1474 IRVINE AND STEELE: CLX1V.-The Constitution of Polysaccharides. Part I . The Relationship of -Inulin to Fyuctose. By JAMES COLQUHOUN IRVINE and ETTIE STEWART STEELE. IT has been shown by one of us in the course of previous publica-tions* that a general method for determining the structure of both di- and poly-saccharides is opened out through the constitu-tional study of methylated sugars. Although the programme of research contemplated in this laboratory has been definitely stated on more than one occasion, we are aware of the fact that other workers have entered this field. It is thus necessary again to point out that the systematic investi-gations on methylated sugars which have been carried out here for the past twenty years were conducted essentially with the object of rendering possible the extension of our work on obvious lines.to the more attractive problems presented by the complex carbohydrates. The preparation of as large a variety as possible of alkylated aldoses and ketoses and the elucidation of their struc-ture provided for purposes of identification the substances which we anticipated would be encountered in solving the constitution of the compound sugars. The principle involved is a simple one in that it is generally possible to substitute all the free hydroxyl groups in a carbo-hydrate or its derivatives by stable methoxyl groups and sub-* References to the use of elkylated sugars are given in the bibliography attached to “The Simple Carbohydrates and the Glucosides” (E. F. Armstrong 3rd edition) and the principles involved are fully described in the Biochemische Zeitschrift 1909 22 357 THE CONSTITUTION OE’ POLYSACCHARIDES.PART I. 1475 sequent hydrolysis yields a methylated sugar or sugars. Deter-mination of the number and position of the alkyl groups in each of the hydrolytic products thus gives direct evidence as to the linkage of the constituents in the parent complex. The general method has already been applied in this laboratory t o the con-stitution of natural and synthetic glucosides (Purdie and Irvine, T. 1903 83 1021; Irvine and Rose T. 1906 8.9 814) to disaccharides (Purdie and Irvine Zoc. cit. and T. 1905 87 1022; Irvine and Dick T. 1919 115 593; Haworth and Law T. 1916, 109 1314; Haworth and Leitch T. 1919 115 809) and also to a typical polysaccharide (Denham and Woodhouse T.1913 103, 1735; 1914 105 2357). Recent developments in the cheinistry of the sugars have added greatly to the complexities involved and incidentally have furnished ample justification of the policy which restrained us from the premature study of the polysaccharides. It is now recognised that a hexose can react n o t only as a butylene-oxide, but also in the more reactive forms provisionally termed “ y-sugars” (Fischer Ber. 1914 47 1980; Irvine Fyfe and Hogg T. 1915 107 524; Irvine and Robertson T. 1916 109, 1305; Cunningham T. 1918 113 596). The chief weight of evidence is in favour of the idea that these isomeric forms of the hexoses possess an ethylene-oxide structure but no rigid formula can yet be applied to all examples and the possibility that an aldo-hexose may react as a propylene- amylene- or hexylene-oxide must also be kept in view.Evidently the constituent sugars of a di- or poly-saccharide may be present in any of the structural forms mentioned above and thus the evidence afforded by the hydrolysis of the unsubstituted complex may be misleading. The case of sucrose may be quoted in illustration. The sugar on hydrolysis yields glucose and fructose of the ordinary type but it has been shown from the study of octamethyl sucrose that the fructose constituent is present in the ‘‘ y-form ” (Haworth and Law loc. cit.). It follows that the complete constitution of a compound sugar, from the disaccharides up to the polysaccharides must include the identification of (1) the constituent sugars (2) their stereochemical form (a or fl) (3) the hydroxyl groups involved in the coupling of the constituents and (4) the position of the internal oxygen ring in each sugar.Determination of factors (3) and (4) demands the introduction of non-hydrolysable residues into the molecule, and it is in this connexion that alkylated sugars play their most useful part. Taking tbe above considerations into account we have resumed 3 1 1476 IRVINE AND STEELE: the study of the constitution of cellulose and have also extended our work to starch and inulin t h e results obtained in the last example being now submitted. The experimental methods employed follow