1362 FARMER AND INGOLD: CXLVII1.-The Chemistry of PoZycycZic Structures in Belation to tlzei?. Homocyclic UnsatuT-ated Isomerides. Part I. Some Derivatives of cyclo-Pentene and dicy clo Pentane. By ERNEST HAROLD FARMER and CHRISTOPHER KELK INGOLD. PERKIN and Thorpe (T. 1901 79 729) showed that when ethyl aa’-dibromo-~&dimethylglutarate is condensed with ethyl malonate in the presence of an excess of sodium ethoxide there is formed a remarkable yellow sodium compound to which hitherto the formula I has been ascribed. Numerous derivatives of this sub-stance were prepared both by hydrolysis and by alkylation and subsequent hydrolysis and many of the products so obtained were subjected to oxidation and reduction. Much more recently one of us in conjunction with Prof.J. F. Thorpe prepared from ethyl ad-dibromocyclohexane-1 1-diacetate a second rather similar yellow sodium compound (11). This also yielded a large number C(CO,Et)*Q:C(ONa) *OEt CMe4(Co2Et)- c THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC. 1363 of hydrolytic products analogous broadly speaking to those of the gem-dimethyl series but exhibiting a number of very striking differences which were studied in some detail (T. 1919 116 320) and correlated in a definite manner to the constitutions of the substances concerned. I n the course of neither of these investigations has there been discovered any fact which could reasonably be regarded as casting doubt on the bridged structures assigned to these sodium com-pounds and to their more immediate derivatives.On the contrary, much evidence confirming these structures has been accumulated. The principal item consists of course in the mode of formation of the sodium compounds. Thus when e thy1 aa'-dibromo-@/3-di-methylglutarate is treated with two molecular proportions of ethyl sodiomalonate there is formed the sodium compound of an ester, which as it can readily lie alkylated cannot but have the structure 111. This structure is quite inevitable and is strictly analogous to that of ethyl ethoxycaronate (IV) which is formed by the action of sodium ethoxide on the dibromo-ester. The ester (111) on being treated with sodium in xylene or with an excess of alcoholic sodium ethoxide loses one molecule of ethyl alcohol and is converted into the yellow sodium compound which therefore must have either formula I or formula V.Of these the former alone is capable of C( CO,Et)*CH(CO,Et) y( C@,Et)*OEt CMe2<bH.C0,Et CMe2<CH*C0,Et (111.) ( IV. ) C- -C(CO,Et) :C(ONa)*OEt CMe,/ \GO C?O,Et int,erpreting t,he many decoinpositions of the substance. Quite recently doubt has been cast by Toivoaen (,4nnulen 1919, 419 176) on the bridged constitution which has hitherto been assigned to these compounds. When the sodium compound (I) is hydrolysed by acids it yields first a dibasic acid (VI) and finally, a monobasic acid (VII). This monobasic acid was obtained by \c/ (V.) (CO,Et)*F:C(GNa)*OEt CMe,< C(CO,H)*YH*CO,H CMez<&CO,Et)*CO CH-- co (VII.) 3 D" 1364 FARMER AND INGOLD: Toivonen in the course of some experiments on the oxidation of isodehydrofenchoic acid and as a result of his investigations he formed the opinion that not only this acid but all the compounds of the series including the yellow sodium compound itself were unsaturated substances containing the cyclopentene and not the dicycZopent,ane ring system.Thus according to Toivonen the sodium coinpound would be represented by the formula VIII the dibasic acid by IX and the monobasic acid by X. Toivonen found + C(CO,Et)====~*OO,Et C(CO,H)=FH CMc2<C[ C( ONa) *OEt]*CO -+ CMe"cH(co2H)*co that the oxidation of isodehydrofenchoic acid by alkaline per-manganate proceeded in two stages. I n the first place a diketonic acid (XI) was produced. This then underwent internal condensa-tion under the influence of the alkali and gave Perkin and Thorpe's acid with the elimination of one molecule of water.Toivonen represents this reaction as follows : (XI. 1 (X.1 We therefore have two methods of formation of one and the same substance which are exceedingly difficult to reconcile with one another. I n fact one must either assume that the conversion of the ester (111) into the yellow sodium conipouiid instead of being a simple Dieckniann condensation is a remarkable change involving the rupture by alcoholysis of t,he cyclopropane ring, possibly as in the following scheme: C( CO,Et)*CH( CO,Et) C( C0,Et) (OEt)*CH( C0,E t)z @Me2<bH*c*,Et * c'Te2<CH,*C02Et C(C0,Et) (OEt)* 7 K*C02Et ~ CH (C0,Et)--CO -+ CMe,< C(C0,Et) =y*CO,Et CMe~<CH(CO,Et)*CO or regard t'hc iiit.erna1 coiiilcnsal ioii of di~netJ~yldiket~ohesoic acid (XI) as taking place in two st,agcs.111 the first place t,here mus THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC. 1365 be formed a cyclic aldol condensation product (XII) which is then dehydrated across the cyclopentane