STRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS. 305XXXII1.-Strychnine, Berbeqine, and Allied A tkaloidsBy WILLIAM HENRY PERKIN, jun., and ROBERT ROBINSON.I.--St r y c hnine and Brucine.SEVERAL years ago (Trans., 1889, 55, 63; 1890, 57, 992) one of uscarried out a systematic study of the alkaloid berberine, andsucceeded in obtaining a series of degradation products, theinvestigafion of which made it possible to suggest it constitutionalformula for berberine, and this formula, except for slight modifica-tions in minor details, is still accepted as correct. The intention atthat time wslrs to use the experience obtained in order to attackthe problem of the constitutions of several other alkaloids, andafterwards to attempt their synthesis, and a series of experimentson cryptopine" (Proc., 1891, 7, 166) and on other opium alkaloidswere commenced.Owing, however, to the necessity for completing other investi-gations, these researches had, for the time, to be reluctantly putaside.I n recent years the difficult problem of the constitution of brazilinand haematoxylin made it imperative that we should very carefullystudy the constitution and nature of the groupings contained innatural products generally, and particularly in the alkaloids, and,in reviewing the work which had been published on strychnine,brucine, berberine, corydaline, and allied alkaloids, we were ledto certain conclusions as to their constitutions which we think maybe worth recording, and the validity of which we propose to test byexperiment. Although the alkaloids, strychnine and brucine, havebeen the subject of detailed investigations, especially at the handsof Tafel and Leuchs, very few deductions have been made as totheir constitutions.It is well known that strychnine, C21H220,N2,while containing two atoms of nitrogen, is only a mon-acid base,yielding salts, such as the hydrochloride, C2,H,202N,,HCI + l&H20,with only one equivalent of the acid. It is also known that thealkaloid is a tertiary base, and that it does not contain methoxy-groups. When strychnine is treated with alkalis (Loebisch andSchoop, Mowtsh., 1886, 7 , 75; Tafel, A n d e n , 1891, 264, 49), itsuffers hydrolysis with the addition of a molecule of water andformation of strychnic acid (and isostrychnic acid, p.317) :C 2 P 2 2 0 2 N 2 + H 2 O = C 2 P 2 4 O 3 N 2 .Strychnic acid is an imino-carboxylic acid, and a t the same time* Owing to the generosity of Messrs. T . and 13. Smith, of Edinburgh, who havesupplied me with considerable quantities of this very rare alkaloid, this investigatioiiis being continued.-W.H.P., jun.VOL. XCVII. 306 PERKIN AND ROBINSON :a tertiary base, since it yields metallic salts and, when treated withmethyl iodide, is converted into methylstrychnic acid methiodide,C2,H,,0N(MeI)(:NMe)-CO2H (Tafel, Annalen, 1891, 264, 59), andthe presence of the imino-group is further demonstrated by theformation of a nitrosamine, C,,H,,ON(:N*NO)*~,H, when theacid is treated with nitrous acid. These results, together with thefact that strychnic acid is readily converted into strychnine byheat, led Tafel to suggest that the relationship of strychnine tostrychnic acid is probably that represented by the scheme :co N i C,,H,,O<&and is similar to that existing,H,*YONH-COIt can scarcely be doubtedand the non-basic propertiesCO H = NiC,,H220<N~ ,between $-isatin and isatinic acid :A $!6H4.70EH2 CO,H 7that this representation is correct,of the grouping 0CO.N: affords anexplanation of the fact that strychnine, although it contains twonitrogen atoms, is only capable of combining with one equivalentof an acid.Further confirmation of the presence of the grouping=CO*N: in strychnine is obtained from the study of the productswhich are formed when the alkaloid is treated with variousreducing agents.When strychnine is reduced with phosphorusand hydriodic acid, it is converted into a substance, C21H2sON,,called desoxystrychnine (Tafel, Annden, 1892, 268, 245), and thissubstance, the importance of which is emphasised in the followingpages, evidently has the formula (A):tA.1 (R. )since, like strychnine itself, it is a mon-acid base, which, onhydrolysis, yields