 (94) 84.5 (84'3) 78.0 (77 -5) 72-8 (71) 58 (59) 51 (51) 29.5 (29) 18.5 (19) - (38.5) 8% (9.5) 4 '0 1 -05 0 -35 97 -7 (96 -8) 88 *O (?) (88 *7) 84 -5 (83 '5) 73-3 (71) 58 (?) (60) 51 (51) 18-5 (19) 94 -0 (95) 79 *o ('19) 2 g?) 10 -5 (9 '5) 3 -75 1 -05 0 -25 97.5 (97) *93'8 (94.1) 88.0 (3) (88'8) "84 *4 (84 '2) 78 -2 (78 %) 73'05 (70.6) *58 -0 (59 -9) 50.3 (50'5) 40 '0 (?) (40 '3) "30'1 (30.6) 18.5 (19) "9 *7 (9 ' 5 ) "3.9 1-05 "0 *30 * The error may be best estimated from the portion marked XII P1.3 ; the points in it should lie on an horizontal line since they are derived from fhe fir& difYerent.ia1 which was here drawn as tt straight line. VOL. LVII. TS PICRERING A STUDY OF cut on prolongation the second differential lines will not meet at these points and may not meet a t all. The majority of the curves appear to be tangential to each other but this is not always so, and consequently the meeting points of the second differential lines may give erroneous results.The position of the changes has there-fore been determined from an inspection of the first differential diagram. Those which I consider to be best established are marked by an asterisk. The differentia-tion of all the densities for different temperatures was worked out at first entirely independently of each other. Each set of results stand-ing by itself there were no grounds €or regarding any particular points as anomalous and consequently the drawings were made with much less smoothing than at present. Yet the positmion of the changes deduced from them agrees in almost every particular with those de-duced from the later drawings. The former values are those given in Iirackets.The indication of changes at 88 and 40 per cent. (where, however other results prove such changes) and the low position of the change at 73 per cent.,* are the chief points where any material differences exist. An investigation of a portion of the results at 18" to test the pos-sibility of drawing the first differential in sections other than those here given was made by drawing the portion between 35 and 81 per cent. in sections of which the middle points coincided with the points of change given in the table. Omitting three determinations which are evidently erroneous it was found that the average error of the differential points according to this second drawing was 0*006008, as against 0.000004 according to my original drawing and that the arrangement of + and - errors told against its correctness even more than the increased magnitude of the error.Its adoption would suppose the existence in 23 determinations of 16 errors greater than the liberal estimate of 0*000008 given on page 75 as a maximum. A similar comparison was also made by deducing equations from the points themselves instead of taking drawn curves but the increase of error was even greater according to this method. The portion of the figure examined i t must also be noticed is one in which the changes are admittedly but feebly marked if not as regards their existence, a t any rate as regards their exact position. The conclusions which I draw from these density results taken by khemselves may be briefly stated as follows :-The direct first differ-ential coeficient forms a continuous but complicated figure which * The determination of the position of the change here is difficult owing to the general change in the direction of the differential in its neighbourhood.This portion has been plotted out on many different scales and the mean result taken. The values given in brackets are of some interest THE N-4.TURE OF SOLUTIONS. 79 can only be drawn in separate sections it does not follow from this that each of these sections represents a definite and distinct curve ; but in the fact that each of these sections-differing considerably from each other as to their curvature- on further differentiation gives a straight line and in the fact that the results at different tem-peratures all indicate changes at the same points we have strong presumptive evidence that such is the case but this presumption can only be converted into proof by finding that other independent proper-ties and curves of a different nature exhibit similar changes at these same points.My conclusions as to the nature of the first differential of the den-sities are in direct opposition to Mendel6eff’s. He first represented it as a continuous curve (Ber. 1886 379) but subsequently (Zcit. physikal. Ohem. 1 275) represented it as rectilineal Fig. 2 D (p. Sl), being a carefully enlarged reproduction of his later drawing. Above it, Fig. 2 C I have plotted the values which he quotes and of which Fig. D wa8 given as a representation. That these points show the existence of more or less abrupt changes of curvature in certain regions is evident but I am unable to see how they can be made to harnionise w i t h Fig.D.* Mendelkeff’s values (deduced from the observations of other physi-cists) are given in Table IV together with a slight modification of them introduced so its to make them more strictly comparable with my own (the want of analogy being merely due to the fact that the actual differences are greater than with mine) and these are plotted i n Fig. E Plates 1 and 2 where they show such a very good general concordance with my own results? that they may be taken together with the latter as affording a further illustration of the effect of temperature on the figure. Contraction on Mixing Xzdphuric Acid and Water. At the suggestion of Professor Foster I calculated from the den-sities a t 18” the contraction which occurs on mixing the acid and water in various proportions (Table V).* Mendelkeff inferred from his mathematical theory of the effect of hydrates in solution that the first differential would be rectilineal and subsepently quoted experimental results in support of this conclusion (see Trans. 1887 779). A study of his drawing of his own results with alcohol leads to the same conclusions as in the case of his sulphuric acid diagram. A reproduction of Mendelkeff’s first curvilinear drawing of the sulphuric acid differential will be found in the PhiZ. Mag. 28 33 ; it forms 8 striking contrast with his second drawing and also with the experimental points. t. The point a t 575 per cent. alone seems to be somewhat erroneous.G SO PICKERIKG A STUDY OF FIG. 2.-Diffcreritiution of t h e Densit.ies of S u l p h u r i c A c i d S o l u t i o n s . ~crcelltngc Ir,s 04. 20 40 60 80 100 ZU 4u 60 80 100 Percentage H,S04 THE NATURE OF SOLUTIONS. 8 1. FIG. 2.-Differentiation of t h e D e n s i t i e s of S u l p h u r i c Acid Solutions. Percentage H2SO+ 0 20 60 80 10 82 PICRERING A STUDY OF The figure which these results form is as in t'he case of the den-Fig. 3 A represents it sities too large f o r convenient manipulation. FIG. 3.-Contractioii o n F o r m a t i o n of S u l p h u r i c Acid Solutions. Per cent. H2S04. 0 10 20 30 40 50 60 70 80 90 100 - *001 - -002 - '003 - ,004 - -005 - -006 0 10 20 30 40 50 60 70 80 90 1CO Pcr ccnt.H,SO, THE NATURE OF SOLUTIONS. 83 on a small scale. It is a continuous though irregular curve, attaining a maximum at 68 per cent. and exhibiti ~g a few move or less sudden changes of curvature of which one at about 50, and one somewhere below 1 per cent. are the most marked. The first differential obtained directly is illustrated by Fig. 3 B and, like that of the densities can only be drawn in separate sections. A second differentiation was performed on this figure (Table VI) arid the results are depicted side by side with the second differential of the densities in Plate 3. The general similarity,* though not identity, of these two figures is very striking and having been obtained from two figures (the first differentials) which differ entirely in their general character afford very strong evidence that the rectilinetil nature of the figure and the position of the changes which they both indicate are true characteristics of the properties i n question and not the chance outcome of the taste of the draughisman.The extension of the curvature of the end portion of the contrac-tions from 0 per cent. to solutions of greater strength than in the case of the densities is but the result of fewer points having been taken in the former the curve here and that at the 100 per cent. end splits up into separate lines when all the determinations are plotted out but the results are not illustrated here. The drawing of the portion X is very doubtful. The similarity of the two figures would be somewhat increased by taking the contractions on the forma-tion of l C.C.(Columns 5 and 10 Table YI) instead of on l gram: the results would at 0 per cent. occupy the same position as thme for 1 c.c. whilst at the 100 per cent. end they would occupy the position indicated by Pig. F in Plate 3. The position of tlie changes of curvature indicated by the contrac-tions and densities at 18" are as follows :-Densities. Contractions. 97.2 97.6 93.6 93.6 84.5 85.5 78-0 78.8 72.8 74.0 58.0 60-0 Densities. Contractions. 51.0 47-0 29.5 31.0 18.5 19.0 4.0 3.9 1.05 0.95 The concordance bet,ween the two sets of results is very good and mould no doubt have been better had the contractions been plotted out on as many different scales and been worked up with the same care as the densities.Two points of importance may be noticed. (1.) The constituent * In working out tliese results I was (and still am) in ignorauce that such a similarity should exist 84 PICKERINO A STUDY OF curves of the first differential of the contractions are different in nature from the corresponding ones of the densities since they do not meet tangentially. (2.) The maximum of contraction does not occur, as is generally supposed at the dihydrate 73 per cent. but at about 68 per cent. and does not coincide with the position of any sudden change of curvature.* It will also be noticed that the changes in the first differentials are more clearly marked with the contractions than with the densities but I doubt whether this advantage would not be more than compensated by the increased calculations and consequent chances of mistakes which the former necessitate.Construction of a Table of Densities. The determinations which had been made of the densities of sul-phuric acid afforded the means of constructing a fuller and more accurate table than any which existed. The special advantages which they offered for this purpose were (1) that the composition of the acid had been determined by a more certain method than mere analysis ; (2) that the very numerous solutions taken were of a com-position very near to round percentage values ; ( 3 ) that such small corrections its were necessary to reduce them to such could be applied with the greatest accuracy now that the rate of change with the composition (that is the first differential) was accurately known.The various steps in the construction of this table (Table VI1) were as follows :-(1.) The densities at 17.925' were correctled so as t o refer to water as 0.998712 (p. 71). (2.) The percentage strength of the solutions was altered in ac-cordance with the determinations based on the freezing points by being raised 712 millionths (p. 73). (3.) The few obvious errors which the first differential diagram had revealed were corrected. (4.) The values for round percentages at 7*978" 17.925" 28.064", and 38.203" were deduced from the experimental values by the help of the first differential figures (Plates 1 and 2). (5.) To detect and correct any arithmetical errors which had been made in doing so each of the four sets of results was differentiated, and the resulting figures compared with those in Plates 1 and 2.(6.) The results at the two higher. temperatures did not include determinations above 9.6 per cent. at less than 1 per cent. intervals, iior did they extend down t o percentages below 20. The former were supplied by plotting out the densities from 96 to 100 per cent. and taking readings at the required points from the curyes thus obtained. The * The position of this maximum is known to alter with the temperature THE NATURE OF SOLUTIONS. 85 densities below 20 per cent. were supplied by a more doubtful pro-cess :.-The differences between the densities a t 17.925" and 7.978" (an interval of 9.947") were plotted out and two curves drawn by their means ; a smoothed curve (1) utilising only the differences between 29 and 20 per cent.together with that of 0 per cent. ; an irregular curve (2) following the irregularity of the points from 29 per cent. downwards ; the differences ( d ) between these two curves (0.00009 of 'the density at most) were taken a t intervals throughout. Then the differences between the densities at 38.203" and 17.925" (an in-terval of 20.278') were plotted out these giving a curve 1' similar to 1 by assuming the differences between curves 1 and 2 to be propor-tional to the intervals of temperature we get the curve 2' for these 20.278 last determinations by multiplying the differences d by __- 9.947 and adding the result on to the readings of curve I' ; in this way the densities at 38.203" were deduced from those a t 17*925" and the densities a t 28.064" from those a t 7.978".The values thus obtained cannot be very exact but I think they are probably right to the fourth decimal place. (7.) The densities of each solution a t the four temperatures were plotted out on a scale which showed 0~00@02 of the densities and 0.02" (the largest scale which could be conveniently adopted) ; from the smoothed curve representing them the rate of change was de-termined and used f o r the arithmetical reduction of the densities at 7.978" to 8" 17.925" to 18" 28.064" to 28" and 38.203" to 38". The values for the other temperatures (every degree from 40" to 0") were got by taking readings directly from the curves these would be less ;iccul*ate than the former. (8.) As the extrapolation of these curves was extended as much as 8" below the last determination the values a t these ext'reme tempera-tiires cannot be regarded as more than approximate.Their accuracy was improved by differentiating all the values a t O" comparing the tlifferential with the figcires in Plates 1 and 2 and introducing such corrections as increased the smoothness of the curve. (9.) Finally as a check on the readings taken from the curves, the values for each solution at different tempcrntures were diffcr-entiated and such slight corrections made as would render them more regular. The accuracy of the values may be thus summed up :-The percentages represent the truth within about 1/10,000th of their total value their relative accuracy however is probably 100 times greater.The density values a t and near 38" 28" 18" and 8" are more accu-rate than a t other temperatures but with certain exceptions th 86 PICKERING A STUDY O F actual errors are probably nowhere greater than one or two units in the fifth decimal place. The exceptions are (1)‘all the solutions a t tem-peratures below about 5” or 6” (2) solutions of a strength of 30 per cent.* downwards a t temperatures above 20”. Here the errors may affect the fourth decimal place. About one quarter only of the full table is reproduced here ; some of the values which are utilised in Part VI are not contained in it, and these are given in Table VII A. PART III.