POLYMEBXSATION OF ~43LUCOSAN. 2903 CCCC1V.-Polymerisation of i3-Glucosan. The Cons-titution of Synthetic Dextrins. By JAMES COLQUHOUN IRVINE and JOHN WALTER HYDE OLDEAM. ONE method of appzoach to the constitutional problems of the polysaccharides is through the study of anhydro-hexoses and the discovery by Pictet and his collaborators that p-glucosan can be p r e p d conveniently by the dry distillation of starch acquires a special importance in this connexion. 1 6-Anhydroglucose has in this way been rendered available in quantity and ifs properties have been examined in detail by Pictet and others. One of the most striking of the new observations is that the compound is readily polymerised the reaction n(C6H1005) + (C6H1,05) pro-ceeding readily when p-glucosan is heated either alone or in the presence of catalysts.Thus when fused with platinum black, glucosan is converted into an amorphous powder to which the formula (C6HI0O,) applies and this displays the general properties of a dextrin yielding glucose on hydrolysis (Pictet Edw. Chim. A& 1918 1 226). Pictet finding platinum black uncertain in its action improved the method by using zinc chloride as a catalyst (&bid. 1921 4 788) and he also varied the procedure by con-ducting the polymerisation under both reduced and increased pressure. Four definite compounds were obtained under these conditions : Polymeride. C.1,. PIWSW0. 1. Diglucomn ........................... + 28.2' 16 mm. 4. Octaglucosan ........................ + 72.8' 13-2 ,) 2. Tetraglucosan .....................+ 11 1.9' 1 atmos. 3. Hexaglucosan + 94.1' 4.6 1 ) ..................... Although there seems no theoretical limit to the number or variety of possible polymerides the above list includes only compounds in which glucosan molecules may be regarded as having become associated in multiples of two. The mode of affachment of the parent molecules has hitherto save in one case remained obscure, but Pictet has recorded the abnormality that the polymerides yield only diacetates or dibenzoates in place of the tri-derivatives to be expected. By arrangement with Professor Pictet we have been engaged on the constitutional study of the polyglucosans and take this opportunity of expressing our thanks for his courteous permission td extend his work. The completion of the investigation which was commenced four years ago has been delayed in consequence of the complexity of the results and in the meantime Pringshei dm IRVINE AND OLDHAM POLYME~USATION OF ~-GLUCOSAN.has described the application of the methylation process to tetra-glucosan (Ber. 1922 55 3001). No difficulty was apparently experienced by him in obtaining complete alkylation and on hydrolysis of the product both tetramethyl and dimethyl glucose were obtained. This is a striking result but a,s shown by Ring-sheim it is in itself insufficient to discriminate between two a,lternative structures for tetraglucosan. In order to obtain an adequate view of the mechanism of the polymerisation of glucosan it is necessary to study the constitution of a variety of polymerides displaying a progressive increase in molecular complexity and this we have accomplished.In repeat-ing Pictet's experiments little success was attained in using platinum black as a catalyst and after numerous attempts to find a superior reagent we adopted the method of heating glucosan a t 250" in the presence of zinc dust. The polymerisation was conducted in a vacuum the residual air having been washed out with hydrogen and under these conditions the change took place without charring or alteration in weight.* As the zinc dust employed contained a trace of chloride it is possible that the latter is the functional catalyst as we find that zinc chloride exercises a powerful poly-merising effect on glucosan and its derivatives the reaction in some instances being violent.For many reasons we prefer the use of metallic zinc and although polymerides differing from those described by Pictet are produced we have continued to employ the process as it proved satisfactory for large-scale working and gave uniform results. It may be remarked however that the relative yield of the different dextrins is affected not only by the catalyst but also by the temperature by the duration of heating, and by the scale of working. In order to obtain polymerides of high molecular weight it is unnecessary in our experience to work under positive pressures, and all our preparations were carried out a t 15 mm. By means of fractional precipitation from aqueous solution the polymerides were separated into three main fractions which in order of increas-ing solubility showed the following progressive diminution in * It may be mentioned that glucosan is much less stable at high tem-peratures particularly in the presence of acids than the method of preparing the compound would suggest.Nevertheless Venn ( J . Text. Id. 1924 15T, 414) having found that the yield of glucosan from cellulose is greatest when the acidity of the distillate is lowest states that this result is opposed to our views as to the origin of the hexosaa. It would have been surprising if any other result had been obtained and Venn's observation which amounts to no more than the statement that the best yields of glucosan are obtained under the most favourable experimental conditions has no bearing on the mechanism of tlie reactions in which gluoosan is fonnocl THE CONSTITUTION OF SYNTHETIC DEXTIUNS.2906 specific rotation. Dextrin I + 83-9"; Dextrin 11 + 60.7" ; Dex-trin 111 + 29.9". In marked distinction to the products described by Pictet all the dextrins yielded a definite triacetate and the presence of three hydroxyl groups per C unit was confirmed by methylation. The convenient solubilities of these trimethyl dexfrins rendered possible the determination of their molecular weights in benzene solution and in this way the parent compounds were characterid as containing respectively seven four and three glucosan units. Trimethyl Dextrin I 1446 1428 Heptaglucosen Methylated product. Mol. wt. (found). Mol. wt. (calc.). Parent compound. 