ring : (VII.) Both alternatives appear almost equally extraordinary. The subject therefore obviously required fresh investigation and some months ago we were asked by Prof. J. F. Thorpe to under-take this work and there have now been obtained r3sults which show unquestionably that for the series of compounds with which we are here concerned the bridged and not the unsaturated, constitution is correct. Turning first to the facts which have already been placed on record one observes two reasons why the formula suggested by Toivonen cannot be regarded as adequate.The first concerns the dibasic acid the bridged formula for which is VI the double-bonded formula IX. That the alternative formulze XIII and C( CO,H)*QH G(CO,H):F*CO,H CMe < C(CO,H)*CO cMe~<CH2-C0 (XIII.) (XN.) XIV are incorrect is proved by the fact that the acid is a tauto-meric substance ; it readily gives a coloration with ferric chloride, and its ester can be alkylated. Now an acid of the formula IX, in which as an examination of structural models shows the carboxyl groups are actually further apart than is the case with such compounds as trans-hexahydroisophthalic acid or cis-hexahydro-terephthalic acid would not be expected to form an anhydride. Actually however the acid does form an anhydride with the greatest of ease a fact which is in full accordance with the bridged-ring formula (VI).The second point arises in connexion with the ester produced by methylating the yellow sodium compound with methyl iodide. The bridged and unsaturated structures for this substance are shown in formulae XV and XVI. On treating with alcoholic potassium hydroxide there is formed the lactone of a hydroxy-tribasio acid the produotlioln of which izlvoilves (a) the hydrolysis of all three carbethoxyl groups to carboxyl; ( b ) the loss of one carboxyl group by elimination of carbon dioxide; ( c ) fission with the addition of water in the immediate neighbourhood of th 1366 FARMER AND INGOLD: ketone group which thus becomes a carboxyl group; (d) a further fission with the addition of water.Now whether formula XV C(C0,Et) *C)Me*CO,Et C(CO,Et)==$WO,E t CMe,< C(CO,Et)*CO I C~xe2<cMe(C0,Et)*C0 (XV.) (XVI.) or formula XVI is adopted process ( a ) can only take place in one way process ( b ) in two ways and process ( c ) in two ways. I f one adopts the structure XV the addition of a further molecule of water process (d) must involve the fission of one of the cyclo-propane bonds. On the other hand if formula XVI be accepted the addition must take place a t the double bond since the product is fully saturated. This might occur in two ways. On considering the combinations of these possibilities one observes that as many lead to the same formulz there are but six possible structures for such a hydroxy-tribasic acid derived from an ester of the constitution XV and six from an unsaturated ester of the constitution XVI.The proper-ties of the lactone (Toc. cit.) show however that in the correspond-ing hydrolxy-acid (a) the hpdroxyl group is in the y-position with respect to one of the three carboxyl groups; (6) that no two carboxyl groups are attached to the same carbon atom; (c) that the two carboxyl groups other than that to which the hydroxyl group is in the y-position are attached to two carbon atoms directly united. These conditions reduce the number of possible formulze derived from the structure XV t<o three one of which is the accepted one and the number from XVI to one only namely, XVII. It will be seen that this differs from the customary formula This could take place in six ways.C(C0,H)*CH,*C02H CMe,/ C! Me, \ Z . c O * O (XVIT. ) (XVIII.) (XVIII) only in the position of the single methyl group. The difference however is an important one as the lactonic acid was found to exist in two forms namely a cis- and a trans-form (or meso- and racemic) each of which yields its own anhydride that of the trans-lactonic acid passing on distillation into that of the cis-lactonic acid. This property is characteristic of substances of the type of s-dimethylsuccinic acid and clearly proves that there exists in the molecule of the lactonic acid a free open-chain succinic acid residue in which both the carboxyl-bearing carbon atoms are asymmetric. This condition is fulfilled by formula XVIII but formula XVII is obviously incorrect as it lacks one of the necessary asymmetric carbon atoms in the succinic acid group.One mus THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC. 1367 therefore conclude that the unsaturated structure XVI for the methylation product of the yellow sodium compound is incapable of interpreting the properties of the products obtained by hydrolysis. Toivonen however was more deeply impressed with the behaviour of the monobasic acid on oxidation. Perkin and Thorpe had shown (Zoc. cit.) that when treated with alkaline