-ELECTRIC CONDUCTIVITY. Crompton’s examination of F. and W. Kohlrausch’s determinations of the electric conductivity of sulphuric acid solutions (Trans.1888, 116) led him to conclude that these results when differentiated a first time gave a continuous but irregular curve ; and when differ-entiated a second time a series of straight lines meeting at the percentlages of 84.4 73.1 47.6 18.5 and 3.5 corresponding to the hydrates with 1 2 6 24 and 150H20 thus confirming the four changes indicated by Mendelkeff’s density curve and showing in addition a fresh change a t 24H20. I certainly think that these results if taken by themselves would scarcely warrant such conclusions though perhaps they might in some points acid confirmation to Mendeleeff’s bydratcs assuming as Crompton did at the time that these were already well established. Crompton’s results I found were also vitiated in parts by various arithmetical mistakes and I have therefore reproduced his table here (Table VIII) in a corrccted form.Fig. 4 A gives an illustration of my drawing of the direct first differential drawn as in the case of the densities with a bent ruler and having the separate sections prolonged. Fig. 4 B represents the second differential deduced also 2irectly from the experimental values. This appears to me to be made up of three straight lines and two curves and the breaks indicated by it correspond in one case only (the monohydrate) with any mentioned by Crompton. The portion VI does not contain sufficient points to admit of a second differentiation but the position of the first differential point at 99.7 per cent. indicates the possibility of another change between 99 and 100 per cent.? Besides the breaks shown by the direct * The actual determinations a t 38” and 28’ extended down to a 20 per cent.solution but the possible liberation of air bubbles rendered the results below 30 per cent. somewhat doubtful (see p. ‘74). t The values a t the higher percentages (marked by an asterisk in the table) were selected by Crornpton from W. Kohlrausch’s determinations. I have ex-amined the other determinations made by the same physicist but without gaining any further information from them THE NATURE OF SOLUTIONS. 87 FIG. 4.-Conductivity o f S u l p h u r i c Acid Sol.ttions. Direct First and Second Differentials. Per ecnt. I12S04. 10 20 30 40 50 60 50 80 90 100 10 20 30 49 50 f;0 ‘70 80 !)O 100 Per cent.HoS04 85 PICKERING A STUDY O F second differential there appears t o be another at about 3.5 per cent., since by differentiating the first differential curve I I always got a second differential consisting of two straight lines meeting at this point as shown by I',* Fig. 4 B. Considering the paucity of the determinations and the severe strain put upon them bytbo direct differentiations I think that they are fairly conclusive in favour of the second differential being recti-lineal and Crompton has done good service by drawing our attention to this fact.? The positions of the various changes which I consider t o be shown are the following although the indications can in no case be taken as 99.5 96.8 94* 84* 63 (or perhaps '73$) 37.5 9*5* and 3.5 per cent. Those marked by an asterisk are I think the best established.I am glad t o be able to state with Mr. Crompton's authority that, he agrees with me in the criticism which I have here given of these conductivity determinations and that he considers that the interpre-tation which he originally put on them must naturally fall through, now that it appears that Mendelheff was mistaken in his statemeut,s as to the densities for Crompton never regarded the conductivity results as sufficient to stand by themselves but only as affording additional proof to a theory which had (as he then imagined) been already satisfactorily established by other facts. proof : PART 1V.-H EAT CAPACITY. The unit heat capacity of a body is measured by the number of units of heat required to raise a unit mass of it 1" ; its specific heat is the ratio between its heat capacity and that of some standard substance under the same conditions the variable standard watei', being generally selected.The two terms are precisely analogous to " density '' and " specific gravity ; " the former measures a property of the substance the latter measures the difference between the pro-perties of two substances. The uuit of heat being the amount necessary to raise 1 gram of water from 0" to lo it is most important in ordinary calorimetry to * Placed one division of the paper higher than it should be. The numerical values are given in Table IX. -f The curvature of line I1 might perhaps be reduced to straightness if we had su5cient points to indicate changes at 9 and 19 per cent.such as my density results show. 2 According to the first differential the position of this change might bc 63 o r '73 per cent. the second differential favours the formcr view ; but a small error in one determination only would alter the case THE NATURE OF SOLUTIOKS. 89 know the heat capacity of water a t different temperatures. Such determinations as have been made of it show unfortunately very large discrepancies Regnault's results which are known t o have been miscalculated being generally taken. The value at lS" deduced from the four most concordant sets of determinations available,* is 1-006 ; in the present work I took 1.005 by mistake. The most accurate method of general applicability known for the determination of the heat capacity of weak solutions consists in dissolving some strorig solution of the same substance in water at two different temperatures.The difference in the heat evolved is equal to the difference between the total heat capacities of the reagents (water and strong solution) and that of the resulting weak solution. H - Hr = (U - V)(t' - t ) . To a-pply this various quantities of a sulphuric acid solution of 58 per cent. strength? were dissolved in water at 15" and 21" ( t and t ' ) : the rise of temperature noticed being r and r r in the two cases W being the weight of solvent water w that of the strong solution, wf the weight of the apparatus and c c' c" and C the heat capacity of the water the strong solution the resulting weak solution and the apparatus respectively we get-[(W + W)C" + w ' c ] ~ ' - [(W + w)c" + w'cI]r = [(WC + Zl"C1 + W C I ) - (w'c + (W + w)c"](t' - t ) , whence c, - (Wc + wc')(t' - t ) + wrc (rr - r ) - (W + w)(r' - r + t' - t ) A drawback to the general use of this method is that the heat capacity of the strong solution must first be known; but it need be known approximately only since an error of 1 per cent.in its value will introduce an error of only 0.0005 in the heat capacity of a 5 per cent. solution and will affect all the results regularly in inverse pro-portion to the strength of the solution obtained. The method assumes that the heat capacity of the liquids con-cerned is the same at the lower as at the higher temperature. This is not the case; but the difference is very small. Judging by Marignac's results the difference at temperatures 6" apart would be otily about &,th for most 5 per cent.solutions and the error f Regnault's recalculated by Bosscha Baumgartner's v. Miinchhausen's and + A solution of this strength gives a resulting solution of the maximum strength Henrichsen's. with the smallest possible rise of temperature. (See Laiidolt and Bornstein Tubellen 1'76. 9 0 PICKERING h STUDY OF thereby introduced would be almost entirely counterbalanced by the heat capacity of the water varying in the same direction as that of t’he solution. The determinations were made in the same way as those men-tioned in Part V. As it was practically impossible to take exactlg the same weight of the strong solution a t the two temperatures, the results obtained were plotted out into curves and the readings of these curves at points near those at which the actual experiments existed were taken.Two determinations at least were made near each point. Two opposite thermGmetric errors of 0.0005” in the determination of Y and r’* wonld cause an error of 0*00017 in the heat capacity found. With sulphuric acid the rapid increase in the heat evolved with the amount of strong acid used renders it impossible to plot out the values €or this evolution on a sufficiently open scale and the error may be doubled from this cause alone. This source of error was reduced in the present case by taking the readings for r (and r’), not only from the curves directly representing them but also from curves in which the values for r/w were plotted out.The error, therefore in the heat capacities might probably amount to 0*00025,-f-and judging by the regularity of the results themselves it never appears to exceed this. The determinations mere extended to a 12 per cent,. solution by a second series of determinations in which the strong acid was dis-solved in a 4.6 per cent. acid instead of in water. The heat capacity of this solvent acid had been determined by the first series and the two series were made to overlap. Results obYnined. The results with the weaker solutions are given in Table X t,he experimental values being quoted in the first four columns. The values for r and r’ for a round number of grams of solution w as determined from the curves directly representing these quantities, are given in the sixth and eighth columns whilst the nzean of these values and those deduced from the r/w and r’/w curves are given in the seventh and ninth columns.With the stronger solutions (Table XI) the values for !r and r’ were deduced solely from the T/ZV and r’/w curves.$ JE These of course are not dependent on the differences between two experi-f The average difference between Thornsen’s duplicate determinations is about Anyone plotting out these values will notice irregularities in the case of rlw mental values only. 0003 THE NATURE OF SOLUTIOKS. 91 Fig. 5 A illustrates the cliagram representing these results. It coiisists of four distinct curves of which TTI a n d I11 arc the best, established ; these bend in opposite directions and do not appear to meet a t all.The same want of continuity seeins to exist between curves I and 11 whilst there are not enough determinations (only two) to settle the nature of curve IV (the dotted line) a i d whether it touches cui’vc 111 or not. There are 4 9 l l (three of series 1 and FIG. &-Heat C a p a c i t y of S u l p h u r i c Acid S o l u t i o n s . Per cent. II,SO,. Fig. A. I ; 8 10 40 ti0 80 100 Per cent. H2S0,. Fig. B. eight of series 2) and 2 experimental points on these four con-stituent curves respectively. The break at 4 per cent. depends on a for about 3 grams and #/w foi* about 8.5 grams wliicli arc somewhat too great to be attributed to error. I n my drawing I have erred on the safe side by smoot,li-in9 out these irregularities.The values in brackcts in the table show what differ-ences their not being smoothed out would makc $1 2 PICRERlSG A STUDY OF clifference of 0.0013 or five tiriics the experimental error bcsiclcs of coiirsc being establislxxl by the general ciircction of the curves on either side of it tlie exact position of the change near 1 1wr cent. camlot be cietermiiied nncl iucleed the recognition of m y change at all hcrc is a difficult matter; it is best donc by carefully drawing CIWYC TI arid producing i t t o 0 per cent. wlien i t will be found that a l l tlie points below 1 per cent. lie above the prolongation.* Tltc first tliffereritisl obtaiiiccl directly (Tables X and XI) is sliowii i i t Fig. 6 B tlie clinnges arc fairly well mnrlied but the n:~turc of tlie constituent curve is soniewliat doubtful owing to the curvnturc h i n g very small i n comparison with tlic cxperimcntnl emor.A second differentintion (Table XI1 and Fig. 6 C) reduces these curves to straight lines. Fig. 6 A represents tlic first differential obtained from the smoothed curve Pig. 5 A (Table XIII) i t differs from the direct first djfierential in representing tlie constituent portions I and I11 as straight lines instead of curves ; this is but a natural result of t!ie process of smoothing and of the smallness of the differcnces on which tlie curvature in this case depends. FIG. 6.-Heat C a p a c i t y of Weak S u l p h u r i c A c i d S o l u t i o n s . from tlie latter. First cliff ereiitial from tile curve uiid from the experiments. Second differential dc - ~-dP - *005 - *01 - ‘005 Fig.I-; - .o* d?C + -002; 0 - _ cPp - ’002.; 2 4 6 8 10 Per cent. H,S04. * I n this and otlier cases it is sometimes advantagc~ous to drilw some prticuIar ciirve on a inore o p i scale t l ~ n could be clone accurately by plotting out the erperiiiieiits tliemelves. This may be effected by drawing the curve first on a small scalc~ tiiliing readings froiri this plotting these out on a large scaic and $ran-iiig a sniootliccl curve tlirougli them THE NATURE OF SOLUTIONS. 93 Whether the heat capacities require one OT two differentiations to reduce them to straight lines must therefore be regarded as some-what uncertain the evidence being in favour of the latter view. The positions of the changes are-9.85 4.0 and 1.0 per cent., of which the first two are very clearly marked, A point of some interest azises as t o these curves.The heat capacity must be a continuous function of the strength of the solution and hence the drawings here made of the results cannot be correct at the points near 1 and 4 per cent. where the curves are represented as not meeting each other. I think it probable that the main curves I IT and 111 are connected by short curves of a totally different nature resulting from and represeiiting a state of passage between the different conditions a-hich exist in the main curves. This is supported perhaps by the fact that there is one experimental point at 4.1 per cent exactly between the two curves. No such intermediate condition is noticed in the density or heat results bnt in the heat capacity we are studjing a property of a iiature totally different from that in these other cases; the density is a property undercfized conditions and the heat of dissolution is a measure of a portion of the potential energy in the substance also under $zed conditions whereas the heat capacity is the work which has to be expended on the substance in order t o alter its conditions.The smallness of the difference between the sum of the heat capacities of the strong acid + that of the water (or 4.6 per cent. acid), and that of the resulting weak acid is noticeable and might excuse a superficial observer in concluding that no chemical action occurred. In series I we find-Per cent. H2S04. c”.X c + c’.+ Diff. 2.67 0*9&084 0.98396 -0.00312 5-06 0.96125 0*96510 - 0.00385 while in series TI the cases of maximum difference are-Per cent.H2S04. c’’.” c -t c’.t Diff. 8.42 0.93631 O*c336ll + 0~00020 9.55 0.92668 0.92739 -0*00071 Calculating pure H2S04 (heat capacity 0.3332) and water as the constituents the diff ercnces though greater are still small-* Heat capacity of resulting solution. t IIeat capacity of components. vor,. r,vII. € 94 PICKERIXG h STUDY OF Per cent. H,S04 c”.* c + c‘.t Diff. 3 0.97812 0.98475 - 0~0@66S 6 0-9.5452 096469 - 0.01017 9 0.93137 0.94454 -0.01317 12 0.90862 0.93438 - 042576 Fig. 5 I3 (p. 91) represents the heat capacities of strong solution:. used in the subsequent heat of dissolution determinations as deducecl from Berthelot’s table (Mec.Chim. 1,496) which collects the values of Thomsen Marignac and Pfaundler. Beyond illustrating the general irregularity of this property as a function of percentage composition. and the probability of the more marked changes occurring a t about 10 50 and i 3 per cent. .they can afford no definite information as t o the nat,urc of the figure through lack of sufficient points. P A R T V.-HEAT OF DISSOLUTION. I n 187.5 Bei~tliclot (Ann. (=him. Phys. [5] 4 446) published his determinations of the heat evolved when solutions of different strengths of the c’onimoner acids and alkalis are diluted with an excess of’ water and concluded t h a t there were signs of changes near certain i)o:ntq indicating in an appi-oximate manner the existence of hydrates ill solution.His determinations were numerous but would not st’carn to bo acrurate enough for settling the question, even if sufficient clatn had bccn quoted to permit of their recalculatioii into a form s;nit:tlile f‘or mathematical analysis. Thomwn afterwards (Tlzewnochem. Untersuch. 