desoxystrychnic acid (B).This imino-acid has properties very similar to those of strychnicacid, and its whole behaviour indicates that the characteristicgroups N: and *CO*N: of strychnine have undergone no changeduring the reduction to desoxystrychnine. When desoxystrychnineis treated with sodium and alcohol, it yields strychnoline (I), the*CO*N: group becoming *CH2*N:, but electrolytic reduction pro-ceeds further and causes the addition of another two atoms ofhydrogen, and dihydrostrychnoline (11) results (Tafel, Annalen,1898, 301, 324 and 326):(1.) (11.)Lastly, when strychnine itself is reduced electrolytically, it yieldSTRYCHNINE, BERRERINE, AND ALLIED ALKALOIDS.307tetrahydrostrychnine (111), the group *CO*N: being reduced toCH,(OH)NH:, and this readily loses water with formation ofstrychnidine (IV) (Tafel, Annalen, 1898, 301, 303) :NiC,,(111.) (1V. 1One of the most characteristic properties of strychnine is theease with which it is nitrated, since warming with very dilutenitric acid (5 per cent.) converb it into dinitrostrychnine hydrate,C,,EI&O,N,(NO,),,H,O (Tafel, Annalen, 1898, 301, 299).This behaviour, and also the fact that strychnine is very readilysulphonated (Leuchs and Schneider, Ber., 1909, 42, 2681), maybe taken as proof of the presence of at least one benzene ring, andit will be shown later that all the evidence points to there beingonly one such ring in the molecule.There can, furthermore, belittle doubt that this benzene ring forms part of a quinolinenucleus, and, although no known quinoline derivative has so farbeen obtained from strychnine, there is ample indirect evidence ofthe presence of such a nucleus.Tafel (,4nnaZen, 1898, 301, 336) investigated the action first ofdilute and then of strong nitric acid on strychnine, and showedthat, under the conditions he employed, the alkaloid is nitratedand suffers degradation with the formation of dinitrostrychol-carboxylic acid, (NO,),C,H,N(OH),*CO,H, and this importantacid, when heated a t ZOOo, loses carbon dioxide, with the formationof dinitrostrychol, (N02)2CgH3N(OH),.There can be little doubt,as, indeed, Tafel has suggested, that this latter substance is adinitrodihydroxyquinoline, and we are at present engaged in itsinvestigation, not only with the object of proving this point, butalso, at the same time, of determining the relative positions of thesubstituent groups in the quinoline nucleus. Still more conclusiveevidence of the presence of the quinoline nucleus in strychnine hasbeen obtained in the following way. Strychnine combines withmethyl iodide to yield strychnine methiodide, C2,H220,N2,MeI,and this, on treatment with silver hydroxide, or barium hydroxide,yields methylstrychnine, and, since this substance shows all theproperties of a betaine, there can be little doubt that its formationis to be represented by the scheme:MeI:NIC,,H,,O<$O -+ Me0H:N iC,,H220<NH C0,H -+PoMeN i C20H220\NH4'Methy Zutrychnine.Methylstrychnine has all the properties ofVOL.XCVII.a secondary base; it308 PERKIN AND ROBINSON :yields, for example, a nitrosamine, and reacts with methyl iodideto form dimethylstrychnine :9 MeNiC H 022 “MeTafel (BnnaZen, 1891, 264, 43) has already pointed out howremarkably the properties of dimethylstrychnine resemble those ofLT-niethyltetrahydroquinoline :CH2/\/\CH, 1 1 ‘\ / \ P H 2NMe.and dimethylaniline. Like these substances, dimethylstrychnineyields a green nitroso-derivative, and condenses with benzaldehydein the presence of zinc chloride with the formation of the leuco-baseof a green colouring matter, which closely resembles malachite-green and the green colouring matter obtained from N-tetrahydro-quinoline under the same conditions ; furthermore, it combineswith diazobenzenesulphonic acid to yield a yellow azo-dye.Becauseof this behaviour, Tafel draws the conclusion, not only that strych-nine contains a quinoline nucleus, but also that the :NMe groupin dimethylstrychnine and the :NH group in methylstrychnine,and therefore the :N-CO- group in strychnine, is combined by onelinking direct to the benzene ring, and that strychnine must there-fore contain the grouping:AI I\/\N*CO*.The