320 INaOLD AND THORPE THE FORMATION AND XXXIII.-lhe Formation and Stability of spiro-Compounds. Part II. Bridged- spiro-compounds Derived from cycloHexane, By CHRISTOPHER KELK INGOLD and JOCELYN FIELD THORPE. INTRODUCTION. (A> General. IN Part I. of this series (Beesley Ingold and Thorpe T. 1915, 107 1080) a comparison was drawn between the conditions of formation and the stability of the bromo-esters of PB-dimethylglu-taric acid and cyclohexane-1 1-diacetic acid. These two acids con-tain respectively the structures I and 11 and the object with which (1.1 (II.) these comparisons were made may be indicated by reference to Figs. (i) and (ii) which are drawn to correspond with structures I and 11. The question discussed was whether the forcing apart of the valpncies (a) and ( b ) of a carbon atom from an inclination of 2 tan-12/?2 to one of $?r radians which according to Baeyer’s strain theory necessarily accompanies the production of the cyclohexane ring would cause the other two valencies (c) and (d) tol approach one another.They might i t was suggested approach one another in such a way as to divide the remaining space into three equal angles as in Fig. (iiA); or alternatively they might be quite unaffected by the straining of (a) and ( b ) and remain as Fig. (ii B) shows a t the normal angle of 2 tan-lJ% It was pointed out that if the first of these hypotheses that represented in Fig. (ii A) were correct and side-chains attached t o (c) and (d) were in closer proximity when (a) and ( 6 ) were bound in a cyclohexane ring than is the case in Fig.(i) where (a) and ( b ) are free then the elimination of groups or elements such as for example hydrogen and bromine as hydrogen bromide from the side-chains of substitution products of cyclohexanediacetic acid should proceed with greater readiness than when corresponding derivatives of dimethylglutarie acid are employed. Further the ring compounds formed its a result of the elimination might be expected t o possess a greater general stability in the former case than in the1 latter. If on the other hand the second alternative STABILITY OF SPIRO-COMPOUNDS. PART 11. 321 for which Fig. (ii B) is drawn is the correct one there should be no difference in the two cases. Experimental evidence was as a matter of fact clearly in favour of the first hypothesis.Thus for example trains-cyclohexanes~n'ro-cychpropane-1 2-dicarboxylic acid which contains the carbon -2p-Rj.(i) 4 skeleton IV was found to possess a distinctly greater stability than trans-caronic acid represented by the outline 111 : CH2*CH 71) CH2<CH,*cH,>:<c~n, (111.) W.1 The present paper extends the comparison t o substances of a more complex type which contain a cycbbutane ring joined rather to th 322 INGOLD AND THORPE THE FORMATION AIUD cyclopropane ring by two carbon atoms common toq both. The struc-tures of these compounds are indicated by the outlines V and VI : (1) (1) CH C - p CH,=Cf12>C<F-$h CH2<CB2*CH2 ( 6 ) C-C(3) (4) CH 3>e<b-c(3) (v.)'4' w.1 It is apparent that the presence of the cyclobutane ring in struc-tures V and VI will make a considerable difference in the kind of effect that one might expect to' observe.This can best be under-stood by representing the strained valencies according t o a graphical niethdd which depends for its rational basis on the following con-siderations. I n Fig. (iii) which is drawn t o correspond with formula 111 the axis. - .. . . carbon atoms of the c?yclopropane ring are represented as points at the corners of an equilateral triangle. Now in the case of such a compound as ethane it can scarcely be doubted that the resultant o f the attracting forces which bind the two carbon atoms together, that is to say the vslency is directed along the straight line joining the centres of the carbon atoms.With cyclopropane however the case is different. For here according t o the strain theory we have each one of the carbon atoms of the ring reacting on those of its valencies which participate in the ring in such a way as to t e n d to make them emanate from the atom in directions inclined to one another a t an angle of 2 tan-ld/2 I f the carbon atoms were entirely successful in bringing about this result then since tw STABILITY OE SPXRO-CO~~PO~IXS. PART n 323 valencies emanating from the same carbon atom a t an angle of 2 t a n - l d y h a v e to reach the two remaining corners of the equi-lateral triangle it is clear that the ouly way in which they could do so is by describing curved paths. The simplest curve which in these circumstances a valency could deecribe would be the arc of a circle each of the terminal tangents of which makes an angle of tan-1JZ with the median through the corresponding corner of the equilateral triangle.This path is drawn in the case 01 the valency ( 3 l ) in Fig. (iii). It represents a state in which the carbon atoms would be entirely free from distortion of the kind we are consider-ing the whole bleing borne by the valencies; that is t o say the physi-cal stress would reside solely on the inter-atomic medium in which the forces of valency are propagated. The valencies will 'however, on their part tend t o reduce their potential energy by shortening their paths and if t l ~ e y completely succeeded would stretch them-selves along the straight lines joining the carbon atoms.This would throw the whole of the strain back on the latter sinoe now the initial directions in which the ring-valencies leave a carbon atom are inclined not a t 2 t a n - I J z but at ~ / 3 . It is probable, therefore that an equilibrium of stresses will be set up and that the valencies will occupy paths which li0 between the straight lines and the limiting ascs the arrangement being of such a kind as t o cause part of the strain to' be taken up by the distorted carbon atoms and part by the bent valencies. Experimental support can he claimed for this view. For in Part I. there were recorded experiments which yidded decided evi-dence that part a t least of this kind of strain is actually taken up by the carbon atoms. In particular it was shown that when two valencies of a carbon atom are bound in a cyclohexanel ring groups attached to the other two valencies apparently take up an altered relative position.It is obvious that this could never happen if the strains existing in the cyclohexane ring resided solely on the valen-cies participating in the ring none a t all being borne by the carbon atom wliicli carries the side-chains. Similarly strains existing in one ring of a sp'ro-compound could not possibly make themselves felt in the second ring unless communicated by a spirane carbon atom which itself is in a state of strain. According to the expri-nients described in Part I. such communication of strain a c r w the spirane carbon atom appears actually t o take place and one can only conclude therefrom that the spirane atom itself is by no means in an unstrained condition.If it be true that part; a t least of the strain of the cyclohexane ring is actually taken u p by the carbon atoms then regarding the matter from the graphical point of view we may say that th 324 INGOLD AND THORPE THE FORMATION AND valencies of the cyclohexane ring and presumably of any other alicyclic ring occupy paths which are distinctly flatter than the limiting arcs the terminal tangents of which meet a t the normal angle 2 tan-1J2. On%he other hand it will be shown in the present paper that it is more difficult to close the cyclobutane ring in the formation oE certain bridged-spko-compounds of the type VI (p. 322) than it is in the formation of corresponding bridged-ring compounds of the type V (p.322). Further the former when obtained are less stable than the latter in certain specific positions entirely in accord with the views here put forward. I f the stresses existing in the cyclohexane ring can be communicated to the cyclobutane ring or to side-chains attached to the cyclopropane ring before the cyclo-butane has been closed then it follows that the valencies of the cyclopropane ring as well as the spirane carboneatom must take part in the transmission of stress. With entirely unstrained valen-cies this is inconceivable and one is therefore forced to the conclu-sion that the valencies themselves are strained. The graphical aspect of this is that the valencies concerned and therefore probably all ring-bound valencies must occupy paths which are not straight lines which are in fact distinctly curved.The curvature of such a path will as we have already seen be less than that of the limiting arc the terminal tangents of which meet a t 2 t a n - l d z and the path may therefore be regarded as lying somewhere between a straight line on the one hand and a limiting curve on the other. Exactly where the position of equilibrium of a particular valency lies and what the precise shape of its true path is it is impossible a t present to say; but the view that the true equilibrium-path lies somewhere between the rectilinear path which represents the limit-ing case in which the valency is unstrained and the curved path which stands for the other limit. in which the whole of the distortion resides on the valency and none1 a t all on the carbon atom affords a simple and so far as it goes a fairly adequate hypothesis in regard to the facts observed.I f we accept the point of view that the more strained a valency is the more prone it is t o break then among the factors which determine the equilibrium of strain-distri-bution will certainly be found the number of groups attached to tho carbon atoms concerned; for the tendency which ring compounds have to undergo fission between quaternary carbon atoms is well known. The electrochemical character of the substituents will of course; be another determining factor. Thus it would seem that there are three causes affecting th STABILITY OF SPIRO COMPOUNDS. P*AIlT 11. 325 stability of any valency which participates in a fully reduced alicyclic structure : (1) The number of carbon atoms in the ring, (2) The character and mode of attachment of any attached rings, and (3) The number distribution and character of the substitucnt groups.I n regard to the first of these causes Baeyer’s simple conception has had a very great degree of success in explaining the broad facts; so much so that one cannot but accept it as an approximation to the truth. I n its elemental form however it does not consider the question as t o whether the atoms or the valencies are the seat of the postulated strain and consequently is not in a position to take into account the mode of operation of causes (2) and (3) which present us with two unsolved problems. The series of which this paper is Part 11.is an attack on the former of these. I n regard to the latter all that can at the moment be said is that the present considerations lead to the view that whilst causes (1) and (2) deter-mine the maximal strain o r curvature which any particular valency can be called on to bear cause (3) operates in such a way as to settle exactly what’ fraction of that maximum it shall bear. The uncertainty which surrounds the operation of cause (3) con-stitutes a difficulty which one meets with in planning experiments with a view to study the mode of action of cause (2). If however, one is careful only t o compare substances in which similar and similarly situated substituent groups are present the strain on a valency may fairly be taken t o be measured by the1 greatest strain which thatvalency could be called on t o bear in the limiting equili-brium.It will actually of course be just a fraction’of this but if the groups are alike i t will be the same fraction for all the substances. I n such cases therefore the strain existing on a valency is from the graphical point of view measured by the degree of curvature of the limiting curved path. Fig. (iii) p. 322 is an application of this method to the cydo-propane ring structure I11 (p. 321) and has already been men-tioned. The limiting curve for the valency ( 3 l ) is drawn. Its terminal tangents make angles of tan-1 d3 with the corresponding medians and its curvature taking the side of the triangle as unity, is d 5 - i J T . Applying the same method to the c?/cZohsxane-s~.rocycZopropane structure IV (p.321) we obtain Fig. (iv) (p. 322) in which for reasons referred to a t the beginning of this paper and given in detail in Part I. the terminal tangents of the limiting curve of the VOL. cxv. 326 INCOLD AND THORPE THE E'ORMATION AND valency ( 3 l ) are drawn t o make angles with the corresponding medians less than the normal value t a n - l f i by say I. radians. The curvature of this curve is less than that of the curve in Fig. (iii) by about -$(7 - 2 dz)p a figure which may be taken as a measure of the increase in stability of the bond ( 3 l ) which was proved experimentally in Part I. The structure V (p. 322) is interesting as exhibiting a difference in exactly the opposite sense. The fact that two of the valencies of the carbon atom (1) are held in a cyclobutane ring will according to the views put forward in Part I.tend to make the initial directions in which the other two valencies leave the carbon atom spread out in such a way as to foran angles with one anotherandwiththecyclo-butane-ring-bound valencies greater than would be the case if the cyclobutane ring 'had not been closed. This will involve) an increase in the initial divergence of the valencies (1 :4) and (1 5 ) . The limiting arc between the carbon atoms (5) and (1) is therefore drawn (Fig. v p. 322) to suit terminal tangents inclined to the corresponding medians a t angles which are greater than t a n - l d 2 by,say p radians. The curvature is therefore greater than J2-gJ3by about i ( 7 - 2 d q q . The point of present interest is of course to see how a bridged-ring compound of the type V compares with a bridged-spko-com-pound of the type VI (p.322). It will be observed that in the latter case the limiting curve for the vnlency (5 1) cannot possibly be symmetrical. For whilst the carbon atom (5) is on account of the cyclohexane ring tending to reduce the angle which the ter-minal direction of the curve makes with the median $0 the value tan-1dZ-p radians the carbon atom (l) as a result of its being involved in the cyclobutane ring is endeavouring to enhance the corresponding angle a t that end of the curve to t a n - l J 2 f q radians. The true curve therefore will lie between the dotted arcs of curvatures J2-4d73-p/6(7-2d6) and approximating to the inner arc near the carbon atom (5) and to the outer near the carbon atom (1).Such a path it is clear from Fig. (vi) (p. 322) must contain a region of comparatively great curvature and we may therefore expect the bond (5 1) in bridged-spire-compounds of the type V I (p. 322) t o be noticeably less stable than the same bond in corresponding compounds of the bridged-ring series V. The effect just noticed will of course be by far the greatest which the cyclohexane ring could be expected to have on the attached dicyclic system. The reactions on the bonds (1 2) and (1 4) will 0- * J 3 + Pl6(7 - 2 d 6) STABILlTY OF SPIRO-COMPOUNDS. PART II. 323 be next in importance but of the second order. It is however of interest briefly to examine them. The fact that the limiting curve of the valency (5 1) is on the whole depressed below the arc which corresponds with the limiting curve of the same valency in Fig.