JAPP AND LANDER: SYNTHESIS OF PENTACARBON RINGS. 123 VIII. -Synthesis of Pentacarbon R.iugs. Pa9-t I. Anhydracetonebenzil and its Homologues. By FRANCIS ROBERT JAPP, F.R.S., and GEORGE DRUCE LANDER, B.Sc. ONE of the chief objects of the systematic study of the condensations of certain a-diketones (benzil and phenanthraquinone), and of a keto- alcohol (benzo’in), which has occupied one of us, along with various1% JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS. collaborators, more or less continuously from the year 1880 down t o the present time, has been the synthesis of cycloids. The various “ condensations in the ortho series ” which, shortly before this work was begun, had been studied by various investigators, especially by Ladenburg and von Baeyer, indicated a possibility of the occurrence of similar condensations in the case of aliphatic chains containing two ketonic or alcoholic functions in the a-position. I n passing, we may point out that the analogy of the a-position in aliphatic compounds to the ortho-position in the benzene series was not quite so obvious then as it is now.A t that time, Ladenburg’s ‘‘ prism ’’ formula could still be seriously put forward as a satisfactory expression of the reactions of benzene ; and in this formula the ortho-carbon atoms are not directly united. Every condensation, therefore, which bore out the foregoing analogy was a fresh argument against the ‘‘ prism ” formula. Indeed, it is cumulative evidence of this character, rather than any definite disproof, that has caused the ‘‘ prism ” formula to be withdrawn from discussion.The a-diketones and the a- keto-alcohol just mentioned were selected for study, in the first instance, on account of the ease with which they could be obtained. A further advantage was that the products of the various reactions were crystalline solids and, consequently, easy to purify. The phenyl and phenylene groups, to the presence of which in the molecule the latter advantage was largely due, took no direct part in the reactions ; so that, in reality, the problem resolved itself into the relatively simple one of ascertaining the nature of the condensations of the groups -CO*CO- and -CHOH* GO- with various compounds. Against these advantages must be set the deterrent effect which the somewhat formidable-looking formulae of the resulting compounds probably exercised on the majority of readers.That this object of the synthesis of cycloids has been amply realised may be seen from the following list of the various classes of compounds obtained in these reactions : 0xaxoles.-1. By the interaction of a-diketones with aldehydes and ammonia (Trans., 1880, 3’7, 669 ; 1881, 39, 225).-2. From benzo‘in and nitriles (ibid., 1893, 63, 469). From a-diketones, aldehydes, and ammonia (ibid., 1882, 41, 146, 157, and 323; 1886, 49, 464 ; 1887, 51, 552 and 5 5 9 - b . Te.r.tiary lmidaxoles and Quc6te.r.- nary Imidaxolium Compounds. From a-diketones and primary amines of the formula R*CH,*NH, (ibid., 1895, 67, 32). PurfurarLs.--By the action of hydriodic acid on the condensation pro- ducts of a-diketones with ketones (ibid., 1890, 5’7, 662).fifidoles. -By the condensation of benzoi’n with primary benzenoid amines (ibid., 1894, 65, 889). Iinidaxoles.-a. Ordinary (secondary) Inaidazoles.JAPP AND LANDER: SYNTHESIS OF PENTACARBON RINGS. 125 Axines.-By the action of ammonia ( 2 ) on benzoin and (2) on U- diketones (ibid., 1886, 49, 528; 1887, 51, 98). To the foregoing list might have been added : Zcctones and pywho- Zones. A pyw4oZe has also been obtained in an investigation the results of which will be published shortly. I n some of the foregoing condensations, a union of carbon with carbon occurs, but only a t one point: the closing of the chain is effected by some other element -nitrogen or oxygen. I n the present and the two following communications, condensations are described in which t,he union of carbon to carbon takes place a t two points, both carbonyl groups of an a-diketone attaching them- selves to carbon atoms of an aliphatic chain and in this way leading to the formation of cycloids containing only carbon in the ring.These compounds are formed by the condensation of benzil with ketones or ketonic acids of the formula RCH,*CO*CH,R’, in which R’ may be hydrogen, or alkyl, or carboxyl, or *CH,*COOH. The condensation of benzil with acetone was first studied by Japp and Miller (Trans., 1885, 47, 21). Acting on these substances with a small quantity of potassium hydroxide, they found that the aldol C,H,* ~(O.H).CH,*CO CH, C,H,*CO condensation compound, acetonebenzil, , was formed. When treated with an excess of the alkali, this compound parted with a molecule of water, yielding anhydracetonebenzil C17H,,0,. From a study of the oxidation of anhydracetonebenzil, they concluded that a closed chain of carbon atoms had been formed during the condensation, and they inclined to ascribe to the compound the constitution C H 5~ *C<gEt>CO