closely the lines already laid down in the methylation of cellulose. It will be recalled that by subjecting cotton cellulose to the action of methyl sulphate and sodium hydroxide solution Denham and Woodhouse (loc.c i t . ) obtained a substance possessing the composition of a trimethyl cellulose from which they isolated a well-defined crystal-line trimethyl glucose as one of the hydrolytic products. The research in question is important as it showed that complete methylation can be effected by means of methyl sulphate in cases where the insolubility of the carbohydrate under examination prohibits the use of methyl iodide and silver oxide as the iii et hyla t ing reagents. Although inulin as the most widely distributed reserve material derived solely from fructose is a compound of considerable import-ance nothing is kno'wn regarding its exact constitution beyond the fact that it is non-reducing and yields fructose on hydrolysis.Even the question of its empirical composition has been debated, as the results of elementary analysis do not agree exactly with the figures required for a compound (C6HI00&. A review of the literature shows that these variations in composition are small, and are doubtless to be attributed to imperfect washing of the samples and the method' of drying adopted. It is now shown that inulin purified and dehydrated as described in the experi-mental part is essentially a polyanhydrofructose with the formula given above. This view does not ignore the presence of the small quantities of inorganic constituents usually associated with inulin, the removal of which is sol difficult as to suggest that they form a minute but definite part of the molecular complex.As it is possible that in the past several closely related poly-saccharides have been included under the general name of inulin, it is necessary t o specify the origin and treatment of the material used in the course of the present research. The inulin employed was prepared from dahlia tubers the standards of purity adopted being that the compound should be white should give less than 0.2 per cent. of ash on ignition be free from any action on Fehling's solution and display the constant' specific rotation of - 35.0° on successive " crystallisation " from water. So far as the methylation process is concerned inulin possesses a marked advantage over celluloss or starch in that it is soluble in aqueous sodium hydroxide.When this solution was treated with methyl sulphate methylation proceeded normally but did no THE CONSTITUTION b~ POLYSACCHARIDES. PART I. 1477 extend beyond the stage at which dimethyl inulin was the essential product. A second treatment with the methylating mixture had very little effect on the iiiethosyl content of the syrup thus obtained and judging from the consistent physical constants dis-played by the product of different preparations it was evident that dimethyl inulin [C,H,O,(OMe),], is a definite compound. I n order to substitute the remaining hydroxyl group recourse was had to the silver oxide method of alkylation. Dimethyl inulin mixes freely with methyl or ethyl alcohols giving a colloidal solution, which is not coagulated by the addition of methyl iodide.By warming such a solution with silver oxide further methylation was effected but the process was tedious owing to the colloidal nature of the material being manipulated. The final alkyl-ations were as usual conducted in methyl iodide solution, and in this way trimethyl inzclin [C6H,O2(OMe)& was obtained. It was not found possible to increase the methoxyl content beyond this stage a result which shows that the series of processes did not result in appreciable degradation, hydrolysis or oxidation of the polysaccharide. Trimethyl inulin is a viscous colourless syrup soluble in organic solvents generally and behaving like a glucoside towards Fehling’s solution. The com-pound could not be crystallised and although in small quantities it may be distilled from a metal-bath at 1960/0.15 mm.the process is wasteful and the further examination was conducted on undistilled material. As in the case of dimethyl iiiulin there can be little doubt that the substance is a definite chemical individual, the protduct of different preparations in which the experimental procedure was varied showing identical physical constants. It is important to note the marked alteration in optical activity which occurs during successive methylation. Whereas dimethyl inulin, like inulin itself is lzevorotatory the introduction