9 ) 9 9 II 874 816 Tetraglucoean Y Y 3 9 III 595 612 Triglucosan This does not exhaust the list of polyglucosam as we have also obtained other isomerides and it is evident that both odd and even numbers of glucosan molecules are capable of forming poly-merides.Further it is clear that the tetraglucosan examined in the course of the present investigation is Merent from that described by Pictet and subsequently examined by Pringsheim. The distinction is shown in the specific rotations of the compounds, in the m. p. of the triacefates prepared from them and most emphatically in the different behaviour of their methylated deriv-atives on hydrolysis. The methylated dextrins which may now be termed hepta-, tetra- and tri- (trimethyl glucosan) respectively were examined so as to give an insight into the mode of attachment of the constituent hexose chains the information being derived by identifying the methylated glucoses formed on hydrolysing each compound.In order to indicate the significance of these results it may be recalled that although it is cwtomary to distribute the hydroxyl groups equally among the C units of a polysaccharide or allied compound, this allocation is based on an assumption and should be subjected to experimental test in each case. Thus cellulose and hexa-amylose (Irvine Pringsheim and Macdonald J. 1924 125 a) which are isomeric with the dextrins now under consideration give tri-methyl derivatives and these in turn yield on hydrolysis 2 3 6-tri-methyl glucose and no other sugar. It follows that the hydmxyl groups in the parent compounds are uniformly &tributed Le., each C unit carries three hydroxyh arranged in the same positions.Were this not the case isomeric trimethyl glucoses would be formed or alternatively a mixture of methylated sugars such as tetramethyl and dimethyl glucoses giving the same average composition. In marked contrast to natural polysaccharides the syntheti 2906 II(vME AND OLDEAM POLYMERISATION OF p-OLUCOSAJY. dextrins afford rstriking examples of compounds constituted on an entirely Merent model in that the hydroxyl groups are not attached uniformly to the individual C units. We have already shown (J. 1921 119 1744) that when P-glucosan is subjected to consecutive methylation and hydrolysis it gives 2 3 5-trimethyl glucose and on the basis of analogy it might reasonably be expected that the same end-product would be obtained from a polymerised glucosan.Such is not the case. Each trimethyl dextrin was converted into the corresponding methylated methylglucosides and thereaffer into the constituent sugars but the product in place of being homogeneous consisted in each case of di- tri- and tetra-methyl glucose mixed in proportions which gave the analytical mes required for a trimethyl glucose alone. The individual sugars were separated and two of them were characterised as 2 3 5 6-tetramethyl glucose and 2 3 $trimethyl glucose but the constitution of the dimethyl sugar is still uncertain and two alternatives are possible. Although in this section of our work we did not conduct the separation of the above sugars on quantita-tive lines good reasons exist for the belief that hepta- tetra- and tri-(trimethyl glucosan) all give the same methylated glucoses as the phyaical constants of each mixture were identical.It is necessary to emphasise that the tetramethyl glucose isolated in these experiments is a genuine scission product of the methylated dextrins and does not originate in any molecular cleavage during the methylation process. This possibility wits carefully excluded, and the result in itself disposes of the idea that the polymerisation of glucosan is merely the union of intact molecules in pairs. The process is evidently complex and consists essentially in the form-ation of glucosidoglucosides which show a general structural resemblance with the constitutional type.ascribed by Hess to cellulose. The development of structural formulae for the poly-glacosans demands however a knowledge of the relative propor-tions of the different methylated glucoses into which they can be transformed. A considerable advance was made by conducting all the operations from the polymerisation of the glucosan to the separation of the methylated glucoses finally obtained on strictly quantitative lines. For this purpose large quantities of material were required and a substantial simplification was effected by methylating the total polymerised product and separating the isomerides by vacuum distillation without isolation of the parent compounds. The first fraction consisted of monomeric trimethyl glucosan whilst the second was a viscous syrup which was shown to be di (trimethyl glucosan).The remaining syrup which con-stituted the largest fraction was practically non-volatile a t 2OO0/0. !CHE CONSTITDTION OF SYNTHETIC DEXTRZNS. 2907 mm. and on cooling it solidified to a clear red glass. As fhis product contained all polymerides higher than the dimeride it is termed poly (trimethyl glucosan). Di(trimethy1 glucosrtn) was converted into the methylglucosides of the constituent sugars the products being formed in the pro-portions shown below Yield %. Calc. for Found. equal mols. flDimethyl methylglucoside. 46-6 47.1 Di(trimethy1 glucosan) Tetramethyl methylglucoside. 53.4 52.9 These proportions were confirmed from the yields of the corre-sponding sugars when as before 2 3 5 6-tetramethyl glucose was isolated together with a dimethyl glucose.An equally sigdicant result was obtained by similar treatment of poly(trimethy1 glucosan) three glucosidic products being obtained in place of two. Yield %. Calc. for Found. equal mob. f 2 3 5 6-Tetramethyl 35-5 35-3 Poly(trimethy1 glucosan)+ 2 3 5-Trimethyl methyl- 33.5 33.3 ,/ methylglucoside. \ glucmide. \ Dimethyl methylglucoside. 26.0 Monomethyl methylglucoside. 5.0) 31'4 As shown by the analytical figures quoted in the experimental part the small amount of monomethyl methylglucoside is attribut -able to incomplete methylation and may be added to the yield of the higher homologue. Discussion is simplified by tabulating the significant facts show-ing the mutual relationships between glucosan and the polymerides now described.[LID of [alv of Mol. wt. of parent cp compound. trimethyl derivative. (water) (MeOH) p-Glucoaan ... -65.4" -53.2" 212 (204) Diglucosan ... - +48.3 418 (408) Triglucosan ... +29-9 +52% 595 (612) Tetraglucosan +60-7 f-68.7 873 (816) Heptaglucosan +85-8 +89.4 1446 (1428) Pol yglucosan - I -yo Yethoxyl content of derived Methylated sugars glucosides. produced. 50.9 51.2 - glucose 2 3 5-Trimethyl glucose. 2 3 5 &Tetramethyl 1. Dimethyl'glucose. ( 2 3 5 &Tetramethyl glUCOSe. 2 3 5-Trimethyl glucose. 49-7 i Dimethyl glucose. 49.6 S f 50-4 50.6 The hove teth- tri- and di-methyl glucosea in exactly molecular proportions. Including Pictet's results a series of polymerides from mono-fo octa-glucosan is now complete with the exception of the penta-form.It will be observed from the table that polymerisation alters the sign of rotation which increases in the dextro sense, with the molecular magnitude this possibility having been fore-VOL. CXXVII. 