3 1-114) went afrebli 0 ; cia tlie same ,ground,§ and also investigated the cases of many salts as ~wc.11. His conclusions-that the action was regular and that liydr-itcq Iiigher than those known in the solid form did not exist in ~,lritio~i-wei~e in direct opposition to Berthelot’s and ham gained n rc1;idy acceptaiice amongst chemiets.11 The points deter-miner1 oil ewh curve rnrely exceeded six7 (sometimes there were only ThoriIwBii’s vonclnsioiis are indeed rcmarkable. * He:] t cripn(*ity of resulting solution.t Heat (1;LI)ncity of comlmnents. 1 The i i i i l i:il t emperatitres sometimes varied 4’ or 5” and the heat capacitics used w01 e v ~ ~ y uiiccrtain. $ He Ii**rl piwiously (Bv. 18’70 496) investigated sulphuric acid as aleo had Hess Gwl~ 1111 Abria Farre and Silbermann and especially Pfaundler (Bey., 1869 122). 11 “ TIIP gt ni~nlly nilopted ‘ hjdrate theory ’ of solution can scarcely be expected to fiurviv~. I r r disscrninntion cf Thornsen’s researches ” (P. Muir Elements 01 Thermol I/m isfry 16’7). 7 And t l i ( w points were' least suited for showing irregularities since they cor-responded to 4rnple molccular proportions (see a remark of Crompton’s PTOC. THE NATURE OF SOLUTIONS. 95 three or four) a number scarcely sufficient for determining even the general character of the reaction and yet in many of the cases where they appeared sufficient they showed that this general character was eminently irregular.In 10 cases out of 53 the value of the reaction changes its sign in two others the irregularity is palpable and even in those actions of greatest apparent regularity it is found that no simple equation can represent all the results. The equations which express the results with strong solutions generally give values too low when extended to weaker ones* (see Chem. News 54 215) ; nor can they be extended in the opposite direction with any better success. Thus Thomseu has extended his curve in the case of copper chloride so as to give the heat of dissolution of liquid CuC12,2H,0 ; if he had extended it further still so as t o get the value for CuCl, itself he would have recognised the inadmissability of such a process ; for the value obtained (9333 cal.) gives as the heat of formation of liquid dihydrate from its liquid components 1434 cal.whereas this quantity should be nearly identical with that of its formation in the solid state from the solid components which direct’ experiment gives as 3540 ca1.t With calcium chloride the case is still more striking. Thomsen’s determinations give-CaCl,,lOH,O + mH20 m m + 5-08 2,508, from which the heat of dissolution of liquid CaCl,,GH,O should be 11801 cal. instead of 6640 cal. as my direct determinations prove it t o be and that of liquid CaCl should be -3460 cal. a negative quantity whilst even the soZid salt gives a positive evolution of as much as 19250 cal.Thomsen’s results with sulphuric acid are of special importance, not only from his having studied this acid more fully than any other substance (12 points were determined !) but that the seeming regu-larity of these results is for ever being urged against the hydrate theory of solution. The language in which he expresses his con-1888 38 of whish Thomsen’s values for SOs and SO, iH,O form cxcclient illustra-tivns ; the former lies within experimental error on the extension of his curve for H2S0 + .nH,O the latter is very far removed from it). * Thus Thomsen’s equation for sulphuric acid is inapplicable beyond 19H,O, that for nitric acid beyond 5H20 and that for hydrochloric acid seems to be inap-plicable everywhere.t If the heat of fusion of the hydrated salt is (us Person concluded from tlie instanjes which he examined) equal to the sum of those of the anhydrous salt and water present the identity would be absolute. I think it probable that in many cases this equality does not hold good (indeed there are instances in which it cer-tainly does not) but the discrepancy could hardly make as much difference as in the instances quoted above. T3 96 PICKERING A STUDY OF clusions as to the curve representing them is of a most positive and uncompromising character. " One sees," he says (111 8) " that the fixed points can be united by a regular curve and that throughout, there are nowhere any signs of an irregularity such as would indicate the formation of definite hydrates." He then gives the equation for the hyperbola representing this curve and a table showing the near agreement of the found and calculated values for H,SOI + sH20, when the values for x are 1 2 3 5 9,19 and 1600 and then proceeds, '' One sees that the differences are very small and are positive or negative indiscriminately." There is therefore no doubt hut that the heat evolved on diluting liquid sulphuric acid with water is a continuous function of the water used and excludes absolutely the acceptance of definite hydrates as existing in the solution." The concordance of the found and calculated values for these seven points is indeed remarkably good but even if it were absolute i t could scarcely warrant siich a conclusion especially when it is remem-bered that the portion where this agreement exists is but little more than one per cent.of his whole curve,? and that throughout the remain-ing ninety-nine hundredths of it there is no a,greement at all the differences reaching as high a value as 683 cal. If any conclusions at all can be drawn from Thomsen's work they must certainly be in favour of the probable irregularity of the actions in question. * If the differences between the calculated values and the experimental points are tabulated for all of the latter the indiscriminate arrangement of the + and -errors is by no means apparent. They are-x = 1 2 3 5 9 19 49 99 199 399 '799 1599 Diff. = - 3 + 14 - 30 - 27 + 66 - 59 - 544 - 683 - 636 -467 - 180 + 17 Had Thornsen extended his determinations to more dilute solutions he would have found still greriter differences.My value for 2 = O( (determined by the extension of an experimental curve which reached as far as 0.04 per cent. or 15,000 H,O) compared with the value calculated by his equation gives a difference of + 3233 cal. f- Measured along the abscissae which represent the number of molecules of water added to H2S04 the concordance is exhibited as far as the 19th H,O the curve extends to 1599 H20. Measured along the curve itself the relative propor-tions of the regular and irregular parts would depend on the scale used with the scale adopted by Thomsen in Fig. AB Table I vol. iii the regular part would measure 1.8 per cent. of the whole figure. It is but fair t o state that Thomsen afterwards (Zoc.cit. p. 56) suggests that this aberration which must invalidate his conclusions may bc due to the contraction on mixing but he offers no evidence in favour of such a view and no explanation of why the effect of this action should be appreciable only in those cases (weak solu-tions) where it occurs to the smallest estcnt THE NATURE OF SOLUTIONS. 97 Graphic Representation of Zesults. The question as to the plotting of the results becomes of paramount importance when a mathematical analysis of the curves obtained has to be performed. The drawings which I give in Fig. 7 refer t,o calcium chloride, and are better for the purpose of illustration than the more com-plicated results with snlph&c acid ; the heat very similar in the two cases.FIG. 7. curves however are When the densities are plotted against percentage composition (A) or when the molecular volumes are plotted against molecular composition (B) rectilineal or parabolic figures are obtained in each case differing only in distribution of the experimental points along them (the same six experiments are plotted in all the figures) ; but if we plot the densities against molecular composition (C) or the molecular volumes against percentage composition we get a compli-cated hyperbolic figure most unsuitable for mathematical analysis. Yet such is the method of plotting which has heretofore been adopted in the case of all heat results ; the heat of dissolution or formation, of a $xed weight of the dissolved substance (the so-called " molecular " heat of dissolution but molecular in a sense different from that of molecular volume since it bears no reference to the molecular com-position of the solutioib) being plotted against the molecular com- 98 PICKERING A STUDY OF position of the solution.Fig. 7 F (p. 97) represents the heat of‘ dissolution thus plotted; if inverted it will represent the heat of formation (Fm) as plotted by Thomsen. As any quantity expressing the heat of dissolution in a manner corresponding to “molecular volume’7 has no meaning (such as molecular weight + heat of dissolution) we must abandoii the repre-sentation according to molecular composition altogether. I plotted out in the first instance the heat of dissolution of various solutions containing gram-molecular proportions of the salt against percentage composition (E) .On differentiating t,his however no satisfactory results could be obtained another apparently continuous curve being produced. Mr. Crompton then suggested that instead of plotting the heat of dissolution of gram-molecul& proportions we ought to take the heat of dissolution of 100 grams of the solution (Fig. D). Satisfactory results were obtained in this way. Similar results would no doubt be arrived a t by taking the heat of formation of 100 grams of the solution ; such a curve instead of falling from a maximum down to zero would start at zero (at the point where no water was present) rise to a maximum and then fall back again to zero. Method of Experimenting. Two classes of determinations were made :-1.“ Mixing ” experiments in which 420 C.C. of the solutions were mixed with successively equal volumes of water. A detailed descrip-tion of the new mixing calorimeter devised for this purpose and the method of using it will be found in the Phil. Mag. March 1890. 2. “ Dissolution ” experiments in which 4 to 30 grams of a strong solution were dissolved in 600 C.C. of water (ibid.). Calculation of the Results. To be comparable all the results must be calculated out for the same final dilution this must be not less than that obtained in the determination with the weakest solution used (0.04 per cent.). Theoretically the dilution should be infinite and the heat evolved f o r infinite dilution is in this case easily obtained by plotting out the results (for dilution down to 0.04 per cent.) against percentage com-position and extending the curve as far as 0 per cent.All the present results are expressed for infinite dilution. Let t and t’ be the initial and final temperatures in any determination, where w grams of solution is mixed with W grams of water the water equivalent o f the apparatus being w’cI and the heat capacity of the resulting solution being c“ then the heat developed by dilutin THE NATURE OF SOLUTIOXS. 99 100 grams of the solution to the same extent as in the actual experi-ment (D'OO observed) will be-100 - l O O [ ( W + W)C" + W'CI](tl - t ) D - -__I___ _-__. 100 To reduce the values for* D'OO in the various mixing experiments to what they would be if the final strength of the solution was the same as in the last determination of the series we must add to each the heat evolved on diluting the (about) 200 grams of weak solution obtained with a further equal volume of water (determined iri the iiext experiment) then that evolved on diluting the resulting (about) 400 grams to a similar extent and so on till the litst stage of dilution is reached.'I'hesc values are then plotted ont and the heat evolved on diluting the last solution with infinity of water is found and the whole results recalculated with this as the zero point. Thus the heat of disnolution of any solution to infinity (DGO) or even to a state of considerable dilution depends not only on the actual dctermina-tion with that solution but on all the other determinations with weaker solutions increased in geometrical proportion." The correc-tion " for infinite dilution t o be applied to the observed values in the case of sulphuric acid is particularly large with an 18 per cent. solution it amounts to 656 cal. that is twelve times the quantity measured (47 cal.) in the actual determination. Jt is evident from this how inadvisable i t is to extend the mixing experiments to solutions of greater strength than is absolutely necessary. With the dissolution experiments the " correction to itifinity " is relatively small since the strength of the solution obtained in any determination is far less (2.5 to 0.25 per cent.) than in a mixing experiment; but even here it attains very large proportions some-times amounting to nearly 3(300 cd. and sometimes exceeding the evolution measured in the determination itself.The correction* has to be deduced from the curve representing the results of the mixing experiments very carefully enlarged in the manner indicated on The so-called " molecular '' heat of dissolution is deduced from that of 300 grams by multiplying the latter by the molecular weight of the anhydrous substance and dividing by the percentage composition of the soluiion. p. 92. X DZa = D'W obs. + ( D'dp"w?) in which D F is the lieat of dissolution W of 100 grams of a solut,ion of the strength of that resulting in the determination, ;tnd W and w the weights of water and strong solution taken respectively 100 PICKERIKG A STUDY OF Eqwriwental Error. The average difference in the readings of the heat evolved as given by the two thermometers used together in the mixing calorimeter, was O*OUO8',# which would represent a difference of only 0.13 cal.in the two values for DlG0 in the one determination ; but the values for D g would contain an accumulation of similar errors from all the determinations with weaker solutions so that it is impossible to estimate the probable accuracy of these latter values except by com-paring them with the dissolution experiments (see below). In the dissolution experiments the average difference in the inde-pendent duplicates was found to be 7.6 cal. (in the corrected values for DF) so that the mean error of a single determination is 5.4 cal., and that. of the mean of two determinations is 3.8 cal.,? the probable errors being two-thirds of these quantities.Two and often three, determinations were made a t each point. This error is certainly very small since an error of only 0*001" (0.1 mm. of the thermometric column) in the rise of temperature measured would on tbe average make s difference of 4.1 cal. in each result and the errors entailed in the application of the large correc-tion for dilution to infinity could not from a priori considerations be placed at less than double this value. Yet the smallness of the actual error is well established by some 150 determinations in the present case, and in other cases whei-e the correct'ion t o infinity is smaller and the weight of solutions taken can be larger the mean error deduced from some 200 experiments is less than half the value here given.Besides the magnitude of the " correction " and the small weights of strong acids which have to be taken,$ there are other special sources of error in the present case owing to the rise of temperature and consequently the rate of cooling being more than usually great, and owing to the tubes from which the acid has been emptied having to be weighed out on a delicate balance as well as may be between the one minute intervals when the thermometer is read and while often, a drop of the hygroscopic acid is running down the outside of the tube. In differentiation we depeiicl on the relative accuracy of consecutive observations and the error in comparing these would not I think be much greater than in comlmring duplicate experiments with the same solution the only cause (provided the compositions of the solutions are known correctly§) which could make i t larger would be X Deduced from more determinations than the present paper contains.t Varying from 6 to 2 cd. with the strength of the solution dealt with. $ To reduce the great rise of temperature. tj Errors in the percentage composition of R magnitude such as are shown to be pro-bable from the irrzgularities in the first diff erentiels of the densitics ~ . v ~ u l d scarcc1. THE NATURE OF SOLUTIONS. 