of a third methyl group alters the sign and trimethyl inulin is dextrorotatory ([a]’; + 55.6O in chloroform). On hydrolysis by heating a t looo with 1 per cent.oxalic acid, trimethyl inulin was converted into trimethyl fructose the polari-metric record of the change showing a smooth unbroken curve. When the methylated ketose was isolated and purified by vacuum distillation it was a t once evident that the product belonged to the y-series. The sugar was dextrorotatory ([a] + 30.5O in water), reducedl potassium permanganate instantaneously in the cold and also although more slowly Fehling’s solution and an ammoniacal solution of silver nitrate. As this is the first occasion on which a trimethyl y-fructose has been obtained it was necessary for the purposes of identification to convert the compound into 1478 IRVINE AND STEELE: methylated sugar already known and characterised. This was effected by condensing the compound a t 30° with methyl alcohol containing 0.25 per cent.of hydrogen chloride and niethylating the trimethyl y -methylf ructoside thus produced by means of the silver oxide reaction. It is to be noted that the above process gives a mixture in unknown proportions of the a- and P-forms of tetra-methyl y-methylfructoside so that direct comparison with other preparations of the same compound was a t this stage impossible. By hydrolysis however tetramethyl y-fructose was produced and this proved to be identical with the form of tetramethyl fructose isolated from sucrose (Toc. cit.). As the sugar is a liquid and so far as known gives no crystalline derivatives the identification rests primarily on the physical constants determined. These are quoted in the following table and compared with the values given by the two forms of tetramethyl fructose already known Compnriso?z of Tet m m e t ltyl Fru c t oses.A . R. C. From inulin. From sucrose. fructoside. From @-methyl-Liquid b. p. 148*5"/10 mm. Liquid b. p. 154"/13 mm. Solid m. p. 98-99' 1 -4554. no 1.4545. -[a];5 (permanent) + 15.6'. .. + 14.04' ..................... - 20.2" in ethyl alcohol , +32.gg*.. + 31.7" ..................... - 20.9" in water. Reduces KMnO,. Reduces KMnO,. Stable towards KMn04. There can be no doubt as to the identity of products -4 and B and their differentiation from C. In confirmation samples of tetramethyl y-fructose from inulin and from sucrose were dissolved in methyl alcohol containing 0.25 per cent. of hydrogen chloride, and the changes in rotation observed at frequent intervals as the formation of the corresponding methylfructoside proceeded.The speed of reaction in these parallel experinients was idenEica1 a t 30° and the end-points coincided thus confirming the identity of the sugars used. Discussion of Results. The bearing of our combined results on the constitution of inulin may be seen from a survey of the reactions described in the present and related researches (p. 1479). I n series A and G the y-fructose component remains throughout in the y-form whilst in B and D the stable types alone are omr ative . It is evident from the above that the structural relationshi THE CONSTITUTION OF POLYSACCHARIDES. PART I. 1479 J. A . B. C. D. J. J. J. J. inulin fructose sucrose (a or P ) + + J/ + INULIN + Fr?cctose +- SUCROSE + Glucose Tetra-acetyl Reptamethyl Methylglucoside Dime t h y 1 Trime thy1 Tetra-ace tyl Oc tame t h yl Te t rame t h yl inulin methyl f ruct oside sucrose me thylglucoside J.4. J. Trimethyl y-fructose Trimethyl y-methylfr uctoside Tetramethyl y-me th ylfructoside J. J. J. Methyl-fructoside Tetramethyl methylfructoside Tetramet h y 1 fructose between sucrose and inulin is a close one. Both compounds are non-reducing undergo hydrolysis with extreme ease and under ordinary conditions yield the form of fructose melting at 112-114O' and displaying [a] - 93O after mutarotation. We have carried out test hydrolysis of the inulin used in our experiments, and were able to isolate a yield of 66 per cent.of the theoretical amount of crystalline fructose showing the above constants. This result is however utterly misleading from the structural point of view as the fructose component is present in the isomeric y-form. Moreover the yields obtained in the first three stages of A and the complete uniformity of the tetramethyl fructoses pro-duced in pracesses A and C show that all t h e fructose residues in inulin belong to t h e y-series. Again the hydrolysis of trimethyl inulin gives an excellent yield of trimethyl fructose the weight of lower-boiling