consideration of the properties of strychnidine leads toexactly the same conclusions.This substance is produced, asexplained on p. 307, by the reduction of the CO- group instrychnine to -CH,:NiC,,H2,0<~0 t o N iC,,H,,O<~Hzand this process converts a substance with the properties ofacetanilide into one which again exhibits exactly the behaviour ofdime th y laniline or N-me t h yl t e t rah ydroquinoline towards reagents.The most valuable evidence as to the internal structure of theother portion of the strychnine niolecule is obtained from theconsideration of the properties of an important monobasic acid,C,,H,,O,N,-CO,H, which is produced when strychnine is oxidisedby chromic acid (Hanssen, B e y ., 1884, 17, 2849; 1885, 18, 777and 1917; 1887, 20, 451). This acid is also obtained under thesame conditions from brucine, an alkaloid which contains twSTRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS. 309methoxy-groups and has properties so exactly similar to those ofstrychnine that there can be no doubt that it is dimethoxy-strychnine. Since the two methoxy-groups disappear during theformation of the acid C,,H,j0,N2*C0,H from brucine, it followsthat the benzene ring of the quinoline nucleus in the two alkaloids,brucine and strychnine, is broken down during the formation ofthis acid, evidently in the following manner :C CNow the acid C1,H,,O2N2,CO2H is a derivative of carbazole,because it yields this substance on distillation with zinc dust, andwe therefore arrive at the conclusion that the molecule of strychninemust contain the two residues :It has already been shown that the quinoline nucleus containsa benzene ring,-and therefore, in order-to account for the largenumber of hydrogen atoms in the strychnine molecule, it is neces-sary, as it appears to us, to conclude that not only the pyridinering of the quinoline nucleus, but also the carbazole section of themolecule must be almost completely reduced.Adopting a line ofargument which we have employed on previous occasions (Trans.,1890, 57, 1004; 1902, 81, 238; 1908, 93, 491), we find that thesefacts afford a basis on which it is possible to build up constitutionalformulae for strychnine and its derivatives which, we are convinced,must a t least be very near the truth.The two residues just figured contain C22, and as the formula ofstrychnine is C21H2202N2, it follows that these two residues mustbe fused together in such a way that one carbon atom a t least iscommon to both.It is clear that the basic nature of strychnine isnot due to the nitrogen atom of the tetrahydroquinoline nucleus,because of its union with the CO group; it must therefore be dueto the nitrogen atom of the carbazole residue. I f we now attemptto construct a formula for strychnine on the assumption that thetwo residues are united in such a way that one carbon atom iscommon to both, we have to remember that strychnine is a tertiarybase, and the nitrogen of the reduced carbazole residue musttherefore be rendered tertiary by union with a carbon atom of thequinoline nucleus, whilst the CO group unites with the carbazole31 0 PERKIN AND ROBINSON :nucleus.This would lead to formuke which clearly camnotrepresent the skeleton of strychnine.It s e h s to us that the only possible alternative is to assumethat, in the fusion of the two nuclei, two carbon atoms becomecommon to both,* and that the skeleton, which now contains C30,is completed by the introduction of an additional carbon atombetween the CO group and the basic nitrogen atom,+ yielding, inthe first place:/--\\-// \\-//\/\I 1\/\/ N*CO-CII,N<Further fusion may now take place in two ways, and two only,and leads us to the following alternative expressions for theskeleton of strychnine :/\/\I l l/-\/\/\-/ NC O N \ I \ co \/ \ /* If the two nuclei are fused together in such a way that three carbon atoms arecommon to both, then two additional carbon atoms must be introduced, and weobtain expressioiis such as the following for the skeleton of strychnine :/-\ ;-p 0\/'$'\c/\/\I l lIn such cases the carboxylic group in strychnic acid methiodide would be so farremoved from the iodine atom that the formation of a