C,H,* GO Y They were led to adopt this view by observing that the compound, oxidation with chromium trioxide in acetic acid solution, yielded on an acid, C1,Hl,O3, to which they assigned the formula of a P-benzoyl- hydrocinnamic (desylacetic) acid.Desylacetic acid has, however, since been prepared 5y Victor Meyer and Oelkers, and we find that the two substances are quite distinct. The argument put forward in favour of the above formula must therefore be withdrawn, although the con- siderations which led Japp and Miller to reject an open chain formula for the compound are still valid.Japp and Burton pointed out (Trans., 1887, 51, 420) that the closed chain might equally well consist of five carbon atoms. They suggested that the first step in the transformation of acetonebenzil into anhydr- acetonebend was $he occurrence of an aldol condensation with the126 JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS. second methyl group of the acetone, yielding the hypothetical inter- mediate compound which would then part with water, forming anhydracetonebenzil. The results of the present investigation lead us to conclude that, in this dehydration, a hydroxyl group is eliminated along with hydro- gen from an adjacent carbon atom, and that anhydracetonebenzil is a diphenylcyclopentenono Z of the formula C',H,*C----- CH I >GO.C,H,* C(OH).CH, The only points in the behaviour of the compound which are not in entire accordance with this formula are that it yields neither an acetyl derivative nor an additive dibromide. The inability to form an acetyl derivative is, however, due t o the circumstance that acetic anhydride has a dehydrating action on anhydracetonebenzil, con- verting it into the compound C,,H,,O,, which was obtained by Japp and Burton (Zoc. cit.), by boiling anhydracetonebenzil with dilute sul- phuric acid. And as regards the non-formation of a dibromide, the merely negative evidence of this fact is outweighed by the positive evidence of the oxidation with sodium hypobromite.Treated with this reagent, anhydracetonebenzil gives, as we find, an almost quanti- t ative yield of Japp and Davidson's desyleneacetic acid, C6H5* C:CH* COOH C,H,* CO I 9 thus proving the presence of double bonds in the molecule of the compound. . This result may also be taken as indicating that desylene- H- C- COOH C,H,* C* COB C,H, acetic acid has the configuration il We do not lose sight of the fact that the formation of desylene- acetic acid could be even more simply explained by assigning to C H *C:CH*CO*CH, ; C,H5*C0 anhydracetonebenzil the open-chain formula I but this formula is excluded by various considerations. Thus anhy- dracetonebenzil yields no acetic acid when oxidised by a mixture of potassium dichromate and dilute sulphuric acid, whereas acetonebenzil readily yields acetic acid under these conditioris (Japp and Miller, ZOC.cit.). But the strongest evidence against the open-chain formula is afforded by the experiments which we are about t o describe. As already mentioned, Japp and Miller, by oxidising anhydrace-JAPP AND LANDER: SYXTHESIS OF PENTACARBON RINGS. 127 tonebenzil by heating it with chromium trioxide in acetic solution, obtained an acid of the formula C,,H,,O,. We find, however, that this acid is not the primary product of oxidation : when the process is conducted in the cold, simultaneous oxidation and hydration occur, C,H,*y( OH) CH, COOH C,;H,* C(OH).COOH 3 is formed. and dipheny Zdihydroxpglutm-ic acid, " I . On heating this acid for some time it decomposes, parting with carbon dioxide and water, and yielding Japp and Miller's acid, which has the C H *C:CH, I C,H,* C(OH)*COOH * formula of an isocinnc~meny~?~~andelic cccid, When boiled with fuming hydriodic acid, or wvth fuming hydro- chloric acid, diphenyldihy droxyglutaric acid also parts with carbon dioxide and water ; but the carbon dioxide is, in this case, furnished by C,H, *aH* CH, COOH, C,H;CO the other carboxyl group, and desylacetic acid, is formed, together with a small quantity of its dehydration C R *C-CH, diphenylcrotolmtone, 11 I The mechanism of the C,H,*C CO \/ product, two pro- 0 cesses may be represented as follows........................................... C~,H,*C(OH)~*CH,*jCOOH~ P c&,* 5: : CH, I ............. ............... ! -3 C,H,.C(OH)-COOH C,H,* C(0H) COOH Action of heat. C,H,* V*CH,-COOH C,H,* CH* CH,* COOE, + I C6H5* C*OH C6H,* co Action of hydracids. a portion of the desylacetic acid, while still in the enolic form, elimi- nating water, and yielding diphenylcrotolactone. The necessit'y of accounting for the formation of desylacetic acid C,H5mCH.CH(OH)--COOH C,H,*C( OH)*cOOH excludes the other possible formula, I , for diphenyldihydroxyglutaric acid. But, quite apart from the question of the exact constitution of this acid (dthough we regard this as settled), the oxidation of anhydracetonebenzil to a dibasic acid containing the same number of carbon atoms as this compound itself is a process