distillate then isolated being less than 4 per cent. of the material treated. This leads to the second conclusion that i n u l i n i s an aggregate of y-fructose residues each lzetose molecule having lost two hydroxyl groups in t h e formation of t h e poly-saccharide.I f this were not the case the hydrolysis of trimethyl inulin would have given a mixture of sugars ranging from dimethyl to tetramethyl fructoses in place of a trimethyl fructose alone. According to one alternative the polysaccharide may be regarded as a poly-merised anhydrocy-fructose in the formation of which the reducing The above conception admits oft two interpretations 1480 IRVINE AND STEELE: group of the fructose residue takes part in the dehydration and is thus eliminated. c6H1206 C6H1005 (C6H1005)~ y-Fructose. Pol ymerised anhydro-y -fructose. The combined results of several researches so far unpublished, show that the most probable formula for anhydro-y-fructose is either CH,*OH /c5! I 0' I \ 'C-AH~OH \ AH.OH/ ~ H - O H / dH,-/ CH2- ' / 04-\ \CH \ 6H*OH ) O 6H-OR >.(1.) (11.1 or Both structurw admit' of polymerisation on the lines suggested by Pictet and his collaborators to give a comphx carbohydrate, and although the formuke need not be further discussed a t this stage it may be mentioned that I is regarded as more probable than 11. The alternative view of the structure of inulin involves the idea that y -fructose molecules are condensed together in such manner that each ketose component loses two hydroxyl groups one of which is the reducing group in the condensation. CH,*OH j CH,*OH I I ~ H ~ O H j &H.OH 1 ~ H - O H I bH*OH I CH2 ; CH, / i CH CH, ~ H - O H ~ H ~ O H ~ H ~ O H ~ H - O H \C--@j- C- ' >o 0 1 PH ~H,*OH j ~H,*OH A ....o/ . . . . . . . - - ..... - . . . ; . . . \o .._._._..._... \ I bH (The dotted lines indicate the cleavage of the molecule on hydrolysis. THE CONSTITUTION OF POLYSACCHARIDES. PART I. 1481 In order to satisfy all the conditions including the fact now established that inulin contains three hpdroxyl groups in each unit of six carbon atoms the number of ketose residues necessary to form a symmetrical molecule must be a multiple of two. The physical properties of inulin show that the compound is not a, disaccharide and taking the next simplest case our results would be explained by the formula shown on p. 1480 which permits of expansion to a hexa- or octa-saccharide by the addition of coupled ketose residuesl a t the etheric linkages marked A and B .The formula on p. 1480 equally with the first alternative dis-cussed involves that! in each c unit of inulin the same hydroxyl groups are unsubstituted and that two of these groups are different from t3he third. This is consistent with the methylation of the com-pound in definite steps. It also demands that only one form of trimethyl fructose should be produced from trimethyl inulin and this is again in agreement with the experimental evidence. The formula suggmted is of course capable of considerable modifica-tion as any part of the carbohydrate chain may be lengthened by coupling the reducing group of one ketose residue with one of the primary alcohol groups of the next. Part of the inulin molecule would in such case contain the system OR *CH2*C- -CH* [C H*OH],- CH,*O*C-CH [CH*OH],* CH,*O* I'd C'H,*OH I'd 0 I It is of course unlikely that inulin possesses a structure so simple as that of a tetrasaccharide but the fact that trimethyl inulin is perceptibly volatile a t 196O/0*15 mm.suggests that the molecular weight of the polysaccharide is much smaller than is generally imagined to be the case. The high molecular weights quoted in the literature are discordant and can have little significance. At the present stage it is premature to give a decided opinion on the relative merits of tbe two alternative formuh for inulin now proposed and the research is being continued. It is also our intention to attempt the synthesis of sucrose from the form of fructose now shown to be present in inulin.E x P E R I M E N T A L . Purification of Inulin. Crude inulin prepared from dahlia tubers was boiled with char-coal until colourless and separated from the filtrate by freezing. 