betaine would appear to be outof the question (p.307).P For, if not, then the only other possible way of uniting the :N*CO* group withthe rest of the molecule, in order to make the basic nitrogen atom tertiary, is theconstruction of a four-carbon ring, which is highly improbableSTRYCHNINE, BERBERIKE, AND ALLIED ALKALOIDS, 31 1Formula I contains a seven-membered ring, and it is veryunlikely that such a ring would be readily produced by theelimination of water from strychnic acid ; on the other hand, suchelimination leading to the formation of a six-membered ring isquite usual, and for this and other reaso-ns we prefer formula 11.If this formula be accepted ils the skeleton of strychnine, therecan be little doubt that desoxystrychnine, C2,H2,0N,, the productof the reduction of strychnine with phosphorus and hydriodic acid(p.306), must be written:Desoz ystr ychniii e.and this formula is in complete agreement with the properties ofthis substance.It exhibits desoxystrychnine as a derivative of tetra-hydroquinoline, accounts for its conversion into desoxystrychnicacid, and for the behaviour of this acid, as well as of desoxy-strychnine itself, towards methyl iodide, nitrous acid, etc.Similarly, the highest product of the hydrogenation of strychnine,namely, dihydrostrychnoline, C,,H2*N2 (p. 306), will be representedby the formula :CH, CH2/\/\CH /\CH, 1 I bH, bH CH2 \/\/ \/ \/\F H YH YH2CH CH,, CH, N-- rBih ydrostrychmline.and the fact that further hydrogenation, without reduction of thebenzene ring, is impossible is well indicated by this structure.On the basis of skeleton formula, 11, it is possible not only toexplain the recent important results of Leuchs (Ber., 1908, 41,1711), Leuchs and Schneider (ibid., 4393; 1909, 42, 2494), andLeuchs and Weber (Ber., 1909, 42, 3703), but also to deduce aconstitutional formula for strychnine which cannot be very farfrom the truth312 PERKIN AND ROBINSON :Leuchs and his collaborators find that strychnine, on oxidationin acetone solution with permanganate, is converted into a keto-dibasic acid, strychninonic acid, which has the composition,NiC17Hls( :N*CO)(CO)(CO,H),.On reduction with sodium amalgam, this acid yields the corre-sponding secondary alcohol dibasic acid, strychninolic acid,Ni ClfHI8( :N*CO)(CH*OH)(CO,H),, and this substance, on treat-ment with dilute potassium hydroxide, is decomposed, yieldingglycollic acid and strychninolone, C19HlS03N2, a substance whichpossesses neither acid nor basic properties.An exactly similarseries of products was also obtained from brucine under the sameconditions.There can be no doubt, that the formulze for strychninonic andstrychninolic acids suggested by Leuchs, which represent theseacids as containing a tertiary nitrogen atom, are correct, and theabsence of basic properties is amply explained by the presence ofthe two carboxyl groups. When strychninolic acid loses glycollicacid under the influence of alkali, the substance produced,strychninolone, is devoid of both acid and basic properties, andit is therefore obvious tha8t, during its formation, mutualneutralisation has taken place between the two carboxyl groupsand basic nitrogen atoms.Since two such atoms are required forthis process, it follows that the section :N-CO*CH,*N$: of theoriginal strychnine skeleton must have lost the grouping *CO*CH,*,and this is therefore the source of the glycollic acid::N*CO*CH,*N: + 2HzO = :NH -+ C?O,H*CH,*OH + HN:.The cause of the non-basic nature of one of the nitrogen atomsin strychnine is thus removed, and the two :NH groups are nowboth basic and free to combine with two carboxyl groups to yielda neutral substance, such as strychninolone.The problem which still remains is to discover the process bywhich the structure : c c\/Ccan be broken up in order to supply two carboxyl groups in suchpositions that they can combine with the two :NH groups toyield two neutral :N*CO* groupings.The following appear to bethe only two ways in which this can be satisfactorily accomplished STRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS. 313&'\/\C/ YO,H I I ' \/\F\P \P\ N H Q $! $! -+c:C CO,Hc : ? NHC0,H.C--C\/CICO,H\ -3NH 