(v) necessitates that the true path of the same valency in Fig. (vi) is, even near the carbon atom (l) rather less divergent from the recti-linear path than is the case in Fig. (v). This will involve a small effect on the valency (1 :3) in the direction of an increase in the curvature of its limiting curve. The bond (1 :2) should therefore be slightly mcre strained and less stable in the series V I than in the series V. The effect of the strains in the cyclo’hexane ring on the bridge-bond (1 4) is also of the second order the depression of the bond (5 1) being the determining factor. I n this case the fact that the bonds (1 5) (1 4) and (1 2) have their curves in different planes makes the geometricai construction more complicated.It would appear however that the effect should be in the direction of a decrease of stability although of course a very slight one. There remains for consideration the bond (2 3). The effect in this case will be still smaller of the third order in fact and prob-ably beyond the limits of detection. It should be in the direction of an increase in stability. It may be stated here that the results of the experiments described in this paper are in the most complete accord with all the above conclusions. The must striking fact which emerged during the experimental study was the marked decrease in the stability of the bond (5 l) of compounds of the type VI when compared with corresponding sub-stances of the type V (see sections C D and G).The fission of this bond was brought about with great ease in certain substances of the former type by alkaline reagents which appeared to be without effect on the latter. This is in agreement with the first of the main conclusions reached as a result of the conception of flexible valencies. The agreement is particularly interesting in this case, for whereas in the comparison of the cychprupane derivatives of types I11 and I V (p. 321) carried out in P a r t I. it wits found in agreement with the requirements of theory that the cyclohexane ring had a stabilising influence in the bridged series V and VI the hypothesis leads us t o anticipate exactly the reverse. Definite experimental evidence (section B) was also obtained on the effect on the bond (1 2) of the presence of the cyclohexane ring in the spiro-compounds.The expected effect is in the nature of an increase in strain that is a decrease in stability and is small in magnitude. Actually we were not able to find a reagent whicb 0 32s INGOLD AND THORPE THE FORMATION AND woulcl break this bond in a compound of the type V I (p. 322) and would not also break the same bond in one of the type V. This was a t any rate in part due to the fact that i t did not appear to be possible t o break the bond (1:2) in the series V I without first breaking the less stable bond (5 1). This so far as it goes, is valuable as i t indicates that the effect of the cyclohexane ring on the bond (1 2) is of a smaller order than the effect on the bond (5 1).Fortunately there is however a more precise and more delicate method of experiment. The cyclohexaneup*rodicyclopen-tane structure V I was in our exEeriments obtained from cyclo-hexane-1 1-diacetic acid by a method which involved first the clos-ing of the cyclopropane ring and then of the cyclobutane ring by the establishment of the bond (3:4). Now the1 additional in-stability which we have been led to expect the bond (1 2) in the spiro-structure V I to exhibit is due t o the tendency which there is owing to the cyclohexane ring f o r the arrgle between the bonds (1 :4) and (1 :5) to increase. It is obvious that this tendency will have greater effect if the bond (3 4) has not been established than would be the case if i t had for in the former case the tendency will n o t be resisted by the strained bonds of the cyclobutane ring.The result. will be to magnify the' effect by drawing further apart the carbon atoms (3) and (4) t o an extent which might quite well be sufficiently great to make i t appreciably more difficult to close the cyclobutane ring in the series V I than in the series V. A set of comparative experiments made in order t o discover whether sucli an effect could be detected revealed its existence very clearly. The interest of these experiments lies in the fact that they show that the stresses in the cyclohexane ring have been communicated to the bond (1 2) a phenomenon which would seem necessarily to involve the assumption that the bond (5:1) as well as the carbon atoms (5) and ( l ) are in a state of strain.With regard to the bond (1 4) in the structures V and VI (p. 322) there is available quite a good method of experiment since sodium amalgam readily breaks this bond in certain substances belonging to these types without attacking in the slightest degree any of the other bonds in the molecules (see section E). The reduc-tions occupy several hours and can readily be conducted a t con-stant temperature and under comparable conditions. By making comparative1 experiments along these lines it was discovered that, although the reactions proceeded in much the same way in the two cases there was a considerable difference in the reaction velocities, the reductions proceeding more rapidly in the bridged-spiro-series VI than in the bridged-ring-seriag V.This is in agreement with the conclusions already reached on theoretical grounds (p. 327) but it STABILITY OF SYIRO-COMPOUNDS. P,4RT 11. 329 llas also some interest from another point of view. It might be thought possible that valencies are so extremely elastic as to be capable of stretching as we11 as of bending. I f that be so the valency (1 :4) ought t o be less stretched out in the structure VI in which the cyclohexane ring is tending t o reduce the angle a t which the bonds (5 1) and (5 4) emerge from the spirane carbon atom (5) than in the structure V which contains no cyclohexane ring. Any such difference in the degree of elongation of the bond (1 4) ought 50 manifest itself as a marked difference of stability since the forces between two carbon atoms must vary as some power, probably a high power of the distance between them.As a matter of fact the difference of stability which the reduction experiments reveal is in precisely the opposite sense from that which this view requires. This appears to us t o show that no appreciable lengthen-ing occurs and that probably all carbon-to-carbon valencies have the same fixed length. It may also be noticed that the effect observed is the opposite t o what would be expected if the bonds occupied inflexible skraight lilies. For in this case the tendency which the cyclohexane ring has t o reduce the angle between the bonds (5 1) and (5 4) (compare Part I.) should operate in such a way as t o increase the stability of the bond (1 4).These facts, therefore constitute a strong argument in support of the conception of flexible valencies. There remains to be mentioned finally a set of experiments which were undertaken in order to obtain information regarding the bond (2 3). This bond according to the conclusions expressed on p. 327, should be practically unaffected by the strains in the cyclohexane ring. Actually we were n o t successful in finding a reagent which would attack this bond in a compound of the type VI even when the stability of the bond was reduced by the introduction of an alkyl group a t the carbon atom (a) without first attacking one of the cyclopropane bonds adjacent to the1 spirane carbon atom. However the experiments were interesting as confirming our pre-vious experiences regarding the instability of these bonds and were in harmony with the conclusions already reached in regard t o the insignificance of the effect which the cycluhexane should have on the bond (2 :3).I n order t o avoid confusion in what follows it ought perhaps t o be pointed out that the bonds (5 1) and (5 4) are equivalent in the carbon skeletons V and VI (p. 322) and only become dissimilar when substituent groups are introduced unsymmetrically. This is actually the case with the corripounds with which we have experi-mented and i t so happens that the bond split in all the substances which were subjected t o fission by alkalis is according t o the metlior 330 INGOLD AND THORPE THE FORMATION AND of numbering employed throughout the bond (5 4) not the bond (5 1).This it is clear invalidates nothing and in fact the agree-ment would be formal as well as actual if in the formulae which follow we had numbered the dicyclopentane ring the other way round. The objection t o doing this is that one would then meet with the same lack of formal consistency in regard to the bonds (1 2) and (3 4). (B) Formation of the Bridgehhq Systems V and V I (p. 322): Closing of the cycloBzctane Bond (3 4). Some time ago in a paper published by one of us in conjunction Gyitli W. H. Perkin (T. 1901 79 729) it was shown that when the dibromo-ester of PB-dimethylglutaric acid (VII) was treated with ethyl malonate and sodium ethoxide condensation took place in two stages the ultimate product being an insoluble yellow sodium deriv-ative of ethyl dimethyldicyclopntanonetricarboxylate (IX).It was (VII.) (VII1.)-(IX. 1 therefore to be expected that if cyclohexane-1 1-diacetic acid were used in place of dimethylglutaric acid a similar condensation would ensue1 : (XI. 1 (2) crr,<C H,* C Ha>(1! c( CO E t) $! C( ONa) 0 E t CH,*CH k<b( CO$t)*cOn (4) (XII.) The sodium derivatives IX and XI1 clearly belong respectively to the types V and VI (p. 322) and it was therefore decided to 9ake their formation and decompositions sub jechs of comparison. It was found possible t o prepare the sodium spirmompound (XII) direct from ethyl dibromocydohexanediacetake (X) and ethy STABILITY OF SPIRO-COMPOUNDS. PART 11. 331 sodiomalonate under the same conditions as are employed in pre-paring the sodium ring compound (IX) from ethyl dibromodi-methylglutarate (VII) .Direct comparison of the two condensations is however complicated by the fact that whilst ethyl dibromo-dimethylglutarate (VII) can readily be obtained in a state of purity by distillation the corresponding cyclohexane derivative (X) decom-poses when distilled. We were therefore forced to use a crude bromination product containing only 76-80 per cent. of the dibromo-ester- Nevertheless the difference between the two cases is significant. Two experiments one with the bromo-ester of dimethylglutaric acid and the other with the bromo-ester of cyclo-liexanediacetic acid carried out under comparable conditions (except that sufficient excess of the crude cyclohexane ester to allow for the impurities was used) gave in the first case a 62 per cent.yield of the sodium compound (IX p. 330) and in the second a 13 per cent. yield of the sodium sera-compound (XII p. 330). I n order to make the comparison more definite the tetraethyl ester XI (p. 330) was prepared in a state of purity. The internal condensations of the esters V I I I and XI (p. 330) with alco-holic sodium ethoxide whereby alcohol is eliminated and the bridged structure formed were then carried out in a series of experiments made under comparable conditions. Several pairs of parallel experiments were1 made in which the two esters were boiled with alcoholic sodium ethoxids under standard conditions for dif-ferent lengths of time a t the end of which the insoluble sodium compounds’were collected and weighed.The results are given in the experimental part of this paper (p. 359). The figures for the yields in the two cases lie upon fairly smooth but widely divergent curvea (Fig. vii). These curves clearly show that the velocity of formation of the bridged-ring structure is much greater than the velocity of formation of the sp.ro-compound. The times required for a 20 per cent. conversion are in the ratio of approximately 10 ik~ 1. An examination of the curves shows that if x is the yield (per cent.) and t the time the quantity - a/ax(dx/dt) is almost invariable with time and has a definite positive value for each curve, being about 0.97-1 hours for the! bridged-ring compound and a b u t 0‘33-1 hour for the1 bridged-spire-derivative.The velocity constants for the’formation of the sodium compounds are therefore approxi-mately in the ratio of 3 t o 1. The best yields obtainable in the two cases are 77 per cent. and 38 per cent. respectively but a longer time is necesqary to produce1 a 38 per cent. yield of the spiro-com-pound than is required to obtain a 77 per cent. yield of the ring compound. The fact appears to be that the spko-ester XI (p..330) will only condense with itself when present in the sodium ethoxid 332 INGOLD AND THORPE THE PORMATIOX AND in considerable concentration. For whereas under the experimental conditions which give the best yield (77 per cent.) of the sodio-ester, IX the tetraethyl e,ster VIII (p. 330) almost entirely disappears from the reaction mixt,ure when the ester X I is treated so as t o produce a maximal yield of the sodio-spiro-ester (38 per cent.) then, 100 90 80 70 60 u c: 2 50 t 40 30 20 10 0 0 2 4 6 S 10 12 hours although longer time is allowed in this case the conversioii is found t o be only partial about 15 per cent.of the tetraethyl ester being recoverable. Yet prolonging t.he time of resctioii does not appear appreciably to increase the yield of sodium compound or reduce the quantity of tetraethyl ester recovered. These experiments clearly show that i t is more difficult t o estab STABILITY OF SPIRO-COMPOUNDS. PART 11. 333 lisli the bond (3 4) o l the cyclobutane ring to form tlie spire-conl-pound XI1 than it is in tlie formation of the ring compound I X .Indeed the low yield of the former compound was one of the greatest practical difficulties of tlie research. (C) Hydrolytic i7econiliositioii.r of tJie Bridyerl-Riiig l l m i z w t i c e s I S (znr? X I 1 (p. 330) Strtbility of tJhe cyclol'ropane Bond' (4 5). The first evidence of a difference of stability between compounds o l the bridged-spiro-series V I and those of the bridged-ring series V (p. 322) was obtained in the coarse1 of some experiments on the hydrolysis of the sodium spiro-compound XI1 (p. 330). The efect of hydrolysing agents on the sodium compound IX has already been described in the paper with W. H. Perkin (Zoc. cit.), and the general coiiclusicii reached was that the end-product of the action of alcoholic potassium hydroxide was a mixture of the dibasic and monobasic acids XI11 and XIV and that the same two acids (XIII.) (XIV.) were formed by the action of dilute sulphuric acid.The dibasic acid XI11 when heated with water in a sealed tube was found t o yield the monobasic acid XIV. On experimenting along these lines with the yellow sodium spiro-compound XI1 (p. 330) it soon became. apparent that the. spiro-conipounds were behaving very differently ; the hydrolytic products were therefore investigated in some detail. When the yellow sodium splil-o-compound XI1 was treated with cold dilute acid a colourless solid ester C19H2607 (XV) was pro-C(C0,Et) * yIi*C'O,Et C 5 H 1 0 C < & ( p ~ 2 ~ q . co (XV.) duced. This ester could be transf orined by cold alcoholic potassium hydroxide into the yellow potassium salt XVI and thence by acids into the colourless acid-ester XVII.C( CO,H)-yH* CO,Et (CO,K)-$X C(OK)* QEt ".Hio:4oo,Et) co c5a10:C<($C0,Et). co (XVI.) (XVII.) Regarding tile question as t o which of the three carbethoxyl groups has been attacked by the reagent i t is clear t h a t it cannot be that attsched t o the carbon atom (2); for if it were the forma-O 334 INGOLD AND TRORPE THE FORMATION AND tion of a yellow insoluble ptassio-compound so very similar to the original sodium compound would be exceedingly improbable. The fact that it is the 1-cerbethoxyl group and not the 4carbethoxy1, which has been attacked follows from the production by further hydrolysis of the acid-esters XVlII and XXI about the constitu-tion of which no doubk exists.The first product of the action of boiling alcoholic potassium fiydroxide on either of the compounds XVI or XVII was a colour-less very readily soluble potassium salt of tho ethyl dihydrogen ester XVIII : (XVIII. ) The constitution of the acid-ester XVIII follows from the fact that it is obviously formed by the hydrolysis of the metal-substi-tuted carbethoxyl group in the yellow potassium compound of the diethyl hydrogen ester. The other possible formula (XIX) of the diethyl hydrogen ester would give an ethyl dihydrogen ester of the constitution XX. The formula XX was however easily shown to (CO,Et)*C K* C0,Et C(CO,Et)* CH*CO,H (XIX.) (XX). CbH J 4 ( C * f l )-&I ~~5~~o:c<~(co2H)-Lo 1 be incorrect by an experiment on the effect of treating the substance with acetyl chloride.With this reagent it readily yielded an anhy-dride which on treatment with water furnished the original acih-ester. These facts clearly favour formula XVIII in which the free carboxyl groups are attached to contiguous carbon atoms and rules out the alternative formula XX. We must also regard XVII as the true formula of the diethyl hydrogen ester since a substance of the formula XIX could not possibly yield the ethyl dihydrogen ester XVIII. The acid-ester XVIII can also be produced from the triethyl ester XV (p. 333) by hydrolysis with hydrochloric acid. Indeed, up to this point hydrolysis with alcoholic potassium hydroxide and with hydrochloric acid proceeds along the same lines. The first product of the action of boiling hydrochloric acid on the ester XV is the diethyl hydrogen ester XVII (p.333) which by continued action of the same reagent is converted into the ethyl dihydrogen ester XVIII (above). The further action of hydrochloric acid causes the substance to decompose simultaneously in two ways losing in the one case a carboxyl group and in the other a carbethoxyl group. The groups eliminated will of course be those in the positions (2) and (4) con STABILITY OF SPIRO-COMPOUNDS. PART 11. 