perman-ganate it was converted into aa-dihydroxy-88-dimethylglutaric acid (XXII). Toivonen proved that the reaction could be carried out a t the ordinary temperatures and as it is capable of being easily explained along conventional lines if the unsaturated struc-ture be assumed he saw in it conclusive confirmation of this method of formulation.The successive stages are as follows: (XX.) (XXI.) GMo2<C(OH)2*CO2H + ~ae,<C02H C H,*C02H CH,*CO,H (XXII.) (XXIII.) I n practice between four and five atoms of oxygen are taken up, and the product is a mixture of dihydroxydimethylglutaric acid (XXII) and as-dimethylsuccinic acid (XXIII) . I n spite of the apparent simplicity of this explanation the issue is not in reality quite so clear. For as was shown in the paper by Ingold and Thorpe (Zoc. c i t . ) the bridge-bond in such dicyclo-pentane systems is in a condition of great strain and may become -in fact in certain circumstances it undoubtedly does become-the most unstable part of the molecule giving rise to reactions the similarity of which with the reactions characteristic of unsaturated compounds is very striking.This state of strain owes its origin t,o the fact that in both the rings to which the bridge-bond is common, the internal angles are considerably less than the normal angle at which two free valencies are inclined (Ingold and Thorpe Zoc. cit.). The case of carone is very different. Here the internal angles between the valencies in the two rings separated by the bridge differ in opposite ways from the normal angle of inclination of carbon valencies. There is therefore a mutual accommodation existing between the strains which react on the carbon atoms terminating the bridge rendering the latter stable to a consider-able degree."he theory of the matter may readily be placed o 1368 FARMER AND INGOLD: a quantitative basis by a simple calculation on lines indicated elsewhere (Zoc. c i t . ; note this vol. p. 603). Therefore Toivonen’s argument that because carone dimethyldzcycloheptanone (XXXI, p. 1370) can be oxidised by permanganate to a cyclopropane deriv-ative namely caronic acid (XXXII p. 1370) the acids derived from dimethyldicyclopentanone should behave similarly with this reagent cannot be regarded as being in the least degree convincing. On a priori grounds i t might be very difficult to discover an oxidising agent capable of attacking the four-membered ring at the carbonyl group and yet leaving the somewhat unstable bridge-bond intact. On the other hand should such a substance as caronic acid be obtainable by oxidation under special and care-fully regulated conditions the circumstance could not but be regarded as the clearest possible proof of the dicyclic constitution of this series of compounds.This proof we have now been able to The general plan pursued in our experiments was its follows. If the alternative formulz VI and I X for the dibasic acid and the two formulm XV and XV1 for the methylation product of the yellow sodium compound be examined it will be noticed that in the one case a carboxyl group and in the other a methyl group, is in a different position relative to the gem-dimethyl group in the two alternative formulz. It should be possible to ascertain the true positions of these groups by oxidation. Thus for example, whilst the ester (XV) after hydrolysis and oxidation might perhaps yield some derivative of PP-dimethylglutaric acid an ester the formula of which is XVI should give derivatives not of PP-dimethylglutaric acid but of aflfl-trimethylglutaric acid.There are similar differences in the oxidation products t o be expected from dibasic acids having the structures V I and IX. A number of interesting results have already been obtained in this field but it has become apparent that an extended investigation is necessary both in the series with which we are here concerned and in other related ones. It is therefore our desire to place on record a t the present time only a limited number of experiments on the oxida-tion of the dibasic acid (VI) which however supply singularly convincing evidence regarding its structure.I n the first place when an aqueous solution of t,he dibasic acid is titrated with cold alkaline permanganate a sharp end-point is reached after three atoms of available oxygen have been taken up. The resulting solution contains aa-dihydroxy-B8-dimethylglutaric acid and oxalic acid the reaction supply. CgH1O0 + 3 0 + 2HZO = C,H,,O + C,H,O, being apparently quantitative. The behaviour of the dibasic aci THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC. 1369 with permanganate is therefore very similar to that of the mono-basic acid (VII). The remarkable fact regarding the product'ion of this dihydroxydimethylglutaric acid by the action of cold alkaline permanganate on these acids is that the fission of the bridge-bond