5 2908 IRVINE AND OLDHAM POLYMERISATION OF ~I-GLUCOSAN. seen through our studies of inulin and starch. The optical effect of methylation is also consistent with previous experience but the essential fact which emerges from the results given in the table is that mono- and di-glucosan are unique members of the series. The former gives rise to only one methylated sugar (2 3 5-tri-methyl glucose) and the latter yields a mixture from which trimethyl glucose is absent.Thereafter in the higher polymerides 2 ; 3 5-tri-methyl glucose is again encountered as an end-product and in the case of the mixed polyglucosans this sugar is present in equi-molecular proportion with the higher and lower homologues. This represents an average result to which all polymerides higher than diglucosan have contributed and the discussion can therefore be focussed on the three types represented by (a) mono- ( b ) di- and (c) ply-glucosan. Mechanism of the Polymerisation. The initial step of the polymerisation can be traced from the s-cant fact that 2 3 5 6-tetramethyl glucose and a dimethyl glucose are invariable products from all the polyglucosans. It follows that the fist action is the conversion of glucosan into glucose one molecule of which condenses with a second molecule of glucosan so that once the process is initiated it is catalytically continued.The experimental conditions employed in the poly-merisation are favourable to this cycle of reactions and no other explanation seems possible. If this be correct the dimeride should differ from the monomeric parent in having the hydroxyl groups unequally distributed in the ratio of four in one glucose residue to two in the other. The results obtained show that this con-sideration applies quantitatively. The precise way in which glucose condenses with a molecule of glucosan must nevertheless remain unknown until the constitution of the dimethyl glucose isolated from diglucosan has been established and despite laborious investigation this has not been solved.As shown in the experi-mental part however the sugar must be either 2 3- or 2 5-di-methyl glucose and as the latter alternative is more strongly supported it is provisionally adopted leading to the following structure for diglucosan : n 7c-, OH)*CH( OH)-CH*CH( OH)*CH,*OH B. Glucose residue. A. Gluco- &&due THE CONSTITU!FION OF SYNTHETIC DEXTRINS. 2909 Speculation on the next stage of the polymerisation is guided by the fact that all polymerides above diglucosan give 2 3 5-tri-methyl glucose. This can originate only from the molecular fragment B and not from A as otherwise the methylated sugars obtained from triglucosan would be two molecules of tetramethyl glucose and one molecule of monomethyl glucose. The attach-ment of the third glucose residue is thus desnitely restricted to position 6 of residue By giving the following constitution for tri-glucosan : r1 i 0 I rYH 6 YH-OH r -1 LC!H I YH-MH*CH( OH)*CH(OH)=CH*CH(OH)* H F 0 (IH~OH I As the yields of methylated sugars from tetra- and hepta-glucosan have no quantitative significance it is inaddable to discuss the further steps involved in the polymerisation but the examination of the mixed polyglucosans contributed valuable information.On occasions these higher glucosans formed as much as 75% of the total material polymerised so that they may be regarded as repre-sentative of the whole reaction. Inspection of the experimental details will show that this material not only gives the three methylated sugars required by the above formula but does so in precisely equimolecular proportions.This result has been verified on more than one occasion; it cannot be regarded as adventitious and it disposes of the possibility that the trimethyl glucose originates in a simple polymeride of glucosan in which the molecular con-stitution of the parent molecule is preserved. It is nevertheless, conceivable that molecules of monomeric glucosan may become associated either together or with simple polymerides and the existence of such compounds as hepta- and octa-glucosan indicates that this should not be ruled out. Further the tetraglucosan examined by Pringsheim yielded no trimethyl glucose but gave rise to the same sugars as we have now shown to originate from diglucosan. Taking a general survey of the results it is clear that the polymerisation involves reactions of two types one involving association and the other condensation.For example, when diglucosan is formed it may either condense with an addi-tional molecule of glucosan to give the trimeride which in turn, by further condensation yields a tetrameride or alternatively two dimeride molecules may become associated to give an entirely 5 ~ 2910 IRVINE AND OLDHAM POLYMERISATION OF ~-GLUCOSAN. different type of tetraglucosan. The inclusion of Pringsheim's results with our own renders this view more than speculative and a similar complexity may accompany each stage of the ply-merisation. Our results do not provide conclusive evidence on this point and reveal only the primary nature of the polymeris-at'ion.It is of special interest to note that two-thirds of the triglucosan formula consists of a residue present in the accepted formula for maltose a remarkable similarity in structure con-sidering the drastic conditions employed in the preparation of the trimeride. Triglucosan may in fact be regarded as qEucosan mdtoside. Further the P-configuration of the parent glucosan is preserved in the higher polymerides despite the pronounced changes in rotation which accompany their formation. The view now put forward demands that the condensation type of polymerisation is dependent on the presence of free hydroxyl groups and is supported by the fact that trimethyl glucosan was recovered unchanged when heated with zinc dust in an ex-hausted sealed tube at 250" for 10 hours.Under similar conditions the use of zinc chloride as a catalyst resulted in profound decom-position but on limiting the reaction to 3 hours at lW" the rotation altered from laevo to dextro owing to conversion of the glucosan into trimethyl glucosidotrimethyl glucose. Apnlicdility of the Methybtion Process to the Structural Problewis of Carbohydrates. The criticism has been put forward that our method of deter-mining the structure of carbohydrates although diagnostic in the case of simple sugars may not be applicable to the '' closed-chain structures " represented by polysaccharides and similar compounds. It is dif6cult in view of the mass of consistent results obtained with many types of non-reducing carbohydrates to find any justification