335 tiguous to the carhnyl group and the products must therefore possess the constitutions XXI and XXII: (CO,H)* H*CO,H C H lo :Q<~(co~L1)-~!? C(CO,Et)* 0 C,H,o C<ZH-& (XXI.) (XXII.) These formulze are in complete accord with the properties of the substances.The acid-ester XXI folr example melted without decomposition and gave no coloration with ferric chloride showing that the free carboxyl group is not adjacent tu the carbonyl group, and that the carboxyl group which was attached to the carbon atom (2) in the acid-ester XVIII has been removed. The dibasic acid XXII on the other hand melted with decomposition gave a crim-son colour with ferric chloride and when treated with acetyl chloride gave an anhydride from which the original acid could be regenerated by treating with water. These facts clearly establish formula XXII. The formation side by side of the acids XXI and XXII does not constitute quite the final stage of the hydrolysis by hydrochloric acid of the ethyl dihydrogen ester XVIII (p.334). For the acid-ester XXI on prolonged boiling with hydrochloric acid loses its 4-carbethoxyl group yielding the monocarboxylic acid XXIII : (CO,H)*CH, C5Hl*:C<~H-bo (XXIII.) The dibasic acid XXII on the other hand was not changed appre-ciably even on boiling for eighty-seven hours with hydrochloric acid. The final product of the action of this reagent on the acid-ester XVIII was therefore a mixture of the dihasic acid and the mono-basic acid (XXII and XXIII). Further the dibasic acid XXII was readily converted into the monobasic acid XXIII by heating for a few minutes a t 200° with water. All the acid-esters (XVII XVIII and XXI) of the series reacted a t this temperature with water giving the same monobasic acid usually in very good yield. The neutral triethyl ester XV (p.333) however required the presence of a trace of an acid such as acetic or hydrochloric acid in the'water. The presence of a small quantity of butyric acid even was found to be quite sufficient so that in the cases of the acid-esters the reaction is in all probability autocatalytic the catalyst being the hydrogen ions produced initially by the electrolytic dissociation of the acid-esters themselves and in the later stages of the reaction by the dissociation of the mono-carboxylic acid XXIII or of carbonic acid. In t,he case of the o+ 336 INGOLD AND THORPE THE FORMATIOW AND triethyl ester i t is necessary artificially to introduce some hydrogen ions in order t o start; the reaction. Alkaline hydrolysis of the acid-ester XVIII (p.334) proceelded in quite a different direction the product being an acid of the formula C,,H,,O,. This substance is a dibasic acid. It is not an aldehyde, and contains no lactone ring. It is therefore a fission product formed by the breaking of one of the bonds (1 :a) (1 :4) (1 :5) or (4 5) of the dicyclic system : One of the three carboxyl groups originally present in the molecule has been lost and since the substance gives no colour reaction with ferric chloride i t is to be presumed that i t is tlie carboxyl group attached to the carbon atom (2) which has disappeared. The ques-tion as to which of the four possible bonds has been btroken is settled very clearly by the properties of the substance. Thus fission of the cyclobutane ring a t the Eond (I 2) should yield either a derivative containing an open-chain acetoacetic acid residue or a hydroxy-com-pound capable of forming a y-lactone according to the way in which the elements of water are added t o the molecule a t the point of fission.The substance actually obtained was found t o be remark-ably stable towards boiling alkalis and showed no tendency to pass i n b a lactone. On the other hand although a tram-acid i t forms an anhydride with the greatest ease. For this reason i t may be safely assumed that the 1- and the 4-carboxyl groups are attached t o contiguous carbon atoms and t h a t the bridge-bond (1 :4) has remained intact. The bond which has been broken is therefore one of the cyclopropane bonds (I 5) and (4 5) and since the substance is not a y-hydroxy-acid i t must have one of the two following formuke : (XXIV.) (XXV.) Although both these formulae are in harmony with the properties of the substance there can be little doubt that formula XXIV and not formula XXV is correct for this reason The monobasic acid XXIII (p.335) does not undergo fission with alcoholic potassium hydroxide. The fission therefore seems to be connected with tli STABILITY OF SPIRO-COMPOUNDS. PART 11. 337 quaternary carbon atom in the position (4). If this be so it is reasonable to assume that splitting takes place a t a point adjacent to this carbon atom. An indirect. but interesting confirmation of this conclusion will subsequently be referred to (p. 345). The substance to which the formula XXIV has been assigned separates from water with two molecules of water of crystallisation.The anhydrous compound when heated a t 250° was found to pass into an anhydride which gave with water a new dibasic acid also of the composition C,,H,,O,. This did not take up wate? of crystallisation and melted with tho immediate elimination of water-vapour. On boiling with hydrochloric acid i t was quantitatively converted into the isomeride previously mentioned. These relation-ships clearly indicate that geometrical isomerism of the cis-traits-type is here being encountered both the acids C,,HI6O having the structure represented by the f mmula XXIV which clearly requires the existence of this kinu of ibomerism. The relationships between the various substances obtained by the hydrolysis of ethyl cyclohexanesphodicyclopentanonetricarboxylate and of its sodium derivative are collected together for convenience in table I.In order to study more closely the contrast presented by the hydrolytic reactions in the dimethyldicyclopentane and the cyclo-hexanwyirodicyclopentans series a number of direct comparative experiments were made in order to determine the relative speeds a t which the acid-esters of the two types decomposed in the presence of alkali. - The substances chosen were the ethyl dihydrogen ester XXVI," and the analogous substance in the bridged-CH,>,<F(CO,H)-FH*CO,H CH C(CO,EL)*CO (XXVI.) spir-o-series namely the acid-ester XVIII (p. 334). The bridged-ring ester XXVI decomposes under the prolonged action of boiling alcoholic potassium hydroxide yielding the monobasic acid XIV (p.333). The bridged-spiro-ester XVIII on the other hand, undergoes fission with the same reagent giving the cyclohexyl-cyclobutane acid XXIV (p. 336). There is however a remarkable difference in the ease with which the two reactions proceed. Thus in one pair of parallel experi-ments made under comparable conditions the acid-ester XVIII of the bridged-spbo-series gave an 85 per cent. yield of the fission * This substance was not isolated during the earlier research. Its properties and mode of formation are therefore given in a note at the end of the experimental part of this paper TABLE I C0,Et +- 4-cold EICL cold NaOH C,Hlo: CO,Et (Yellow FeCl crimson.) KOfI in I MeOH cold J. CO,K (M. p. 46-47', short C0,Et (Yellow FeCl violet.) (Gum, CO,H H*CO,H > C 5 H o C < ~ ~ o KOH in EtOH hot 60,Et (M.p. 206 dec. FeC1 violet. co*o*co CO,H -4 C~H~,, C H - ~ < ~ ~ ~ ~ > C O ~ ! C ~ H ~ ~ cH;c<;tij>c 0 C,H o c<X-X~ C, iO,Et (tmna M. p. 206' dec.) (cia M. p. 145' dec.) (U. p. 126".) ''zH GO H I C0,H I co*o*co (M. p. 155" ; anilic acid m. p. 202' dec. ; anil m. p. 199".) (M 340 INQOLD AND THORPE THE FORMATION AND product whilst on the contrary the bridged-ring acid-ester XXVI was recovered unchanged to the extent of 93 per cent. only a trace of the1 monocarboxylic acid being isolated. No fission pro-duct of the dimethyldicyclopentane series was isolated in the course of these experiments. These experimental comparisons are interesting as showing the extraordinary facility with which the bond (4 5) in the bridged-spiro-series is broken.They have however interest from another point of view. For if they had not been made i t would have been poasible to advance an explanation of the difference in the ease of fission of the bond (4:5) in the t'wo series based not on the] strain effects of the cyclohexane ring but on steric hindrance caused by the attached groups. It has already been noticed that i t appears t o be necessary to have a quaternary carbon atom in the position (4) in order to bring about the fission of the bond (4 5) by alkalis. One might assert therefore that in the bridged-syirol-series the C,HIo group attached to the carbon atom (5) has the effectl of preventing the elimination of the 4-carbethoxyl group.The carbon atoms (4) and (5) therefore both remain quaternary in the presence of the alkaline reagent and splitting occurs between them. One would have to assume of course that although the steric effect of the C5H,,, group attached to the carbon atom (5) is sufficiently powerful to prevent the atkack of the reagent on the 4-carbethoxyl group yet for some reason unknown i t does not inhibit the attack of the same reagent on the bond (4:5). I n the bridged-ring series on the contrary the (CH,), group attached to the carbon atom (5) might not have any appreciable steric effect. Splitting therefore might not t'ake place in this case owing to the fact that when the acid-ester XXVI (p. 337) is treated with alcoholic potassium hydroxide the 4-carbethoxyl group is so quickly eliminated that the carbon atom (4) becomes tertiary before the reagent has had time to react appreciably on the bond (4 5).We were of the opinion a t one time1 t-hat there might be some truth in this way of explaining the phenbmena and it7 was the desire to test this hypothesis that! furnished our chief motive for undertaking the comparative experiments with the acid-esters XXVI and XVIII. It will be seen however that the experi-ments effectively dispose of this explanation since they show that the 4-carbethoxyl of the acid-ester XXVI is not a t all readily eliminated under the experimental conditions employed. We consider this to be strong evidence that the bond (4:5) in the bridged-spire-series is actually under considerable strain much more so than in the bridgsd-ring series and that the fission reac-tions are not to be accounted for as secondary phenomena due t ST.4BILITY OF SPIRO-COMPOUNDS.PART 11. 341 steric hindrance or ot,her such causes. Several other examples of the splitting of the bond (4 5) in the bridged-sp’ro-series will sub-sequently be given (sectmiom D and G). (D) Hydrolytic Decoinpositioiis of the Methylation Products of the Bridged-ri?iy Derivatives IX and X I 1 (p. 330) Stability of the cycloiPropane Bond ( 4 5 ) and of the cycloBiitatze Botzd (2 3). I n the paper on bridged-ring derivatives to which reference has already been made interesting results were obtained by methyl-ating the yellow sodium compound IX (p.330) and subjecting the methyl derivative to alkaline hydrolysis. It was found that the entrance of the methyl group a t the carbon atom (2) in XXVII created a point of instability between the carbon atoms (2) and (3) and that fission took place with the1 formation apparently of the cyclopropane acid XXVIII which t-hen underwent a second fission giving the dibasic lactonic acid XXIX. C(C0,Et)-CMe*CO,Et CH, ,C( CO,H)*CHMe*CO,H CHs\,C/ I -+ C ’ CH/ \c(co,Etj*ko CH,,’ \~H*CO,H (XXVII.) (XXVIII.) CH C(CO,H).OHMe*CO,H CH,/ \CH,*CO% --3- \(/\-. (XXIX.) The sodium spiro-compound was quite readily methylated by means of methyl iodide. When the ester thus obtained was treated with alcoholic potassium hydroxide the principal product was a dibasic lactonic acid evidently a double-fission product, which however did not appear to be constituted analogously to the lactonic acid XXIX.It would appear that. there are two quite probable ways in which hydrolytic action might proceed. I n the first place the attack on the ester XXX mightl commence a t the bond adjacent to the C(CO,R)-CHMe-CO,H / I C(CO,Et)*CMe*CO,Et C,H,,:C / I I -+ c,Hlo:c -‘C(CO,Et)*ko \~H~CO,H (XXX.) (XXX I .) \CH2*CO* 0 C(CO2H)*CHMe*CO2H 4 C,H,,:C’ ‘‘-\ (XXXII 342 INGOLD AND THORPE THE FORMATION AND methylated residue and give as the first product a spirocycZo* propane acid XXXI which is similar to XXVIII and would ultimately yield a lactonic acid XXXII strictly analogous to XXIX. On the other hand we know that the cyclopropane bond (4 5) is a very vulnerable point in the molecule of the unmethylated ester XV (p.333) and it can scarcely be supposed that the entrance of a methyl group a t the carbon atom (2) would stabilise it to any marked degree. If in spite of the weakening of the bond (2:3) by the methyl group the bond (4:5) still remains the most readily attacked part of the molecule we should expect from the behaviour of the unmethylated ester to obtain a cyclohexyl-cyclobutane acid XXXIII which would then split again this time across the bond beside the methylated residue giving ultimately a dibasic lactonic acid of the formula XXXIV. C0,Et CO,H . CO,Et (XXX.) C0,E (XXXIII. CO,H /C-CH Me CO.0 ,//--3 C,H,,:CH I ____- -CH*CO,H (XXXIV.) In seeking evidence to enable us to decide between the formulae XXXII and XXXIV we made a study of the conditions of anhydride formation of the substance.It will be noticed that both formulse represent substances which as they have carboxyl groups attached to contiguous carbon atoms ought easily to form anhydrides. A substance of the formula XXXII would however, belong to the type of ad-dimethylsuccinic acid which yields two stereoisomeric anhydrides corresponding with the two stereoisomeric acids (Bone and Perkin T. 1896 69 266). The trans-acid on treatment with acetyl chloride gives a trans-anhydride which on distillation passes into the anhydride of the &-acid as shown in Scheme 1. - HzO Heat + H2O trans-Acid Tz trans-anhydride -+ cis-anhydride 7- cis-acid. Scheme 1. The lactonic acid XXIX (p.341) to which XXXII is strictly analogous actually* does form two anhydrides related to one +H@ - Hz STABILITY OF SPIRO-CIOMPOUNDS. PART II. 343 another and to their acids in this way. Consequently a substance of the formula XXXII should do the same. On the other hand in formula XXXIV the bond uniting the two carboxyl-bearing carbon atoms forms part of a ring and hence such a substance would be expected to dehydrate in the manner customary with carboxylated alicyclic compounds the trans-acid, in which the carboxyl groups are on opposite sides of the ring, being incapable of forming its own anhydride but passing on dehydration directly into the anhydride of the &-acid as indi-cated in Scheme 2. - HzO +H20 trans- Acid -+- cis-anhydride zz cis-acid.Scheme 2. - TT29 Experiment showed that the lactonic acid obtained by the hydro-lysis of the ester XXX (p. 342) actually dehydrated in accordance with Scheme 2. The original product of hydrolysis was the tram-acid. It did not eliminate water a t the melting point but was readily dehydrated by acetyl chloride. The anhydride formed was the same substance whether tbe product of dehydration was dis-tilled or not and on treatment with water yielded the cis-acid. This substance melted with the immediate elimination of water vapour and was instantly dehydrated by acetyl chloride. These facts point most distinctly to formula XXXIV (p. 342) as repre-senting the true structure of the stereoisomeric acids. It is obvious however that we have by no means exhausted all the possible formulae in the above considerations since either of the intermediate compounds XXXI and XXXIII (pp.341 342) could split and take up water in a variety of ways. Indeed besides XXXI and XXXIII there are a number of other formulze which the single-fission product might have. When however all these possibilities are examined it would appear that there are only two formulae besides XXXII and XXXIV (pp. 341 342) which fulfil the following conditions relating to the tram-lactonic acid isolated : ( a ) That it is a dibasic lactonic acid of the composition C13H1806, forming in neutral solution a silver salt C13H1606Ag2 and in alkaline solution a barium salt (C,,H,,O,),Ba,. ( b ) That ih free carboxyl groups are attached to contiguous carbon atoms.We regard the behaviour of the substance on dehydration as a proof of this. ( c ) That it is a y-lactane. Experiment showed that there was an exceedingly strong tendency for the lactone ring to be formed, and in spite of many attempts we were unable to prepare the free hydroxy-acid 344 INUOLD AND THORPE THE FORMATION AND The two formulz which along with XXXII and XXXIV fulfil the above condit,ions are XXXV and XXXVI. C,H,, $ -F(CO,H)* CH Me*CO,H C,H,, $l--$!