that can be explained only on the assumption that a closed carbon128 JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS.chain, which was formed in the original condensation, has been opened during the oxidation, Isocinnamenylmandelic acid yields an acetyl derivative. When heated above its melting point, this acid parts with 1 mol. of water, forming a compound ClGHlZO,. By partial redaction by boiling it for a few minutes with fuming hydriodic acid, it is converted into isopJLenethyl- naundeZic acid, I C H *CH*CH, C, H5*C(OH) *COOH' By the partial reduction of anhydracetonebenzil with hydriodic acid, Japp and Burton obtained a compound, C17H140, melting a t 110", which yielded a hydrazone, and therefore contained the original car- bony1 group of the anhydracetonebenzil. We find that this compound C H -C.CH, C,H,- C-CH, has the constitution of a dip~henylcyclopentenone, 11 >co.That the foregoing change in the position of the double bonds has faken place during the reduction, is shown by the fact that the com- pound yields, on oxidation with sodium hypobromite, diphenylmaleic acid, which, when liberated from its salts, changes into the very - C,H,*C.CO C,H;C:CO characteristic anhydride, 11 >o. The hydrocarbon, C17H18-(m. p. 47"), obtained by Japp and Burton by the complete reduction of anhydracetonebenzil with hydriodic acid and amorphous phosphorus, is a diphen?/ZcycZo~ntane, C,H,*YH*CH, C,H,* C H* CH, >CH,. The various compounds obtained by Japp and Burton (Trans., 1887, 51, 431) by the condensation of benzil with homologues of acetone of the general formulae CH,R'*CO*CH, and CH,R*CO*CH,R must be regarded as homologues of anhydracetonebenzil.The constitution of those members of this class which are formed from symmetrical homo- logues of acetone follows, as a matter of course, from that of anhy- dracetonebenzil it'self ; thus the condensation product of benzil with diethyl ketone (Zoc. cit., p. 438) is di~zethylccnhydrcccetonebenxil, C H 5~ -C= "("H3)>,0. C,H,- C(0H) CH(CH,) But in the case of the compounds obtained from unsymmetrical ketones there are two possibilities ; thus, in nzethyZanh?/dracetonebenxiZ (formed from b e n d and methyl ethyl ketone), the methyl group may replace hydrogen either in the methylene or in the methenyl group of anhy- dracetonebenzil, lending respectively to the formulsJAPP AND LANDER: SYNTHESIS O F PENTACARBON RINGS.129 The following considerations appear to us to decide in favour of Formula I. Von Baeyer's well-known test for non-saturation in organic com- pounds-namely, the rapid change of the colour of permanganate to brown when a drop of permanganate solution is added to a cold solution of the substance in presence of excess of sodium carbonate (Annalen, 245, 146), is not applicable to compounds in which an ethylene group has no hydrogen attached to it, as in this case the action of per- manganate on the substance in the cold is very slow. Von Baeyer mentions, as examples of this limitation, dimethylmaleic acid and A1 tetrahydrophthalic acid (Annulen, 1889, 252, 207); and we find that the same holds good of diphenylmaleic acid, desylenemalonic acid, and dibenzoylstilbene, all of which are stable towards permanganate in the cold.I n this way, it is possible to decide between two competing formulae for an unsaturated compound, one of which contains hydrogen attached to the ethylene group and the other not. Thus, in the case under discussion, if methylanhydracetonebenzil has the constitution represented by Formula I, it should be as easily attacked by per- manganate as anhydracetonebenzil itself; if, on the other hand, its constitution corresponds with Formula 11, it should be as stable towards permanganate as dimethylanhydracet80nebenzil. Experiment showed that the former was the case." The times required for the complete reduction of the permanganate were : with anhydracetone- benzil, 30 seconds ; with methylanhydracetonebenzil, 30 seconds ; with dimethylanhydracetonebenzil, 5 minutes.Amylanhydracetonebenzil, which was also tested, required 18 minutes, the somewhat more slug- gish action being due to the larger molecule, although the compound obviously belongs to the same category as methylanhydracetonebenzil. We had not a specimen of ethylanhydracetonebenzil; but it would doubtless behave like the other two monnlkyl derivatives. The monalkyl derivatives of anhydracetonebenzil prepared by Japp and Burton would therefore be formulated as follows. C,H,* C CH i >GO C H .C-- CH C,H,* C(0H) CH(CJ3,) >" Met h y lanhy drace tonebenzil. (m. p. 179"). 5~ C,H,* C(OH)*CH(C,H,) E th ylanh ydrace tonebenzil. (m. p. 156"). * The experiments were carried out in the way recommended by von Baeyer for testing non-acid substances (Annulen, 252, 286) ; to a solution of the compound in pure alcohol, a little of an aqueous solution of sodium carbonate was added, and then a drop of the permanganate solution.VOL. LXXI. K130 JAPP AND LANDER : SYNTHESIS O F PENTACARBON KIhGS. ">co C H *C-- - C,H,* C(OH)-CH(C,Hll) Amylanh ydracetonebenzil. (m. p. 150.5"). 