7 $!02HF)NH-c\/CCc c (7 cI f now the hydrogen atoms and the carboxyl and other groupsin strychninonic acid, NiC,:H,,(:N*CO)(CO)(CO,HZ,, are filled intothe first formula of scheme I, it will be found that the completeexpression must contain one double linking, whereas, if the sa'meprocess is carried out in the case of scheme 11, two double linkingswill be found to be necessary. This difference is due to the factthat, in developing scheme I from the skeleton formula of strych-nine, one closed ring suffers fission, whereas in the case of scheme I1two closed rings suffer disruption314 PERKIN AND ROBINSON :I f we consider, in the first place, the development of a formulafor strychninonic acid on the basis of the first skeleton formulaof scheme 11, we arrive at expressions of the types:CH CH,' I ' I\/\/c\/co y 7 p 2 HC 0 N-CH*[ CH2],* C0,H\/CH2CH, CH2/\/\ /\ I I FH YH2and CHCH :CH*C02HIt can be shown in several ways that such formulae cannotrepresent strychninonic acid, and the same is true of any otherformulae built up on the basis of the first skeleton formula ofscheme 11, but only one reason for this need be given here.Strychninonic acid is obtained from strychnine by a simple per-manganate oxidation, and it must therefore be possible to recon-struct the formula of strychnine in a comparatively simple mannerfrom that of strychninonic acid, I f we select formula I, andattempt to construct from it a formula for strychnine, we arriveat the expression:C!H CH,/\/\c/\c"I I 6 hH2CH2 \/\/ \/ .\ / \ 7 Y*OHYH,CO N-CH CH,which cannot be correct, because oxida'tion with permangitnatewould attack the double linking combined with the benzene ring,and it would therefore be impossible to obtain from this expressionthe formula for strychninonic acid from which it was derived, andthe same line of argument applies equally to all other formulaederived from the first skeleton formula of scheme 11.Whilst, then, scheme I1 is out of the question, careful con-sideration shows that scheme I leads t o the following expressions forstrychninonic acid and strychnine STRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS.315g CH, CO,H CH, C H ,/ \ (‘(b’. I CO,H CH I CH, CH I CH2fYH $932.C O N--CH CH, CO N--CH CH,d\/\A/ \ / \ 7 7H YH,\/ \/CH, CH*OH\/ ‘\\/CH, COStrychrtinonic acid.cStrychnine.which seem to us to be in every way satisfactory, since theyaccount in a comparatively simple manner for all the knownreactions of these substances. From the several possibilities wewere led to select the positions assigned to the double linkings inthe strychnine formula for various reasons, and of these the follow-ing need only be discussed. I f the formula assigned to strych-ninonic acid is correct, the formation of this acid by the oxidationof strychnine clearly points to one of the double linkings occupyingt.he position marked a.The position b was selected for the seconddouble linking, because this linking must be situated in a, stableportion of the molecule, otherwise it would also suffer oxidationunder the conditions employed in the preparation of strychninonicacid, and experience of heterocyclic systems has shown that thedouble linkings in nuclei, such m :/\ -y 7- /\ -y y- or co N- OH-C N-are not readily oxidised. St.rychninonic acid is a keto-dibasicacid, and there can be little doubt that the keto-group is producedby the oxidation of a secondary alcohol group; in other words,the molecule of strychnine contains a :CR(OH) group (compareLeuchs, Ber., 1908, 41, 1711), and the difficult problem remainingto be solved is that of assigning the correct position to this group.It cannot occupy the position 9, because this would makestrychninonic acid a derivative of benzoylacetic acid, and f and eseem also t o be out of the question, for the reason that hydroxy-groups in these positions would be y- and 6- to one of the carboxylgroups in strychninonic acid, md this acid does not appear tohave any tendency to yield a, lactone.So far no definite experi-mental evidence is available to enable us to select with anycertainty either of the remaining positions d or e, but we hav316 PERKIN AND ROBINSON :chosen c as the rbsult