101 the different strengths of the final solutions and consequently their different heat capacities and the difference i n the correction for infinity ; but the quantities taken in duplicates generally differed so much that there would often be a greater difference in final drength with the duplicates than with consecutive determinations.Such a difference in tihe quantities taken though it increases the divergence of the duplicate results from the mean tends to bring that mean nearer the true value. Results obtained. Of the two series of mixing experiments (Table XIV) the first and fuller one was owing to the frequency of the changes exhibited with very weak solutions alone available for extension 5 3 as t o get the value for infinite dilution. A curve obtained by plotting out the values for D T / p was in t'his and other cases found more adapted for extension t o 0 per cent. than the D&!" curve itself. In the calculation of the last determinations i n both series errors were subsequently discovered (indicated in the table) but as the difference made by them was far within the limits of experimental error it was not thought worth while to recalculate the results in consequence.The stronger solu-tions used in these experiments were made up independently of and therefore had not quite the same composition as the solutions obtained in the various preceding determinations ; to apply the small corrections necessary on this account the results had first to be calculated on the assumption that the two solutions in question had been identical (values given in the column headed DT approx.) the correction determined from the curve which these values formed and then the various results recalculated. Leaving these for the present the dissolution experiments (Table XV) will be examined first.On a small scale they form a curve of great seeming regularity very similar to that shown in Fig. 7 D (p. 97). The difliculty of treating them on a scale commensurate with the experimental error is con-siderable owing to the rapidity of the increase and the magnitude of the quantities measured. They were generally treated in three over-lapping portions drawn to different scales the total length of which was s3me 13 feet. The two extreme portions of wliat I consider as being the smoothed curve representing these results are illustrated though very imper-fect,ly by Fig. 8 (next page). The whole figure is continuous throughout, have an appreciable effect on the heat of dissolution results since the error thereby caused in the plotting of the results is partially counterbalanced by an opposite error introduced into the calculation of the magnitude to these results 102 PICRERING A STUDY OF except in the vicinity of 20 and 50 per cent.where the discontinuity must I think be due to some error. Like the density curves it can only be drawn in separate sections but uulike these the sections all cut each other on prolongation. The average divergence of the experi-FIG. &-Heat of D i s s o l u t i o n o f S u l p h u r i c Acid Solutions. Per cent. H,S04. Fig. A. Pol 82.5 8 i -5 90 92.5 95 100 5 10 12 -5 15 17.5 20 22.5 Per cent. H2S04. Fig. B. mental points (the means of two or three determinations) from the curve drawn was almost exactly what it should be according to the estimate of the error deduced above from the difference between duplicates and varied from 6.5 cal.with the stronger solutioiis (100 to 80 per cent.) down to 2.1 cal. with the weaker ones* (30 to 3 per cent.). There are on the average about six mean experimental points on each section. Of the various changes that a t 9 per cent. # Somewhat larger errors may be noticed in this and in other cases where two solutions of nearly the same strength have been used. But in such instances the second debermination was made owing to there having been some doubt a8 to the accuracy of the first they would therefore lend to a wrong estimate of t.he average error THE NATURE O F SOLUTIOYS. 103 is perhaps the most marked some change a t about 99.5 per cent. is established by the highest two points (containing six determinations), but these are hisufficient to permit of any drawing bcing made of thc cmve on which they lie.On differentiating this smoothed curve ( T d h XVI) thc results FIG. 9.-Heat of D i s s o l u t i o n of S u l p l i u r i c Acitl S o l u t i o n s . First Differentials and Second Diffcrcntid. Per cent. II,SO,. - _ d D dP 10 20 30 40 50 60 70 so 90 100 ffir A 10 20 30 40 50 60 70 SO 00 100 Per writ. I12S0, 104 PICKERING A STUDY OF illustrated by Fig. 9 B (p. 103) were obtained each separate section of the original figure yielding either a curve or a straight line as a first differential. Two details may be mentioned :-(1) that a drawing of the original curve may be made obliterating the change a t 36 per cent. but that the differential from this drawing (given at the end of Table XVI) still shows a change here ; (2) that no change a t 84 or 73 per cent.-the composition of the mono- and di-hydrate-is noticeable i n the experimental curve and only becomes apparent after differentia-tion.The first differential curve was then differentiated (Table XVI) and a discontinuous figure which was rectilineal throughout was thereby obtained (Fig. 9 E p. 103). The comparative magnitude of the changes at the higher percentages is the chief point to be noticed respecting it. On reconsidering these results grave doubts were entertained as to the correctness of the conclusions drawn. It seemed possible that there might be some unknown sources of error in the experiments, and that the splitting up of the curve was but a consequence of the manner in which the figure bad been drawn.To settle this point the results were again plotted out but on a scale about half as open as before and the curve was drawn without paying any attention to the position of the changes previously noticed and purposely smoothing out all seeming irregularities. The only part which could not legitimately be smoothed over was the change at 9 per cent.; the curve therefore starts at this point. The readings of this curve will be found in Table XVII and the diffe-rential deduced from them is illustrated by Fig. 9 C. I t certainly offers ample justification for the interpretation given to the large scaie drawing ; for in spite of the excessive smoothing it still shows the majority of the changes which the latter had shown ; their exact position has in some cases been slightly altered by the smoothing, and in four cases-namely at 97.5,* 88,s 84 and 36 per cent.-they have been entirely obliterated but the general agreement is far better than could have been expected.Table B p. 109 gives the position of the changes shown by the two drawings. The application of a direct differentiation to the experimental values proved a matter of great dif€iculty and no success was obtained till the error was diminished by treating them in two series (last two columns of Table XV).? The result of this treatment how-* Of these two however there were some indications ; those a t 84 and 36 per cent. were very feebly marked even in the large curve. t See footnote p.74 ; there are here so many pairs of differential points deduced from determinations wit.h solutions of nearly the same strength that I have marked them A and A' &c.; A being a differential deduced from two deter THE NATURE OF SOLUTIONS. 105 ever is to diminish the curvature of the constituent parts of the differential and to obliterate two of the more feebly marked changes at 84* and 97.5 per cent. Fig. 9 A (p. 103) illustrates the figure thus obtained.? On applying a second differentiation to this (Table XVIII), a rectilineal figure was the result closely resembling the second differential obtained from the smoothed experimental curve (Fig 9, E) except in the parts between 80 and 88 per cent. and between 94 and 100 per cent. where the nature of the first differential was uncertain and the straight lines here drawn to represent i t gave two horizontal lines on the second differentiation.No illustration of this second differential is here given. The remaining drawing D in Fig. 9 is worthyof special attention. It shows the results of pushing the smoothing process still further than was done in what I have termed the ultra-smoothed curve, which gave the differential Fig. 9? C. If instead of taking readings from the ultra-smoothed experiment,al curve at every 1 or 2 per cent., we take them at considerable distances say at every 5 per cent. greater smoothing will be produced in the differential obtained from them ; but though the majority of the changes are thereby obliterated the magnitudes of those near certain points are intensified and we get Fig.9 D which cannot. be drawn except as three distinct curves, one of which bends in the opposite direction to the other two. The mixing and dissolution experiments have to be treated sepa-mtely for where they overlap (5 to 18 percent.) there is a consider-able difference between them. This difference reaches as much as 16 cal. but as it mould be caused by a single error of 0.06 cal. or 0.0003° in the experiment with the weakest solution (0.08 per cent.) -supposing all the other determinations t o be absolutely correct-it must be considered to show a remarkable degree of concordance.: In dealing with these mixing experiments the rapid increase in the heat evolved rerideis it necessary t o plot them out to several different scales.The most extensive of these is illustrated by minations marked A and A’ bcing one from one determination marked A and one marked A‘. * The irregularity of the points near this percentage indicated the probability of some change. -f The line XVIlI is of course not established and is only of use in showing the existence of a change above 99 per cent. The drawings of the various smoothed experimental curves do not extend beyond this percentage. $ Better concordance might be obtained by re-calculating the mixing experi-ments starting with the strongest solution and working downwards instead of starting with the weakest solution and workiug upwards as in the first case. The actual value to be taken for the strongest solution would hare to be deduced from the results of the dissolution experiments a considerable number of these bcing performed a t the point in qucstion.But there were not enough points to definc it 106 PICKERISO A STUDY OF Fig. 10 A :* the changes which occur in it of wliicl~ the best marked one is a t 4 per cent. are of the same character as those in the region FIG. lO.-l€eat o f D i s s o l u t i o n of S u l p h u r i c A c i d S o l u t i o n s . Citl. 350 300 250 A 200 150 100 50 Mixing Experiments. 0.05 0.10 1.5 *20 Per cent. H2S0,. Fig. B. 1 2 3 '1 5 ci '7 H 9 Per cent. &SO4. Fig. B. of stronger solntious. The general changc in direction at the extreme end is ve1.y remarkable it is i1lustr:Ltecl on an eiilarged scale in Fig. 10 13. The curve here is of such a nature that i t cannot be drawn coiiveniently with a bent rider ; and more satisfuc-tory results mere obtained by plotting out the values for D?/JI as sliown in Pig.11. The results of the secoiici series of experiments are distinguished from the others by crosses ; the point a t 0.88 pel. cent. appears to be somcwhat anomalous ; but, whether wc include i t or omit it (as has here beeu done in making the present drawing) it will be apparent that the inclination of the curve becomes very abrupt somewhere about 0.4 per cent.,t and that * Thc liigliest clctenninatioii that at 18 pc" cent. is too far from the next lower one (9 per cent.) to bc of any service in drawing the curves. t It may be licrc noted that this rcxnarkable changc could not liave been revealed by Tliomscn's determinations sincc his results endcd M itli a 0.34 per cent.solu-tion. The importance of cbrtcncling any series of determinations to the greatest, tlilution possible is rendered rcry evident by tlic present rcsults. T!icre arc two sliglitly different drawings of this curvc below 0.16 per cent. j both are treated in Tablc XX. This figure is well woi-thy of attention. THE NATURE OF SOLUTIONS. 107 FIG. 11.-Heat of D i s s o l u t i o n of S u l p l i u r i c Acid S o l u t i o n s . Calories + percentage strength. Cal. 0.1 0'2 0 . 3 0.4 0.6 0 . 6 0 . 7 0.8 0.9 1.0 1.1 1.2 Per cent. H,S04. about 0.1 per cent. it alters its direction again bending in the oppo-site direction. If this form of plotting the results be extended to stronger solu-tions the curve again bends upwards presenting on the whole the same general appearance as Fig.7 E (p. 971 the values for DF/p being of the same nature as those for the so-called molecular heat of dissolution. The differentiation of the mixing experiment curve (Table XIX)* is illustrated in Fig. 12 (ncxt page). The first differential A is composed of well-marked curves ; whilst the second one B is rectilineal within experimental error. The comparative magnitude of the changes with very weak solutions may be judged by the portions 111 and IV which are reproduced as dotted lines in Fig. 9 B and El. On a larger scale a certain amount of curvature in the second differential line I is notice-able but this is probably due to error. A difference in the nature of the curvature of line V in the first differential as shown by the mixing or dissolution experiments will also be noticed and is no doubt due to the rapid accumulation of small errors in the former.The number of changes which existed with weak solutions showed that it would be useless to perform a direct differentiation of the experimental values owing to the comparative paucity of the detey-minati on s . * The values for the weakest -solution were deduced from readings of the D?/p curve 108 PICRERING A STUDY OF FIG. 12.-Heat of D i s s o l u t i o n of S u l p l i u r i c Acid Solutions. First and Second Differential from the Smoothed Curves. Per cent. H,SO,. dD 1 2 3 4 5 G 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Per cent. H2S04. The positions of the various changes show1 by the heat of dissolu-tion results are collected in the following table B.The ultra-smoot.hed curve gives approximations only the results obtained from an ex- -amination of the smoothed experimental curve (together sometimes, with that of the first differential deduced from it) are more trust-worthy than those from the direct first differential. The mean does not include the values in the first column. The concordance of the position of the changes as determined by the two methods of analysis is very good; yet the heat results generally cannot be regarded as very satisfactory when considered by themselves and looking at the number of determinations which were made and the care which was taken about them they proved somewhat disappointing.Perhaps such a result should have been anticipated owing to the thermometer being an instrument 50-100 times less delicate tha THE NATURE OF SOLUTIONS. 109 TABLE B.