5~ E X P E RIME N T A L . Prepration of An~Lydracetonebenxil---The following mode of prepara- tion differs from that originally employed by Japp and Miller only in the fact that the ingredients are heated during the reaction. I n this way, a great saving of time is effected, whilst the yield is not diminished. Two hundred grams of finely-powdered benzil, 125 grams of pure acetone (Kahlbaum's acetone (( from the bisulphite compound " was used), and 3 C.C.of a 33 per cent. solution of caustic potash were introduced into a flask, and constantly shaken until the benzil had all dissolved. The liquid became slightly warm during this process, owing to t.he formation of the aldol condensation compound, acetone- benzil. Fifty grams more of the caustic potash solution were then added, and the liquid was gently warmed for about half-an-hour on the water bath, shaking it from time to time. Hot water was then added to remove the potash, and the organic substance was washed with hot water. When cold, the solidified product was ground in a mortar, washed with a small quantity of ether to remove dark-coloured impuri- ties, and recrystallised from benzene. The substance thus obtained is yellow, but is pure enough for the study of its reactions, and even for analysis, so that it is not necessary to subject it to the troublesome and wasteful process required to obtain it in a colourless state (cJ: Japp and Miller, Zoc.cit., p. 27). Cryoscopic determinations of the molecular weight, using benzene as a solvent, gave the following results. The yield was 150 grams. Weight of Weight of Mol. substance. solvent. Depression. weight. I ............... 0.1255 22.49 0.105O 260 I1 ............... 0.3050 21-75 0.245 280 C,7Hl,0, = 250. Action of Acetic Anhydride on Anl~ydraceto.lzebenziZ.--Ten grams of the substance were boiled with 30 grams of acetic anhydride and 5 grams of fused sodium acetate for 4 hours, the latter addition being made because Japp and Miller had obtained no definite result with acetic anhydride alone.The only product, however, was the compound C34H2402, prepared by Ja pp and Burton by boiling anhydracetonebenzil with dilute sulphuric acid. It was deposited from benzene in forms indistinguishable from those of the compound C34H2402, and, like theJAPP AND LANDER: SYNTHESIS OF PEXTACARBON RINGS. 131 latter, melted a t 195-200°, evolving gas. It is formed according t G the equation 2C1iH,,0, - 2H,O = C,,H2,02, and the double molecular weight was assigned to it to account for the fact that, when heated, it parts with 1 mol. of carbon monoxide, yielding the compound C33H240* (Trans., 1887, 51, 426). Our cryoscopic determinations, made with a benzene solution, confirm this conclusion.Weight of Weight of Mol. substance. solvent. Depression. weight. I ............... 0.1995 32.46 I1 ............... 0.2715 21.65 C,,H,,O, = 464. Prepccmtion of Diphenylcyclopentenon~, compound was obtained by Japp and Burton 0*065O 463 0.130 4'73 C,H,* f*UH, C,H5* C*UH, >CO.-This " - by boiling anhydracetone- beniil for a few minutes with excess of fuming hydriodic acid. The following method is more economical and gives a better, although by no means satisfactory, yield. Forty grams of anhydracetonebenzil were boiled with 160 grams of glacial acetic acid, 9 grams of amorphous phosphorus, and 4-5 C.C. of fuming hydriodic acid (sp. gr. 1-96} for 4 hours, using a reflux condenser, and adding a few drops of water from time to time, whenever vapours of iodine appeared in the flask.The liquid was filtered hot and diluted with water. The semi-solid mass which separated was dissolved in ether ; the solution was shaken first with aqueous sulphur dioxide and then with sodium carbonate, dried with calcium chloride, after which the ether was expelled and the residue distilled under reduced pressure. It passed over at 250-260' under a pressure of 18-20 mm. The solidified distillate was recrystallised from alcohol. It formed pale yellow needles, melting at 110', as described by Japp and Burton. The yield of pure substance was only 12 grams. C,H,* CH*CH, C,H,* CHGH, Preparation of Diphenylc yc lopentane, I >CH,.-This compound was obtained by Japp and Burton by heating anhydracetone- benzil with hydriodic acid and amorphous phosphorus in a sealed tube.We find, however, that the reduction may be equally well effected by boiling these substances in a flask fitted with a reflux condenser, Ten grams of anhydracetonebenzil, 150 grams of hydriodic acid (sp. gr. 1 T), * Japp and Burton state that the substance, C,,H,,O, crystallired from benzene, melts, after expelling the benzene of crystallisation, at 162-163". This melting point is too low ; probably the benzene was not entirely expelled. The substance is deposited from alcohol in lustrous crystals, containing no solvent of crystallisation and melting a t 175". K e132 JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS. and 20 grams of amorphous phosphorus were employed, and the boiling was continued for 5 hours. The product was purified by distillation under reduced pressure and subsequent recrystallisation from alcohol (cf.Trans., 1887, 51, 423). The yield was small : from the foregoing quantity of anhydracetonebenzil only 1-8 grams of pure recrystallised hydrocarbon melting a t 47' were obtained. Cryoscopic determinations of the molecular weight, using benzene as a solvent, gave the following results. Weight of Weight of MoI. subs trtnce. solvent. Depression. weight. I. ......... 