3 I 1482 IRVINE AND STEELE: Thereafter the material was ‘‘ recrystallised ” several times until the action on Fehling’s solution had entirely disappeared after which it was transferred to tall cylinders and shaken with cold distilled water. When the inulin settled the wash-water was syphoned off and the treatment repeated the process being con-tinued for a week. This method of washing proved to be quite as effective as dialysis in yielding a product giving the minimum of ash on ignition. After filtration the moist inulin was spread on plates to ensure that uniform hydration was obtained before commencing the dry-ing process.In this condition the material contained 60 per cent. of water and in quantities of 200 grams was shaken with 25 per cent. aqueous alcohol. After settling the dilute alcohol was poured away and 50 per cent. alcohol substituted. This in turn, was successively replaced by 84 96 and 98 per cent. alcohol. Exactly similar treatment was then given with mixtures of absolute alcohol and ether until finally pure ether was used as the washing agent. After filtration the inulin was kept in a high vacuum until constant in weight. This somewhat elaborate method of dehydra-tion appears to be necessary in order to obtain inulin as a fine, white mass of uniform microscopic appearance. The compound was dried at 50°/80 mm.elver phosphoric oxide but it was found that owing to surface attraction of moisture it was extremely difficult to obtain constant weighings. A dry sample gave for c = 2.7624 [a] - 34*21° in water. Hydrolysis of Inulin. This reaction was repeated for reasons stated in the introduc-tion. Using Wohl’s methbd (Ber. 1890 25 2107) in which inulin is heated with very dilute hydrochloric acid the yield of solid fructose varied considerably but the average result of a series of experiments was that 200 grams of dry inulin gave directly 45 grams of crystalline fructose and a further 11.5 grams were obtained from the syrupy by-products. On the other hand when the hydrolysis was effected by dilute oxalic acid the proportion of crystalline fructose was greatly increased the mean result being a yield of 132 grams of crystallisable sugar from 200 grams of inulin.Preparation of Dime t h y l Inulin. As the methylation of inulin presents some unusual features, an account is given of a typical experiment. Thirty-two grams of finely powdered pure inulin (2 mols.) were dissolved in 40 C.C. o THE CONSTITUTION OF POLYSACCHARIDES. PART I. 1483 45 per cent. sodium hydroxide by heating in water a t 60-'70°, and after cooling 80 C.C. of methyl sulphate (3 mols.) and 140 C.C. of 50 per cent. sodium hydroxide (total 6 mols.) were run simultaneously into the solution which was maintained a t 3 5 O . This addition extended over three hours the mixture meanwhile being vigorously stirred and the alkali kept in excess.There-after the temperature was slowly raised to 7 5 O and finally to looo for thirty minutes. Carbon dioxide was then passed through the liquid for a prolonged period in order to destroy the bulk of the sodium hydroxide and without removing the suspended solids an approximately equal volume of 88 per cent. alcohol was added. After again passing carbon dioxide and allow-ing to stand it. further quantity of inorganic salts was deposited. These were separated by filtration drained and washed with rectified spirit. Dilute sulphuric acid was then added to the filtrate until it was exactly neutral when the aqueous alcohol was distilled off under diminished pressure. The bulk of the product was contained in the residue but some remained behind with the inorganic salts so that both portions were extracted several times with boiling absolute alcohol.This solvent however takes up sodium inethyl sulphate which is formed in considerable amount as a by-product of the reaction and consequently an extraction with boiling chloroform was carried out thus leaving the sodium salt undissolved. On concentration an amber-coloured syrup was obtained which was subjected to a second methylation. The same proportions of reagents were used and the crude syrup was isolated in the manner just described. Thereafter it was further purified by boiling repeatedly with ether to dissolve any niethylated fructose which might have been formed owing to hydrolysis. The undissolved syrup was then boiled in chloroform solution for three hours with decolorising charcoal.The clarified solution when dried over magnesium sulphate gave on removal of the solvent a clear amber syrup amounting to 78 per cent. of ths weight of inulin taken. This on further drying in a vacuum-oven a t looo, became so brittle that it could be powdered (Found (3-50.55; H=7.51; OMe=35-8; ash=1*76. [C,H,O,(OMe),~ requires C = 50.53 ; H =7*37 ; OMe =z 32.7 per cent.). Dimethyl inulin although sparingly soluble in cold water gives a faintly opalescent solution in hot water. The aqueous solution behaved as a glucoside towards Fehling's solution and reduced aqueous potassium permanganate rapidly but not so quickly as a true y-sugar. For c = 1.845 [a]; - 42.1" in chloroform. 