for this objection; but as at present much reliance is placed on the validity of the methylation process as a means of determining constitution the occasion is opportune to take into account the essential requirements of the method.The principles developed in this laboratory can be applied to the structural problems of carbohydrates provided the following primary conditions are satisfied (1) that methylation does not alter the configuration of a sugar; (2) that methylation does not disturb the positions in which sugar residues are attached to each other or to other groups; and (3) that under the conditions employed in methylation and in hydrolysis non-glucosidic met hoxyl groups do not migrate from one position to another in a sugar chain. In default of direct experimental evidence to the coiitrary THE CONSTITUTION OF SYNTHETIC! DEXTRTNS.291 1 and in view of the following observations it may be claimed that the above requirements are satisfied. According to circumstances, the methylation of carbohydrates as practised by us is effected either by the silver oxide reaction or by the methyl sulphate method, used independently or in succession. It has been adequately proved that these alternative methods when applied to the =me compounds give the same methylated sugars the only distinction being that the alkaline reagent acts upon reducing sugars to @ve a larger excess of the corresponding p-glucoside. Configuration is therefore affected but only so far as the reducing group is con-cerned. The one aspect of conQuration however which is utilised in our deductions is that of the non-reducing groups and it h a long been known that the silver oxide reaction yields derivatives which retain the confirmration of the parent compound.The conversion of d-dimethoxysuccinic acid into d-tartaric acid (Purdie and Barbour J. 1901 79 972) is a convincing but by no means unique example of this regularity. In addition mono- di- tri-, and tetra-methyl glucose all of which were prepared by the silver oxide reaction have been demethylated and converted into glucose phenylosazone showing the correct optical activity. That the second and third requirements of the method are fulfilled is shown in numerous ways. For example Haworth and Leitch subjected maltose and cellobiose to identically the same methylating treatment yet isolated isomeric trimethyl glucoses in the two investigations.This result cannot be reconciled with the idea that molecular linkages are-altered by methylation or that the methyl groups fail to retain their positions. The evidence is equally clear in the case of the more unstable types of carbohydrate deriv-atives such as y-glucosides and the existence of isomeric tetramethyl glucoses and of the corresponding tetramethyl fructoses which reflect accurately the essential properties of the compounds from which they are derived may be quoted in illustration. In this connexion it is important to note that with the exception of the glucosidic alkyl group (which in any case is not concerned with our structural studies) the stability towards acids and alkalis of the methyl groups in a sugar chain is remarkable.Concentrated sodium hydroxide at the boiling point has no effect on a fully methylated hexose and in our experience the elimination of the methyl groups from an alk-ylated reducing sugar has been accom-plished only by such processes as boiling with concentrated hydriodic acid heating under pressure with glacial acetic acid saturated with hydrogen bromide or in one example,* by prolonged action with * This result which is unpublished was obtained with a dimethyl galactose which yielded a monomethyl galactosazone. A similar irregularity is reporte 2912 IRVINE AND OLDHAM POLYMERISATION OF ~-GLUCOSAN. the phenylosazone reagents. This stability is far removed from that encountered in cases where methyl groups have been shown to migrate or to enter an irregular position (see Kubota and Perkin, this vol.p. 1889). It is possible that the frequency with which 2 3 6-trimethyl glucose has been obtained from different polysaccharides may have suggested the idea that the reagents had converted definitely isomeric compounds into a common form so that the same sugar would inevitably be produced in all cases. The results of the present investigation go far to refute this remote possibility. The only practical distinction between the alkylation of a polysaccharide and that of a simple sugar is that; owing to solubility diflticulties and to steric hindrance it is necessary in the former case to use the methyl sulphate method throughout and to repeat the rnethyl-ation many times.This treatment may conceivably affect the degree of polymerisation of a compound but as the discussion does not involve this point it may be focussed on the question if the repeated use of the alkaline reagent gives results divergent from those obtained by restricted treatment with silver oxide and methyl iodide. For this purpose we have selected p-glucosan as a test substance and after ten methylations by means of methyl sulphate obtained a normal yield of the same crystalline trimethyl glucosan formerly prepared by the alternative method (Irvine and Oldham Zoc. cit.). On extending the process until a total of twenty treatments had been given the same product was again obtained. The result shows that even when the methyl sulphate process is repeated as often as in the case of cellulose starch and glycogen, the methyl groups enter the same positions in glucosan as they do when the silver oxide reaction is used only once and further, by Freudenberg and Hixon (Ber.1923 56 2119) and confhned by Levene and Meyer (J. BioZ. Chem. 1924 59 i 145) who found that a mono-methyl mannosediacetone was completely hydrolysed by treatment with very dilute acid a property which indicates that the compound waa a y-methyl-mannoside dimetone. Should this prove to be the case rearrangement may have taken place between a methyl group and an hopropylidene group during either methylation or hydrolysis. Numerous examples are now known of the reversible displacement of bopropylidene and methyl but the reaction proceeds in acid solution whereas Freudenberg’s process does not admit of this condition.Apparently the configuration of mannose is conducive to irregular results. It may be recalled (Irvine and Paterson J. 1914 105, 9 15) that one hydroxyl group in mannitol resists methylation completely and that when methylglucosamine (a-amino-methylmannoside) is acted upon by dilute nitrous acid not only is the amino-group removed but the methyl group which is otherwise