0.1475 15.34 0,225" 209 I1 ....... , . 0.1025 18.05 0.1 35 206 C17Hl, = 222. This 1 : 2-diphenylcyclopentane ought to exist in two modifications, cis and tyans, but we failed to detect, among the products of the reduction, any form other than the foregoing. Oxidation of Dip~nyZc/cZopentenone with Sodium Hypobrornite.Formation of Bipihen y lrna Zeic A cid. -Twelve grams of fine1 y-powdered diphenylcyclopentenone, the first reduction product of anhydracetone- benzil, were mixed with a solution of 36 grams of bromine in excess of strong caustic soda, and the mixture was stirred by means of a mechanical stirrer for 72 hours; the unchanged substance, which was separated by filtration, weighed 9 grams. The alkaline filtrate was saturated with sulphur dioxide, acidified with dilute sul- phuric acid, and extracted with ether. Aqueous sodium carbonate removed no organic acid from the ether, which was then shaken with sodium hydroxide solution; the latter, on acidifying, gave a yellow precipitate ; this, on recrystallisation from benzene, was deposited in thick, yellow needles, with a slight greenish fluorescence, and melted at 156-157'.These are the properties of diphenylmaleic anhydride, into which diphenylmaleic acid spontaneously changes on liberation from its salts. Analysis gave figures agreeing with the expected formula C,,H,,O,. Found : C = 76.50 ; H = 4.11. Calculated : C = 76.80 ; H = 4.00 per cent. From the 3 grams of diphenylcyclopentenone used up in the oxidation, 2 grams of pure anhydride were obtained. Allowing for unavoidable loss in purification, this may be regarded as a quantitative yield. Oxidation of Anhydracetonebenxil with Sodium Hypobrornite. Formation of De Zeneacetic Acid.-Ten grams of finely-powdered anhy- dracetonebenzil were shaken with a solution of 20 grams of brominein an excess of a 15 per cent.solution of After remaining for about an hour in the cold, with frequent shaking, the liquid was almost filled with a colourlese, crystalline substance which sodium hydroxide.JAPP AND LANDER: SYNTHESIS OF PENTACARBON RINGS, 133 enclosed particles of unaltered anhydracetonebenzil. On adding a little water and warming to about 40°, the substance dissolved. The liquid was filtered, saturated with sulphur dioxide, and acidified with dilute sulphuric acid, which produced a microcrystalline precipitate of an organic acid. This acid was extracted with ether and the ethereal solution shaken with a solution of sodium carbonate. After this treat- ment, the ether left practically no residue on evaporation. The organic acid, reprecipitated from the Carbonate solution, and purified by re- crystallisation from benzene, melted at 168 O; recrystallised, it melted at 1 4 2 O , and, after resolidification, again at 168'.It was indistin- guishable from a specimen of desylenaacetic acid prepared by Japp and Davidson (Trans., 1895, 67, 138) by heating desylenemalonic acid, and, on analysis, gave figures agreeing with the formula of desyleneacetic acid. Found : C = '76.03 ; H = 4-87. Calculated for C,,H,,O, : C = '76.19; H: = 4.76 per cent. Japp and Davidson also observed two melting points for desylene- acetic acid-15O0 and 168O-but gave it as their opinion that the lower melting point was merely that of an unstable crystalline form of the substance, I n spite of this, it is stated, in the abstracts of the paper which appeared in the Bericlhte (Referute, 1895, 465) and in the Bulletin (Trccvaux ktrangers, 1895, 1039), that stereoisomerides were observed. It is clear, however, from the above changes in the melting points- from the lower to the higher on melting and resolidifying, and from the higher to the lower on recrystallising-that the phenomenon is due merely to dimorphism.The form of lower melting point does not contain benzene of crystallisation. An attempt to obtain a stereoiso- meride by heating the substance in a sealed tube with glacial acetic acid saturated with gaseous hydrogen chloride gave no result ; only unchanged substance was recovered. Oxidation of Anhydrucetonebenxil with Chromium Trioxide. Formation of Dipphenyldihydroxyglzctccric Acid.