3 I* 1484 IRVINE AND STEELE: I’reparatioit of Trimethyl I?LuZin.I n tha further methylation of dimethyl inulin it was found necessary to adjust the procedure according to small variations in the composition of the material used. When the methoxyl con-tent corresponded exactly with that required for a dimethyl inulin the compound gave a clear solution in methyl iodide. The presence however of even small quantities of lower methylated compounds affected this solubility to such an extent that the addition of methyl alcohol was necessary. I n this event the alkyl-ation was conducted in methyl-alcoholic solution by the addition of methyl iodide and subsequently silver oxide the process being repeated until the product was freely soluble in methyl iodide alone. The final alkylation was then carried out in the absence of any extraneous solvent.On the other hand when a niethylated inulin contained more than 32 per cent. of methoxyl the compound was soluble in methyl iodide and one methylation was then suacient to give a trimethyl inulin. Twenty-five grains (1 mol.) of dimethyl inulin were dis-solved in 100 grams of methyl iodide at the boiling point. When a clear solution was obtained 63 grams (2 mols.) of silver oxide were gradually added and the alkylation continued by boiling under reflux for eight hours. The product was isolated by extract-ing with hot alcohol the soslvent removed and the residue extracted with a large excess of boiling ether. The ethereal solution was heated with charcoal to remove traces of dissolved silver dried Over anhydrous sodium carbonate and the solvent removed, teaving a clear viscous syrup.Even after two further methyl-ations in methyl iodide solution the methoxyl content did not ncrease above the value quoted below so that the formation of trimethyl inulin represents the liinit of the reaction. The cpm-plete drying of the product presented difficulties. When heated at 80°/8 mm. the syrup darkened and developed acidity which resulted in hydrolysis. It was thus necessary to dehydrate the material slowly at 65O/ 150 min. (Found C = 53.05 ; H = 7.64 ; OMe =43*83. [C6H,0,(OMe),] requires C =52’94 ; H = 7.84 ; Trimet hyl in din is a colourless syrup resembling in appearance anhydrous glycerol a t loo. The compound mixes freely with alcohol chloroform or acetone but is sparingly soluble in ether or in water.A significant fact is that solution of inethylated inulin in organic solvents removed all associated mineral matter so that. the compound then left no ash on ignition. Trimethyl inulin has no effect on boiling Fehling’s solution but like the parent OMe= 45.5 per cent.). THE CONSTITUTION OF POLYSACCRARIDES. PART I. 1485 polysaccharide is very readily hydrolysecl by heating with dilute acids and it is likewise unaffected by potassium permanganate solution. NO crystallising medium could be found for the com-pound and it is doubtful if it forms true solutions in any solvent. For c=1*980 [u] +55*6O in chloroform; c=1*3707, [u] +50a340 in ethyl alcohol. Experiments on a small scale showed that trimethyl inulin can be distilled under low pressures but the process is wasteful owing to the ready tendency of the compound to generate traces of organic acids.This occasions some hydrolysis and trimethyl y-fructose thus contaminates the distillate. The first material to distil boiled a t 126-132°/0.15 mm. was acid to litmus and reduced Fehling’s solution in the cold. These properties together with the mobility of the syrup and its action on potassium per-manganate solution which it reduced instantaneously showed that the product was trimethyl y-fructose (Found OMe= 4 2.92. Calc. OMe=41.88 per cent.). For c=2*58 [u] + 31*01° i,n ethyl alcohol. The fraction of higher boiling point (b. p. 196O/0*15 mm.) was a viscous syrup soluble in water and organic solvents! generally. Although the material effected some reduction of Fehling’s solution on heating the behaviour of the compound towards this reagent was essentially that of a glucoside (Found C=51*09; H=7-62; OMe=41*8.