remarkably stable also is eliminated. In con-sequence the aIkylated mannoses have not been applied by us to structural studies THE CONSTITUTION OF SYNTHETIC DEX'I"S. am 3 that aIkyl groups show no tendency to migrate to the 2 3 6 - p i -tions. The dextrins described in the present paper were 'also subjected to repeated methylation yet the sugar finally obtained consisted exclusively of 2 3 5-trimethyl glucose and no trace of the 2 3 6-isomeride was detected.Precisely the converse applies when cellulose is subjected to parallel treatment as 2 3 6-tri-methyl glucose alone constitutes the final product. It is difficult to imagine that these sharply differentiated results obtained under duplicate conditions are due to the vagaries of the reagents rather than to inherent structural differences in the parent compounds. Clearly however the silver oxide reaction when applied to polymerides tends to lower the degree of polymerisation zt8 indi-cated by a change in solubility and an alteration of the rotatory power in the direction of that of the parent unit.This phenomenon has already been encountered in the investigation of methylated inulin and has again been observed when for comparative purposes, the mixture of higher polyglucosans was subjected to this reaction. E x P E R I M E N T AL. Polymerisation of p-Glucosan.-AS the polymerisation of @-glucosan gives rise to a variety of komerides it is necessary to specify accurately the experimental method adopted in preparing the polymerides now described. In the h t series of preparations, small quantities (7 g.) of glucosan were used in each experiment and this weight of material together with 0.1 g. of zinc dmt was intro-duced into a 300 C.C. distilling flask the neck of which was sealed by a cork carrying a manometer. To the side limb a T-tube ww attached provided on one branch with a tap (A) leading to the water pump and on the other with two taps (B and C).These were placed closely together so that the enclosed volume was small, C being connected to a supply of pure dry hydrogen. The flask, which with the exception of the side fittings was immersed in an rtsbestos air-bath provided with windows was exhausted and the residual air washed out with hydrogen. After again evacuating the temperature of the bath was raised to 250" the tap A being closed when the glucosan began to melt M otherwise the compound volatilised unchanged. After about 15 minutes the melt became viscous and as frothing ensued the tap A was opened at intervals. In approximately 30 minutes from the start of the reaction the frothing usually became most severe and it was then necessary to close A and C and open B momentarily.A bubble of hydrogen was thus introduced which served to break the film of glucosan and thereafter both C and A were rapidly opened and closed. In from 60 to 75iminutes from the time the glucosan wa8 thoroughly melted 2914 IRVINE AND OLDHAM POLYMERISATION OF ~-GLUCOSAN. frothing became sluggish and unless A remained open for a con-siderable time ceased entirely. At this stage heating was stopped and the flask allowed to cool the vacuum being maintained. No loss of weight was recorded under the above conditions and no charring took place. Direct Isolation of Polyglucosans. The product of several preparations was extracted with the minimum quantity of hot water and on the addition of rectified spirit to the united solutions a dark-coloured precipitate was formed.This was removed and absolute alcohol added to the filtrate. The dextrin then separated as a fine powder except in casa where too much water was present when a sticky precipitate was produced. In such an event the liquid was decanted the residue dissolved in a small quantity of warm water and after removal of the solvent dried a t 100"/15 mm.; the product could then be readily powdered. The less-soluble dextrin obtained as above is referred to as " Dextrin I." The mother-Iiquor which had deposited Dextrin I was concentrated and the addition of alcohol repeated until no further precipitate was formed. In this way the material was separated into two further portions (Dextrins I1 and III) each fraction being relegated to its class on the basis of sol-ubility and specific rotation.Heptuglwsan.-The material was redissolved in hot water and boiled with charcoal a treatment which removed colouring matter and also diminished the yield considerably. Finally the compound was precipitated with alcohol and dried in a vacuum. Dextrin I was thus obtained as a slightly hygroscopic powder which although perfectly white gave a pale yellow solution in water. Analysis of different preparations gave C 44.5 44.45; H 6.4 6-2; ash 0.266 (C6H1,,05 requires C 44.4; H 6.2%). Dextrin I is insoluble in organic solvents with the exception of acetic acid but is freely soluble in water giving a non-reducing dextrorotatory solution. This value was obtained in different preparations and when a specimen of the material was fractionally precipitated with alcohol the most active fraction showed [.ID + 85.8" thus indicating the uniformity of the compound.Despite this observation and the fact that dextrin I was afterwards shown to be heptaglucosan the application of Karrer's method of determining the molecular magnitude of polymerised units by analysis of the sodium hydroxide compounds gave inconclusive results which lay between those required for For analytical and reference purposes the triucetak was prepared Emmination of Dextrin I . [.ID + 83.9" for c = 2-08. (c6H1005)3 and (c6H1005)4 THE CONSTITUTION OF SYNTHETIC DEX!L"S. 2915 by boding the dextrin until dissolved with excess of acetic anhydride containing zinc chloride.The product wm isolated by precipitation with water washing with ether and solution in a large excess of hot aholute alcohol from which it separated as a fine white powder. After three such treatments the m. p. was constant (142") (Found : C 49.95; H 5.6; CH3-C0,H 65.6. C,,H,,O requires C 50.0; H 5-55; CH,-CO,H 62.5%). The triacetate is insoluble in water, cold a.bsolute alcohol or ether soluble in other organic solvents including aqueous alcohol [.ID in 50% alcohol + 85-1" for c = 2.2315. When heated for 2 hours at 70" in methyl alcohol which was nearly saturated with hydrogen chloride the dextrin was com-pletely converted into methylglucoside. Initially excess of B-methylglucoside was formed but on continuing the reaction for 24 hours the equilibrium mixture was obtained from which pure a-methylglucoside was isolated in the usual manner (m.p. 164"; OMe 15.1%; [.ID in water + 158.4"). Tetraglucosan.