-The conditions essential to success in the following experiment were discovered only after considerable expenditure of time and material, and a very slight deviation from them suffices to reduce the yield-at no time good-to the vanishing point. Fifty grams of anhydracetonebenzil were dissolved with the aid of heat in 350 grams of glacial acetic acid and the solution cooled.The beaker containing it was kept immersed in cold water and the solution stirred by a mechanical stirrer, while a solution of '75 grams of chromium trioxide, in a sufficiency of glacial acetic acid, was added in small portions at a time, so as to avoid any appreciable rise of temperature. A slight separation of solid substance, which afterwards re-dissolved, occurred on the addition of the chromium trioxide. After the stirring134 JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS. had been continued for 24 hours, the beaker was removed from the water and the solution allowed to sta,nd a t the ordinary tempera- ture for 5 days, after which it was ponred into excess of water, which occasioned the separation of a flocculent substance; this was collected in a filter, washed with a little cold water, and dissolved in alcohol with the aid of a gentle heat.Alcoholic caustic soda and afterwards aqueous caustic soda were then added, until the liquid was distinctly alkaline; this caused a separtion of solid matter. Warm water was then added to dissolve the salts of organic acids, and the liquid was filtered from unaltered substance and chromium componnds. From the filtrate, dilute sulphuric acid precipitated an oily acid, which was extract,ed with ether, again removed from the ether with sodium carbonate solution, reprecipitated, and re-dissolved in ether.The ethereal solution wa,s evaporated to a small bulk and excess of benzene added; this occasioned the separation of the new acid in needle-shaped crystals grouped in rosettes. (Yield : 11-12 grams.) The substance thus obtained, although otherwise pure, contains benzene of crystallisation, which, apparently, cannot be entirely expelled by heat without a t the same time decomposing the substance. The crystals, freed as far as possible from benzene of crystallisation by long exposure t o the air, were therefore dissolved in ether, and t o the solu- tion light petroleum was carefully added. I n this way, the compound was obtained in crystals which, when rapidly heated, melted a t 120° with evolution of gas.On analysis they gave figures agreeing with the C,H,*~(OH) *CH, COOH C,H5* C(OH) *COOH formula of di~~en~Zdih~drox?/glzctaric acid, 0.1427 gave 0.3367 CO, and 0-0640 H,O. C = 64-35 ; H = 4.98. 0.2209 ,, 0.5219 CO, ,, 0.1007 H,O. C=64*43 ;H=5.07. Cl7Hl,O, requires C = 64-55 ; H = 5.06 per cent. The silverr. salt was obtained as a white precipitate by dissolving the It was dried in a acid in dilute ammonia and adding silver nitrate. vacuum desiccator. 0.3070 gave, on ignition, 0.1263 Ag. C17H,,06Ag2 requires Ag = 40.75 per cent. When the oxidation of anhydracetonebenzil is allowed to go on in the cold for a fortnight, instead of for 5 clays, or when the liquid is heated, isocinnamenylmandelic acid (m.p. 1 60°), the oxidation product obtained by Japp and Miller, is formed. This is not due to a further oxidation, but to a splitting off of carbon dioxide and water from diphenyldihy droxyglutaric acid (see following paragraph). Action of' Heat on D~p~n?/ld~~ydoxygZut~~~ic Acid. Formatioiz of Ag = 41-14.JAPP AND LANDER: SYNTHESIS OF PENTACARBON RINGS. 135 Isocinnamenylrnandelic Acid-Diphenyldihydroxyglutaric acid decom- poses when kept for some time a t looo, melting, with evolution of gas, and then resolidifying. The best yield of the product of transforma- tion by heat is obtained when air is excluded during the process, and the temperature is not allowed t o exceed 105'. The operation mas conducted as follows.Ten grams of diphenyldihydroxyglutaric acid were introduced into a tubulated flask and heated at 105" by means of a glycerol bath, the flask being attached to a filter-pump and exhausted during the entire process. The fusion, frothing, and resolidification of the substance proceeded gradually, from the outer portions inwards. The end of the reaction was shown by the rise of the mercury in the gauge of the pump. The product was dissolved in hot benzene, from which it separated in tufts of needles melting at 160' with evolution of gas. The total yield from 40 grams of diphenyldihydroxyglutaric acid, treated as above, was 18 grams of recrystallised substance. On analysis, it gave figures agreeing with the formula of isocinncmengl- C H *C:CH, C,H,- C( OH) *COOH ' mandelic acid, I Found : C = 75.56 ; H = 5.38.Clal- culated : C = 75.59; H = 5.51 per cent. An experiment, in which a weighed quantity of diphenyldihydroxy- glutaric acid was heated in a Sprengel vacuum, pumping off and mea- suring the carbon dioxide evolved, and determining the loss in weight of the heated substance, showed that the decomposition took place according to the equation C,7H,,06 = Cl6Hl,O, + CO, + H,O. Isocinnamenylrnandelic acid is identical with the acid obtained by Japp and Miller by the oxidation of anhydracetonebenzil (Trans., 1885, 