-Positio?t of the Changes in the Heat of DissolzctiorL Czcrues. From the 1st diff. of tlie ultra-smoothed curve (approximate). the balance on which the densities depend and to the fact that the results of the individual determinations cannot be util ised directly as the densities can without the application of complicated calculations and corrections. Many of the changes are so feebly marked that much uncertainty must prevail as to their existence and position ; it was only by the expenditure of an amount of time and trouble of which the present account can give but a faiiit idea that any definite .conclusions could be arrived at.It is n o t difficult to see the reason of this. The magnitude of the individual changes appears to bear no relation to the magnitude of the total heat evolved or to the rapidity with which this increases with the strength of the solution. Tliese last two peculiarities of the reaction have the sole effect of renderiiig the experimental error almost double what it would be in any ordi-nary case and of necessitating the adoption of methods (differentia-tion in alternate series) which tend to obliterate the sharpiiess with which the changes of curvature are marked. Table C (p. 110) contains the values for the so-called molecular heat of dissolution (D; ). The strengths entered are either such as corre-spond t o the hydrates which exist according to the present work or such as will permit of a comparison of my own determinations with those of Thomsen and Pfaundler (Thermockein.Untersuch. 3 54 and * Very uncertain. There might be a change on each side near 76 pcr cent., instead of one a t 76 per cent. t Probably rather higher. Pickering. ( d ) From '' dissolution " experiments. (112) From I f the composition of the acid as given by the freezing point determinations be taken, for D',"O for 100 per cent. mill be 50 cal. too high that for a 50 per cent. solution 18 cal. to 112 PICKERING A STUDY OF TABLE D.-Heat Evolved on Addition of Various Proportions of Water to Sulphuric Acid Solutions at 17.91”. Jakresb. 1869 15.32). Table D gives the values for the heat evolved on adding various proportions and also successive proportions of water to the strong acid.Where there might be a doubt as t THE NATURE OF SOLUTIONS. 113 whether the values were deduced from the mixing or dissolution experiment the letters rn and d have been inserted. The error in these absolute values is of course very different from that of the relative values of successive determinations which has heretofore been alone considered. A difference of 0.001” or 0.2 cal. in the determination of the value for a 0.4 per cent. solution would make a difference of 50 cal. in the heat of dissolution of H,SO, ibself ;* yet the concordance of the mixiug and dissolution experi-ments showed that the actual error accumulated as far as 18 per cent,. could not have amounted to more than one-seventh of this so that thc value for HzS04 is in all probability right considerably within 50 cal.To compare my values with Thomsen’s and Pfaundlcr’s these lattei. must all be increased by 2124 and 3411 cal. respectively the heats of dissolution of the weakest solutions (0.34 and 4.377 per cent.) which these two physicists obtained in their final experiments. Pfaundler’H values it will be seen agree very satisfactorily with my own thew being but one instance in which the difference between our results exceeds 100 cal. whereas Thomsen’s results are all considerably lower the difference amount’ing to as much as 1112 cal. in the case of the pure acid. This difference I imagine may be caused by errors due to Thornsen’s mixing calorimeter or to the acid which he used not being as strong as H,SO /.It will be seen by the footnote at the bottom of Table C (p. 111) that my values are probably somewhat high but the excess would be almost exactly counterbalanced by taking 98 instead of 97.82 as I have done for the molecular weight of HzS04 Pfaundler and Thomsen took the former of these values. The manner in which the results are stated in Table D does not show the discrepancy between Thornsen’s results and those of Pfaundler and myself so forcibly. A considerable difference will be noticed however in the case of H2SO4799HZ0 800E,O-Thomsen’s last determination. There is certainly I think an error here and it mill affect all his values for DZ (see Trans. 1889 325). It will be seen from the upper portion of the last column of Table D that the heat evolved f o r H,O on the addition of successive portions of water diminishes till 199H,O have been added that it then increases and again diminishes.From the last portion of this column if will also be seen that the heat evolved on adding the first * I f it were determined by mixing the acid with consecutively eqzcal volumes of water as in the “ mixing ” experiments the error of 0.2 cal. a t 0.4 per cent. would make an error of 500 cal. in the value for H,SO,. j- He says nothing about the preparation or analysis of it and his values are such as would be obtained with a commercially ‘‘ pure” acid of 97-98 per cent. Pfaundler took pains to prepare his acid by crystallisation 114 PICKERING A STUDY OF 4 mol. of water t o I12S0 is considerably less instead of greater than that evolved on the addition of the second +H,O as i t should be if the curve mere regular throughout.Thus there are at least two portions of the curve which even on a cursory inspection show that it does not possess that general regularity which the supporters of the physical theory of solution are so anxious to attribute to it. PART VI.-EXFANSION BY HEAT. A. As a F u n c t i o a of Percentage Composition. The density determinations made supply the means of calculating the expansion of the solutions by heat. In doing so we depend on the comparison of two series of results at diferent tcrnperatures and, hence the conclusions drawn may be regarded as almost independent of those already drawn from the densities themselves where we depended on the results of one series at the same temperature.If s1 and s2 be the densities at t and t, the expansion or increase in volume will be 5 - 1 the volume at the lower temperature tl being taken as unity. In this way I have calculated the expansion in several cases but have preferred to take for the purposes of analysis, the complementary property of the decrease of density s1 - s2* the figures in the two cases are of a similar nature but the latter present the advantages of (1) an error in the densities causing an error of the same magnitude throughout ; (2) an error in the density deter-minations having generally a smaller effect (actual and relative) on the results ; (3) of necessitating fewer calculations and possibilities of mistakes.The solutions used a t 17.925" had not the same composition as those used at the other temperatures so that if we take the actual experimental values we can only get the expansion for about lo", from 28,064" to 38.203" ; for about Z O O from 7.978" to 28.064"; and for about 30° from 7.978" to 38.203". These values are given in Table XX. To utilise the results at 17.925" the values as reduced t o round percentages (and also t o whole degrees of temperature) given in Tables VII and VIIA must be taken. From these we get three series for a rise of lo" two for a rise of 20" and one for a rise of 30" ; they are given in Table XXI and will probably yield more uiiiforin results than t)hose derived from the unreduced densities since in the reduc-tion a few obvious errors were corrected (see p.70). In the case of the numerous results with strong solutions and for s2 * This divided by the density a t the higher temperature gives the expansion results below 20 alone sufficiently also contains the inter vals . Fig. 13 gives a THE NATURE OD' SOLUTIONS. 115 percent. the reduced densities at 8" and 18" are trustworthy for t'he present purposes. Table XXI values for the expansion proper for the three 10" representation of some of the results obtained the signs of the values for the decrease in density having bcen changed to + so that the figures may be more easily compared with those foy the expansion proper. A C and D give the decrease i n density for FIG. 13.-Decrease of Density a n d E x p a n s i o n of S u l p h u r i c Acid Bolut i o n s o n b e i n g h o a t e d 30° 20' a n d 10".Per cent. H2S04. 0 10 20 30 40 50 60 70 80 90 100-~ ~.,-0 10 20 30 40 50 60 70 80 90 100 Per cent. H2S04. intervals of 30° 20° and lo" deduced from the reduced values ; E and F (which should be raised $ and 1 division of the paper respectively), the values for the other intervals of 10" ; G (which should be raised 18 div. of the paper) shows the expansion proper for 10"; and B (which should be raised + div.) gives the decrease in density for the 30" interval as deduced from the unreduced density values. All ths figures resemble the first differential of the densities in being continuous thronghout,* and yet necessitating their being f Except in some parts of B where the discontinuity is probably the result of errors 116 PICKERINQ A STUDY OF drawn in sectiong when a bent ruler is used for that purpose.The separate sections are indicated in the present illustration in the cases of A B and the end portions of D only.* The agreement of the experimental points with these drawings is somewhat better than in the case of the first differential of the densities (Plates 1 and 2 p. 184) when the same scales are used. Both as regards the position of the changes and the nature of the constituent curves the various figures resemble each other so closely, that I have analysed by differentiation that only which refers to the 30' interval supplementing it in the case of very strong and weak solutions by a similar examination of Fig. D. The direct differentiation (Table XXI and Fig.14 A) gives a fignre composed of three straight and two slightly curved lines there being three parts of it in which the results are too doubtful to permit of FIQ. 14.-Expansion of S u l p h u r i c Acid Solutions. First and Second Differentials. tEE 10 20 30 4Ll 50 60 70 80 90 100 "' Per cent. H2S04. -0 for Fig. D - '025 0 10 20 30 40 50 60 70 80 90 100 Per cent. H,S04. any drawing being made. Indeed it could hardly be expected that the decrease in densities would stand a direct differentiation for the original values here are merely the differences beheen two experi-mental results and even in the case of the 30" interval they are only comparable in accuracy with the first differential of the densities.A differentiation of the smoothed curve (Fig. 13 A p. 115) led to better results (Table XXII and Fig. 14 B) and supplied sufficient data * I n plotting out the results from the unreduced densities considerable assist-ance may be derived from a study of Plates 1 and 2 which will indicate whicli determinations probably cont,ain exceptionally large errors THE NATURE OF SOLUTIONS. 117 for the drawing of those parts where the direct differentiation had failed. The whole of the figure from 50 to 88 per cent. is rectilinear, the other portions of it being made up of gentle curves. A second differentiation (Table XXII) reduces all these curves to straight lines, and gives the figure illustrated by C Fig. 14. There are of course, non-comformable points between most of the lines constituting this figure.The first differential of the curve for the 10" interval for weak and strong solutions will be seen from Fig. 14 D to be made up of sundry straight lines the portion 111 does not appear to apply quite as far as 0 per cent. the last few points in the diagram exhibiting an irregularity which indicates the occurrence of some change at 0.3 to 0.4 per cent. ; but this is simply a suggestion. Table E (p. 118) summarises the information as to the position of the changes derived from a study of the various results. The numbers given at the bottom of the columns refer to the weight which has been assigned to the figures in that column in deducing the mean : the weighting has been made in proportion to the interval of teni-perature to which the expansion applies the results drawn from a direct differentiation the differentiation of the smoothed curve, and also from the inspection of the original curve count as indepen-dent observations so that the value attached to the results from that curve in which the changes have been established by differentiation may amount to two-thirds of that attributed to the evidence derived from a mere inspection of all the other curves together.This is the only instance in which I have attributed any weight at all t o conclusions drawn from an experimental diagram without confirming the changes shown in i t by differentiation. Such changes as are specially doubt-ful a r e enclosed in brackets and to these only half the value of the other points has been attributed.The general concordance of the results is certainly satisfactory. Very little weight can be (and is) attached to the indications between 89 and 50 per cent. for the 10" intervals. Nearly all the results sugget;t a change somewhere about 62 per cent. where none of the other properties studied have shown one ; this is veyy probably due to the results in this neighbourhood being affected by some error in the composition of one o r two of the solutions (p. 76). Due probably, to the same cause we have the fact that a change at 57 per cent. is unmarked in most of these results although it has been established by nearly all the other properties investigated. It is to be noticed that a change at 89 per cent. is well established by all these results, although of the density results that at the lowest temperature was the only series which indicated it.There are several points of interest besides the position of the changes TABLE E.-Position of the Changes shown by the Cwuesfor the Expansion From the reduced values. Frcni tlic unreduced ~aliics. The d u e s in this Table refcr t o percentages expresscd nccording to the results of the THE NATURE OF SOLUTIONS. 119 The three 10" curves when compared together show that the effect of temperature on this property is largely dependent on the composition of the solution taken. At about 57 per cent. the expansion from 8" to 18" (Fig. 13 L) p. 115) shows a depression whereas that from 18" to 28" (E) shows an elevation and that from 28"-38" (F) shows a uniform increase so that from 8" t o 38" the rate of expansion of this solution will increase wit'h the temperatine." With a solution of about 70 per cent.strength the reverse will be the case. All the peculiari-ties exhibited by these three curves are equally shown by the corre-sponding curves representing the expansion proper one of which is illustrated by Fig. 13 G. There is 2 well-marked maximum of expansion which i n all cases appears to be situated at 86.3 per cent. ; a percentage be i t specially noted which corresponds with no definite hydrate and with no sudden change in the curve. The comparatively high points shown by the difference of density at about 33 per cent. become points of maxitna in the expansion proper. The position of this second maximum is not so clearly defined as that of the first and appears to vary somewhat with the actual temperature.From 8" t o 18" i t is situated at about 33 per cent. from 18" to 28O at 34 per cent. from 28" to :38" at 3 2 O ; but these variations are very doubtful. I n any case i t is not situated at a point corresponding to any abrupt change in curvature. With the expansion from 18" to 2 8 O there is further a slightly marked maximum at about 57 per cent. It may also be noticed that the change at the monohydrate (84.5 per cent.) is very feebly marked in the original curve and becomes pronounced only when we come to the first differential. That the constituent curves of the figure representing the expan-sion require two differentiations in some cases is I think fairly certain especially when it is remembered that the small magnitude of the original quantities dealt with would favour their differentiation into practically straight lines and that in spite of this six out of the ,eleven portions constituting the first differential (Fig.14 B) exhibit a decided curvature. These constituent curves do Eot appear to meet tangentially and consequently the straight lines forming the second differential do not meet. at the points of change. B. As a Function of the Temperature. The above results have shown that the expansion is sometimes greater and sometimes less as the temperature is higher according * These features are barely risible in the rough sketch given in Fig. 13 120 PICRERING A STUDY OF to the strength of the solution. A few determinations were made with a view of gaining further information on this point.Three solutions containing 84 81.1 and 41.8 per cent of acid respectively were taken and their densities determined a t every 2" between 6" and 38". The first of these solutions has a composition nearly corresponding with that of the monohydrate the composition of the second coincides with no particular hydrate and the third solution was taken simply at random as an instance of a weaker solu-tion. The weights of the contents of the bottle when differentiated give the figures in Fig. 15.