-This substance was produced in greatest yield when the polymerisation of p-glucosan was carried out in quantities of 20 g. at a time a 600 C.C. flask being employed. The subsequent procedure was as described and after removal of dextrin I the more soluble products were fractionally precipitated with alcohol. Fractions showing similar rotatory power were mixed and again precipitated so that by prolonged repetition of this process the total product was ultimately separated into two portions which could not be further sub-divided. These showed respectively [a], + 60-7" and + 29.9" in aqueous solution and are indexed it5 dextrin II and dextrin 111.Both substances were h e white hygroscopic powders and had the same com-position as the parent glucosan. Dextrin 11 when acetylated as already described gave the same triacetate as dextrin I (Found: C 50.2; H 5.6; CH3*C0,H 63.6%). The specific rotation and melting point also agreed within experimental error and on hydrolysing the acetate with sodium hydroxide dextrin I was regenerated (Found C 44.5; H 6.1 ; [.ID + 81-5" for c = 2-14 in water. Dextrin I requires C 44.4 ; H 6.2% ; [.ID + 83.9" for The solid was covered with methyl alcohol and hydrogen chloride passed in from time to time until the dextrin dissolved after which the solution was diluted with more alcohol and heated at 70", polarimetric readings being taken every 15 minutes.In one hour the constant value [.ID + 94.5" (calc. on the weight of glucoside formed) was reached and the product was then isolated as usual. Examination of Dextrin I I . c = 2.08). Dextrin I1 was readily converted into methylglucoside. 5 F 3916 EWE AXD OLDHAM POLYMERISATION OF p-a~ucosm. A crystalline mixture of a- and p-methylglucosides was thus obtained from which the pure a-form was separated by crystallisation from alcohol (Found m. p. 165-166"; [.ID + 157.2"; OMe 15.6. a-Methylglucoside requires m. p. 165-166"; [.ID + 157.5"; OMe, 15.9%). When the above reaction was conducted a t No in place of 70° and was arrested after 45 minutes (3-methylglucoside was the chief product and the a-form was present only to the extent Triglucosan.-In all respects save aolubility and specific rotation dextrin I11 resembled dextrin 11.Like the latter it yielded apparently the same triacetate as dextrin I (Found C 50.1; H 5.6; CH,-CO,H 62.3%). The specific rotation in aqueous alcohol was however + 74.2" in place of 78.9", but this discrepancy appears to be due to a trace of impurity as, on treatment with sodium hydroxide dextrin I was regenerated. Dextrin I11 was also converted into methylglucoside in the usual way. The total crystalline product isolated showed [.ID + 92-9" for c = 1 in methyl alcohol and gave OMe 15.3%. On crystallis-ation from absolute alcohol pure a-methylglucoside displaying the standard constants was obtained. In this case also when the reaction wm restricted to heating a t 50" for 1 hour p-methyl-glucoside was the chief product.of 20%. Examination of Dextrin I I I . Methylation of the Individual Dextrins. As the methods adopted for the methylations have been described in previous papers from this laboratory only s i e c a n t results are now quoted: Heptag1ucosan.-The action of silver oxide and methyl iodide on a methyl-alcoholic solution was definitely arrested at the stage where two alkyl groups had been introduced per C unit. Product a colourless glass emily powdered. [.ID in chloroform + 76.0" for c = 0.6035 (Found C 50-65; H 7.4; OMe 32.4. C,H1,O requires C 50-5; H 7.4; OMe 32.6%). The methyl sulphate reaction was more effective repeated treatment with the reagents under conditions similar to those employed with inulin giving a product readily soluble in organic solvents and showing [aID in chloroform + 89.4" for c = 2.745.The methylation was however still incomplete but approximated to the trimethyl stage [Found C 53.0; H 7.5; OMe 40-9. (C9H1605),. requires C 52.9; H 7.8; OMe 45.5%]. No further purification was possi-ble as the compound which was isolated as a glass readily con-vertible into a white powder was exceedingly soluble in organic solvents with the exception of light petroleum. The molecular Methylation of Dextrin I THE CONS-ON OF SYNTHETIC DEX!L"S. 2917 weight determined in benzene by the crymopic method was 1446, whereas a hepta (trimethyl glucosan) C ~ O require;s 1428. As contrary to expectation methyl alcohol containing 1% of hydrogen chloride proved to be without action on the compound, simultaneous hydrolysis and condensation was effected by heating for 15 hours a t 70" with alcohol nearly saturated with the gas.No charring resulted from this drastic treatment and the specifx rotation showed the usual characteristic rise and fall during the change. ([.ID + 75-0"-+ 93-7" -+ 87.5"). The mixed glucosides were isolated by vacuum distillation as a colourless 'syrup (OMe 50.4%) and were hydrolysed therea,fter by means of 3% aqueous hydro-chloric acid to give a solution of methylated glucoses showing [.ID + 68.4". A syrup was finally obtained on distillation showing OMe 38.6% and yielding crystalline tetramethyl glucose on extrac-tion with boiling light petroleum. The rema.ining sugars were di-and tri-methyl glucoses in llnknown proportions.Tetraglucosan.-As a result of three methylations by the methyl sulphate method 18 g. of the dextrin were converted into 16-5 g. of a viscid syrup showing [.ID + 66.6" for c = 3.3335 in chloroform and having OMe 39.5%. Two further methylations raised the methoxyl content to 41.6% and four subsequent treatmenfg gave the maximum value of 43.4%. The product was a glass possessing the customary solubilities [Found C 53.0; H 7.8; OMe 43.4. (C9HI6OJn requires C, 52-9 ; H 7-8 ; OMe 45.5%]. Methylution of Dextrin I I . Solvent. C. [QIrD. Methyl alcohol ..................... 1-3625 + 68.7' Acetone .............................. 1.968 70.5 Chloroform ........................ 