of a comparison of strychnine with quinineand other natural products in which a similar grouping occurs.This argument, based on lactone formation, might appear toapply also to a hydroxyl group in the position c, but theexamination of a model shows that a hydroxyl group in thisposition is too far removed from the carboxyl group to makelactone formation probable.The formula which we have suggested for strychninonic acidleads to the following expressions for strychninolic acid andstrychninolone :CH, CO,HI I FH bHCH,/\A/ CO,H\/\P\/\A\/ \/7 FH SH,CO N-CH,CH,CH, CH-OHandCH2J’lrychninolic mid. Strychninolone.and these also appear to agree in a satisfactory manner with theproperties of these substances so far as they have been investigated.A possible objection to the formula assigned to strychninoloneis the stability of this substance towards oxidising agents, whichmay not be considered compatible with the presence of the :CH*OHgroup, but, in our opinion, this argument does not carry muchweight.It has already been pointed out (p.309) that brucine is di-methoxystrychnine, and the positions of the methoxy-groups inthe benzene nucleus seem to be fixed by the observation of Leuchsand Weber (Ber., 1909, 42, 3709) that brucinolone is readilyoxidised by nitric acid, with elimination of the methoxy-groups andformation of a quinone which crystallises in red needles, andyields a q u i d on reduction with sulphurous acid. Since ano-quinone would hardly be produced under these conditions, it isprobable that f i e substance is a p-quinone of the constitution :CH*OH8 $: YHC;H,CO-CH-CH CH2\/CHSTRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS. 317and its formation is a strong indication that the methoxy-groupsoccupy the position assigned to them in the fallowing formula forbrucine :Me0 CH,CHC CH CH,\/\A/\/\M e O r f: QH YH,CON-CH CH,Brmine.One other very interesting point in connexion with strychnineand brucine which we have already mentioned (p.305) is thebehaviour of these alkaloids on hydrolysis. When strychnine isdigested with sodium methoxide or barium hydroxide, it yieldsstrychnic and isostrychnic acids :N ~ c , , H , , o ~ ~ ~ -+ NX,,H,,O<~; CO H ,and these isomeric acids are converted by the action of heat intothe isomeric alkaloids , strychnine and isostrychnine.Brucineexhib?ts an exactly similar behaviour. Strychnic and isostrychnicacids resemble each other so closely in all their reactions that itmight at first sight appear that they were simply stereoisomericmodifications of the same substance. The conversion of these acidsinto the isomeric strychnines is, however, scarcely in accordancewith this view, and it is far more probable that the isomerism isdue to a difference in the position of one of the double linkings inthe molecule, probably in the sense represented by the formulze:CH, CH/\/\AC CH CH, \A/ \/\/\N H f.’H 5!H2C0,HN-CH CH,\/’ \/CH, CH*OHCH, CHCO,H &-CH, CH,CH, CH*OH\/ \/S’tyycknic acid. so Sty y c h ic nc id.If the formulae suggested for strychnine a nd brucine be examinedwith the view of discovering some reason for the extremely poisonou318 PERKIN AND ROBINSON :nature of these alkaloids, it would seem that the only section towhich this property can be ascribed is the grouping (C) :t:/\ -7 7-co N-(C. 1 (D.)containing the two nitrogen atoms.Schotten (Ber., 1888, 21,2244) has called attention to the fact that a-ketopiperidine(a-piperidone) (D) has poisonous properties of the same kind aathose exhibited by strychnine and brucine, and it is not a t allimprobable that the introduction of the second nitrogen atominto this molecule may have the effect of much intensifying thesepoisonous properties.We are at present engaged in synthesising substances containingthe above di-nitrogen group, and propose to have these examinedin order to find out whether they have poisonous properties similarto those of strychnine and brucine.11.