* The scale is somewhat too open for the experimental error (which here depends on the correctness of the Table XXIII embodies the results. FIG. 15.-Weights of a t to. First Differential. dzo df actual temperatures as well as on the weighings) and the points f o r the 81.1 per cent.solution have not been inserted the crosses repre-sent the mean points given in the table. The three lines are all con-tinuous and straight within the errors of the experiments from which it would follow that the change of the density with tempera-ture is represented in each case by a single parabola of the second ordert as far as the determinations extended. . * Differentiation of the densities or volumes would give lines differing only in position and inclination (not in form) from those derived from the weights ; but they would contain the errors due to the determination of the capacity of the bottle at each temperature. -f Mendel6eff drew the same conclusions from his determinations with alcohol solutions between 0' and 30".The points determined with each eolution however, never exceeded five THE NATURE OF SOLUTIONS 121 60 per cent. That this however is not the case with all solutions there are theoretical considerations (given below) as well as direct evidence to show. I n constructing the table of densities of solutions of the acid the four results witch each solution were plotted against tem-perature and it was found that whilst with all solutions of 100 t o G3 and from 43 to 20 per cent. the results lay in an apparently uniform curve from 63 to 43 per cent. they could only be depicted by two different curves bending in opposite directions or by two different straight lines* (the latter with solutions from 53 t.0 43 per cent.).This indicates the existence of more or less abrupt changes a t certain temperatures with some solutions and the instances found were too numerous and consistent and the differeiices too great (sometimes O*OOOG of the density or 0.76") to admit of their being explained away by experimental error. The differential obtained in one of these cases (for a 60 per cent. solution) is given in Fig. 15 E ; the three points of which A is one cannot be represented by the same straight line the error here being but one-fifth of what it is with the lines for 84 81 and 42 per cent. solutions an instance is also given in the figure of the differential obtained from a case (20 per cent.) where the four experimental points lay on one curve, and here we get an uniform straight line.? It was certainly unfortunate that the solutions selected for the fuller examination of the effect of t,emperature on density should not have included any in that region where the effect was subsequently found to be irregular.The different inclinations of the various lines in Fig. 15 confirm the previous conclusion that the rate of expansion is sometimes greater and somelimes less a t the higher temperature according to the strength of the solution examined. 20 per cent. 122 PICKERTSG A STUDY OF PART VII.-DISCUSSION OF THE RESULTS. The persistency with which the methods here adopted for ex-amining curves has been misunderstood," renders i t necessary for me to say a few words on the subject even a t the risk of repeating myself.The method does n o t consist in fitting sundry equations on t o the curves and on the strength of the concordance observed to conclude that these latter are continuous or otherwise. It is quite true that, if a curve differentiates into a straight line after a certain number of differentiations an equation of a certain form must represent that, curve and i f i t yield several straight lines there must be as many different equations applicable to different parts of i t ; but it is one thing to find equations empirically and prove (?) their truth by x display of those often most fallacious of arguments known as tables of " found '' and " calculated " values arid another thing to apply an ordinary process of mathematical analysis t o them letting them speak for themselves and tell us whether they are continuous or not.On the former of these methods I would place absolutely no reliance, and so far have I been from making use of it that I have not found the equation for any single curve here depicted. The mathematical argument on which the method depends is that a curve if it be continuous will on differentiation give either a draight line or another continuous curve ; whereas if i t be not con-tinuous but be made up of different curves i t will yield a series of straight lines or curves. Of the dangers attending differentiation and the consequerit un-certainty of any conclusions depending entirely on this process I have already said enough and it is only necessary to remind my readers here that the recognition of the changes of curvature rarelr depends on differentiation only but that these are generally notice-able in the experimental curves themselves.On the other hand it must be admitted that conclusions drawn solely from the experi-mental curves would be equally unsatisfactory; they would bc dependent on the taste of the draughtsman and on the drawing of ;i particular portion of a figure in one section-which without the sub-sequent evidence supplied by' differentiation would be no proof t h t the section drawn is a single curve-but iii no case have such unsup-ported conclusions been drawn. I n some instances of course the experimental curves migli t show changes of such abruptness that a mere cursory inspection would IE sufficient to prove their existence but it would be highly improbable that any of the properties here studied would do so ; these properties This I think is an incontestable fact.* See Nature 40 343 THE NATURE OF SOLUTIONS. 123 must be continuoils functions of the percentage composition and, whatever changes they show will be of a somewhat uncertain nature, since they must be due to changes in the unstable and dissociating liquid hydrates present. The main point of my opponent’s attack will I imagine be directed against the closeness with which my drzwings follow the experi-mental values. That I have ever gone within the limits of known o r probable errors I must deny and I have often left a considerable margin beyond these limits that there may be other unknown sources of error is of course quite possible; but ought we to conjure up unknown possibilities in order to reduce our results to regularity in an investigation the sole purpose of which is to settle whether they are regular or not? Surely the only rational and scientific method of procedure is to draw the curves in the first place so as to allow not much more error in the points than we know must exist, and then ascertain the results; if such a drawing leads us to obviously incorrect conclusions or to conclusions which are not con-cordant in the various cases investigated then and then only shall we be justified in assuming the existence of unknown errors.That a second differentiation deals with quantities which are in most cases within the limits of the error of the experimental points, is quite true (and this is the reason why the second differentiation can hardly ever be performed on the experimental values themselves), but a curve which is drawn to represent a series of points smooths out the errors of each individual point and diminishes this error i n proportion t o the number of points available for the drawing just as the mean of many repeated determinations gives a result which is more accurate than each separate determination.Near the point at which two curves meet the differences in their readings will of course be very small and at the point of junction, nil but the existence of the two curves does not depend on the differences between the points at or near their junction but on t h e position of the many points (generally 6 or ‘7) on either side of the junction.The question so often asked-on what differences do these changes depend ?-is one to which it is impossible to givc a satisfat,-tory answer a small difference would often obliterate certain par-ticular changes (as with some of those in the heat of dissolution, results which were obliterated by differentiating. the values in two series) ; a greater amount of smoothing will obliterate others (the ultra-smoothed curve) ; but smooth over and mutilate the results a s much as we can we still have left some changes which gain but greater prominence as the minor ones become effaced. Such a con-clusion is drawn from st study of the most (seemingly) regular portion of the most regular curve dealt with-the heat of dissolutio 124 PICKERING A STUDY OF curve-a similar examination of those rspresenting the other pro-perties would but enforce this conclusion more strongly.The multiplicity of the changes will,* no doubt prove a stumbling block to the acceptance of my conclusions in some minds. T’he objection may be raised on either mathematical or chemical grounds. Every one is aware of course that a curvilinear figure may be regarded as being made up of a number of straight lines of segments of circles or of parabolas provided we split it up sufficiently. But this is no argument that some particular figure may not in reality be made up of independent parabolas. We can as a matter of fact, string a series of true parabolic carves together so as to make a con-tinuous and apparently regular figure but no one could say that such a figure was not made up of parabolas because any Curvilinear figure can be regarded for certain purposes and within certain limits, as being made up of such.The question must be settled by iiispect-ing the curves and deciding whether the number of parabolas into which they split up according to my interpretation at all approaches that theoretically infinite number into which any curvilinear figure. whatever its real nature might be would split up ; whether it is at all possible that a mere mechanical and chance distintegration of curvilinear figures of such very different magnitude and such totally different forms as those representing the densities dilatation heat of dissolution and heat capacity could result in the production of the same number of parabolas in every case and a11 locating the changes at the same points.The impossibility of such a chance result will be apparent I think to anyone who gives the matter a fair consideration. The chemical grounds on which the objection may be raised arc still weaker. What reason have we f o r asserting the improbability of sulphuric acid forming seventeeri different hydrates iii solution ? At present we know nothing whatever of the nature or number of hydrates which may exist in the liquid condition but we do know that there are cases in which some four or five solid hydrates of the same salt have been isolatedf- (and that within comparatively narrow limits of composition) and i f so many compounds can exist in the solid form where they are in a definiie and stable condition thc probability is I think in favour of the existence of a still 1;~rger number in solution where they are no doubt in a partially dis-sociated condition.* And has done so. Arrhenius Phil. Hag. 28 37. t With sodium carbonate we have eighc; with basic ferric sulphate there is evidence of as many as fourteen (Trans. 1853 lSZ). The isolation of dilFerent hydrates has nearly always bezn the result of chance experiments ; tb sjstemxtic attempt to obtain fresh ones would no doubt enlarge the number known rery considerably THE NATURE OF SOLUTIONS. 125 It is necessary to draw special attention to the definite and specific evidence of change in a reaction or property deducible from changes of curvature such as the present work reveals as opposed to the loose and insufficient evidence which it is sometimes argued is s'iown by an alteration in the general inclination of a figure.If we stmart with a curvilinear figure and prove that on differentiation it ultimately yields say two different lines which are straight within the limits of experimental error we prove that within these same limits it consists of two different parabolas-two regular and definite figures which we may legitimately conclude represent two regular conditions or actions of the solutions. But a mere change in the general direction of the figure such as occur a t points of maximum o r minimum elevation can prove no specific changes a t all for it is perfectly obvious that such maxima and minima mny occur in the middle of the most regular portions of t'he whole figure.The position of these specific changes is unaltered by the manner in which the results are expressed and also by the conditions (such as temperature) under which the determinations are made,* and this permanency is one of the st'rongest arguments in favour of their reality; whereas, with mere general changes of direction we find the reverse to be the case ; the position of a maximum alters according to the mode of expression adopted-as with the densities the contraction the first differential and second differential of the densities,? and we may almost add the expansion by heat where the maxima occur a t 97.5 68 73 49 and 86 (also 33) per cent. respectively-and its position varies also with the temperature as it mnst in the case of the densities,$ and has been proved (Mendelheff Ber.1886 387) to do in the case of the contractions. Agreement of t h e Results obtained from Difcrent Sources. The establishment of the particular changes of curvature which I consider exist must depend mainly on the concordance of the resulttj obtained from the different properties studied. These are * Unless of course we alter the temperature so much that a hydrate is entircly destroyed or a lresh one makes its appearance. t The maximum in the case of the rectilineal second differential must correspond to a specific change ; from the examples a t our disposal I should imagine t,hat it also does so in the case of the first differential. I t s doing 80 with the sulphuric acid densities explains how MendelBeff by the inordinate smoothing of this curved figure was led accidentally to name some of the hydrates which actually do exist.j My results show that the variation is inappreciable from 8" to 38" but if the tcnperature were to be raised sufficiently high to destroy all compounds in solution, the maximum would be altered to 100 per cent. consequently its position must be variable. VOL. LVLI. 12 6 PICKERIKG A STUDY O F collected in Table F (next page) ; the changes which are doubtful are enclosed in circular brackets and are allowed but half value in deducing the mean ; the specially doubtful ones are enclosed in square brackets and are not included a t all i n the mean nor in the following discussion. N.D. signifies that there were insufficient data to show whether a particular change existed or not.Of the 17 changes shown there is one (that at 0.1 per cent.) which could have been indicated hy the heatl of dissolution results only ; of the remaining 16 we have-1 shown in t,wo cases. 