144 JAPP -4ND MURRAY : SYNTHESIS OF PENTACARBON RINGS. X.-Sy.nthesis of Pentacarbon Rings. Paiht 111. Condemation of Benzil with Lavulic Acid. By FRANCIS ROBERT JAPP, F.R.S., and THOMAS SMITH MURRAY, D.Sc. HUGO ERDMANN (Annwlen, 1889, 254, 187 ; 1890, 258, 129) has shown that, in the condensation of benzaldehyde with lsvulic acid, the benzal group attaches itself to the lsvulic acid chain either in the p- or in the &position, according to the nature of the condensing agent employed. Thus either /3-benzallsvulic acid, C6H,* CH:C<gg2:FdoH, or 6-benzal- laevulic acid, C6H,* CH:CH* CO*CH,* CH,* COOH, is formed, accord- ing as the dehydrant is fused sodium acetate or dilute caustic alkali. We find that, in the condensation of benzil with laevulic acid under the influence of caustic potash, both carbonyl groups of the benzil, and both the /3- and the &groups in the lmwlic acid chain, take part in the condensation.Two isomeric acids are obtained, the formation and constitution of which may be represented as follows.JAPP AND MURRAY: SYNTHESIS OF PENTACAREON RINGS. 145 C,H,* 70 + C H 3 > ~ ~ - H,O C,H,* CO CH, CH,*COOH >GO and C,H,* C=CH >GO I - C,H,* Y(OH)*CH, - C,H,*C L__ C CH,.COOH C,H,*C(OH)*CH CH,*COOH a-AnhydrobeiizillEvulic acid. &Anhydrobenzillfevulic acid (known only in its salts). the latter substance being formed in subordinate quantity. The condensation is analogous to that which occurs in the formation of anhydracetonebenzil and its homologues (see Part I. of this series of papers), the same cyclopentene group being formed in both cases.The relationship of the two anhydrobenzillaevulic acids to anhydracetone- b e n d might be indicated by naming them anhyd9*acetonebentxilethyZoic acids. The formation of the two acids might be accounted for by sup- posing the first stage of the reaction to consist in a di-aldol condensa- tion, leading to a hypothetical intermediate compound of the formula >CO C,H,* C(OH)*CH, I c,H,- C(OH)-CH CH,.COOH ' which might then part with a molecule of water in two different m a p , yielding the two anhydrobenzillsvulic acids. The sodium salt of the P-acid is instantaneously oxidised by perman- ganate in the cold, whilst that of the a-acid is only slowly oxidised. This behaviour is in accordance with the foregoing formulz, in which the P-acid contains, and the a-acid does not contain, hydrogen directly attached to the ethylene group.The a-acid is stable in the free state, the P-acid, when liberahed from its salts, spontaneously parts with water, changing into the lactone, (See Part I.) CGH5* C:CH I >GO C,H,* C CH 0-co ' X H , The action of the a-acid with hydroxylamine is abnormal, an addition of hydrogen, effected by the excess of hydroxylamine, accompanying the formation of the oxime : C,,H,,O, + NH,OH + H, = C19H180,(N* OH) + H,O. The action of the P-acid with hydroxylamine was not studied for want of material. Both the a-acid and the lactone, when boiled for a short time with fuming hydriodic acid, are converted into the same diphenylcyclopen- tenonethyloic acid, C6H5* !?cH2>co , a change in the position C,H,* C*CH * CH,* COOH VOL.LXXT. L146 JAPP AND MURRAY : SYNTHESIS OF PENTACARBON RINGS. of the double bonds accompanyin: the reduction. That this change occurs is shown by the formation of this compound both from the a-acid and from the lactone of the P-acid, and also by the fact that diphenyl- cyclopentenonethyloic acid yields, on oxidation with sodiuiii hypo- bromite, a mixture of cliphenn?/lmuleic and diphenyllfumuvic cccicls. We do not, however, lay much stress on this last argument, as anhydrace- tonebenzilcarboxylic acid, in which the double bonds are not in the position which they occupy in diphenylmaleic and diphenylf umaric acids, gives a mixture of these two acids on oxidation (see preceding paper). The formation of the fumaroid form in these cases is anomn- lous; one would expect that, in the opening out of the closed chain, only the malenoid form would be produced.Diphenylcyclopentenonethyloic acid yields an oxirne. I f a-anhydrobenzillaevulic acid is boiled with hydriodic acid and amorphous phosphorus for some hours, the diphenylcyclopentenonethy- loic acid, which is first formed, is further reduced and a t the same time parts with carbon dioxide, yielding methyldiphenylcycZo~)eiztane, 2>CH,, C,H,. CH* CH C,H,*CH*CH CH, I a homologue of the diphenylcyclopentane obtained from anhydracetone- benzil. Only one of the possible stereoisomeric forms of this hydro- carbon was observed. EXPERIMENT A L. Prepration of a- A?