is not due to oxidation but is of a purely hydrolytic character.There is therefore no analogy to fission of the double bond in unsaturated substances by permanganate the first stage of which involves the addit'ion of two hydroxyl groups. The fission of the bridge must be assumed to be due to the addition not of ZOH but of H-OH otherwise it is not possible to account for the structure of the product. The oxidation for example of the dibasic acid is therefore to be represented as follows: C(OH),-CO,H ' C(OH),*CO,H Cnle2<C H,*CO* CH(OH)*C02H * cMe2<CH2*CO*CO* CO,H (XXV.) (xxvr . + CMe,<c(oH)2*Co2H CH,*CO,H + CO,H*CO,H (XXII.) If the process is conducted with care no dilnethylsuccinic acid is formed. A precisely analogous scheme involving an intermediate hydrolytic product similar to XXIV and two intermediate oxida-tion products like XXV and XXVI may be considered as repre-senting the course of the oxidation of the monobasic acid-only in this case four atoms of oxygen are t,aken up as the final product., which replaces oxalic acid in the scheme outlined above is not formic acid but carbon dioxide.The formation of oxalic acid along with dihydroxydimethyl-glutaric acid when the dibasic acid is oxidised affords an interest-ing confirmation of the existence originally of the bridged struc-ture. For if the double-bonded formula IX for the original acid were correct the production of oxalic acid as a main product would be impossible. Four and not three atoms of oxygen would be taken up the successive changes being represented as follows : C(C0,H) =QH C( OH) (CO,H)-$XC-OH Cn4e~<CH(C02a)-C0 -+ CMe2<CH (GO,*)--CO -3 (XXVII.) C(OH),-CO,H C( OH),* CQ H (XXVIII.) (XXIX.) C(OH),*CO,H C(OH),*CO,H CMe,<CH(CO,H) -+ CMe2<CH,*C0, 1370 FARMER AND INGOLD: I n this method of representation the intermediate products XXVII to XXX are strictly analogous to the substances XIX to XXII which figure in the corresponding scheme (p.1367) for the oxidation of the monobasic acid whilst the direct elimination of a carboxyl group as carbon dioxide from the acid XXX appears to be the only method of accounting for the unsubstituted methylene group in the final product. We are unable to see any plausible alternative mechanism whereby the f orniation of oxalic acid along with dihydroxydiniethylglutaric acid by the oxidation of an unsaturat,ed substance having the formula IX might be explained.Owing to the curious hydrolytic action on the bridge-bond to which reference has just been made it does not appear possible to obtain cyclopropane derivatives by oxidation with perman-ganate. We have experimented with many other oxidising agents under a variety of conditions and have discovered two reagents by means of which it is possible to produce caronic acid from the dibasic acid which therefore must have the bridged structure VI. These are hydrogen peroxide and potassium ferricyanide. I n the former case the conditions necessary in order to obtain a good yield of caronic acid appear rather difficult to determine but with cold ferricyanide the oxidation proceeds very smoothly and if a little of the reagent is added each day is complete in rather more than a week when an excellent yield of trans-caronic acid can be extracted from the solution.The direct comparison wit'h the case of carone asked for by Toivonen now becomes possible: (XXXI.) (Carone.) C(CO,H)*~H*CO,E co CH-- CMo,< I K3WCNk (XXXII.) (Caronic acid.) -i If this experiment places as we believe i t does the bridged constitution beyond doubt there appears to be no alternative to accepting the mechanism proposed on p. 1365 for the internal con-densation of Toivonen's dimethyldiketohexoic acid (XI). Such a reaction is not however entirely without precedent for the inter-mediate aldol-condensation product (XII) is a cyclic pinacoline alcohol very similar in constitution to the dimethylcy clohexanol (XXXIII) the dehydration of which has been investigated by Meerwein (Annulen 1914 405 129) who found it to take place across the ring the final product being isopropylcyclo THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC.1371 pentene (XXXV). pound (XXXIV) is formed : We assume that intermediately a bridged com-(XXXV. ) The difference between this case and that with which we are con-cerned lies in the fact that the hydroxy-compound (XII) owing presumably to the presence of the ketone and carboxyl groups, undergoes dehydration spontaneously or a t any rate under very mild conditions. For the same reason the reaction stops a t the first stage the presence of the carboxyl group makes the analogous isomerisation impossible.Although no reaction resembling this isomeric change has been observed amongst the derivatives of gem-dimethyldicydopentarie there has been noticed (Ingold and Thorpe Zoc. cit.) in the cyclohexanespirodicyclopentane series a