-Berb erine, Corydatine, and A Zlied AIlcaloids.The constitution of berberine, C2,,H1,0,N, is generally acceptedas being represented by the formula:0-CH,and this formula is based on the investigation of the long seriesof products which result from the degradation of the molecule byoxidation with permanganate (Trans., 1889, 55, 63; 1890, 57,992).The position of the methylenedioxy-group * is determined by thefact that hydrastic acid (I) and o-aminoethylpiperonylcarboxylicanhydride (11) are found among the products of this oxidation :C0,H/)c)>cH2 Co"a(/ TR-COf)O>CH, CH,*CH,\/O ()CO,HMe0M~O/\CO,H(1.1 (11.1 (111.)* It is unfortunate that this group should have been wrongly placed in theoriginal papers.-W.H. P., junSTRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS. 319On the other hand, the fact that hemipinic acid (111) is producedin considerable quantities during the oxidation of berberine doesnot definitely fix the positions of the metho,xy-groups, since theformula for berberine :0-CH, /\A CH I Iwould also account for the formation of hemipinic acid onoxidation.On carefully considering this matter, it appeared to us that theonly ox7dation product of berberine which is able to afford definiteinformation as to the position of these methoxy-groups is berberal(Trans., 1890, 57, 1000 and 1062). This substance, on hydrolysis,yields $-opianic acid and o-aminoethylpiperonylcarboxylic an-hydride, and conversely it may be synthesised by simply heating$-opian ic acid and o -aminoe t hy lp ip er on y 1 car b o x y li c anhydride a t180O. When this experiment was described (Zoc.cit., p. 1079),this important synthesis was assumed to take place according tothe equation :and it was suggested (p. 1002) that the constitution of berberalmust be represented thus:Me0M~O/\CHO 1 ICO~N~CH,~CH) ,-- cq;>c€€,.\/ \/The formula, for berberine itself, given at the commencementof this section, was largely based on this constitutional formula forberberal. $-Opianic acid was first obtained as the result of theseexperiments on the oxidation of berberine, and, as it is difficultto prepare in any quantity by this process, and no other methodof preparation has yet been discovered, the mechanism of itscondensation with basic substances has not been investigated indetail.If, however, the formule of $-opianic acid is compared withthat of opianic acid, it will be seen that they are both o-aldehydo320 PERKIN AND ROBINSON :acids, and differ only in the positions of these groups relative tcthe methoxy-groups :Me0 Me0MeO/\CHO MeO/\CO,H!,)CO,H (,!CHO '+-Ottiaiiic acid.Opianic acid.and as this is the only difference in constitution, it may be safelyassumed that they will behave in an exactly s i m h r manner whentaking part in reactions characteristic of o-aldehydo-acids.Important evidence in support of this view has already beenobtained, since it was shown in the earlier papers (Trans., 1890,57, 1081) that opianic acid condenses with w-aininoethylpiperonyl-carboxylic anhydride to yield a substance :which has properties exactly similar to those of berberal, and wastherefore named isoberberal.During recent years opianic acid hasbeen the subject of detailed investigation, and the results whichhave been obtained necessitate a modification of our views as tothe constitution of isoberberal, and conseque.ntly of berberal andof berberine itself. Liebermann (Ber., 1886, 19, 2284; 1896, 29,175) has shown that opianic acid reacts with aniline in the coldto yield anilino-opianic acid, and expressed the opinion that,during this process, the opianic acid reacts as a derivative ofhydroxyphthalide in accordance with the equation :Me0 Me0 CO'M ()/\A + NH2*C,H, = 1 I 0\/\/CH *NH*C,H, V\PCH*OHand from the study of this and many other similar condensations,it is clear that, in condensations of opianic acid with basic sub-stances, it is always the carbon atom of the aldehyde group whichis attached to the nitrogen atom in the final product.There canbe little doubt that a similar process takes place when opianicacid reacts with w-aminoethylpiperonylcarboxylic anhydride to yieldisoberberal, and the constitution of this substance is therefore notthat originally assigned to it (p. 319), but must be modified to:Me0 COMeO(\/ '0 ,COf)O>c.,yCH*N*CH,*CH,\ A / / V0woBerbera1.STRYCHNINE, BERRERINE, AND ALLIED ALKALOIDS. 