47, 30), and regarded by them as /3-benzoylhydrocinnamic (desylacetic) acid. The lower melting point (152') which they found is accounted for by the fact that, in the method of preparation employed by them benzoic acid is simultaneously formed, and it is difficult to free the substance from it, except by repeated recrystallisation.A speci- men prepared by them, which we examined, contained a little benzoic acid, but was otherwise indistinguishable from that just described. Owing to the; circumstance that Japp and Miller erroneously as- cribed to this acid the const'itution of desylacetic acid, and that Victor Meyer and Oelkers, who afterwards prepared the true desylacetic acid, were apparently unaware of the existence of an acid for which this constitution had already been claimed, and therefore made no compa- rison of the two acids, the properties of both-often of a somewhat contradictory character-figure side by side in Beilstein's Hcindbuch(3rd ed., vol. ii, p.1713) as those of desylacetic acid. As it would be impossible for any one, without studying the original memoirs, to assign these properties to the compounds to which they belong, we append, in tabular form, a comparison of the two acids. 1. Crystallises from benzene in tufts of needles, melting a t 161". 2. The solution in sodium carbonate is instantaneously oxidised by per- manganate in the cold. 3. Not reduced by sodium amalgam in alkaline solution. 4. When boiled for a few minutes n-ith fuming hydriodic acid, is reduced t o isophenethylmandelic acid (v. infm). 5. Does not interact with phenyl- hydrazine or with hydroxylamine. 6. Yields, with acetic anhydride, a monacetyl derivative (v. infra). 7. Heated above its melting point, it parts with water yielding a compound C,,H,,O,, ni.p. 118-120" (v. Cnfm). Desylacetic Acid. CRH5* vH-CH,*COOH C6H5* co 1. Crystallises from benzene in lustrous 2. Sodium salt stable towards per- octahedra, melting at 162". nianganate in the cold. 3. Reduced to By-diphenyl-y-hydroxy- butyric acid, which when liberated from itssalts, formsthe lactoneC,H,*CH*CH, (To be described in a subsequent paper.) 4. Not reduced by this treatment. Longer boiling, however, reduces it to By-diphenylbutyric acid. (To be de- scribed in a subsequent paper.) 5. Yields, with phenylhydrazine, anilinodipheiiylpyrrholone (Klingemann). 6. Action not studied. 7. Yields diphenylcrotolactone, m. p. 151 -5" (Klingemann). The barium salt, (C16H130,),Ba,2H,0, given by Beilstein under desylacetic acid, was prepared by Japp and Miller, and is a salt of isocinnamenylmandelic acid.The fact that isocinnamenylmandelic acid is not reduced by sodium amalgam, shows that the unsaturated group, the presence of which is proved by the behaviour of the acid towards permanganate, is not in the up-position relatively to the carboxyl group. Action of Hydi-iodic Acid and of Hydrochloric Acid on Diphnyldihy- droxyylutaric Acid. Formation of Desylacetic Acid and Diphenylcroto- lactone.---Five grams of diphenyldihydroxyglutaric acid were boiled with excess of fuming hydriodic acid (sp. gr. 1.96) for 5 minutes. The ethereal solution of the product was shaken with aqueous sulphur dioxide, washed with water, and then treated with a solution of sodiumJAPP AND LANDER: SYNTHESIS OF PENTACARBON RINGS.137 carbonate, which extracted an organic acid, whilst a neutral substance remained, dissolved in the ether. The reprecipitated acid was purified by recrystallisation from benzene, from which it was deposited in the characteristic lustrous octahedra of desylacetic acid, melting a t 16 1". The yield was 0.7 gram. When dissolved in sodium carbonate and treated with a drop of permanganate in the cold, no action occurred, showing that it was free from the isomeric isocinnamenylmandelic acid. Analysis gave figures agreeing with the formula of desylacetic acid. Found : C = 75.54 ; H = 5-57, Calculated for C16H1403 : C = 75.59 ; H = 5.5 1 per cent. It was in every respect indistinguishable from a specimen of desylacetic acid prepared by Knoevenagel's method (from sodium deoxybenzoin and ethylic iodoacetate, hydrolysing the ethereal salt thus obtained).The ethereal solution which remained after the desylacetic acid had been removed by sodium carbonate, left on evaporation a neutral substance which crystallised from benzene in tufts of needles melting a t 150". It was identical with Klingemann's diphenylcrotolactone (m. p. 151*5"), which he obtained by the dehydrating action of heat on desylacetic acid (Anncden, 1892, 269, 134). A supersaturated solution of the substance in benzene crystallised immediately on adding a crystal of diphenylcrotolactone. A simultaneous formation of desylacetic acid and diphenylcrotolactone, by the action of hydriodic acid on desylene- malonic acid, was observed by Japp and Davidson (Trans., 1895, 67, 136).As the action of hydriodic acid on diphenyldihydroxyglutaric acid was thus not a reduction, but merely an elimination of carbon dioxide and water, it seemed probable that hydrochloric acid would have the Eame effect. This was found to be the case. Diphenyldihydroxyglutaric acid, boiled with fuming hydrochloric acid for 10 minutes, gave desylacetic acid, which, after purification, crystallised in the charac- teristic octahedra and melted a t 160". An acid melting at 185" was simultaneously formed, but in quantity too small for further examina- tion. Porma- tion of Isop?