( CO,H)~CH,~ CO,H O*CO*CH O*CO*CHMe (XXXV.) (XXXVI.) Of these the first XXXV possesses an anhydride-forming group exactly similar to that of the lactonic acid XXXII (p. 341). A substance of the formula XXXV ought therefore for reasons given when XXXII was considered to behave like aa'-dimet.hylsucQnic acid and like the acid XXIX (p.341) and form anhydrides in accordance with Scheme 1. The formula XXXVI on the other hand does not contain two asymmetric carbon atoms in its anhydride-forming group and is therefore out of the question. It will be seen that of the four possible formulz XXXIV (p. 342) is the only one which accords with the facts of the case, namely that the dehydration of the substance proceeds according to Scheme 2. It may be added that the cis-lactonic acid is converted by boiling hydrochloric acid into the tram-isomeride thus completing the cycle of transformations which in Scheme 2 is only fragmentary. The trans-lactonic acid XXXIV (p. 342) was not the only pro-duct obtained by the alkaline hydrolysis of the methylated ester XXX (p.342). There was always formed side by side with it a somewhatl smaller quantiky of a dibasic acid of the composition C13H1806. Both the dibasic acid and the lactonic acid appeared to be end-products of the reaction; that is to say they were quite stable towards the reagent used in their preparation. The dibasic acid had properties practically identical with those of the dibasic acid C,,H1606 (XXIV p. 336) which was obtained by the alkaline hydrolysis of the unmethylated ester XV (p. 333). It has there-fore' without. much doubt been formed in a manner precisely analogous to that' in which the acid XXIV was produced and has the structure shown in the formula XXXVII. CO,H C0,H (XXXVII.) Like the parent substance XXIV the homologous acid XXXVII was isolated in cis- and trcrm-forms t,hat originally obtained being the trans-form STABILITY OF SPIRO-COMPOUNDS.PART 11. 345 'l'he formatimi side by side of the acids XXXIV (p. 342) a d XXXVII (p. 344) is readily explained if we accept the view put forward on p. 342 that the cyclopropane bond ( 4 5 ) of the methylated ester XXX is the first point in the molecule attacked by the alkaline reagent. For if this is so t"he substance XXXIII which is first formed may undergo disruption beside the methylated residue in two ways corresponding with the two hydrolytic decom-positions of ethyl acetoacetate. It may either split the cyclo-butane ring between the carbon atoms (2) and (3) to give the lactonic acid XXXIV as shown on p. 342 or itl may split between the methyl-bearing carbon atom and the attached carboxyl group, giving the acid XXXVII.C( CO,H)-$!Me'CP,H C( CO,Et)*C Me* CO,F,t ~ H o ~ < t ) ( C o 2 E t ) . ~ o -+ cAo:CH'' C(C0 H)(OH)*CO (XXX .) (XXXIlI.) /C(CO,H)- -7HMe -+ C,H,,:CH I C( CO,H )(OH) CO (XXXVII.) It is interesting once again tto refer to the parent substance of which the acid XXXVII is the methyl derivative. It was noticed on p. 336 that formulce XXIV and XXV were equally in harmony with t.he properties of the substance but that for reasons there given formula XXIV was t o be preferred. We have just seen that the formula which follows from this for the methylated substance enables us to explain the simultaneous production of this com-pound and of the lact'onic acid XXXIV in a very straightforward manner.The alternative formula XXXVIII for the methylated C0,H C,H,, C H C<<g:A>C H Me (XXXVIII.) dibasic acid strictly analogous to the formula XXV has not this advantage. Such a substance could not be produced side by side with the lactonic acid XXXIV (p. 342) exceptl as a result of the simultaneous occurrence of two totally different sets of reactions ; also the lactonic acid which one might expect to be produced along with a compound of the formula XXXVIII would have properties diffe'rent from those which t>he lactonic acid isolated was found to possess. We therefore think that we were right in selecting formula XXIV rather than XXV. The results of tQhe experiments on the alkaline hydrolysis of the &0$ 346 INQOLD AND THORPE SPIRO-COMPOUNDS.methylation product of the yellow sodium sp*ro-compound are summarised in table 11. These "experiments have an interest inasmuch as they confirm, and even emphasise the remarkable instability of the bond (4 5) in the bridged-spiro-series. For in spite of the fact that t-he entrance of the methyl group a t the carbon atom (2) of the methylated ester XXX creates a point of instability between the carbon atoms (2) and (3) the reagent commences its attack not a t the bond ( 2 3) but a t the bond (4 5 ) . (E) Reduction of the Monocarb oxykated Bridged-ring Derivatives Sta'bility of the Bridge-When the bridged-ring acid XIV is reduced by sodium amalgam, there is formed a cydopentane acid XXXIX which contains two more atoms of hydrogen than the original acid.The reduction product is a ketonic acid and on further reduction yields the corre-sponding hydroxy-acid XL. The fact t'hat the ketonic acid has been formed by the addition of hydrogen a t the bridge-bond (1 4) is shown (Zoc. cit.) by the production on oxidising with nitric acid of BB-dimethylglutaric acid and as-ctimethylsuccinic acid. The reduction is therefore to be represented thus: XZV and XXZII (pp. 333 and 335). bond (1 :4). CO,H C0,H CO,H b CH-CH FH3>C< CIH-CiH, C4>c<Y-QH? cH CH*CO + CH3>C<CH2,b0 CH CH CH,*CH*OH (XIS-.) (XXXIS.) FL.1 The remarkable feature of this reaction is that the bond (1 4) is actually more susceptible of attack by the reducing agent than is the carbonyl group and that consequently the bridged hydroxy-acid XLI cannot be isolated.(XLI.) (XLII,) A series of experimente using the monobasic acid XXIII yielded precisely comparable results. I n spite of careful search among the products of reductions carried out under varying conditions no bridged hydroxy-acid of the formula XLII was isolated. The firgt product of the action of sodium amalgam on the ketonic acid XXIII (p. 335) was a substance which contained two atoms of added hydrogen. It did not appear to react with acetyl chloride, but readily gave a semicarbazone. On subjecting it to furthe 0 U-0 V E ;\g '3 .. 0 m 54 X-u" 348 INGOLD AND THORPE THE FORMATION AND reduction by sodiuiri mialgam two more atoms of hydrogen were taken up and there was formed a subst'ance which gave an acetyl derivative on treatment with acetyl chloride.The successive reduc-tions are theref ore apparently analogous in the dimethyldieyelo-pentane and cyclohexaiiespirodieydopentane series and in the latter case may be represented thus : C0,H CO,H CO,H (XXIII). (XLIII.) (XLIV.) The fact that i t was realJy the bond (1 4) and not the bond (4 5) or the. bond (5 l) which had been broken by the reducing agent was clearly proved by the manner in which the reduced sub-stances behaved with oxidising agents. These experiments are dealt. with in Section F. I n our earliest experiments on the reduction of the bridged ketonic acid XXIII we used conditions which were known to give a good yield of the reduced acid XXXIX when applied to the reduction of the bridged ketonic acid XIV.As a result we obtained a product of indefinite melting point which proved to be a mixture of the ketonic and hydroxy-acids XLIII and XLIV. It was therefore apparent thatl the' reduction was proceeding more easily in the bridged-sp-ro-series than in the bridged-ring series. I n order to establish this point more definitely a series of com-parative experiments were instituted. I n the first place the bridged ketonic acid XIV (p. 333) was reduced under standard con-ditions for different lengths of time and the products were isolated. They were in general a mixture of three acids XIV XXXIX, and XL. The proportion of hydroxyl group in this mixture was determined by est'imating the acetic acid obtained by acetylation and subsequent hydrolysis.This method was found t o give good results when applied to the pure hydroxy-acid XLIV. I n this way a certain time of reduction was discovered during which no appreciable quantity of hydroxy-acid was produced. The hydrogen content was determined by combustion and it was found that the formation of hydroxy-acid began to be appreciable after the addi-tion of about 1.7 atoms of hydrogen to the molecule. A similar set of experiment6 with the bridged ketonic acid of the spiro-series showed that the production of hydroxy-acid in this case became appreciable only after the addition of about 1.9 atoms of hydrogen to t8he molecule. The two bridged ketonic acids XIV and XXIII were t.hen reduced under the same standard conditions for a certain length of time the same in both cases sufficiently short to ensur STABILITY OF SPIRO-COMPOUNDS.PART II. 349 that in neither case would any measurable amount of hydroxy-acid be formed. The products were then isolated and the hydrogen contents determined by combustion. Several pairs of experiments were made with different? lengths of time and the results obtained 2.2 2.0 1.54 1.6 1.4 & 1.2 B 1.0 B 5 0.8 0.6 0.4 0.2 0.0 0 2 4 6 8 10 12 Aours are given in the experimental part of this paper (p. 375). The figures lie fairly well on smooth but widely separated curves (Fig. viii) and graphically interpolated they show t h a t if we take the time required for a 50 per cent. conversion that is the time during which the molecule of the bridged acid takes up on 350 INQOLD AND THORPE THE FORMATION AND a t a n of hydrogen as the standard of comparison then this time, in the case of the spiro-acid XXIII is about 0.55 times as long as in the case of the ring derivative XIV.That is to say the periods of half-change are in the ratio of approximately 1.8 1. As a check on this result the reduced acids XXXIX and XLIII (pp. 346 and 348) were prepared in a state of purity and further reduced under standard conditions for different lengths of time. The figures obtained in these experiments (p. 375) lie with slight irregularities on one and the same curve (Fig. ix). The periods FIG. (ix). Reduction of the ketone group. 1.4 1.2 1.0 x 2 3 0.8 9 0.6 0.4 0.2 0.0 0 2 4 6 8 hours of half-change are therefore as 1 1 as nearly as the experimental figures can be interpreted.Thus there appears to be a very real difference in the ease of reduction of the monocarboxylic acids of the bridged-ring- and bridged-sp-yo-series. It will be noticed that the difference is in the sense anticipated from theoretical considerations (see Section A). It also would appear to be of the correct order of magnitude. For whilst the first-order effect that on the bond (5 l) is manifested by a reaction which proceeds at a considerable speed in one series but does not go a t all in the &her SQ fsr qt STABILITY OF SPIRO-COMPOUNDS. PART II. 35 1 can be detected both the second-order effects that is those on the bonds (1 2) and (1 4) exhibit themselves experimentally as moderate differences in reaction-velocities which are finite quantd-ties in both the series.The third-order effect that on the bond (2 3) was not detected experimentally. (F) Oxidation of the Fission-Products Derived front tJbe Bridged-spiro-compound X I 1 (p. 330). All the products of fission of the cyclohexanespirodicyclopentane ring structure so far considered belong t o one or other of the three following classes : (1) Substances in which the bridge-bond (1 :4) only has been broken. (2) Those in which the cyclopropane-bond (4 :5) only has been broken. (3) Those in which the cyclapropane-bond (4 5) and the cyclo-butane-bond (2:3) have both been broken. I n order if possible to obtain some confirmatory evidence regard-ing the constitutions of these substances a t least one typical example from each class was subjected to the action of oxidising agents.In class (1) the first substance taken was the cyclohexanespbo-cyclupentanone acid XLIII (p. 348). By far the most suitable reagent in this case is nitric acid. Dilute nitric acid however, appears to have little action on the substance. With 'hot concen-trated nitric acid a dibasic acid of the composition C8H,,0 was obtained. This substance when distilled gave off carbon dioxide with the formation of cyclohexanecarboxylic acid (hexahydrobenzoic acid). The dibasic acid is therefore evidently cyclohexane-1 l-di-carboxylic acid XLV. The same dicarboxylic acid was obtained when the hydroxy-acid XLIV (p. 348) was used in place of the ketonic acid.These experiments arekteresting as leaving little room for doubt that the bond broken in the reduction of the bridged-ketonic acid XXIII (p. 335) was actually the bond (1 4). The example taken from class (2) was the trans-cyclohexylcyclo-butane acid XXIV (p. 336). In this case nitric acid appeared t o be without effect. Warm alkaline permanganate was however, quickly decolorised. The acid product was a liquid substance whic 352 INGOLD SND THORPE THE FORMATION AND distilled apparently without a serious amount of decomposition and was identified as cyclohexanecarboxylic acid XLVI : CO,H (XXlV.) (XLVI.) The member of the series (3) experimented on was the tmtLs-P-cycZohexyI-92-butane lactonic acid XXXIV (p. 342). This acid was also1 unacted on by nitric acid and required an alkaline solution of permanganate kept a t above TOo t o oxidise it a t all rapidly.The product was as in the former case cyclohexylcarboxylic acid : Y c,B,,,:cH&o,H I c' H (0 H ) CO,H (Hydroxy-acid of XXXIV.) (XLVI.) Both these oxidations with periliarigailate were carried out under various conditions and in both cases the products were carefully examined for any traces of polybasic or lactonic acids in which two side-chains might be attached to bhe c!ycZohexane carbon atom. No such products were detected and in view of the formula of the oxidised substances none would be expected. On the other hand, if for example in the procluctioii of the cyclohexylc~clobutane acid XXIV (p. 336) some bond in the dicyclopentane structure other than one of those attached to the spiraiie carbon atom had been ruptured one would expect t o find products with two side-chains among the oxidation products.The cyclohexanecarboxylic acid XLVI obtained in the course of the above experiments appeared on careful examination to be in all cases identical with the product obtained by the reduction of beiizoic acid. The various reactions by which the dicyclopentane ring in the original bridged-spiro-compound XI1 (p. 330) has been broken down forming ultimately cyclohexanecarboxylic acid are suminar-ised for eoiivenieiice in table 111 TABLE 111. C0,Et C0,H tic1 C6HIQ:C<8H.KE2 f-bO,Et (Yellow solid.) (M. p. 236' ; semicarbazone, m. p. 259' dec.) sodium amalgam .I. GOSH . c---y H, C,H,, CH' I C(OH)-Cl> C0,H (31.p. 20Vdec.) (M. p. 136-138' ; semicarbazone, m. p. 310' dec.) (31. p. 12.?