2.900 61.2 Molecular weight determined by the cryoscopic method in benzene solution = 873.Tetra (trimethyl glucosan) C,,H,O,, requires 816. When converted into the corresponding methylglucosides as described in the case of the hepta-isomeride the product on dis-tillation showed in successive experiments OMe 49-6 49.7% and [a],,+ 100.9" in water for c = 1.3016. On hydrolysis.with 4% aqueous hydrochloric acid the permanent value for the specific rotation was + 66-7". The mixture of sugars thus obtained had OMe 38.9% and was separated into (a) a fraction soluble in ether, amounting to about two-thirds of the total and (b) a fraction, consisting of the remainder soluble in acetone but not in ether. From the soluble portion tetramethyl glucose c r y s t a w spon-taneously and was identsed by analysis and by determination of the physical constants.The uncrystallisable sugars were mixed, 6F* 2918 IRVINE AND OLDHAM POLYMERISATION OF ,B-GLUCOSAN. converted into their methylglucosides distilled in three fractions, and analysed. The methoxyl contents in order of increasing volatility were respectively 38.3 44-8 and 53-4y0 showing that the compounds were derived from a di- and a tri-methyl glucose. In no case did any of the sugars give a phenylosazone and the incom-pletely methylated forms were convertible into the stable variety of tetramethyl glucose in 80% yields. It was necessary to emure that the tetramethyl glucose bolated did not originate in degradation during the methylation of the original dextrin. The methylated dextrin was therefore extracted repeatedly with boiling light petroleum but this did not cause any alteration in the methoxyl content.Further when heated at 200°/0.2 mm. the methylated dextrin distilled very slowly but analysis of the distillate showed that no tetramethyl methylglucoside was present. Trig1ucosan.-Twenty grams sub-jected to eleven successive treatments by the methyl sulphate reaction gave 15.3 g. of a colourless glass convertible into a white solid [Found C 52-7; H 7-8; OMe 42.2. (C,H,,O,) requires C 52.9; H 7.8; OMe 45-5y0]. Methylutwn of Dextrin I I I . Solvent. C. [a],. Chloroform ........................ 2.1035 + 53.5" Acetone .............................. 2-048 56-4 Methyl alcohol ..................... 1.753 62.8 The molecular weight determined by the cryoscopic method in benzene was 595.Tri(trimethy1 glucosan) C2,H48015 requires 612. When heated in a vacuum approximately one-half of the material distilled at 180-200"/0~5 mm. but this result was evidently due to depolymerisation as the distillate was a comparatively mobile syrup having nD = 1.4770 and [.ID in chloroform + 23.9". The composition remained unaltered (OMe 44.7 %) The methylated dextrin when subjected to the joint action of methyl alcohol and hydrogen chloride mas converted into the corresponding mixture of akylated methylglucosides the whole of the product being distill-able although over a wide temperature range. No fractionation was attempted the total distillate being hydrolysed in one experiment by b o w with 4% aqueous hydrochloric acid.It is sigm6cant that when hydrolysis was complete the corrected specific rotation of the solution was + 6 5 ~ 3 " ~ showing that the end products were the same its those obtained from the hepta- and tetra-isomerides. As before the sugars thus obtained were divided into ( a ) a fraction soluble in ether which yielded crystalline tetramethyl glucose and a trimethyl glucose and (b) a fraction soluble in acetone but not in ether. The less soluble sugar which was purified by dissolvin THE CONSTITUTTON OF SYNTHETIC DESTRTF-S. 2919 in chloroform and precipitating with ether possessed the com-position of a dimethyl glucose (Olvze 33-4%) but could not be obtained in a crystalline form and failed to give a phenylosazone. Both the di- and the tri-methyl glucose formed above belonged to the butylene-oxide series and polarimetric tests for the presence of 7-forms gave negative results.Indirect Isohtion of Polyglucosans in the Form of the Methylated Derivatives. In place of separating the polyglucosam by precipitation and conducting the methylations in Merelit experiments an alternative is to alkylate the total polymerised material and thereafter to separate the methylated products. The latter method proved to be more effective more economical and more decisive. Thirty grams of glucosan were polymerised as described in each experiment. By extraction of the product with boiling rectified spirit it was possible to remove unchanged glucosan together with the bulk of the dextrins of low molecular weight leaving polymerides of high molecular weight undissolved and in this way it was shown that both types of dextrin yielded normal triacetates.It was, however preferable to methylate the total product directly the methyl sulphate method being used throughout. A comparatively mobile syrup was thus obtained in good yield 80 g. of glucosan giving 73 g. of methylated polymeride. On distillation at the Gaede pump the product was separable as under : Fraction I. B. p. 135"/0-2 mm. 28 grams. Trimethyl glucosan. Fraction 11. B. p. 205-210"/0-2 mm. 17.1 grams. Di(trimethy1 glucosan). Residue. 27.2 grams. Poly(trimethy1 g l u c m ) . In a second preparation where more complete polymerisation was effected and where over-heating during distillation was avoided 60 g. of glucosan yielded 55 g.of methylated product which in turn gave : Trimethyl glucosan .................. 9.41 grams 17% Pol y (trime t hyl glucosan ) ......... 3 1 . 7 7 57.8% The monomeric trimethyl glucosan crystallised in the receiver ; the dimeride was a pale yellow clear viscous syrup (nD 1-4720) whilst the undistillable isomerides consisted of a glass. Di(trimethyZ gZucosan) .-The composition and physical constants of the distilled syrup showed very little variation in successive preparations [Found C 53-0 52-9; H 7-7 7.9; OMe 44.5; N cryoscopic in benzene 418. (C9H,,0,) requires C 52.9; H, 7.8; OMe 45.6%; M 4081. Di(trimethy1 glucosan) ............ 12.98 , 234% Solvent. C. [ . Chlorof o m ........................ 1-95 16 + 46.5" Methyl alcohol .....................1.8790 48.3 Acetone .............................. 2.0040 51. 2920 IRVINE AND OLDHAM POLYMERISATION OF ~-GLUCOSAN. Conversion into methyluted methy,?g~ucosides. This reaction was accomplished by boiling with methyl alcohol nearly saturated with hydrogen chloride the specific rotation corrected for the change of concentration becoming constant a t + 110.0". The mixed gluco-sides were isolated by distillation and although the liquid boiled over a wide range of temperature the product w~ts collected as a single fraction (yield 97% ; n 1.4587). Under these conditions the distillate should have the same composition as trimethyl methyl-glucoside and this was the case (Found C 50.8; H 8.5; OMe, 51-2. CloH,,06 requires C 50.8; H 8-5; OMe 52.5%). Separ-ation of the constituents was effected by dissolving in water con-taining sodium bicarbonate and extracting repeatedly with chloro-form.This treatment removed all glucosides containing four or more methoxyl groups per C unit lower methylated compounds remaining in the aqueous layer. The process can be applied quantitatively and as in the present case yields are of special importance the results of one exact experiment are quoted. 