& y d ro bend Zevulic A cid ccnd j3- An hydro benxi Zlcevu Zo- lactorte.-One hundred and five grams of benzil are dissolved with the aid of heat in 300 C.C.of alcohol ; 58 grams of laevulic acid are added, and the mixture is cooled, shaking the flask all the time to prevent the benzil from separating in large crystals. To the magma thus obtained, a solution of 50 grams of caustic potash in 75 C.C. of water is added, and the whole is heated on the water bath to boiling for about 20 minutes, after which it is poured into water and the solution saturated with carbon dioxide, to prevent the excess of caustic alkali from exercising any decomposing action during the subsequent evaporation. The liquid is then evapo- rated until the whole of the alcohol is expelled; a large bulk of water is added, and the solution, after filtering, if necessary, is precipitated with dilute sulphuric acid, adding the acid gradually to the hot liquid, so as to prevent the simultaneous precipitation of salts.By stirring vigorously and heating on the water bath, the organic acids are obtained as a tarry mass, from which the supernatant liquid can be poured off. On standing for two or three days, this mass solidifies. It is then ground in a mortar and thoroughly extracted with boiling water, t oJAPP AND NURRAY: SYNTHESIS OF PENTACARBON RINGS. 147 remove salts and sulphuric acid, after which it is twice boiled with a solution of sodium carbonate ; this extracts the a-anhydrobenzillaevulic acid, leaving the P-anhydrobenzillaevulolactone as a brown powder, the treatment of which will be described later on.The solution of the sodium salt is then precipitated with diIute sulphuric acid, washing the organic acid thoroughly with water. It is dried, extracted with a small quantity of boiling benzene, which removes most of the brown colouring matter, dissolved in the smallest possible quantity of hot glacial acetic acid, and an equal bulk of benzene added. On standing, the solution deposits the acid in needles. It is sufficiently pure for the study of its reactions, but it was further purified for analysis by recrystallisation from a mix- ture of ethylic acetate and light petroleum. The yield of once c rystal lised substance from the above quantities, conducting the operation as described, ranges from 70 to 80 grams ; but more may be obtained from the mother liquor.a-AnhydrobenziZZ~vzcZ~c acid crystallises in colourless needles, melting with decomposition at 178-179O." It is readily soluble in alcohol, glacial acetic acid, and ethylic acetate; but only sparingly in benzene and ether. It is only very sparingly soluble in boiling water, but may be obtained pure by recrystallisation from a large bulk of this solvent. It gives a brown coloration with concentrated sulphuric acid. Analysis t gave figures agreeing with the formula C,,H160,. 0.3414 gave 0.9230 CO, and 0.1558 H,O. C = 73.73 ; H = 5.07. 0.2436 ,, 0.6602 CO, ,, 0.1114 H20. C = 73.91 ; H = 5.08. C19H1604 requires C = 74.02 ; H = 5.19 per cent. The sodium, potassium, and ammonium salts are sparingly soluble in cold, readily in hot, water. The barium salt, obtained by precipitation from the ammonium salt, crystallised from dilute alcohol and gave figures agreeing with the formula (C,,H,,0,),Ba,5H20. 