~enethylmccndelic Acid.--Five grams of isocinnamenyl- mandelic acidwere boiled with excessof fuming hydriodic acid (sp. gr. 1 *96) for 5 minutes. The product, dissolved in ether and freed from iodine by sulphurous acid, was separated by sodium carbonate into two substances : an acid and a neutral oil, the latter of which was not further examined.The acid crystallised from benzene in small, oblong plates, generally grouped together into rosettes, melting at 134-136'. The solution in sodium carbonate was stable towards permanganate in the cold, showing that the ethylene group of isocinnamenylmandelic acid had taken up Action of Hydriodic Acid on Isocinnamenylinccndelic Acid.138 JAPP AND LANDER : SYNTHESIS OF PENTACARBON RINGS. hydrogen during the reduction. those required for isop?~eneth?/ZmandeZ,ic cccid, Analysis gave figures agreeing with C,H,* YH*CH, C,H,* C(OH)*COOH 0.1214 gave 0.3339 CO, and 0.0688 H,O. 0.1306 gave 0.3592 CO, and 0.0749 H,O.We endeavoured, by carrying the reduction further, to convert this acid into up-diphenylbutyric acid, in order to compare it with the py-diphenylbutyric acid which we describe in a subsequent paper. For this purpose, 10 grams of isocinnamenylmandelic acid were boiled for 5 hours with hydriodic acid (sp. gr. 1.7) and amorphous phosphorus. A considerable quantity of neutral oil was formed, together with a mixture of acids. We could not succeed in separating the latter from one another, owing to the small quantity of substance a t our disposal. Porma- tion of the Monacetyl De~ivatiue.-One gram of isocinnarnenylmandelic acid was warmed with acetic anhydride a t 100" for 18 hours, the excess of anhydride was distilled off under reduced pressure, and the residue twice crystallised from benzene.The acetyl derivative was deposited in tufts of needles melting, without decomposition, a t 145-1 46". Analysis of the substance, dried at 90°, gave figures agreeing with the formula C=75*01; H=6*29. C =75 01 ; H=6*31. Cl,Hl,03 requires C = '75.00 ; H = 6.25 per cent. Action o j Acetic Anhyd?*ide on Isocinnarnen~lman&Zic A c i d ~l,Hl,O, ( %,H:3*)* 0.0856 gave 0.2292 GO, and 0.0445 H,O. C,,H,,O, requires C = 72.97 ; H = 5.41 per cent. Action of Heat o n Isoci.lznc~~~enylmccndelic Acid.-Three grams of isocinnamenylmandelic acid were introduced into a small distilling flask, which was exhausted by a Sprengel mercury pump and heated in a glycerol bath a t 160' until the evolution of gas had almost ceased. The gas, which was removed by the pump during the process, was collected and examined. It was found to be pure carbon dioxide, and its volume was 120 C.C. (standard-dry), or rather less than half of that which would have been evolved had the whole of the acid been decom- posed with elimination of this gas. The substance remaining in the flask was dissolved in ether and shaken with a solution of sodium carbonate, which extracted only a trace of an acid, but caused the separation of oily drops which remained suspended in the aqueous liquid ; addition of benzene readily removed these. The ether-benzene solution was evaporated to a small bulk and alcohol added ; this caused the separation of large, yellow, prismatic crystals. Recrystallised from hot alcohol, the substance was deposited in yellow needles, closely C=73*02; H=5*77.JAPP AND LANDER: SYNTHESIS O F PENTACARBON RINGS. 139 resembling benzil in appearance, but melting a t 118-120'. Further recrystallisation did not remove the colour or alter the melting point. Analysis gave figures pointing to the formula Cl6Rl2O2. 0,1602 gave 0.4769 CO, and 0.0746 H,O. The substance is formed from isocinnamenylmandelic acid according C=81*19; H=5*17. CI6H,,0, requires C = 81.35 ; H = 5.09 per cent. t o the equation and the evolution of carbon dioxide observed is due to another reaction, the product of which is probably the oily substance which remains after the removal of the foregoing crystals; but me could not succeed in isolating any definite compound from this. The compound Cl6Hl,O2 (m. p. 1 18--l2O0) is insoluble in aqueous caustic alkali; but if it is dissolved in alcoholic caustic soda, and the solution evaporated to dryness, the residue is soluble in water. On acidifying, an organic substance is precipitated ; but the quantity a t our disposal did not suffice for further examination. If the compound C,,H,,O, is a lactone, it cannot be formed from isocinnamenylmandelic acid except by an intramolecular change. C16Hl403 = C,6H,,O2 + w, CHEMICAL DEPARTNENT, UNIVERSITY OF ABERDEEN.