1 , three ,, 4 7 four ,, 1 7 five 7 , 7 7 six 7 7 2 , seven ,, each " case )' meaning either the study of a different property or of the same property at a different temperature.* There is not a single iiistance I think where the absence of evidence for any change may not fairly be attributed to the lack of sufficient data and there can be but little doubt but that all the eight sets of results would have shown all the changes had they been investigated sufficiently. The average divergelice of the individual results from the mean, as deduced from the 82 cases available is only 0.388 per cent.there being but nine cases in which it reaches or exceeds 1 per cent. When we consider the extrcme difficulty i n determining the exact, point a t which a change occurs within 1 or even 2 per cent. even where the change itself is undoubted we must admit that this degree of concordance is quite as good as could possibly have been expected : that i t could be accidental is out of the question and we must remember that not only were the properties investigated different, but that the greater part and sometimes all of the solutions used iii investigating one property were distinct from those used in the others the only exceptions to this being the densities of sulphuric acid a t 8" 28" and 38".There can therefore be but little doubt that though there may bc several points where uncertainty still prevails and where future work J I have counted the expansion results as but one case they should parhaps be counted as two since the four series of density determiiiation yield two independent series of expansion values. The density results a t 18' are further confirmed by their treatment in the form of Lontractions (p.83) TABLE F.-Position (in Perceibtaye Values) of the Changes of Cuwature shown Su@huric Acid exmiined. At 38"-203 N.D. Densities. At 28O.064. N.D. CL) The values in this column are 7/10000ths lower than those in Table E (p 118) where the later determinations \ The values for the means are 7 10000th~ higher than they should be according l o the numbers in the first $ 73 or 63 per cent.but very uncertain. 0 This change is too doubtful to be counted. 1' The pooition of this change is defined so much more clearly by the experimental urve of che heat of dissolution ance with the later determinations of the strength of the acid. take; the composition indicated by this to be the more probable one 128 PICKERING A STUDY OF may introduce some modifications the great majority of the changes which I have mentioned as existingare established beyond all reason-able question. Composition of the Hydrates Indicated. Strong presumptive evidence in favour of the changes here indi-cated will be afforded if it be found that they occur a t percentages corresponding to simple molecular proportions.Unfortunately however there are but seven cases from which any conclusion on this head may be drawn cases that is in which the addition of another molecule of water would alter the composit,ion by from 11 to 2 per cent. (marked by an asterisk in the table p. 127).* From these it is seen that the average difference between the mean position found and that calculated for an exact number of water molecules or half molecules (the latter in three cases) is only 0.226 per cent. or about 0.05’7 H20 on the average the grcatest pcrcentage difference (in the case of H2SOa4H20) amounting to only 0.48 or l j l l t h H20.+ Such differences are well within the limits of the experimental error.: That this concordance of position of the changes in curvature with the composition of definite hydrates has not been in any way influenced by a knowledge on my part of the latter I may confidently affirm for throughout nearly the whole of the work I purposely avoided ascertaining the percentage composition of any hydrates the only exception to this being the case of the monohydrate and in spite of my knowledge of its composition and the probability of its causing some marked change I noticed no change a t all Corresponding to it either in the case of the experimental curves for the heat of dissolu-tion or expansion.The changes in the highest portion of the curves are of interest, inasmuch as their discovery was the outcome of theoretical considera-tions as well as of a reliance on the correctness of the results obtained in the other portions of the curve.The former were that just as a weak solution of acid in water contains compounds of the acid molecule with many water molecules so should a weak solution of water in acid contain compounds of a water rnolccule with many acid * The last column contains a few percentages which do not refer to the hydrates indicated by the determinations ; they will be of use in showing the alteration in percentage with molecular corn position. 9 The greatest difference in molecular coinposition is with 3112S0,1€,0 where it amounts to one-sixth H2SO4 but only 0.3 per cent. $ The addition of the results arrived a t in the case of the calcium salts would double the Porce of this conclusion THE NATURE OF SOLUTIONS. 129 molecules for in all very weak solutions the solvent acts as a polymer of its fundamental molecule comparable if not identical with the polymers which constitute a mass of the pure liquidX (Trans.1888, 872 and 1889 23). The practical considerations were that if the second differential were really rectilineal as the bulk of the figure showed it t o be an apparent curvature in any part of it could only be due to the lack of sufficient points to resolve it into straight lines ; this had been proved t o be the case with the curved portion of the second differential near the zero end and it was argued it would probably be the case also with the curved portion at the 100 per cent. end. Looking at the constitution of the various hydrates there are some signs in the extreme caseB of the higher ones bearing a simple relation t o the lower ones thus the water molecules contained in the three highest hydrates are about 500 l,SOO and 5,000 whilst the four compounds richest in acid contain about 36 6 3 and 3/2 H2S04 to every H,O.But such indications must be accepted with great reservation owing to the impossibility of determining the exact molecular composition in the cases in point. Between the less complex hydrates there would seem to be no simple numerical relation at all. The p e a t complexity of the hydrates in very weak solutions is one of the most remarkable facts brought t o light by the present work, and ia such that many chemists will no doubt find difficulty in accepting it.? That there should be such things as chemical com-pounds containing several hundrcd and even several thousands of molecules of one of their constituents all bound together and united (however loosely) with 1 mol.of the other constituent is indeed a most remarkable fact and one which must necessitate no unimpor-tant extension of our views as to chemical action. Yet these changes are amongst the best established in the whole work and they are, moreover precisely similar in character t o those changes in the region of stronger solutions which certainly do coincide with definite hydrates. If the liquid and gaseous conditions be really continuous, it follows I think that the uveruye composition of the aggregates composing a mass of water must be about 1800 H,O and if these * The highest compounds recognised would be less complicated in the case of very strong than in that of very weak solutions; a body consisting of 5000 H,SO,H,O would contain only 0'004 pcr cent.of water and its recognition here would have been practically impossible. t Our opinions are naturally biassed by the fact that our experience is inevitably confined in the majority of cases to stable compounds. Their stability is but a consequence of the simplicity of their structure ; we are here dealing with unstable compounds 130 PICKERINQ A STUDY OF aggregates combine as such with any dissolved substance (as from the heat of neutralisation I argue they must) the hydrates formed will have a corresponding complexity of composition. We should not therefore be surprised at a hydrate containing even 5000 H,O. The existence of such compounds gives I think an explanation of how " trsces " of foreign matter may influence the properties of a body in ft manner altogether disproportionate to the amount present, and how the effect produced may be modified or reversed by altering the proportions of the foreign substance as for instancc in the case of bismuth in gold antimony in copper silicon and carbon in iron &c.The comparative magnitude and frequency of the changes a t the extreme ends of the curves is but a consequence of the method used for depicting the results. Percentage composition has no scientific meaning and is but a convenient mode of expressing results matter must be regarded as consisting of molecules and the real nature of a mixture or compound is indicated by the relative number of the different molecules present a small difference of say 0.7 per cent., such as that from 1 to 0.3 per cent.really means a trebling of the relative proportions of water and acid present and the curve repre-senting any action between the two points should extend over three times the distance of the whole of the rest of the diagram from 1 per cent. upwards. I n Fig. 16 I have given a sketch of one set of my results (the second differential of the densities of sulphuric acid a t 8 O ) plotted out against molecular composition. The relative number of molecules are given a t the top of the plate. The proportions of 1 1 would occupy the centre of the diagram if the investigations had been pushed to the same limits with strong solutions as with weak ones.It has not been practicable to represent any changes beyond the one a t 3.9 per cent. and many of them are seen with such difficulty that their position has had to be specially marked.* The relative magni-tude of the changes at the simpler and more complex hydrates as well as the relative magnitude of the different hydrates amongst the simpler ones is very striking. The most conspicuous are those a t 30 50 84 and 94 per cent. corresponding with hydrates with 13, 5.5 1 and 1/3 H20. It is very evident that this method of plotting out the results would in most cases be very unsuitable for working purposes. Indication of the Monohydrate in Solution. That the monohydrate exists in solution is in my opinion proved The by the fact that we can crystallise it out from solutions.* The constituent lines are numbered so as to correspond with those in P1. 3 THE NATURE O F SOLUTIOSS. 131 Per cent,. H2S04. 84.4 existence of a body in the crystalline form is not however ent>irelg dependent on the stability of its molecules in the liquid condition, but on the ease with wliich these molecules coalesce to form the aggregates of wliich it is composed in the solid condition; it does not follow therefore that the monohydrate must necessarily be conspicuously stable when liquid or that the change in curvature indicating its presence should be more marked th 111 those indicating otlier hydrates in solution. At first sight it appears that such is not the case the experimental curve for the heat of dissolution shows no discernible change a t this point (84.5 per cent.) ; in the expnn-sion curve it is but feebly marked and in the first differential of the densities it is by no means prominent.But i n all these cases it is found that a repeated application of differentiation (wherever the data are sufficient) alters this and shows that the change a t the monohydrate is really a prominent one (see especially Fig. 2 B, p. SO). Its being partially masked in the experimental curves may, I believe be due to the fact that this hydrate is really exceptionally stable while liquid and that just like the other stable liquids water and the pure acid forms compounds containing several funlamental molecules of itself with a single water molecule on the one hand an 132 PICKERINQ A STUDY OF a single sulphuric acid molecule on the other ; the proximity of the minor changes due to such compounds would in the absence of B very large number of experimental points result in the rounding off of the change which shows the monohydrate itself,* Nature of the Various Curves.For the chief conclusions drawn from the present work-the existence of more or less sudden changes of curvature at certain points-it is important to show that the various constituent curves differentiate eventually into lines which are straight within the limits of experimental error ; for their doing so shows that they are each of them single and simple curves within these same limits. But, apart from these practical considerations the nature of the consti-tuent curves raises several important questions of theoretical interest : whether the second differential lines are absolutely straight whether the first differential lines are sometimes so and whether the con-stituent curves of the original figure meet t'angentially or otherwise.These questions can be answered only with great hesitation. That a single differentiation is insufficient to reduce the densities (and contractions) and electric conductivities to straight lines is beyond doubt; with the heat of dissolution the heat capacity and the expansion the case is more doubtful ; but the fact that a consider-able portion of the first differential figure of all these properties is curved renders it most probable that it should be curved throughout, and that the straightness observed in parts is but a consequence of comparative magnitude of the error of the determinations and the small scale on which these have to be plotted in conscquence.Whether the straightness of the lincs composing the second diffe-rentials is merely the result of repeated differentiation o r whether they are straight owing to the nature of the experimental curves from which they are derived (that is owing to these curves being parabolas of the third order) is a difficult question to settle. It is true that the majority of (but by no means all) curves of a gentle nature such as are here dealt with would yield straight lines on repeated differentiation;? but for the following restsonu I am * Since writing the above two changes in the ricinity of the monohydrate have been revealed by a study of the freezing points of solutions of the acid.t As instructive instances of curves which do not do so I may cite the end portions of the first differential of the densities and various other gentle curves, each drawn as a single curve but really made up of two or more independent curves and which on repeated differentiation show their coniplex nature (in-stances in the densities and heat of dissolution have been quoted above) ; and, again curves representing the molecular heat of dissolution plotted against per-centage composition which are not parabolic a t all THE NATURE OF SOLU'l'IONS. 133 inclined to think that the straightness of second differential is really a consequence of the parabolic nature of the experimental curves.