-127"; acetyl derivative m. p. 158-163'.) (31. p. 207' dw. 354 INGOLD AND THORPE THE FORMATION AND (G) Examination of the By-products obtained i n the Prepuratiou of the B r i ~ ~ e d - s p i r o - c ~ r n ~ o u ? ~ XIZ (p. 330) Isolation of Products of Fission Derived from this Substance through Side Reactions. During the preparation of the yellow sodium compound XI1 (p. 330) there was formed a considerable quantity of an oily by-pro-duct from which a number of acids were obtained by hydrolysis with hydrochloric acid. Amongst those isolated were n-butyric acid cyclohexane-1 1-diacetic acid and tram-cy clohexanesp'ro-cy clopropane-1 2-dicarboxylic acid.The formation of these acids is evidently traceable to the presence in the crude ethyl dibromo-cycluhexanediacetate used for the condensation of the corresponding monobromo-ester of the unbrominated ester and of ethyl bromide as impurities. There was also obtained a cyclohexanespirocyclo-propane acid of the formula XLVII. This substance proved to be identical with the acid obtained by the action of acid hydrolysing C(C0,H) *CH,*CO,H C5H10:C<&~.c0s~ (XLVII.) agents on the tetra-ethyl spirocyclupropane ester XI (p. 330) which clearly establishes the constitution of the compound. I n addition to the above-mentioned acids there were isolated two others to which the formula XLVIII and XLIX have been C0,H (XLVIIT.) E t CK- -CO I I C5H1,:CH*U*C02H LH( CO,R)*O (XLIX.) assigned.These are clearly the products respectively of acid and of alkaline hydrolysis of the ethylated ester L. This ester is doubt-(L.) less produced by the action of ethyl bromide on the yellow sodium compound XII and apparently behaves towards alkaline hydrolys-ing agents similarly to the corresponding methylation product XXX (p. 342). We did not investigate the action of acids on the methyl derivative XXX but the corresponding substance XXVII (p. 341) of the dimethyldicydopentane series was found (loc. cit ) to yield with acid hydrolysing agents a monocarboxylic acid to which XLVIII is strictly analogous STABILITY OF SPIRO-COMPOUNDS. PART U. 356 The most interesting of these substances is the fission product. XLIX of the ester L and it is of interest to examine what possi-bilities there are of alkaline hydrolysis of the ester L taking place.No 'hydrolysis to a lactone of hydroxy-ester would be likely to occur in the anhydrous alcoholic solution in which the oily by-product was formed. The oil was however separated from the sdium compound XI1 by means of 95 per cent. alcohol and since some sodium ethoxide would certainly be adhering to the crude sodium compound the alcoholic washings would contain sodium hy.droxide, which must have brought about the fission of the ethylated ester L. The lactonic acid XLIX was found to possess properties practi-cally identical with those of its prototype the fission product XXXIV (p. 342) of the methylated bridged ester XXX. Like the lactonic acid XXXIV it was isolated in cis- and trans-forms the substance originally obtained being the trans-form.One notable point of difference was noticed between these compounds and the methylated lactonic acids previously obtained. The ethylated cis-lactonic acid differed from the trans-form and from both forms of the lactonic acid XXXIV in the fact that it was found possible to isolate from it the free tribasic hydroxy-acid. This substance was, however very unstable. It slowly gave up water when exposed to' air a t the ordinary temperature the product being the cis-lactonic acid. On boiling with hydrochloric acid it was converted into the trans-lactonic acid. The other relationships between these ethyl-ated products may be seen by reference to table IV in which they are shown in relation to the ethylated ester L the decompositions of which form another example of the great ease with which the bond (4 5) is ruptured by alkalis.E X P E R I M E N T A L. (a) Condensation of Ethyl Dibromocyclohexane-1 ; 1-dincetate utith Ethyl Sodiomalonate. cycloHexane-1 1-diacetic acid was prepared for use in these experiments by the method given by Thole and Thorpe (T. 1911, 99 422). Ethyl DibromoqyclohexaneJ 1-diacetate.-The bromination of the acid was effected by the Hell-Volhard-Zelinsky method as described in Part I. of this series. The neutral product contained about 80 per cent. of dibromo-ester 356 INGOLD AND THORPE THE FORMATIOX AND h n ? c f- I 0, A 1 ; \ + I E 'k s L $\2 -4" -I -'2 4 z 0 -+ h ?3 STABLLrPY OF SPIRO-COMPOUNDS.PART II. 357 Ethyl cycloHexunespiro-1-methylcyclopropune-1 11 11 2-tetra-carboxylate (XI p. 330). Forty-six grams of the crude dibromo-ester were added t o a solu-tion in 60 grams of alcohol of 4.6 grams of sodium and 32 grams of ethyl malonate. The solution was boiled for three hours and then p u r e d into dilute hydrochloric acid. The oil was extracted w i t h ether and the extract washed with a solution of sodium carbonate dried and distilled. The estw was obtained as a colourless liquid which on reldistilla-tion boiled at 250-260°/10 mm. On boiling with hydrochloric acid i t was hydrolysed with the formation of the tricarboxylic acid (XLVII p. 354) dealt with on p. 379: 0.1209 gave 0.2708 CO and 0.0849 H,O.C = 61.09 ; H = 7.80. C,,H,,O requires C=61*2; H=7.8 per cent. (b) Formation of the B&dge&spiro-ester and i t s Sodium Compound. The formation of the bridged-sp'ro-compound by the condensation of the above-mentioned tetraethyl ester with itself has already been referred to in the Introduction (Section B). The compound was, however usually prepared direct from the dibromo-ester by treat-ing it with ethyl malonate and excess of sodium ethoaide. Many experiments were1 made in order t o determine1 the best conditions. Ethyl Sodiocyclohexanespirodicyclope~tan-3-one-l 2 4-tricarb-oxylate (XII p. 330). Nine grams of sodium dissolved in 140 grams of absolute alcohol were treated with 30 grams of ethyl malonate. The solution was carefully under-cooled to about 35O and 40 grams of dibromo-ester were gradually added the temperature being kept blow 40°.Half an hour after the addition was complete the liquid was heated on a steam-bath and kept boiling for thirty hours. A t the end of that time the greater part of the alcohol was boiled off and water added to the residue in the flask. The mixture was then shaken vigor-ously and filtered by the aid of the pump. I n these circumstances the whole of the oil precipitated by the water adhered to the solid sodium compound. The filtrate which gave no precipitate on acidi-ficahion was discarded. The purification of the sodium compound was effected by washing on the filter with 95 per cent. alcohol and VOL. cxv. 358 INUOLD AND THORPE THE FORMATION rW'D finally by triturating with the same solvent until the weight of the dry solid was not altered on repeating the treatment.The alcoholic filtrates contained the oily by-product. The sodium compound was obtained as a bright yellow insoluble substance which gave a crimson colour with aqueous ferric chloride containing a trace of alcohol: (Section j p. 377.) 0.3132 gave 0.0590 Na,SO,. Na=6'10. C1QH2507Na requires Na = 5.93 per cent. Ethyl cycl~~exanespiralicyclopen tan-3-one-1 ; 2 ; 4-tm'carb-oxylate (XV p. 333). When the yellow sodium compound was shaken with cold dilute aqueous hydrochloric acid and ether it passed quickly into solution, the yellow colour being discharged. The ethe'real layer on drying and evaporating the solvent yielded a mass of crystals melting a t 46A7O.The ester was exceedingly readily soluble in the usual organic solvents and did not appear capable of being easily recrystallised. It gave a crimson coldur with ferric chloride and on treating with cold aqueous sodium hydroxide yielded the original sodium com-pound : 0.1331 gave 0.3036 CO and 0.0842 H,O. C=62.21; H=7*03. C,,H2,07 requires C=62.3; H=7.1 per cent, (c) Comparative Experiments on the Formationi of the Bridged-ring- and Bridged-spiro-cmpounds (IX and XII p. 330). /3/3-Dimethylglutaric acid was prepared by tlhe mathod of Thole and Thorpe (T. 1911 99 422). Ethyl Dibromodimethy1gEutwate.-The acid was first converted into its anhydride (T. 1899 75 48) which was then brominated (T. 1901 79 776). The dibromocester was redistilled and collected for use in subsequent experimenb a t 182-185O/24 mm.Ethyl 1 3 3-Trirnethzylcyclopr~pan~e-1 1 1 1 1 2-tetracwrboxylate. This wter was prepared by condensing the dibromo-ester with ethyl sodiomalonate in alcoholic solution under the conditions used by Perkin and Thorpe (ibid.) and purified by distillation the frac-tion boiling a t 231-236O/34 mm. being taken as sufficiently pure for the experiments herennder described. Et'hyl trimathylcyclloprolpanetetracarboxylate and ethyl cyclo STABILITY OF SPIRO-COMPOUNDS PART II. 359 h exa nespirum ethyl c y el opr op an etetr a ca r b ox y 1 ate were then employed in a series of experiments which were carried out with the object of determining the relative speeds with which the two esters under-went internal condensation in the presence of sodium ethoxide.The method was as follows One molecular proportion of each ester was treated with two atomic proportions of sodium dissolved in fifteen molecular proportions of absolute ethyl alcohol. The solutions were kept in a thermostat for known lengths of time after which the alcohol was boiled off under diminished pressure and water was added. The precipitates were then collected washed with alcohol dried and weighed. The following percentage yields of the sodium compounds IX and XI1 (p. 330) were obtlained the temperature being 7 5 O : Time (hours) 1.0 2.0 4-0 7.0 10.0 14.0 24.0 TABLE V. Bridged-ring-compound IX. (Per cent.) 49.8 64.7 73.0 76.6 75.3 75.3 -Bridged-qiro -compound XII.(Per cent,) 10.1 17.9 26.5 32.4 32.9 37.7 -(d) Hydrolysis of the Bridged-spiro-ester and of its Pellow Sodium Compound. The remarkable diversity in the characters of the substances which can be obtained by hydrolysing the yellow sodium spiro-compound or the corresponding free ester under different conditions has already been alluded to in the Introduction. The following is a summary of the principal experimental details. Die thy1 Potassium Potassiocyclohexanespirodicyclope n~tan-3-one-1 2 4-t~icarbolxylute (XVI p. 333). When the original yellow sodium compound was left in cont,act with cold alcoholic potassium hydroxide f o r ten hours it gradually dissolved and a canary-yellow potassium salt separated. This compound was found to be insoluble in water but quite appreciably soluble in methyl alcohol.When recrystallised from a large bulk of this solvent it separated in long yellow needles, which gave a violet colour with ferric chloride: 0.1354 gave 0.0580 K2S0,. K = 19.16. C,,H2,O7K2 requires K = 18 -8 per cent. P 360 INQOLD AND THORPE THE FORMATION AND Diet h yl Hydrogen cycloH exanespir~icyclopentan-3-one-1 2 4-tricarboxylate (XVII p. 333). When the potassium salt was treated with cold dilute hydro-chlorio acid a gummy precipit'ate was obtained. This could not be induced to crystallise and on distillation underwent extensive decomposition. It was therefore extsacted with pure ether and, after drying and evaporating the solvent left for some days in an exhausted desimator : 0.1249 gave 0.2751 CO and 0.0730 H,O.CI7H,O7 requires C = 60.4 ; H = 6.5 per centr. This acid-ester is also the first product of the action of boiling hydrochloric acid on the triethyl ester (XV p. 333) as is proved by the following experiment. The yellow sodium compound was boiled with 20 per cent. hydrochlorio acid for one hour. The liquid was evaporated and the residue dissolved in the minimal quantity of water. The hot aqueous solution was rapidly cooled, and the oily precipitate which separated was collected by pouring the liquid through a wet filter. The oil on the filter was then washed through with alcohol and caused to crystallise as corn-pletely as possible from 15 per cent. alcohol. The crystals con-sisted of the ethyl hydrogen ester X X I (p. 335).The ultimate oily residue from these crystallisations was dissolved in methyl alcohol and treated with a slight excess of cold methyl-alcoholic potassium hydroxide. The yellow precipitate which immediately separated was collected and recrystsllised from methyl alcohol. On analysis it gave K=19*03 whilst the free acid-ester obtained on acidification gave C=60*21 H=6.60 per cent. With ferric chloride the acid-ester gave a violet coloration. Cola methyl-alcoholic potassium hydroxide converted it into the potassium compound. It distilled a t about 200-260°/23 mm., with however considerable decomposition. Attempts were made to hydrolyse the gummy distillate both by acids and by alkalis, but no pure substance was isolated from the products. ct=60.07; H=6.50.Ethyl Bih ydrogen cyclolHexanaspirod/icyclo pentan -3 -one-1 2 4-tricarboxylate (XVIII p. 334). Five grams of the yellow sodium compound were suspended in 20 C.C. of 3N-ethyl-alcoholic potassium hydroxide and the mixture was boiled until the yellow colour was discharged. The colourless precipitate which was very hygroscopic was collect'ed as rapidly as poasible and drained on porous porcelain in a desiccator. I STABILITY OF SPIRO-COMPOUNDS. P m T Iz. 361 was then dissolved in a small quantity of water and decomposed with hydrochloma acid. The acid which separated out was re-crystallised from water. The potassium salt XVI may be used in place of the sodium compound in this preparation. The same acid-ester may also be obtained by acid hydrolysis of the triethyl ester (XV p.333) or of its sodium compound. Thus, when the sodium compound was boiled for one hour with 20 per cent. hydrochloric acid and the acid-esters XVII and XXI separated by precipitating them together as an oil in the manner described on p. 360 it was found that in the filt'rate from the oil there were present two crystalline substances. These were isolated by treating the solution after concentration with sufficient con-centrated hydrochloric acid to clear the turbidity. The crystals which were then deposited from solution were separated by fraction-ally crystallising from water into the acid-ester XVIII and the dibasic acid XXII the latter being the more readily soluble. The acid-ester formed long colourless needles which melted and decomposed a t 206O.It gave a bluish-violet colour with ferrio chloride : 0.1298 gave 0-2745 CO and 0.0683 H,O. The mhy&o-ester, C=57*68; H=5*84. Cl,Hl,O requires C=58*1; H=5*8 per cent. co-0-co I was prepared by treat'ing the acid-ester at looo for one hour with acetyl chloride in a closed . flask. The residue obtained on evaporation was crystallised from ether. The crS;stalline anhydride melted a t 126O and was converted into the original acid-ester on boiling with water : 0.0751 gave 0.1705 CO and 0.0385 H,O. C=61*92; H=5.70. (&&&)6 requires C=61*6; H=5.5 per cent. E t hy 1 H ydroyen 5-cycloHexanespirodicy clopen tan-3-one- 1 4-dicarboixylate (XXI p. 335). The formation of this substance has already been alluded to on p. 360. It is best prepared by boiling the triethyl ester (XV, p.333) with 20 per cent. hydrochloric acid for five hours or by boiling the acid-ester (XVIII p. 334) with the same reagent for two hours. In either case the product obtained on evaporation was found to be a mixture of three acids. It was crystallised fro 362 INGOLD AND THORPE THE FORMATION AND the minimal quantity of boiling 50 per cent. ethyl alcohol. The crystals which separated consisted principally of the monobasic acid (XXIII p. 335) and were collected the filtrate being then evaporated until most of the alcohol had been removed. It was then cooled as rapidly as possible and the oil which separated was collected on a wet filber. The filtrate on concentrating and mixing with concentrated hydrochloric acid deposited the dibasic acid (XXII p.335). The oil on the filter was dissolved in alcohol and recrystallised several times from a mixture of alcohol and water. The acid-ester obtained in this way melted a t 104-106° and gave no colour with ferric chloride. It did not decompose appreci-ably when heated to 250O: 0.0887 gave ?