hydrolytic decomposition showing a considerable degree of similarity. The connexion is best exhibited by means of the forniulz expressed below and from these it will be seen that both stages of the above scheme for the dehydration of dimethylcyclo-hexanol have their counterpart. The cyclohexylcyclobutanolone acid (XXXVII) is obviously incapable of eliminating the elements of water and passing into a cyclobutene derivative as the strict analogy requires. TXII.) (VII.) C( CO,H) Q H CO,H+ C,H,,, CH*F(CO,H)--?H, C”H’~:C<k(CO,H)*CO C(CO,H)(OH)*CO (XXXVI.) (XXXVII.) The above considerations have an obvious implication regarding the internal condensation of other diketones and i t is hoped in the future to devote attention to this and similar questions. EXPERIMENTAL. Oxidation of Dimeti~yZdicyclope1Ltanonedicar~oxylic Acid (VI p. 1363) by Cold Alkaline Permanganate Formation of aa-Dihydroxy-&3-dime t hylglutaric Acid and Oxalic Acid. Two grams of the dicyclopentanone acid were dissolved in a small excess of aqueous potassium carbonate and carefully titrah 1372 THE CHEMISTRY OF POLYCYCLIC STRUCTURES ETC. in the cold with a 3 per cent. solution of potassium permanganate. When Chree atoms of oxygen had been taken up the perman-ganate ceased to be decolorised.The solution was then treated with a current of steam and filtered the precipitate of manganese dioxide being extracted with boiling water. The combined filtrates were acidified with hydrochloric acid boiled for a few minutes, and then rendered alkaline with ammonia. On adding calcium chloride to this solution there was obtained a precipitate of calcium oxalate which was collected and washed with dilute acetic acid in order to remove any traces of admixed calcium carbonate. The filtrate from the calcium. oxalate was evaporated to a small bulk, acidified with hydrochloric acid and exhaustively extracted with pure ether. On drying and evaporating the extract there was obtained a nearly quantitative yield of aa-dihydroxy-PP-dimethyl-glutaric acid.The acid prepared in this way melted a t 83-84O, and after recrystallisation from chloroform at 84O (Found : C-44.20; H=6*37. Further proof of the identity of the acid was obtained by means of a direct comparison with a specimen prepared from the hydrogen ester of aa-dibromo-PP-dimethylglutaric acid (Perkin and Thorpe, Zoc. cit. p. 757). Both preparations as well as a mixture of the two melted a t 84O. aa-Dihydroxy-PP-dimethylglutaric acid may also be obtained from the dicyclopentanone acid by oxidation with sodium manganate. Calc. C=43*7; H = 6 - 3 per cent.). Oxidation of Dimethyldicyclopentaizonedicarbozylic Acid (VI p. 1363) 7jy Cold Ferricyanide Formation of trans-Caroimc Acid. Thirty-three grams of potassium ferricyanide and 8 grams of potassium carbonate were dissolved in 140 C.C.of water and one-fifth of this solution was added daily to 2 grams of the dicyclo-pentanone acid dissolved in a slight excess of aqueous potassium carbonate. After the addition of the reagent was finished the mixture was allowed to remain for another period of five days and was then acidified and extracted repeatedly with pure ether. On drying and evaporating the ether there remained a crystalline residue which after washing with chloroform and recrystallising from water melted at 213O [Found C=53*13; H=6-41. Calc. : C=53.1; H=6*4 per cent. 0.0435 required 40.7 C.C. of 0-0135N-Br(OH),. C,H,(CO,H) requires 40.8 c.c.]. There can be no doubt that this substance is trans-caronic acid. It was identified with a known specimen of this acid by direc A NEW TYPE OF COMPOUND CONTAINING ARSENIC.1373 comparison and by the method of mixed melting point. I n addi-tion i t was converted by nieans of hydrobromic acid and by hydro-chloric acid into terebic acid which was similarq identified by comparison with a known specimen and by a mixed melting-point determination. Before we were aware that the product of this oxidation was trans-caronic acid we recrystallised the crude residue from the ether from concentrated hydrochloric acid. This treat-ment caused a considerable degree of conversion into terebic acid. Consequently after several recrystallisations we obtained pure terebic acid apparently as the sole crystalline oxidation product. The extraordinary ease with which this reaction takes place does not appear t o have been noticed previously (compare Beasley, Ingold and Thorpe T. 1915 107 1080). A search for cis-caronic acid amongst the residues of the oxidation proved fruitless. trans-Caronic acid may also be obtained by oxidising the original acid with cold alkaline hydrogen peroxide. We have to thank the Chemical Society for a Research Grant which has defrayed a considerable portion of the cost of this investigation. THE IMTERIAL COLLEGE OF SCIENCE AND TECHNOLOGY, SOUTH KENSINGTON. [Received October 6th 1920.