0.2802 lost, at 150°, 0.0297, and the residue gave 0*0776 BaSO,.H,O = 10.6 ; Ba (in anhydrous salt) = 18.22 per cent. (CIgH,,0,),Ba,5H,0 requires H,O = 10.7 per cent. (C,,H,50,),Ba requires Ba = 18-27 per cent. The process of heating with dilute sulphuric acid to which the freshly-precipitated mixture of anhydrobenzillaevulic acids is subjected ensures that any P-acid which may be present is converted into lactone, ++ By an unfortunate transposition of the figures in writing the preliminary note which we published on this subject (Proceedings, 1896, 146), this melting point was erroneously given as 187-189". .F These analyses were made by Dr.J. Bishop Tingle, by whom the acid was first prepared. (See footnote to preceding paper.) L 2148 JAPP AND MURRAY: SYNTHESIS OF PENTACARBON RINGS. which remains undissolved when the a-acid is extracted with sodium carbonate. The brown powder thus obtained (u. supra) is washed wit11 ether, which removes much of the colouring matter, and is then mrgrstallised several times from benzene; it is thus obtained in well- defined, colourless, flat prisms, or plates, with bevelled edges, melting a t iI51-152O. It is moderately soluble in benzene and alcohol, sparingly in efher, and insoluble in light petroleum. It gives a pale yellow colora- hima, with concentrated sulphuric acid. The yield is small; only 8 grams of the lactone were obtained from 210 grams of b e n d Analysis gave figures agreeing with the formula of P-ccnhyd~o6enxil- hseclolactone, C,,H,,O,.0,1455 gave 0.4183 CO, and 0.0646 H,O. C = 78.40 ; H = 4.93. D.1501 ,, 0.4315 CO, ,, 0.0668 H,O. C = 78.40 ; H = 4.94. C1,H1403 requires C = 78.62 ; H = 4.83 per cent. A cryoscopic determination of the molecular weight, using benzene as a solvent, gave a result in accordance with the foregoing formula. Weight of Weight of Mol. substance. solvent. Depression. weight. 0.1070 23.435 0 . 0 8 O 280 ClgH140, = 290. /l-Anhydrobenzillzevulic acid, when freshly precipitated from solutions of its alkali salts, is soluble in sodium carbonate; but after standing for some time under the liquid, it no longer dissolves, and is found to have been transformed into the lactone. The silver 8aZt of /3-anhydrobenzillaevulic acid was, after several unsuccessful attempts to prepare it from the ammonium salt, obtained by the following method.The lactone was dissolved in alcohol, and alcoholic potash was added so as to leave the lactone in excess. This point could easily be observed, owing to the change in the colour of the solution, which occurred as soon as the potash was in excess, in which case it was necessary to add more lactone. The solution was then evaporated to dryness, the residue dissolved in water, and the liquid filtered from unattacked lactone. From the solution of the potassium salt, the silver salt was obtained as a white precipitate by adding excess of silver nitrate. It was dried in a vacuum and analysed. 0.7908 gave 0.2062 Ag.Ag=26*07 per cent, ClgH,50,Ag requires Ag = 26.02 per cent. Behaviow of a- and P-An?~yd~oEenxilZevzcZic Acids towavds Peiman- gnnate.-The sodium salts of the two acids were employed, that of the 8-acid being prepared from the lactone by the method just described inJAPP AND NURRAY: SYNTHESIS OF PEKTACARBON RINGS. 149 the case of the potassium salt, and recrgstallised several times to ensure purity. Solutions of the two d t s , of equal strength, were. mixed with sodium carbonate, and a drop of permanganate solution was added to each. The change to a brown colour was instantaneous in the case of the /?-salt, but required one or two minutes in the case of the a-salt. This, as has already been pointed out, is in keeping with t h e difference in the constitution of the two acids.Action of Hydroxyhrnine on a-A~hydi.obenxilZevu7/ic Acid.