321Furthermore, since +-opianic acid is so exactly similarly con-stituted to opianic acid, and exhibits in all its reactions so closean analogy with this acid, it cannot be doubted that, when itcondenses with basic substances, it also reacts in it similar manner.It follows, therefore, that in the formation of berberal by thecondensation of +-opianic acid with o-aminoethylpiperonylcarboxylicanhydride, the aldehydic carbon atom of the +-opianic acid becomescombined with the nitrogen atom, and the constitution of berberalmust therefore be represented by the formula:Kerberal.and not by that originally assigned to it (p. 319).*This new expression is in complete agreement with the propertiesof berberal, and its acceptance involves the alteration of theposition of the methoxy-groups in the old berberine formula(p.318), so that the .constitution of the alkaloid must now bewritten (compare p. 319):O-CH,Berberine.* The actual mechanism of the condensation of opianic acid or +-opianic acid withbasic substances is probably not so simple as that suggested by Lieberrnann (loc.cit.). Whes the syntheses of berberal and isoberberal were described (Trans., 1890,57, 1080 and 1082), it was proved that in both cases the first step is the formationof the salt of the acid with the base. Thus, for example, +-opianic acid combineswith w-aminoethylpiperonylcarboxylic anhydride to form the salt,/CHO CO-- 0( Me0)2C6H2 I \C6H2' b H , ,\CO*O*NH2*CH2* CH2/ \O/which, when heated at 180°, yields berberal.Most probably aldoltakes place during this change, followed by rearrangement according,GO-0 co---\CII(OH)*NH*CH,*CH,( Me0)2C6H'2' \/ >CC.~<,">CH~formation firstt o the scheme :\C H /"\ co co---CH-N*CH,*CH~'(MeO)&6H2< >o I 6 Z\O/CHP.The synthesis of isoberberal takes place in an exactly similar manner, and it isvery probable that all such condensations between opianic acid or +opianic acid andbasic substances proceed on similar lines322 STRYCHNINE, BERBERINE, AND ALLIED ALKALOIDS.The formation of berberal by the oxidation of berberine is noweasy to understand, and evidently takes place according to thescheme :O--QH,/\. 0I /0-CH, /\A I 1/\/,\/CO$ / I /\/"02" co -+Me01\/\cH(oH)-N CH,\/ Me0 I \/'(-3%l f e d \/\CH=N ' ;i' CH,Me0CH2(OH)0-y H 2/\aI I -+ / \// \ / ? O Y 0 1M"o\/\ I /\/I I 0 N CH,Me0 CH CH,The salts of berberine are derived from the hydroxyl formulagiven above (compare J.Gadamer, Arch. Pharm8., 1905, 243, 31),but there is reason t o believe that the alkaloid itself exists in thecorresponding aldehyde modification. Gadamer (Chem. Zed., 1902,26, 291) has shown that, when berberine sulphate is treated withbarium hydroxide, it yields a brownish-red, strongly alkaline solu-tion, which doubtless contains the hydroxy-modification of berberine.If, however, this solution is mixed with excess of sodium hydroxide,a yellow modification of berberine is obtained, which apparentlyhas the properties of an aldehyde, and Gadamer has named thismodification b erb erinal. The constitution of this modification,based on the new formula for berberine which we have proposed,will be the following : -0-QH,/)OCH 1 /\/\A/ Meou NHCH, ? IMe0 C H O W(3%Berberinal.Berberinal yields an oxime, reacts with magnesium alkyl iodideswith formation of homologues of berberine (Freund and Beck, Ber.DECOMPOSITION OF DIMERCURAMMONIUM NITRITE HP HEAT, 3.231904, 37, 3336 and 4673), and, when treated with a large excessof alkali, is reduced t o dihydroberberine and simultaneouslyoxidised to hydroxyberberine, the aldehyde grouping being con-verted into -CH,(OH) and -CO,H in the manner chmacteristic ofaromatic aldehydes.The proposed modification of the positions of the methoxy-groupsin the berberine formula receives further confirmation from theconsideration that berberine occurs along with hydrastine inhydrastis canadensis, and the close relationship between thesealkaloids becomes very striking if the new formula for berberinalis placed by the side of that of hydrastine:0-CH, /\d CH I I0-CH, /\oBerberinal. Hydrastine.These alkaloids are, indeed, so closely related as to suggest thathydrastine is either formed in the plant from berberine, or thatthey are both derived from some common parent.THE UNIVERSITY,MANCHESTER