[C,H,O,(OMe),] requires C =52*94 ; H = 7.84 ; OMe=45*5 per cent.). The lack of exact agreement with the calculated figures is readily explained by the presence of a small quantity of dimethyl fructose, and the properties of the distillate show that this is the case. The compound was slowly hydrolysed a t 1 5 O by N / 10-hydrochloric acid, the specific rotation falling during the reaction to +42*4O. Recalculation of the end-value for the weight of hexose formed gives [~t]g+37*8~ which is in fair agreement with the rotatory power of trimethyl y-fructose. Hydrolysis of Trimethyl J n t t l i n .Trimethyl y-Fructose. A 10 per cent. solution of undistilled trimethyl inulin in 1 per cent. aqueous oxalic acid was heated a t looo the progress of the hydrolysis being ascertained polarimetrically . The optical changes observed were regular the specific rotation diminishing frcn +53*5O to 40-9O in eight hours. After neutralising the solu-tion with calcium carbonate the filtrate was decolorised with charcoal and evaporated to dryness under diminished pressure. The rwidue was extracted with ether the solution dried ove 1486 IRVINE AND STEELE: anhydrous sodium carbonate and the solvent removed. A colour-less syrup then remained which was distilled under diminished pressure. A small first fraction was collected a t 127-129O/0*25 mm. but the main fraction which weighed 76 per cent of the trimethyl inulin taken boiled steadily a t 146O/0*37 mm.This proved to be trimethyl fructose [Found C =48’*90 ; H = 7.94 ; OMe=41*49. C,H,O,(OMe) requires C=48.75; H=8.11; OMe=41.88 per cent.]. Trimethyl fructose is a viscid syrup resembling glycerol in appearance and has nD 1.4689. The aqueous solution reduces neutral potassium permanganate solution instantaneously and also Fehling’s solution in the cold giving bright red cuprous oxide. Although the sugar likewise reduces ammoniacal silver nitrate at the ordinary temperature it does not affect mercuric chloride and fails to give Schiff’s reaction. On treatment with phenylhydrazine and acetic acid it yielded a reddish-brown syrup which could not be crystallised and it is impossible to say if the product is a hydrazone or an osazone.Trimethyl y-fructose is dextrorotatory in all the solvents examined : For c = 1.016 [a] + 30*51° in water. c = 1.029 [a] -t 28’18O in ethyl alcohol. c = 1.052 [u]g + 2 6 ~ 6 1 ~ in chloroform. c = 1.084 [a] + 2‘7’77O + + 22’14O in acetone. The above optical values were permanent except in acetone solu-tion and in this case it would appear that the sugar reacted slowly with the solvent. This was supported by the observation that when the compound was dissolved in acetone containing 0.05 per cent. of hydrogen chloride the rotation a t first diminished and then increased rapidly until the final value [a]’,”+60*0° was recorded. This capacity to react with acetone is of importance in giving a clue to the constitution of trim ethyl y - f ructose.Conversion of Trimethyl y-Fructose into Tetramethyl y-Met hylfr imt oside . The formation of the corresponding methylfructoside from tri-methyl fructose takes place at the ordinary temperature. A 2 per cent. soIution of the sugar in methyl alcohol containing 0.25 per cent. of hydrogen chloride was kept for forty hours at 17O. In this time the reducing action on Fehling’s solution disappeared THE CONSTITUTION OF POLYSACCHARIDES. PART I. 1487 whilst the specific rotation a t first diminished and then increased to a constant. The following selected observations show that the speed of condensation is of the same order as that exhibited by y-fructose derivatives generally. Time from contact of solvent and solute.Specific rotation. 1 minute + 18.7" 6 minutes 18-2 60 Y9 24'0 120 1 9 26-1 24 hours 50.3 60 ,9 57-0 (constant) A t 30° the reaction is much accelerated and is complete in nine hours. The acid was neutralised by means of silver carbonate the filtrate evaporated to dryness under diminished pressure and the residual syrup dissolved in alcohol. After treatment with char-coal to eliminate traces of silver compounds the solvent was again evaporated the product extracted with ether and the extract dried with magnesium sulphate. On removal of the solvent, trimethyl methylfructoside remained as a clear syrup which with-out further purification was dissolved in methyl iodide (4 mols.) and methylated by the addition of silver oxide (2 mols.). The alkylation was continued for eight hours and the product was extracted and isolated in the usual manner.On distillation tetra-methyl y-methylfructoside was obtained as a colourless syrup (b. p. 134-135O/12 min. n, 1.4469). After a second methylation under the same conditions the boiling point was 137-138'5O/ 12 mm. and refractive1 index 1.4472. The yield