-Nine grams of the finely-powdered acid were suspended in about 100 C.C. of water, and 4.2 grams of hydroxylamine hydrochloride and 12 grams of caustic potash, each dissolved in water, were added. The mixture was allowed to remain, with occasional shaking, for 3 days, at the end of which time almost everything had dissolved. The liquid was filtered, and carbon dioxide was passed into the clear solution. The dense white, granular precipitate thus produced was filtered off, and washed with a little cold water. It was found to be the potassium salt of an oxime. It was soluble in much water, and apparently unaltered even by long boiling. The oxime mas obtained from this potassium salt by treating it with excess of cold, dilute sulphuric acid and extracting rapidly with ether ; the ethereal solution, after washing it with water and allowing it, t o evaporate spontaneously, left an oil, which was dried in a vacuum oveF sulphuric acid. By dissolving this oil in ethylic acetate and adding benzene, the oxime was deposited in tufts of white needles, which, when dried, had a matted appearance. The substance was several times recrystrallised from the same mixture.It melted, with decomposition. at 122-123O. The analytical figures appeared to point to the formula. which is that of a normal oximeplus 2 atoms of hydrogen. 0,1352 gave 0.3459 CO, and 0.0694 H,O. C=69.75 ; H=5*70. 0.1358 ,, 0.3475 CO, ,, 0.0690 H,O. C = 69.80 ; H=5.65. 0.1053 ,, 0.2693 CO, ,, 0.0550 H,O.C = 69.75 ; H= 5-80, 0.2384 ,, 8-02 C.C. moist nitrogen a t 14' and 760 mm. N = 3-96, 0.2058 ,, 7.04 C.C. ,, ,, ,, 11' ,, 759 mm. N=4.07, C,9H,9N0, requires C = 70.15 ; H = 5.85 ; N = 4-31. The oxime is sparingly soluble in water. Boiled with water, it dis- solves ; on further heating, carbon dioxide is evolved, and a yellowish substance separates, insoluble in alkalis. We were unable t o ohtain this substance in a crystallised state, and did not examine it further. The siher salt of the oxime was obtained as a white precipitate by150 JAPP AKD MURRAY: SYNTHESIS OF PEKTACARBON RINGS, adding silver nitrate to a solution of t,he potassium salt. i t was dried at 100". For analysis, 0.2660 gave 0.0669 Ag. Ag = 25-15, C19H18N0,Ag requires A g = 25.00 per cent.The foregoing sparingly soluble potassium salt was not analysed, but. was doubtless the monopotassium (carboxylic) salt corresponding with this silver salt. The solution in excess of caustic potash must have contained a soluble dipotassium (carboxylic and oximic) salt which was converted by carbon dioxide into the monopotassium salt. The same phenomenon is exhibited in the case of another oximino-carboxy-acid described later on. Reduction of a-Anl~~drobenxilZ~vzLlic Acid with Liydriodic Acid.--In the first experiment which we made t o reduce this acid by boiling it, mith hydriodic acid, great difficulty was found in obtaining the product in a crystallised form. As the substance was an acid, salts of it were prepared ; but these showed equally little disposition to crystnllise.At last it was found that in a specimen of the gummy reduced acid, which had been allowed to stand for some months covered with benzene, and protected from evaporation, a small rosette of prisms had formed. This was freed from the adhering gummy substance, and was used in starting subsequent crystallisations. It was also found that the product was more readily cry stallisable if the boiling with hydriodic acid was not continued longer than was absolutely necessary for the reduction. Forty grams of finely-powdered a-anhydrobenzillaevulic acid were heated to boiling with excess of fuming hydriodic acid (sp. gr. 2.0) for about a minute and a half, shaking the flask continually during the process. The substance melted. It was then poured into water and the liquid extracted twice with ether.The ethereal solution was decolorised with sulphurous acid, well washed with water, and dried with calcium chloride. On evaporation, it left a gum, which was dissolved in benzene, and the crystallisation was started by means of the crystalline substance already mentioned. A mass of hard crystals, embedded in the syrupy benzene solution, was thus obtained. The syrup was removed with the aid of the filter-pump, and the crystals, after washing with benzene, were redissolved in this solvent. The solution deposited 31 grams of almost pure substance, which gave practically no colour with concentrated sulphuric acid, showing the almost entire absence of unchanged a-anhydrobenzillEvulic acid. Recrystallised once more from benzene, it melted constantly a t 126- 127*, and now gave no coloration with sulphuric acid.It forms rosettes of prisms. It is slightly soluble in boiling water and separates The following process gave a good result.JAPP AND MURRAY: SYKTHESXS OF PENTACARBON RINGS. 