*2061 CO and 0.0548 H,O. 0.1200 required 18.62 C.C. of Ba(OH) solution (0-0243N) for C=63.37; H=6*88. Cl4HI8O5 requires C= 63.2 ; H = 6.8 per cent. neutralisation. C,,H,80 (monobasic) requires 18.6 C.C. 5 -cycloHexan~espirod~cyclopen tun- 3- 011 e - 1 2-dicarb ox y lic A cid (XXII p. 335). The formation of this substance as a by-product in the pre-paration of the various acid-esters of the series has already been noticed; It was found to be produced in good yield by boiling either the yellow sodium compound or the acid-ester (XVIII, p.334) with 10 parts by weight of 20 per cent. aqueous hydro-chloric acid for twelve hdurs. As the boiling proceeded oily pro-ducts separated out and subsequently passed again into solution. Then crystals appeared in the boiling liquid. A t the end of the period the mixture was cooled and allowed to remain a t &he ordinary temperature for twenty-f our hours after which practically the w b l e of the organic material had crystallised out. The crystalline mixture consisted of about three parts of the dibasic acid XXII to one part of the monobasic acid XXIII (p. 335). It was boiled with four times its weight of water and the suspension cautiously cooled and quickly filtered.By this means the mono-basic acid was separated almost quantitatively from a solution which in the cold was supersaturated witlh respect to the dibasic acid. The agitation caused by filtering usually caused the filtrate to set to a stiff pasfa of crystals of the dibasic a&d. These were recrystallised from water. The acid separated from water in rosettes of long silky needle-shaped crystals which melted and decomposed a t 2 3 4 O . It gave a deep crimson colour with ferric chloride butl did not appear t-o be acted on when boiled with hydrochloric acid for several days STABILITY OF SPIRO-COMPOUNDS. PART 11 363 0,1087 gave 0.2407 CO and 0.0584 H,O. 0-0715 required 24-65 C.C. Ba(OH) solution (0.0243N) for C=60.39; H=5*97. Cl2H1405 requires C = 60-5 ; H = 5.9 per cent.neutralisation. CI2Hl4~ (dibasic) requires 24.7 C.C. co-0-co I free acid was treated with acetyl chloride a t looo in a closed flask. The solid residue obtained on evaporation was triturated with aqueous sodium hydrogen carbonate and recrystallised from ether. It melted a t 154O and on treating with aqueous sodium hydroxide gave the sodium saltl of the original dibasic acid: 0-1050 gave 0.2515 CO and 0-0518 H,O. @=65*32; H=5*48. C,,H,,O requires C=65*5; H=5*5 per cent. 5-cycloHexanespirod~cyclope~t~n-3-on~e-l-carboxylic Acid (XXIII p. 335). Al mixture of aboutl one part of this acid to three parts of the dibasio acid (XXII p. 335) was formed when either the triethyl ester XV or its sodium compound or the diethyl hydrogen ester, XVII or its potassium compound or the ethyl dihydrogen ester, XVIII was boiled for twelve hours with 20 per cent<.hydrochloric acid. The ethyl hydrogen ester XXI was found to be converted quantitatively into the monobasic acid by boiling hydrochloric acid. The dibasic acid XXII on the other hand did not appear to be affected by this reagent. The dibasic acid when heated above its melting point however, evolved carbon dioxide and from the dark-coloured residue a small amount of monobasic acid could be isolated. A good yield was obtained when the dibasic acid was heated with water a t 200° for about ten minutes. The acid-esters XVII and XVIII also gave excellent yields of the monobasic acid when treated in this way. The triethyl ester XV however required the presence of a trace of acid in the water.A small quantity of hydrochloric acid or acetic acid or even of butyrio acid was found to be sufficient. The most convenient way of preparing the monobasic acid is by heating the yellow sodium compound with a slight excess of dilute hydrochloric acid a t 200O. When preparing considerable quanti-ties however it was found desirable. to drive off as much carbon dioxide and alcohol as possible before closing the vessel. Th 364 INGOLD AND THOR,PE THE FORMATION AND sodium compound in portions of about 10 grams was boiled with ten times its weight of 20 per cent. hydrochloria acid for twelve hours in a strong flask provided with a short reflux air-condenser. Enough aqueous sodium hydroxide was then added t;o reduce the concentration of free mineral acid to 2 or 3 per cent.and the solution was again boiled to expel the air. The flask was then securely corked and immersed' in an oil-bath a t 200° for ten minutes. After cooling to the ordinary temperature the crystals were collected and recrystallised from 96 per cent. alcohol using a little animal charcoal to remove the dark impurities. The yield was 85 per cent. The monobasic acid was sparingly soluble in hot or cold water and in most cold organic solvents but it crystallised well from hot ethyl alcohol in long needles. It melted a t 236" without decom-position and gave no colour with ferric chloride. It was found to be unacted on by boiling aqueous or alcoholic potassium hydr-oxide and by prolonged boiling with hydrochloric acid.Cold alkaline permanganate was however instant'ly decolorised : 0.1337 gave 0.3310 (70 and 0.0860 H,O. C=67*52; H=7.14. C,,H,,O requires C = 68.0 ; H = 7.2 per cent. prepared by boiling the acid with an aqueous solution of semicarbazide acetate for a few seconds. On coding the solution, the semicarbazone separated out ,and was recrystallised from alcohol. It melted and decomposed at 2 5 9 O : C,,H$3N3 requires C!=57.4 ; H = 6.8 per cent. 0.1071 gave 0.2240 CO and 0.0661 H,O. Cr=57-04; H=6*85. trans-3-Hydrox y-4-cyclohexylcyclobutan-2-one-3 4-dicarb oxylic Acid (XXIV p. 336). Five grams of the yellow sodium compound were boiled with 30 C.C. of 4N-alcoholic potassium hydroxide the boiling being. con-tinued for one hour after the suspended matter had become colour-less; or alternatively 5 grams of the acid-ester (XVIII p.334) were boiled with 30 C.C. of the same reagent for one hour. I n either case the product was isolated by evaporating the 9 alcohol and adding water and hydrochloric acid. The acid solution was extracted ten times with its own volume of ether and the extract dried over calcium chloride for at least three days. This was found to be necessary since the acid was present in the ether in its hydrated form which apparently gave up water ix the calcium chloride very slowly and incomplete dehydration interfered wit STABILITY OF SPIRO-COMPOUNDS. PART 11. 365 the subsequent purification. When quite dry the ether was evaporated and the viscid residue triturated with Chloroform. The crystals which were caused to separate by this treatment were drained on porous porcelain and washed with fresh ether.The acid prepared in this way was fairly pure and melted and decomposed a t 203O. It was very readily soluble in water and in all the usual organic solvents except chloroform and light petroleum. I n these solvents it was only sparingly soluble but it did not appear t801 crystallise well from mixtures of solvents. It gave no colour wit'h ferric chloride and did not appear to be acted on by boiling acetyl chloride (compare Part I). The acid was purified for analysis through the hydrate (see below): 0.1331 gave 0.2734 CO and 0.0748 H,O. 0.0352 required 11-25 C.C. Ba(OH) solution (0*02431\7) for Tho hydrated form @12H,,06,2H,0 separated in large dense prisms when the anhydrous acid was dissolved in hot water and the solution cooled.The hydrated acid readily dissolved in dry ether and was much more readily soluble in chloroform than the anhydrous substance. A t looo it evolved water vapour leaving the anhydrous acid in a very pure form melting a t 206O tol a colour-less liquid which evolved steam and after coding set to a solid mass which melted a t about 135O: C=5692; H=6.24. C,,H160 requires C=56*2; H= 6.2 per cent. neutralisation. C,,H,,O (dibasic) requires 11.3 C.C. 0.1648 lost 0.0201 a t looo. The siluey salt was precipitated by silver nitrate from a neutral 0.1435 gave 0.0656 Ag. Ag=45-71. 0.1729 , 0.1929 CO and 0.0473 H,O. C=30*43; €1=3.03. C,,H,,O,Ag requires C = 30.6 ; H = 3.0 ; Ag= 45.9 per cent.The barium salt was precipitated from a solution of the acid 0.1007 gave 0.0595 BaSO,. Ba=34.74. H,O=12*20. C,2H,606,2H20 requires H,O = 12.3 per cent'. solution of the ammonium salt: in water by adding an excess of barium hydroxide: C,2H,,0,Ba requires Ba = 35.1 per cent. cis-3-~y&oxy-4-cyclohexy~~yc~obutan-2-one-3 4-diciwb oxylic Acid (XXIV p. 336). This acid was prepared by dissolving its anhydride (see below) in a slight excess of 4N-aqueous sodium hydroxide and then add ing a slight excess of concentrated hydrochloric acid. The p r e cipitated acid was collected and dried. It was then dissolved in P 366 INGOLD AND THORPE THE FORMATION AND dry ether containing a trace of alcohol and caused to crystallise from this solution by adding benzene.The crystals were finally purified by t,riturating with pure dry ether and again recrystal-lising from a mixture of ether and benzene containing alcohol. The pure cis-acid melted at 145O and rapidly evolved water vapour. It readily dissolved in water or alcohol but was almost insoluble in pure dry ether. Unlike the trams-form it did not appear to take up water of crystallisation: 0.0929 gave 0.1912 CO and 0.0548 H,O. 0,0446 required 14.40 C.C. Ba(OH) solution (0.0243N) for C=56.13; H,=6.55. C12Hl6O6 requires C = 56.3 ; H = 6.3 per cent. neutralisation. C1,HI6O6 (dibasic) requires 14-3 C.C. The Anhydride of the cis-A cid C,H,,:CH*C<-CH2->C0. / C(OH) The tram-acid on heating above its melting point gave off water vapour. The transformation was however by no means complete unless the fused material was raised to 240-250° and maintained a t this temperature until it began to darken in colour.The product solidified on cooling and after triturating with aqueous sodium hydrogen carbonate and drying was recrystallised from dry ether. It separated in large oblique prisms which melted a t 155O. The substance was also recrystallised from benzene. The same anhydride was obtained by heating the cis-acid above it8 melting point: 0.1335 gave 0.2959 CO and 0.0709 H,O. C=60*45; H=5*90. CI2H,,O5 requires C'= 60.5 ; H = 5.9 per cent. -CH2-The cis-Anilic Acid 1 bO,H C~H~~~CH*c<C(OH)>Co (-?O*NHDC6H, This substance was a t once precipitated when the cis-anhydride was dissolved in benzene and treated with a solution of aniline in the same solvent.It was purified by first triturating with ether and then recrystallising from dilute alcohol. It separated in minute crystals which melted and decomposed a t 202O : C18Hz10,N requires C = 65.3 ; H = 6.3 per cent. 0.1145 gave 0-2721 CO and 0.0666 H,O. C=64*81; H=6*46 STABILITY OF SPIRO-COMPOUNDS. PART II. 367 U,H,~:CH ~ p ~ ~ f ! ~ c o . Co j The cis-Ad, C,H,*N-CO The cis-anil was readily obtained by heating the anilic acid a t 21O0 until the evolution of st'eam had ceased. The product was triturabed with sodium hydrogen carbonate and then recrystallised from absolute alcohol. It separated in long silky needles which melted a t 199O: 0.1098 gave 0.2778 CO and 0.0592 H,O. C=69.00; H=5.99. C,,H190,N requires C = 69.0 ; H = 6.1 per cent.(e) C'ompara,tiPie Experimemts with the A cid-Esters (XXVI g. 337, and XVIII p. 334) of the Dimethyldicyclopentane and c y clo H e xa ri espirodic yclopen t a ne Series .* One molecular proportion of each acid-ester was boiled for fifteen minutes with 6 molecular proportions of potassium hydroxide in 3.6N-solution in ethyl alcohol. The bulk of h e alcohol was evapor-ated under diminished pressure and water and excess of hydro-chloric acid were added to the residue. The acid products were then extracted quantitatively with ether. The percentage yields were as follows : TABLE VI. Bridged-ring-acid-ester XXVI. I Bridged-spiro-acid-ester XVITI. icid- Monocarb- Fissio 1 Acid- Products of Fissi.n product I ester loss Of ** products ester recovered.fzL:d. formed. I recovered. carbethoxyl €armed. 1 A / oxylic group. 85 1.5 0 0 0 85 81 4 O r 5 0 78 In all cases a small quantity of gummy material was formed and for this reason the whole of the original material was never accounted for as cryst.alline products. The small quantity of mono-carboxylic acid XIV was readily isolated by reason of its sparing solubility in cold water. The recovered acid-ester XVIII was also quite easily separated from the fission product by means of dry chloroform in which the latter if quite anhydrous is almost insoluble. * See note p. 337. P* 368 INGOLD AND THORPE THE FORMATION AND (f) Prepration and Hydrolysis of the Methylatiom Product of the Ph?ow SoAum spiro-Compundr (XII p. 330).The yellow sodium compound does not react a t all readily with methyl iodide under the usual conditions even a t looo in a close4 flask. If however four or five times the theoretical quantity of methyl iodide is used methylation proceeds rapidly. Ethyl cycloHexanespiro-2-me t hyldicyclopentan-3-one-1 2 4-tri-carboxylate (XXX p. 342). Twenty grams of the yellow sodium compound were heated with a solution of 20 grams of methyl iodide in 100 grams of absolute alcohol a t looo for one and a-half hours in a closed flask which, from time to time was vigorously shaken. The excess of the methyl iodide and most of the alcohol were then distilled off and the residue poured into 400 C.C. of water. The precipitated oil was extracted with ether the extract being washed with water and sodium car-bonate solution dried and evaporated.The oily residue was found to decompose on attempting to distil it under diminished pressure. It was therefore allowed to remain in an exhausted desiccator for several days and then analysed: 0.1109 gave 0.2556 CO and 0.0729 H20. C2,H2,0 requires C=63*2; H=7.4 per cent. The figures quoted are those for one of three closely agreeing analyses. They indicate that a partial conversion into the methyl diethyl ester has taken place : C,,HB07 requires C=62*2; H=7.0 per cent. This is perhaps a natural result of the use of a large excess of methyl iodide in the preparation. C=62*86; H=7.30. trans-lactonic Acid of y-Hydrox~-~-cyclohexyl-a-methyltri-carbaLlylic Acid (XXXIV p. 342). Twenty-five1 grams of the methylated ester were boiled under a reflux condenser with 170 C.C.of 4iV-alcoholic potassium hydroxide for two and a-half hours. The mixture was then cooled and the precipitated salts were collected and drained on porous prcelain in a desiccator. They were then dissolved in water and the solution was acidified and repeatedly extracted with ether. The residue left after drying and evaporating the extract was caused to deposit crystals by triturating with benzene the process being repeated until an ultimate gummy residue was obtained from which no crystals would separate STABILITY OF SPIRO-COMPOUNDS. PART 11. 