(1.) The curvature of the first differential is often greater than that of the experimental curve and the straight lines which it yields on further differentiation cannot therefore be of a result of a gradual straightening process." (2.) The second differential lines are straight in nearly every instance in spite of the great variations in the curvature of the various first differentials which yield them. (3.) The first differential of the densities differentiates into straight lines whereas the experimental curve of the expansions-a curve comparable in its nature and curvature with the density first dif-ferential-differentiates in moat parts into a series of curves. (4.) Wherever a curved second differeiitinl has been obtained its cnrvature has been found tlo be the result of insufficient data.The question has been raised as to what limit we should put to the process of differentiation (Morley Proc. 1888 l27) and Crompton's answer-that " a limit to the diff ereiitiation would necessarily have t o be made according to the nature of the case under investiga-tion and the discretion exercised by the investigator "-has, I think been somewhat misunderstood (see Arrhenius Phil. Mag., 28 37). For the purposes of the present work the limit to be put to the process of differentiation is that at -which we get a rectilineal figure for a straight line is the only form of line which, on mere inspection is evidently regular and which must pi-ove regularity in the curye which yields it. A further differentiation would not add to our knowledge (unless we wished to determine diagrammatically the constants in the equations for the curves) for our series of straight lines would simply yield thereby a series of horizontal lines at different levels and a still further differentiation would yield nothing at all.A practical limit to the differentiation will also be determined in each individual case by the error of the original experiments. Even when we reduce these errors by taking the curve which represents them instead of the actual experimental values there are few cases in which we could proceed beyond a second differentiation so as to obtain results of any value. f A comparison of the curvature of the experimental and differential curves is a di5cult matter for the scales used must bear a fixed relationship to each other in the three cases and the various curves thus produced generally Lare a different inclination to the axis.Should we determine the curvature in the same lengths or between the same percentages ? I f the former the curvature of the first differential would nearly always be greater than that of the experimental curve. The expan-sions and the mixing experiments are the simplest instances to take for such an investigation 134 PICKERING A STUDY OF A difference of considerable importance may be noticed between the heat results and the density results. The former are represented throughout by a series of curves which on production cut each other and from which second differmtials will be obtained which, consequently do not meet a t the points where the change occurs if indeed they meet a t all ; whereas the first differentials derived from t h s densities and consequently the densities themselves form a series of curves which for the most part meet tangentially or very nearly so and the second differentials from these will meet at the points of change.This is certainly not always so but the general teiidency is unmistakable. This difference is no doubt connected with the difference i n the nature of the two properties the former being a condition of the solutions the latter a measure of potential energy possessed by them. The curves for the electric conductivity seem for the most part to be o€ the same nature as those for the density the change a t 9.5 per cent. however is a marked exception.The contraction on mixing, the expansion and the heat capacity on the other hand appear to resemble the heat of dissolution more closely ; the constituent curves certainly do not meet tangentially and indeed in the case of the last do not appear to meet at all though as I have shown above, there must be some connecting link between them (p. 93). InfEuence of Temperature o n the Hydrates i n Solution. The influence of temperature on the first differential of the densities is of considerable interest. Each figure in Plates 1 and 2" (p. 184) should if they were all identical occupy positions five small divisions of the paper below or above its neighbour; but the actual distance between t,hem varies considerably and shows that all the changes of curvature become more marked as the temperature is lower.The upward curva-ture a t the zero end and the whole of the constituent curves up to 73 per cent. show this progressive increase very clearly. The portion between 84 and 94 per cent. is also of special interest it has a decided downwards curvature at 8" which becomes so slight a t 18" and 28" that it is here practically a straight line whilst a t 38" the curvature is decidedly upwards and if we could extend the determinations t o still higher temperatures it would no doubt, increase this portion of the diagram becoming less and less dis-tinguishable from the neighbouring curves. The result of this would be that changes in its vicinity would become less marked and would eventually disappear altogether; the rest of the figure would no doubt become simplified in the same way and would exhibit less * Including that representing the results collected by Mendelkeff THE NATURE OF SOLUTIaNS.135 and less curvature till finally reduced to a horizontal line a t a tem-perature sufficiently high to precent any hydrates a t all being formed. The possibility of making dcterminatioiis would however, cease long before such a temperature was reached. But the different hydrates being of different degrccs of stability, mill not all disappear a t the same temperature and consequently, some solutions will show a more or less sudden alteration in the rate a t which their density is affected by temperature under conditions where others consisting of stabler hydrates will show no such change.The experimental evidence already discussed tends to con-firm this conclusion and me may reasonably argue that the evidence of sudden changes in the rate of increase of the density with tem-peratnre would have been of a more certain character had the method adopted permitted of the examination of very weak solutions the hydrates in such cases being more complex arid presumably less stable than in the case of strong solutions. Further evidence 3s to the existence of such sudden changes bas been afforded by my work on the effect of temperature on the heat of dissolution of salts (Trans. 1887 2'23) which showed that this property and consequently the heat capacity of the weak solutions obtained in the determinations exhibited sudden alterations in its rate of increase at certain temperatures.The highest line in Fig. 15 (p. 1'20) is also of considerable interest in connection with this point it represents the rate of increase in the density of water with temperature and according to my in terpreta-tion of it (details of which cannot be given here) it exhibits abrupt changes a t certain points such as I have concluded would be ex-hibitcd by tlie densities of weak solution and such as the analogy between the constitution of the weak solutions and of the piire solvent (both being niixtiires of molecular aggregates of different degrees of complexity) would lead us to expect. .A consideration of the effects of temperature on the densities must make us recogiiise the possibility of there being yet other hydrates than those a t present discovered hydrates which would only be revealed by experiments at lower temperatures ; it is probable however that t,hese would be confined mainly t o very weak or very strong solu-tions.That the densities at 8" should have indicated (though feebly) two changes which were not found a t higher temperatures supports such a conjecture. Summary and Conclusion. The curves representing the first differential coefficient of the den-sities of sulphuric acid solutions form a currilinear figure and not ,z rectilinear figure as Mendel6cff imagined. This figure thoug 136 PICKERINQ A STUDY OF continuous is mads up of a series of separate curves which on further differentiation yield a series of straight lines, When we plot out the contractions occurring on mixing sulphnric acid and water instead of the densities of the solutions obtained we get a figure totally unlike that formed by the densities on a first differentiation i t gives another curvilinear figure also totally unlike the first differential of the densities ; but on a second differentiation, a rectilineal figure is obtained closely resembling the second differ-ential of the densities and indicating the existence of changes at the same points as the latter does.The electric conductivities resemble the densities in giving straight lines on a second differentiation only as shown by Crompton and they indicate changes at certain points though the position of these changes are according to the. author not all identical with those mentioned by Crompton.The heat capacity is represented by a series of gentle curves which probably require two differentiations to reduce them to a series of straight lines. A critical examination of Thomsen's results on the heat evolved by diluting solutions with successive quantities of water leads to the conclusion that the actions are eminently irregular instead of being regular as he concluded. The heat of dissolution (or dilution) of solutions of different strengths are represented by a series of intersecting curves each of which requires (in most cases) two differeutiations to reduce it to a straight line. The expansion of sulphnric acid solutions through various ranges of temperature between 8" and 38" gave results which formed irregular curved figures and these figures split up on two differentiations into a series of straight lines showing changes at the same points as the other properties did.An examination of three solutions of sulphuric acid a t every 2" showed that the effect of temperature on the densities of these solu-tions was represented by a single parabola of the second order ; butl the less numerous density determinations with other solutions indicated that in some cases the results cannot be represented by less than two different curves more or less sudden changes occurring a t certain temperatures. The points a t which the changes occur are the same whichever of the above properties be taken for their determination this con-cordance placing the existence of the majority of them beyond doubt.The point's a t which they occur moreover correspond to definite and simple hydrates considerably within such limits of error as might reasonably be expected. The author's present determinations lead to the same results THE NATURE OF SOLUTIOXS. 137 The hydrates thus recognised amount to as many as 17 the mono-hydrate being one of them. The highest hydrates contain a very large amount of water ; i n the extreme case as much as 5000H20. There may be some simple numerical relationship between the constitution of the various hydrates but only in very weak and in very strong solutions. As a consequence of the results obtained on analysing the experi-mental values and the curves representing them it follows that these latter are represented by a series of parabolas of the third order, such as s = A + Bp + Cp2 + Dp3 this probably being true in the case of all the properties though sometimes the last constant is so small in comparison with the experimental error t'hat it may be iieglected and the results be expressed by a parabola of the second order.I n the case of the densities arid the electric conductivities the various curves constituting the first differential appear t o meet tangentially and consequently the straight lines constituting the second differential meet at the points of change. This is not the case with the contraction on mixing the heat of dissolution or the expansion by heat the first differential curves (and consequently the experimental curves also) in these cases intersect-ing each other on prolongation.In the case of the heat capacity the constituent curves representing the experimental values do not appear to meet at all. The main facts elucidated by t'his work afford absolute proof that the propert'ies of solutions do not vary regularly with their composi-tion and that the nature of the solution is therefore not merely physical. There can be no doubt but that solutions consist of hydrates of a definite composition though these may always be i n a state of partial dissociation. It may be urged that although this may be the case where the solvent is water in other cases similar compounds would not exist. It is extremely improbable however that actions seemingly so analogous as the dissolution of a substance in water and its dissolu-tion in another liquid should belong t o totally different classes of phenomena the one being chemical the other physical.It is how-ever a question which calk for an answer based on experiment and, as it would be difficult if not impossible t o find a solvent other than water with which solutions might be obtained possessing sufficient differences in density &c. to render them suitable subjects for investi-gation some other means of investigation must be adopted. I hope that the freezing points of solutions of different strengths may affor 138 PICRERING A STUDY OF the requisite information and the resuIts which I have already obtained since writing the above from a study of this property in the case of sulphuric acid justify the expectation that such may be the case. I am fully aware that the present work supplies material for far more discussion than I have ventured to raise. The precise nature of the changes which occur when solutions are diluted has scarcely been discussed at all. What is the meaning of these various proper-ties being representable by a series of parabolic curves of certain orders? Why should the component curves of the density the heat of dissolution and the heat capacity diagrams show such differences as that of meeting tangentially in one case cutting each other in the second and (apparently) not meeting at all in the third case ? How can the fact that the changes occur so sharply at the exact composi-tion of certain hydrates be reconciled with the extent t o which we must I think necessarily imagine that these hydrates are dissociated, even when the water and salt is present in the exact proportions required for their formation (see Trans. 1889 22)? These and other questions I leave for the discussion of abler mathematicians than myself the present work has no pretentions beyond being an experimental investigation. But one word remains to be said a word of thanks to a little girl who bas devoted all her spare time to the laborious calculations involved in the present work to the patience and accuracy of Miss Ethel Willmott all who are interested in the question of the nature of solutions owe much THE NATURE OF SOLUTIONS. 139 s. P . TABLE I.-Densities of Xulphuric Acid Sollitions at 17*925' and First Diferentiation. Alternate series A and 13. TABLE XXIV.-List of the most convenied Scales used in the Worlcing Diagrams. Property. Scale. * All the differentials here mentioned are those obtained directly from the experimental values the scales used for those obtained from readings of curves have been sufficiently indicated in the text p. 68 PLATE 3. SECOND DIFFERENTIAL OF THE FIGURES REPRESENTING THE DENSITIES AN

ISSN:0368-1645    

DOI:10.1039/CT8905700064

出版商:RSC    

年代:1890

数据来源: RSC