was 80 per cent. of the theoretical amount and evidence was obtained that the undistillable by-product consisted of a polymerised trimethyl fructose or of a methylated difructose. The tetramethyl y-methylfructoside isolated as described was a neutral colourless syrup which reduced potassium permanganat e vigorously. Although the material behaved essentially as a glucoside towards Fehling's solution some reducing compound (probably tetramethyl fructose) was present and as repeated dis-tillation failed to' remove this impurity the analytical figures were affected [Found (after two fractionations) C=52.31; H=8-66; OMe= 60.6 ; nD 1.4470 ; (after three fractionations) C = 52.33 ; H= 8-47 ; mD 1.4471.C,H,O(OMe) requires C = 52.80 ; H = 8.80 ; OMe = 62-0 per cent.]. Considering the method of preparation two stereoisomerides would be present in unknown proportion and thus the specific rotation cannot be compared with previous determinations. For c=1-192 [a] + 20'98O in ethyl alcohol 1488 THE CONSTITUTION OF POLYSACCHARIDES. PART I. In preparing tetramethyl y -methyl€ructoside by the above method about 20 per cent. of the crude syrup coluld not be dis-tilled although the reactions were conducted on trimethyl y-fructose which had been subjected to repeated distillation.This residue consisting of a viscous clear syrup was further examined. The material was glucosidic and gave on hydrolysis trimethyl y-fructose which was in turn converted into' tetramethyl y-methyl-fructoside. It woald appear that during the condensation with methyl alcohol some of the trimethyl y -fructose had undergone an extraneous change which is probably auto-condensation or polymerisation. Tetrarnet hyl y -Fruc tose. The hydrolysis of tetramethyl y-methylfructoside was carried out in 0.25 per cent. aqueous hydrochloric acid the concentration of the fructbside being adjusted to 1-0684 so as to render possible comparison with the results obtained in parallel work on the same compound prepared from sucrose.At the temperature of the room the reaction was slow and occasioned a fall in rotation. On continuing the hydrolysis a t looo the activity measured a t 15O increased from +24.30 to +30.7O in thirty minutes but the end-point was difficult to detect, on account of the extreme sensitiveness of the rotation with small fluctuations in temperature. I n a control experiment conducted on the same compound derived from sucrose the permanent value [a] The usual procedure was followed in isolating the sugar which boiled a t 148*5O/ 10 mm. (Found C = 50.88 ; H = 8.57 ; OMe = 53.2 ; n 1.4554. Calc. C=50*85; H=8*47; OMe=52.5 per cent.; n 1.4545). In every respect the sugar showed identical properties and physical constants with the tetramethyl y-fructose obtained from sucrose.After distillation the compound displayed slight down-ward mutarotation the permanent values in water and alcohol, respectively being [a] + 32.9O and 15.5'. + 29.6O was recorded so that the agreement is close. Speed of Condensation of Tetramethyl y-Fructose with Methyl A Icohol. The condensation with methyl alcohol was carried out in conjunc-tion with a duplicate experiment in which tetramethyl y-fructose from sucrose was used. A 1 per cent. solution of the sugar was dissolved in methyl alcohol containing 0.25 per cent. of hydroge THE CONSTITUTION OF POLYSACCHARIDES. PART 11. 1489 chloride and preserved a t 15O polarimetric readings being taken a t regular intervals. Typical observations are given. Time from contact of solvent and solute. Specific rotation. 21.9 Fall 80 9 9 26.3 }Rise Thereafter the solution was heated at 30-40° to complete the reaction the end-point being [a] + 59'9O. With tetramethyl y-fructose from sucrose the minimum rotation recorded is +19-So as compared with +19*l0 above whilst the end-point is [a] + 5 7 * 6 O compared with + 59'9O. Moreover on plotting the specific rotations graphically the curves representing the two parallel reactions were identical within the limits of experiment a1 err or. + 1 minute 4 minutes 14 9 19.1 1024 , 27.9 The above investigation was carried out in connexion with the Carnegie Trust Research Scheme and we desire to express our thanks to the Trust. We are also much indebted to Professor W. N. Haworth and Mr. J. G. Mitchell for access to results recently obtained by them in the study of the tetramethyl y-fructose present in sucrose. UNITED COLLEGE OF ST. SALVATOR AND ST. LEONARD, CHEMICAL RESEARCH LABORATORY, UNIVERSITY OF ST. ANDREWS. [Received October 18th 1 920.