369 The crystals were placed in a twtrtube with just sufficient benzene to cover them. The benzene was then boiled for a few minutes and the suspension filtered while hot.The filtrate on cooling deposited crystals of the lactonic acid. The ultimate gummy residue was esterified with alcohol and sul-phuric acid in the usual way and after adding water the esters were extrachd with ether. From the extract the acid products were shaken out with aqueous sodium hydroxide and again ex-tracted from the aqueous solution after acidification. The residue obtained on evaporating the ether was hydrolysed by boiling hydro-chloric acid. After twelve hours the liquid was rendered alkaline and extracted with ether then re-acidified and again extracted with ether. On drying and evaporating the latter extract a residue was obtained which when treated with benzene yielded a further quantity of the crystalline lactonic acid.The substance after recrystallisation from benzene melted a t 1 7 2 O without decomposition and did not appear to decompose appre-ciably a t 2 6 0 O . It dissolved very readily in water or alcohol and fairly readily in benzene or chloroform : 0.1310 gave 0.2774 C02 and 0.0800 H20. C=57*75; H=6*78. The silver salt was a t once precipitated when silver nitrate was 0*1118 gave 0.0449 Ag. Ag=44*63. C13H1806 requires C=57.8; H=6*7 per cent. added to a boiled solution of the lactonic acid in ammonia : C13H,,0,Ag2 requires Ag = 44'6 per cent. trans- y-Hydroxy-S-cycloi~exyl-a-met~yltm'carballylic ,4 cid. The hydroxy-acid appeared to be stable only in the1 form of its salts and in spite of several attempts it was not found possible to obtain it free. However when the trans-lactonic acid was dissolved in water and slowly titrated with barium hydroxide an end-point correspnding with the neutralisation of three carboxyl groups was obtained : 0.0434 of the trams-lactonic acid required 19.90 C.C.Ba(OH)2 solu-tion (0.0243N) for complete neutralisation. C,,H,,O changing in solution t o C13H2,0 (tribasic) requires 19.85 C.C. The barium salt was prepared by treating a solution of the trms-lactonic acid in water with an excess of aqueous barium hydroxide: 0.1671 gave 0.1202 BaSO,. Ba.=42.34. (C,,H,,O,),Ba requires Ba = 42.0 per cent 370 INGOLD AND THORPE THE FORMATION AND cis-Lactonic A cid of y-Hydroxy-/3-cyclohexyl-a-met hyltricarballylic Acid (XXXIV p. 342). This substance was readily obtained by dissolving its anhydride (see belo,w) in boiling sodium hydroxide solution and acidifying with 'hydrochloric acid.Microscopic crystals separated and were recrystallised from water. The cis-lactonic acid melted a t 1 5 2 O with the immediate eslimina-tion of water vapour. It was more readily soluble in water than the trans-acid : 0.1182 gave 0.2489 CO and 0.0724 H,O. The silver salt was prepared by adding a solution of silver nitrate 0.1029 gave 0'0457 Ag. Ag=44.41. On boiling with concentrated hydrochloric acid the cis-lactonic C=57*43; H=6*81. C,,HI8O requires C =I 57.8 ; H = 6.7 per cent. to a boiled solution of the cis-lactonic acid in ammonia: CI3HI,O,Ag requires Ag = 44.6 per cent. acid was partly converted into the trans-isomeride. cis- y-Hydroxy-P-cycloh ezyl-a-me t hyltricnrbnllylic A cid.Like the tram-modification this substance appeared to be stable only in the form ,of its salts. The cis-lactonic acid on titration with aqueous barium hydroxide gave however an end-point correspond-ing with salt-€ormation in respect of three carboxyl groups : 0.0260 of the cis-lactonic acid required 11.9 C.C. Ba(OH)2 solution (0*02431\7) for complete neutralisation. C,,H,,O changing in solu-tion to C,,H,,O (tribasic) requires 11.9 C.C. The barium salt was precipitated from a solution of the cis-lactonic acid in water by the addition of an excess of barium hydroxide solution : 0.1403 gave 0*1011 BaSO,. Ba = 42.42. (C,3H,707),Ba requires Ba = 42.0 per cent. co-0-co The cis-lactouic Anhydride C,H,,:CH*F--YH ' I 1 MeCH-COO0 The lactonic anhydride was prepared by heating the truns-lac-tonic acid with amsty1 chloride a t looo in a closed flask.The product was evaporated in a vacuum and the residue crystallised from ether STABILITY OF SPIRO-COMPOUNDS. PART II 371 when it separated in small crystals melting a t 170O. be purified by distillation under diminished pressure. acid with acetyl chloride a t atmospheric pressure : It could also The same anhydride was also obtained by boiling the cis-lactonic 0*1170 gave 0.2637 CO and 0.0679 H20. C=61.47; H=6.45. C13H1605 requires C= 61.9 ; H = 6.3 per cent. trans-3 -Hydrox y -4-cychh ex y l- I-me th ylcyclo butcun-2-one-3 4-dicaroxylic Acid (XXXVII p. 345). The portion of the cryfitalline mixture obtained in the prepara-tion of the trans-lactonic acid (p.369) which did not diissolve in the hiling benzene consisted essentially of the cyclobutane acid XXXVII and was recrystallised from aqueous alcohol. The ethereal solution QP the eshrified gummy residue (p. 369), a f h r shaking out the acid products with aqueous sodium hydroxide, was dried and evaporated. The residue on distillation under diminished pressure yielded a fraction passing over a t about 260°/ 25 mm. which was hydrolysed by boiling hydrochloric acid. The product was rendered alkaline and extracted with ether then acidi-fied and again extracted with ether. The latter extract on drying, and evaporating the solvent yielded a residue which deposited crystals of the cyclobutane acid on adding benzene. The crystals were washed with benzene and recrystallised from aqueous alcohol.The trams-cyclobutanel acid melted and evolved steam at 185O, without appreciable discoloration. The liquid on cooling solidified, and on reheating melted at about 140*: 0*1008 gave 0.2132 CO and 0.0618 H20. 0*0600 required 18.37 C.C. Ba(OH) solution (0-0243N) for C=57.68; H=6*81. C,,H,,O requires C=57.8; H=6*7 per cent. neutralisation. C,,HI8O6 (dibasic) requires 18.3 C.C. cis-3-~ydro~-4-cycl<zhexy~- 1-meth ylcyclobutan-2-one-3 :4dicadoxylic Acid (XXXVII p. 345). The cis-acid was prepared by boiling its anhydride (see below) On cooling the solution the anhydride separated in tho The acid melted a t 148O with the immediate elimination of water-It was more readily soluble in water than the trams-modi-with water.pure condition. va.pour. ficat,ion 372 INGOLD AND THORPE THE FORMATION AND 0.0704 gave 0.1497 CO and 0.0430 H,O. C=57*99; H=6*78. 0.0481 required 14.65 C.C. Ba(OH)2 solution (0-0243N) for When boiled for four hours with concentrated ‘hydrochloric acid C13H1806 requires C=57*8; H=6.7 per cent. neutralisation. C13H180 (dibasic) requires 14.7 C.C. the cis-acid was quantitatively convel-ted into the trans-isomeride. C,H,,:CH-O<CHMe>CO. O H ) / Anhydride of the cis-Acid, c0*0*60 A t its melting point the trans-acid evolved water-vapur but the elimination was by no means complete a t this temperature. The melted substance was therefore raised to 230° and maintained a t this temperature until it began to darken in colour. The product was triturated with aqueous sodium hydrogen carbonate dried and recrystallised from dry ether.The same anhydride was obtained by heating the cis-acid above its melting point: 0.0833 gave 0.1886 CO and 0.0483 H,O. It melted a t 158O. C=61*75; H=6*44. C13H1605 requires C = 61.9 ; H = 6.3 per cent. (g) Reduction. of the Monobasic Bridged-spiroacid (XXIII, p. 335) b y Sodium Amalgam. It was found necessary in order to be able to repeat the results, to standardise carefully the method of experiment. The reductions were always carried out with 3 per cent. amalgam which passed through a 10-mesh sieve but not through a 16-mesh sieve. The solu-tions were contained in round-bottomed flasks of capacity two and a-half times the volume of the solution and kept a t a definite tem-perature.During a reduction a stream of carbon dioxide was led into the flask but was not- allowed to bubble through the liquid. These oonditions apply t o all the experiments described in this and the next section. 5-qycloHexanespirocyclopentan-3-one-l-carb~xylic Acid (XLIII, p. 348). Five grams of the ketonic acid (XXIII p. 335) were dissolved in an amount of sodium carbonate sufficient to give a neutral solu-tion and the whole made up t o 200 C.C. This solution was kept a t 14O by immersing it in cold water and reduced under the standar STABILITY OF SPIRO-COMPOUNDS. PART II. 373 conditions by adding 10 grams of amalgam once every half hour until 120 grams in all had been used. Half an hour after the addi-tion of the last of the. amalgam the mercurial layer was removed and the aqueous layer acidified.The oily precipitate was allowed to solidify and was then collected and rscrystallised from dilute alcohol. 'The acid melted a t 136-13So and was found to be very readily soluble in all usual organic solvents except light petroleum; in this solvent as in water it was sparingly soluble: 0.0867 gave 0.2145 CO and 0.0635 H,O. Cz67.49; H=8*14. Cl1HI6O3 requires C = 67.3 ; H= 8.2 per cent. CH(C0,H) p2 The Semicarbaaone C,H,,:C< CH,--CN*NH*CO*NH, The semicarbazone separated when a solution in which the acid and se-nicarbazide acetate had been boiled together was cooled. After recrystallising from alcohol it melted and decomposed a t 210°: 0.1079 gave 0.2262 CO and 0.0757 H20. C=57*17; H=7-80.C,,]EI&N3 requires C = 56.9 ; H = 7-5 pe;r cent. 5 -cycloHe xanespirocyclopen t an- 3- ol-1 - cap6 ox y lic A cid (XLIV , p. 348). Five grams of the above cyclopentanone acid were dissolved in a quantity of aqueous sodium carbonate sufficient to give an approxi-mately neutral solution which was made up to 400 C.C. This solu-tion was kept a t 17O and reduced under standard conditions with 240 grams of amalgam 10 grams being added every half hour. Half an hour after the addition of the last of the amalgam the mercurial layer was run off and the aqueous layer acidified and extracted with ether. The solid residue obtained after the ether had been dried and distilled off was recrystallsised from a mixture of benzene and light petroleum. The hydroxy-acid melted a t 125-127O and was very readily soluble in all the usual organic solvents except light petroleum.It was much more readily soluble in water than was the corresponding ketonic acid : 0.1092 gave 0.2672 CO and 0.0884 H,O. C = 66.93; H = 9.00. C,,HI8O3 requires C = 66.7 ; H = 9.1 per cent 374 INOOLD AND THORPE THE FOBMATION AN.D When the hydroxy-acid was boiled with acetyl chloride for four hours and the solution evaporated there was left a solid residue which was recrystallised from benzene. It melted a t 157-160° : 0.0887 gave 0.2122 CO and 0.0680 H,O. C = 65-24 ; H = 8.52. 0.1552 gavel acetic acid requiring 26.4 C.C. Ba(OH) solution Cl3H2,O4 requires C = 65.0 ; €1 = 8.3 per cent. (0.0243N) for neu tralisation. Cl3H,,O4 requires 26.6 C.C. (h) Comparative Experiments on the Reduction of the Bridged-rin,g- and Bridged-spiro-acids X I V (p.333) and X X I I I (p. 335). The general plan which was followed in these experimelnts has already been sketched in the Introduction (Section E). The most convenient method of preparing the bridged-ring acid XIV was found to be by treatsing the sodium compound IX (p. 330) according to the method (p. 364) used in the preparation of the acid XXIII from the sodium compound XII. The reductions of the bridged-ketonic-acids XIV and XXIII were carried out under t,he usual conditions (p. 372). The experiments were conducted in pairs using 0.80 gram of the acid XIV and 1-00 gram of the acid XXIII. The neutral solutions of the acids were immersed in the same water-bath and treated with 1 gram of amalgam every fifteen minutm 50 long as the experimeat lasted.The aqueous layers were then acidified and extracted quantitatively with pure ether. The solid residues obtained on evaporating the solvent were allowed to remain in an exhausted desiccator for f orty-eight hours. Usually about 0-4 or 0.5 gram was taken and the water formed by combustion deter-mined. I n certain cases the substance was also acetylated and the product left after evaporating desiccated over potassium hydroxide and quantitatively hydrolysed,' the acetic acid being distilled off in a current of steam and estimated by titration with standard alkali. This figure gave the quantity of hydroxy-acid which had been formed in the reduction whilst the water formed on combustion enabled one to calculate the total quantity of hydrogen which had been ir,troduced during the reduction.The accompanying table (table VII) gives the results of these experiments the figures within the brackets (. . . . . .) representing the calculated limits. Another set of experiments similar to the above was instituted, me dried products were then analysed STABILITY OF SPIRO-COMPOUNDS. PART 11. 375 in which the reduced ketonic-acids XXXIX (p. 346) and XLIII (p. 348) were used in place of the bridged-ketonic-acids XIV and XXIII. The quantities taken for each experiment were 0.80 gram of the acid XXXIX and 1.00 gram of the spiro-acid XLIII. The other quantities and conditions of experiment were the same as in the fo'rmer case. In this instance the products were not acetylated and hydrolysed but the water formed on combustion was deter-mined.The results of these experiments are given in table VIII, the figures in brackets (. . . . . .) representing as before the calculated limits. TABLE VII. Time Amalgam (hours). (grams). 0 4 8 2 4 16 6 24 10 40 14 56 (? (03 03 Time Amalgam (hours). (grams). 0 8 16 4 7 28 '2" (a 00 Bridged-ring-acid (0.80 gram). -A-+; e-u 8.- 0 a$ 2% 6.49 6.67 6.85 7.20 7.27 7-46 7.79 8-86 0.00 0 0.30 -0.60 -1.18 -1-30 0 1-62 3 2.16 21 4.00 100 TABLE VIII. Ring-acid (O-bO gram). - Per cent Atoms of H of H in product. introduced. 7.69 0.00 7-92 0.39 8.13 0.75 8-42 1.25 8.86 2.00 Bridged-spiro-acid (1.00 gram).2.2 5 p{ $3 7.22 7-49 7.64 7.97 8.09 8.30 9.09 -0.00 0) 0.58 -0.89 -1.68 -1-85 2 2.30 18 4.00 100) - -spiro-Acid (1.00 gram). - Per cent. 4 Atoms of H of H in product. introdmed. 8.39 0.49 8.51 0.75 8.67 1-10 8.16 0-00) 9.09 2.00) (i) Oxidation of Fissiovn-products Derived from the Bridged-spiro-sdo-ester X I 1 (p. 330). Oxidation experiments were tried with three types of fission-product including t'he spirocyclopentanone acid XLIII (p. 348), the transcyclohexylcyclobutane acid XXIV (p. 336) and the trams-lactonic acid of hydroxycyclohexylmethyltricarballylic acid XXXIV (p. 342). None of these apEeared to be1 acted on by boiling dilute nitric acid. Concentrated nitric acid however 376 INGOLD AND THORPE THE FORMATION AND readily oxidised the first of bhese three acids but did not react with the last two.These however were readily attacked by warm alkaline parmanganatei. Two crystalline oxidation products cyclo-hexanecarboxylic acid and cyclohexanel 1-dicarboxylic acid were obtained in tihe course of these experiments. cycloHexane-1 1-dicarboxylic Acid (XLV p. 351). This acid was prepared by oxidising both the sp’rocyclopentan-one acid XLIII (p. 348) or the corresponding s@rocycZopenhnol acid XLIV (p. 348) with concentrated nitric acid. Ths organic acid (2.5 grams) was warmed with an excess of concentrated nitric acid until most of the red fumes had been evolved. The resulting solution was &en boiled for a few minutes and finally evaporated to dryness.The residue was treated with water and again evaporated. I n this way a semi-solid mass was obtained from which the crystals were separated by spreading on porous porce-lain. After recrystallising from water the acid melted at 207O with the evolution of carbon dioxide and a certain amount of dis-doration : 0.1008 gave 0.2058 GO and 0.0650 H,O. 0.0412 required 19-75 C.C. Ba(OH) solution (0-0243R) for C=55-68; H=7-16. C,H,,O requires @= 55.8 ; H = 7.0 per cent. neutralisation. C8H,,04 (dibasic) requires 19.7 C.C. cyclolirexanecarb oxylic A cid (HexahydTobeneoic A cid). This acid was obtained in two ways: (1) B y distilling cyclohexarrzre-l 1-dicarboxylic acid. The di-carboxylic acid m distillation under ordinary pressure gave off carbon dioxide and yielded a distillate which boiled a t 230-235O and solidified when cooled by ice.The crystals melted a t 18-23O. (Found C= 65.42 ; H =9.51. Calc. C= 65.6 ; H =9.4 per cent.) (2) By 0xidisilz.g cyclohexyl derivatives. Either the trans-lactonic acid of hydroxycyclohexylmethyltricarballylic acid XXXIV (p. 342) or the trams-cycl~hexylcyctobutanone acid XXIV (p. 336) may be used. The organic acid (10 grams) was dissolved in an excess of a solution of sodium carbonate and treated with 30 grams of potassium permanganate. The permanganate was added in successive small quantities sufficient time being allowed between each addition for the solution to become decolorised the reaction being aided by heating. When all the permanganate ha STABILITY OF SPIRO-COMPOUNDS.PART 11. 