151 again on cooling ; readily soluble iii ether, alcohol, or benzene; in- soluble in light petroleum. Analysis showed that the original acid had parted with an atom of oxygen during the reduction, yielding an acid of the formula C19H160,. 0.1281 gave 0.3674 00, and 0.0642 H,O. C = 78.22 ; H = 5.57. 0.1258 ,, 0.3601 CO, ,, 0.0635 H,O. C=78.07; H ~ 5 . 6 1 . C,,H,,O, requires C = 78-08 ; H = 5.48 per cent. We have already given reasons for regarding this compound as The silvey salt warns obtained by precipitating the ammonium salt di~~henylcyclopenteIIo~~ethyloic w i d .with silver nitrate. For analysis, it was dried at 100'. 0.8733 gave 0.0730 h g . Ag=26.71. C19Hl,0,Rg requires Ag = 27-07 per cent. Action of Hyd~*oxylc~rnine on D~p~~enylcyclo~enterw~ethyloic Acid .- 5.8 grams of acid, 5.5 grams of hydroxylamine hydrochloride, and 15 grams of caustic potash were dissolved in water, and allowed to stand in the cold for 3 days. Carbon dioxide was then passed into the solution, when a potassium salt of the oxime was precipitated ; this was washed with cold water, in which it was practically insoluble. A portion of it was dissolved in alcohol, and an alcoholic solution of silver nitrate was added. The gelatinous precipitate of silver salt thns obtained was washed first with alcohol and then with water, and finally dried in a vacuum desiccator for some days.0.4492 gave 0.1165 Ag. The free oxime was obtained by shaking the potassium salt with dilute sulphuric acid and ether, washing the ether with water, drying it with calcium chloride, allowing it to evaporate spontaneously, and, when it had reached a small bulk, adding light petroleum. The oxime separated in minute, white, crystalline warts, melting constantly a t 183--184O, with slight darkening, Analyses agreed with the formula Ag = 25.93. C,,H,,NO,Ag requires Ag = 26-09 per cent. C1,H160,(N0H)* 0.1558 gave 0.4224 CO, and 0-0791 H,O. C=73.94 ; H=5*64. 0.3165 ,, 12.3 C.C. moist nitrogen at 7" and 752 mm. N= 4.66. Cl,H17N0, requires C = 74-26 ; H = 5.54 ; N = 4.56 per cent.Oxidation of Diphenylcyclopentenonethyloic Acid with #odium Hypo- bromite.-As diphenylcyclopentenonethyloic acid had evidently the same relation to a-anhydrobenzillsvulic acid as diphenylcyclopentenone has t o anhydracetonebenzil, and as diphenylcyclopentenone yields, by152 JAPP AND MURRAY : SYNTHESIS OF PENTACARBON RINGS. oxidation with sodium hypobromite, diphenylmaleic acid (see Part I.), it was of interest to ascertain how diphenylcyclopentenonethyloic acid would behave towards this oxidising agent. 5.8 grams of the acid were dissolved in caustic soda, and a solution of 30 grams of bromine in excess of caustic soda was added. The mixture was allowed to stand for 3 days with frequent stirring, after which sulphur dioxide was passed in, the liquid acidified with dilute mlphuric acid, and the organic substance extracted with ether.The ethereal solution was extracted twice with sodium carbonate, and then with caustic soda. The united sodium carbonate extracts, on acidification, gave a pre- cipitate, which was taken up with ether and was left, on evaporating the ether, as a resinous mass containing crystals. After removing the resin with benzene, the crystals were dissolved in ethylic acetate, and benzene was added. A compound separated in needles and appeared to be diphenylfumaric acid, but the melting point was unsatisfactory. As the substance was suspected to contaio diphenyl- maleic anhydride, we treated it again with sodium carbonate. An undissolved residue was filtered off and the acid reprecipitated from the filtrate.It now crystallised from the mixture of ethylic acetate and benzene in small, white needles, melting at 271". The melting point of diphenylfumaric acid is given by Reimer