8.5483 Grams of di(trimethy1 glucosan) gave 4.9161 g. of methylated glucosides extractable with chloroform and 3.5824 g. were retained in the aqueous layer. The less soluble compound when isolated, Ciissolved in ether dried and recovered proved to be dimethyl methylglucoside showing [.ID + 108.4" in methyl alcohol and nD 1.4743 (Found C 48-7; H 8.2; OMe 41.2.Calc. for C,H1,06, C 48-8; H 8.1 ; OMe 4143%). Analysis of the syrup extracted with chloroform having shown that it consisted essentially of tetramethyl methylglucoside together with a little trimethyl methylglucoside the mixture was hydrolysed to give the corresponding sugars. On repeating the extraction with chloroform only tetramethyl glucose passed into the lower layer, whilst trimethyl glucose was retained in the aqueous layer. The tetramethyl glucose when recovered p d e d by distillation and recrystallised as usual displayed the correct melting point and rot-ation (Found C 50.8; H 8-4; OMe 52-0. Calc. for C1&2006, C 50.8; H 8.5; OMe 52.5%). In this way 3.2457 g. of pure sugar were obtained from 4.4717 g. of the mixed glucosides.The small amount of trimethyl glucose isolated shows that this sugar cannot be regarded as a definite molecular product but originates in unpolymerised trimethyl glucosan. Neglecting this by-product and small undistilled residues the relative yields of the hydrolysis sugars calculated as the correspondmg glucosides become : Calc. for equal Dimethyl methylglucoside . . . . . . . . . . . . . . . . . . 47.1% Tetramethyl methylglucoside ...... ......... 53.4 S2.9y0 Found. molecules. 46. THE CONSTITUTION OF SYNTHETIC DEXTIUENS. 292 1 Examination of Methyluted Polyglucosans.-The non-volatile residue left when di(trimethy1 glucosan) was separated by dis-tillation constituted the main product of the methylation of p l y -merised glucosan. After solution in ether filtration and removal of the solvent the mixture of higher polymerides was obtained as a clear glass.The carbon content was low although the composition was uniform in different preparations [Found C 52.4 52.4; H, 7.6 7.8; OMe 44.4 45.2. (C,H,,O,X requires C 52.9; H 7.8; OMe 45.5y0]. As from its method of preparation the material consisted of all polymerides higher than the dimeride no molecular weights were determined. Solvent. C. Calw Chloroform ........................ 1-850 + 63.3" Acetone .............................. 2.266 66.6 Methyl alcohol ..................... 1.765 64.4 Conversion by the action of acid methyl alcoho€ into the mixture of methylated glucosides gave a 95% yield of a colourless syrup, which was distilled under 0.4 mm. and collected in one portion with-out fractionation.The average composition of the total distilla,te should be the same as that of trimethyl methylglucoside but although this held approximately all the values were low (Found C 50.4; H 8.3; OMe 50.6. Calc. for C,,H,O, C 50.9; H 8.5; OMe, 52.5%). On separating the constituent glucosides by extraction of an aqueous solution containing sodium bicarbonate with chloroform, the syrup recovered from the extract weighed 69% of the original total and the remaining 31% was recovered from the water. The former was hydrolysed with 8% hydrochloric acid to give the corresponding methylated glucoses which were isolated in 94% yield. Analysis of the distilled sugars showed that tri- and tetra-methyl glucose were present in equimolecular proportions (Found : C 49.9 ; H 8.2 ; OMe 46.8.Calc. C 49.8 ; H 8.3 ; OMe 4703%). Separation of the sugars mas effected by the method of chloroform extraction 11-176 g. of the mixture giving 5.300 g. of crystdline tetramethyl glucose displaying the correct analytical figures and constants. The less soluble sugar was recovered from the water, distilled as a viscid syrup and identified as a trimethyl glucose (Found C 48.6; H 7.95; OMe 41.4. Calc. for C,H,,O, C, 48.6; H 8-1 ; OMe 41-8y0). The specific rotation in chloroform solution was + 72.5" but in order to identify it completely the sugar was converted successively into the diacetate the mono-acetobromo-derivative and finally into the corresponding trimethyl p-methylglucoside. These steps were controlled by blank experi-ments in which authentic 2 3 5-trimethyl glucose was used and the optical changes were parallel throughout.On nucleation wit 2922 IRVME AND OLDHAM POLYMERISATION OF ~-GLUCOSAN. 2 3 5-trimethyl P-methylglucoside the product crystallised and was purified as usual from light petroleum. The m. p. refractive index and specific rotation in methyl alcohol were correct (Found : C 50-7; H 8.5; OMe 51.9. Calc. for C10H20067 C 50.8; H 8.5; OMe 52.5%). The combined results show that the sugars extract-able by chloroform from aqueous solution are 2 3 5 6-tetra-methyl glucose and 2 3 5-trimethyl glucose in equimolecular proportion. The lower methylated glucoside retained in aqueous solution after extraction of the tetra- and tri-methyl methylglucosides with chloroform consisted of a viscid syrup (nD 1-4779 [a]= in methyl alcohol + 106.2").Although with material of this nature vacuum distillation is difficult this purification was undertaken to obtain more accurate analytical figures. Pure dimethyl methylglucoside was thus obtained as the main fraction (b. p. over 190"/0.4 mm.; nD 1.4743) (Found C 48.55; H 8.2; ONe 40.8. Dimethyl methylglucoside C9HI8O6 requires C 48.6 ; H 8.1 ; OMe 41.8%). The undistillable residue was a glass and consisted essentially of monomethyl methylglucoside (Found C 45.7 ; H 7-2 ; OMe 28.9. Calc. for C.&606 C 46.15; H 7.7; OMe 294%). About 5% of the total sugars formed from the dextrin consisted of this material which almost certainly originates in incomplete methylation. Attempts to establish the constitution of the dimethyl glucose isolated in the present section of the research led to no h a 1 conclu-sions. The sugar failed to crystallise and formed no phenylosazone ; it also failed to enter into condensation with acetone when dissolved in this reagent containing 0.2% of hydrogen chloride but reacted when the concentration of acid was raised to 1.3%. No definite benzylidene derivative could be prepared either from the sugar or from its methylglucoside but the combined results although not conclusive favour the view that of the two alternatives 2 3- or 2 5-dimethyl glucose the latter is more probable. We desire to express our indebtedness to the Department of Scientific and Industrial Research for a Research Assistantship granted to one of us. UNITED COLLEGE OF ST. SALVATOR AND ST. LEONARD, UNIVERSITY OF ST. ANDREWS. [Received October 16th 1925.