377 been added and decolorised the liquid was filtered and extracted with ether to remove any neutral oxidation products. The aqueous solution was then acidified with hydrochloric acid and the add products were extracted with ether. From the residue obtained on drying and evaporating the solvent some gummy material was separated *by distilling under diminished pressure, and the liquid distillate was fractionally distilled under atmo-spheric pressure. I n this way there was obtained a fraction boil-ing at 232-236O which solidified on cooling in ice to a mass of crystals melting a t 21-25O. (Found C= 65.63 ; H = 9.51. Calc. C=65*6; H=9.4 per cent.) The adid prepared by both these methods melted a t a tempera-ture a few degrees lower than the recorded melting point namely, 29O (Lumsden T.1905 87 91). The same experience in regard to it is recorded by Hawmth and Perkin (T. 1894 65 103). It certainly appears to be exceedingly difficult to obtain preparations showing the correct melting point when working with small quanti-ties of material. Crystals' of the acid when left exposed to air rapidly liquefied and it did not appear to be possible to induce the acid t o separate in a crystalline form by cooling a solution in light petroleum below Oo. These observations agree preciwly with the statements made by Lumsden and along with the analytical figurw are regarded as leaving no doubt as to the identity of the substance. (j) Examination of the Oily By-product obtained in the Prepration of the Bridged-spirclsodio-ester XZI (p.330). The alcoholic solution of the oily by-product obtained during the preparation of the yellow sodium spiro-compound (p. 357) was distilled to remove the alcohol the residue dissolved in ether and shaken with water. The ethereal solution was then dried and evaporated the residue being distilled under diminished pressure. Three fractions were obtained (1) below 1ZOo/20 mm., (2) betweea this and 250°/11 mm. and 43) a t 250-160°/11 mm., together with a small dark-coloured non-volatile residue. Fraction (1) .-The first fraction contained practically the whole of the ethyl malonate present in the original oil. It was redis-tiIIed and the small residue of high boiling point was added to the second fraction (2). Fraction (2).-The second fraction which did not appear to be capable of separation into pure compounds by distillation was boiled with hydrochloric acid for forty-eight hours.During the reaotion a strong d o u r of butyric acid was developed. Whe 378 INQOLD AND THORPE THE FORMATION AND the hydrolysis was complete the butyric acid was distilled off in a current of steam and the residual solution boiled with charcoal, filtered and evaporated to dryness. Five acids were isolated from the crystalline residue. cycloHexame-1 1-diacetic A cid-The crystalline residue was recrystallised from the minimal quantity' of 35 per cent. alcohol. The crystals which separated contained thre,e of the five acids the separation from the other two which remained in the mother liquor being very nearly quantitative.The crystals melted at 1 60-200° approximately. They were dissolved in dilute ammonia and after the excess of ammonia had been evaporated were treated in the cold with a solution of zinc sulphate. The pre-cipitate was collecbd washed with cold water and then digested with hydrochloric acid. The acid thus precipitated melted a t 174-179O and was again precipitated as its zinc salt. The free acid was then liberated and recrystallised from water when it melted a t 18l0 and was identified as cycluhexane-1 1-dicarboxylic acid by direct comparison with a specimen of that substance. The preparation together with some small residues obtained subse-quently constituted 15 per cent. of the original crystalline mixture. trans-cyclolHexanespirocyclolpropane-1 2 d i ~ b o x y l i c A cid.The combined filtratles from the zinc salt of cyclohexane-1:l-diacetic acid were concentrated and treated with concentrated hydrochloric acid. The precipitated acids were dissolved in ammonia and after conceatrating the solution treated in the cold with a considkrable excess of a saturated solution of lead nitrate. The lead salts were collected and recrystallised from the minimal quantity of boiling water and then decomposed with a slight excess of dilute nitric acid. This treatment separated the last trace of cyclohexanediacetic acid which along with a small quantity of the tra~s-slrirocyczopropane acid remained in the filtrates from the lead salt. These traces were separated by means of their zinc salts as described above. The combined cycla-hexanediacetic acid-free preparations were extracted twice with three parts by weight of boiling benzene.The undissolved portion melted a t 234-237O and on recrystallisation from dilute alcohol melted sharply a t 237O. It was identified as cycbhexanesp-ro-cyclopropanedicarboxylic acid by direct comparison with a speci-men. The. preparation together with some residues subsequently obtained amounted to 65 per cent. of the original crystalline mixture STABILITY OF SPIRO-COMPOUNDS. PART 11. 379 3-cyclolHexanespiro-1-methylcyclopro~e-1 1 f 2-tricarrb oxyCic Acid (XLVII p. 354). The solut'ion in the 35 per cent. alcohol from which the above two dicarboxylic acids were crystallised (p. 378) was evaporated, and the residue resolved by crystallisation from 50 per cent.alcohol into three distinct fractions (1) a small quantity of a mixture of the same two dicarboxylic acids (2) a much more readily soluble crystalline substance melting a t about 208O and (3) a viscous gum. The first fraction was treated according to the methods already described for the separation of the two dicarboxylic acids. The second consisted essentially of the sp+ocyclopropanetricarb-oxylic acid. It represented about 8 per cent. of the original crystalline mixture and on recrystallising from water melted a t 215O with the immediate eliminatiun of water vapour : 0.1156 gave 0.2386 CO and 0.0675 H,O. 0.0453 required 21.95 C.C. Ba(OH) solution (0.0243,N) for C=56*30; H=6*49. CI2H,,O requires C = 56.2 ; H = 6.2 per cent. neutralisation.C,,H,,O (tribasic) requires 21 *9 C.C. co*o*co CH,*CO,H C-bH I c-co I The Anhydro-acid C1,H,,:C<&H.CO,H or C 5 ~ 0 W & a.co>O. The anhydro-acid was prepared by boiling the free tricarboxylic acid with acetyl chloride for two hours and evaporating the excess of the reagent. The residue on treating with benzene set to a mass of crystals. After recrystallising from the same solvent, the anhydro-acid melted a t 128O. It was immediately soluble in cold sodium hydrogen carbonate solution and could be recovered un-changed by acidifying. On boiling a solution of the anhydro-acid in aqueous sodium hydroxide for a few minutes and then acidify-ing the free tribasic acid was regenerated: 0.1548 gave 0.3418 CO and 0.0823 H,O. C=60.23; H=5.90. C12H,,0 requires @= 60.5 ; H =5.9 per centl.5-cycloHexanespiro-2-e thyldicyclope~tan-3-o~~e-~-cwb oxylic Acid (XLVIII p. 354). The two benzene filtrates obtained in the preparation of the spirocyclopropanedicarboxylic acid (p. 378) were combined and evaporated. The residue which still contained about three part 380 INGOLD AND THORPE THE FORMATION AND of the dicarboxylic acid to one of the bridged-monocarboxylic acid (this was indicated by a titration) was weighed and ext<racted with three times its weight of boiling benzene. The extract was evaporated and the residue treated again in this way the treat-ment being repeated until the residue obtained after evaporating an extract was completely soluble in three parts of boiling benzene. The product was fractionally crystallised from benzene.After a considerable number of fractional crystallisations a preparation was obtained which did not appear to change in melting point when again recrystallised. This substance separated from benzene in long needles which melted a t 191-194O. It was sparingly soluble in water and gave no colour with ferric chloride: 0.0610 gave 0.1537 CO and 0.0432 H,O. 0.0376 required 7.20 C.C. Ba(OH) solution (0.0243N) for The analytical figures especially those for the titration do not correspond with the formula so well as could be desired. The dis-crepancies are however all in a direction that may be taken to indicate that the preparation still contained a small quantity of the spirocyclopropanedicarboxylic acid which owing to the small-ness of the quantity of material it was not possible to rmove.The preparation constituted about 0.1 per cent. of the original crystalline mixture. cf=69.75; H=7.99. CI3Hl8O3 rquires C=70.3; H=8*1 per cent. neutsalisation. C,,H,,O (monobasic) requires 7.0 C.C. trans-lac tonic A cid of y-Hydroxy-&cyclohexyl-a-et h yl t r i c w b d l y l i c Acid (XLIX p. 354). The viscous gum which was obtained as the third fraction in the separation by means of 50 per cent. alcohol mentioned on p. 379 was distilled under diminished pressure. More than three-quarters of i t passed over at.240-250°/40 mm. and this on treatment with benzene solidified to a mass of colourless crystals. These represented about 3 per cent. of the original crystalline mixture the residue in the dist'illing flask which was not further examined accounting for about 1 per cent.The crystalline lactonic acid after recrystallisation from either water or benzene melted a t 149O. It dissolved readily in water and all the usual organic solvents except light petroleum: 0.1228 gave 0.2660 CO and 0.0962 H,O. The silver salt was precipitated by adding silver nitrate solution C=59*08; H=7*09. C,H,,O requires C = 59.2 ; H= 7.0 per cent. to a boiled solubion of the lactonic acid in ammonia STABILITY OF SPIRO-COMPOUNDS. PART 11. 381 0.1063 gave 0.0461 Ag. Ag=43*37. C,,13[,@,Ag2 requires Ag= 43.4 per cent. tran~y-~€ydroll;.~-~-cycl~~exyl-a-ethyltricarbally~~c Acid. This hydroxy-acid (contrast the cis-form) appeared t o be stable only in the form of its salts.All attempts to obtain it in the free state were unsuccessful. However when the tram-lactunic acid was dissolved in water and the solution titrated slowly with barium hydroxide in the cold an end-point corresponding with the neutral-isation of three carboxyl groups was obtained : 0.0516 required 22-40 C.C. Ba(OH) solution (0.0243N) for com-plete neutralisation. C14H2006 changing in solution to CI4HBO7 (tribasic) requires 22.4 C.C. The barium salt was precipitated from an aque'ous solution of the trans-lactonic acid by adding an excess of a solution of barium hydroxide : 0.1678 gave 0.1179 BaSO,. Ba=41*36. (C,,H,90,)2Ba requires Ba = 40.8 per cent. cis-Lactonic Acid of y-Hydroxy-/3-cyclohexyLa-ethyltricarbaUylic Acid (XLIX p. 354). The cis-lactonic acid was readily prepared by heating the free cis-hydmxy-acid (see below) a t looo for a few minutes.When dissolved in water the lactonic acid regenerated the hydroxy-acid which could either be crystallised out or titrated in solution. The lactonic acid could however be recrystallised from a mixture of benzene and light petroleum without any change taking place. It melted a t 198O with the evolution of water-vapoar : 0-1040 gave 0-2244 CO and 0.0664 H20. C,,H,,O requires C = 59.2 ; H = 7.0 per cent. The silver salt was prepared by dissolving either the cis-lactonic acid o r the cis-hydroxy-acid (see blow) in dilute ammonia boiling, and adding an excess of silver nitrate solution : C=58*84; H=7*10. 0-1041 gave 0.0449 Ag. Ag=43*13. On boiling the cis-lactonic acid with hydrochloric acid a partial C,4H,806Ag2 requires Ag = 43.4 per cent.conversion into the tra-rzs-lactonic acid took place 382 INGOLD AND THORPE THE FORMATION AND cis- y -H ydrox y-b-cycloh ex$-a- e t h yltricarb ally lic A cia?. This substance was prepared by boiling the cis-lactonic anhydride (see below) with a slight excess of aqueous sodium hydroxide and acidifying with hydrochloric acid. !C'he cis-hydroxy-acid slowly separated out. The1 substance is very unstable and loses a molecule of water with great readiness giving the cis-lactonic acid. The change was found to proceed with rapidity a t 60° in an exhausted desiccator a t the ordinary temperature and slowly when the acid was exposed t o air under ordinary conditions. I n cold aqueous solution the sub-stance bmehaved in every way as a tribasic acid : 0.1003 gave 0.2045 CO and 0.0666 H,O.C=55.60; H=7*38. 0.0382 required 16.65 C.C. Ba(OH) solution (0'0243fl) for The barium salt was precipitated when the hydroxy-acid was dis-solved in water and treated with an excess of barium hydroxide solution : C,,H,,O requires C = 55.6 ;< H = 7.3 per cent. neutralisation. C,,H,,O requires (tribasic) 16.6 C.C. 0.1057 gave 0.0740 BaSO,. Ba=41-18. When the cis-hydroxy-acid was boiled with 20 per cent. hydro-chloric acid f o r some hours a partial conversion into the trans-(C14H,90,),Ba3 requires Ba = 40.8 per cent. lactonic acid took place. T h' e cis- Lac t o iaic The trans-lactonic acid appreciable extent when co-0-co I Anhydride C,H,,:CH*C- &H I E~.~H*COW did not appear t o be dehydrated to any 'heated above its melting point or when distilled under diminished pressure. It readily eliminated water, however wheln heated a t looo f o r half an hour with acetyl chloride. On evaporating the product and treating the residue with benzene, crystals of the Iactonic anhydride were obtained. After being recrystallised from benzene the substanoe melted a t 154O. It did not dissolve ab an appreciable rate in cold aqueous sodium hydrogen carbonate but on warming with sodium hydroxide both oxygen rings were opened up and the cis-hydroxy-acid was formed: 0*1023 gave 0.2356 CO and 0.0625 H,O. C=62*81; H=6*79. Fraction (3).-The third fraction which was obtained on distilling C,,H,,O requires C = 63.2 ; H = 6.7 per cent STABILITY OF SPIRO-COMPOUNDS. PART 11. 383 the oily by-product (p. 377) boiled a t 250-260°/11 mm. and con-sisted of fairly pure ethyl cyclohexanespiromethylcyclopropanetetra-carboxylate (XI p. 330). When hydrolysed with hydrochloric acid it yielded the tricarboxylic acid (XLVII p. 354) dealt with on p. 379. (k) Note on Ethyl Dihydrogert) 5 5-;T)imethyldicyclopentm-3-me-1 2 4-tricarboxybate (XXVI p. 337). This member of the dimethyldicyclopentane series which was made use of in this research has not been previously described. The yellow sodium compound X (p. 330) was boiled with five times its weight of a '31;2T-solutioii of potassium hydroxide in alcohol until the suspended matter just became colourless. The precipitate was collected and drained on porous porcelain in a desiccator. It was then dissolved in the minimal quantity of water and decom-posed with excess of concentrated hydrochloric acid. The crystals obtained were recrystallised from 15 per cent. hydrochloric acid. The acid-ester melted and decomposed a t 162O and gave a purple colour with ferric chloride. It was readily converted into the mono-carboxylic acid XIV (p. 333) by heating for a few minute a t 200° with a little water. When heated in the dry state it evolved carbon dioxide and steam and from the charred residue a small amount of the acid XIV was isolated. It is perhaps worthy of note that the moiiocarboxylic acid should be obtained in this way but it is doubtless the result of the hydrolytic action of the steam which is evolved as the substance decomposes : 0.1261 gave 0.2453 CO and 0.0599 H,O. C=53*05; H=5.28. 0.1070 required 32.7 C.C. Ba(OH) solution (0.0243fl) for CI2Hl4O requires C = 53.3 ; H = 5.2 per cent. neutralisation. C,,H,,O (dibasic) requires 32.6 C.C. IMPERIAL COLLEGE OF SUIENOE RESEARCH LABORATORIES, CASSEL CYANIDE COMPANY LTD., [Received March 131h 1919.1 AND TECHNOLOGY, SOUTH KENSINOTON. GLASGOW