at 260", and by Japp and Lander at 276" (see preceding paper). Much depends on the rate of heating, as the substance decomposes on melting. Analysis gave figures agreeing with the formula of diphnglfu;maric ncid. Calculated for C16H,,0, : C = 71.64; H = 4.48 per cent. The caustic soda solution gave, on acidification with dilute sulphuric acid, a yellow prccipitate, which was recrystallised twice from benzene, and was thus obtained in rosettes of yellow prisms, with a characteristic greenish fluorescence, and melting at 156.5'. These are the properties of diphnylmccleic nnhydride.Calcu- lated for C,,H,,O,: C = 76.80; H = 4.00 per cent. Reduction of a-An~yd~.obenxiIk~vulic Acid with Hydriodic Acid and Amo~phous Yhosphorus.-The fact that anhydracetonebenzil, which by boiling for a short time with hydriodic acid yields diphenylcyclopen- tenone, can, by more protracted reduction with hydriodic acid and amorphous phosphorus, be converted into the hydrocarbon diphenyl- cyclopentane, led us to try whether an analogous further reduction could also be effected in the case of a-anhydrobenzillEvulic acid. We expected in this way to obtain a diphenylcyclopentanetl~yloic acid. On trying the experiment, however, we found that carbon dioxide was eliminated and methyldiphenylcyclopentane was formed. Found : C = 71.70 ; H = 4.72. Found : C = 76.63; H = 3.93.JAPP AND MURRAY: SYNTHESIS OF PENTACARBON RINGS.153 Ten grams of a-anhydrobeiizillzvulic acid were boiled with 150 grams of hydriodic acid (sp. gr. 1.75) and 20 grams of amorphous phosphorus for 6 hours. Water was added, and et,her to dissolve the organic sub- stance, after which the excess of amorphous phosphorus was filtered off. The ethereal solution, after decolorising with sulphurous acid, was shaken with alkali, which, however, extracted practically nothing from it. On evaporating the ether, a yellow oil remained, which became crystalline on standing. By dissolving the crystalline mass in ether and adding methylic alcohol, the compound was obtained in rosettes of very slender, white needles, which, after repeating this process of crystal- lisation two or three times, melted constantly at 62-63'.The sub- stance is soluble in light petroleum, very soluble in benzene and ethylic acetate, less so in ethylic alcohol and methylic alcohol. Analysis agreed with the formula of metl~~lWip~~enyZcycZo~~e.1atccne. 0.1271 gave 0.4255 CO, and 0.0980 H,O. C = 91.30; H = S.5'7. 0.1243 ,, 0.4166 CO, ,, 0.0955 H,O. c1 = 91.40; H = 8.54. C,,H,, requires C = 91.52; H = 8.48. h'ecluctiorz of ~-~nhydrobenx~~kevu~o~ccctone with Hychiodic Acid.-- This experiment was performed in order to ascertain whether the pro- ducts of reduction of the a-and p-acids were the same or different. Two grams of the finely-powdered lactone were boiled for 2 minutes with excess of the strongest hydriodic acid, and the product of the action was treated exactly as in the preparation of diphenylcyclo- pentenonethyloic acid from a-anhydrobenzillaevulic acid, except that, in order to remove any unchanged lactone or other neutral substance, the ethereal solution of the reduction product was extracted with sodium carbonate and the acid reprecipitated from the carbonate solution. It was again taken up by ether and, on evaporation, remained as an oil, which, on touching it with a crystal of diphenylcyclopentenonethyloic acid, obtained by the reduction of the a-acid, at once began to crystal- lise. The crystals were freed from gummy matter by draining on a porous tile. After recrystallising five times from benzene, they melted at 124-126'. They were indistinguishable from those of diphenyl- cyclopentenonethyloic acid, except that the latter melted 1' higher. 0.1383 gave 0.3953 CO, and OW719 H,O. The a- and the P-acid thus both yield the same clil3JLe~Zylc~clo~ei~~~?~- The conclusions to be drawn from this C = 77.95 ; H = 5.77. C19H1,0, requires C = 78.08 ; H = 5.48. onethyloic acid on reduction. fact are discussed in the introduction. CHEMICAL DEPARTMENT, UNIVERSITY OF ARERDEEN.