1164 O’SULLIVAN : GUM TRAGACANTH. CXXIV. -Gum Trugacanth. By CORNELIUS O’SULLIVAN, F.R.S. I. Introductovy. FOR a number of years, I have occupied myself, amongst other things, with a study of the constitution of gum tragacanth, but the investiga- tion is still not so complete as I should have wished it to be. Recently, however, Widtsoe and Tollens (Ber., 1900, 33, 132) described some of the products of the sulphuric acid hydrolysis of the gum, and this fact leaves me no alternative but to place on record the results I have hitherto obtained, and to indicate the direction in which I am still working. 11. Earlier Investigutions of the Composition of the Gum. Of the proximate composition of gum tragacanth, very little seems to be known, and where the character and composition of the proxi- mate constituents are described, it is difficult to resist the conclusion that our knowledge is of little value.We have to begin with the broad statement that gum tragacanth consists of cellulose, &arch, bassorin, a gum like arabin, a little nitro- genous matter, sugar, and ash. We are not told the nature of the cellulose; we have 110 evidence that the granules contained in the cells are starch granules, for although, without dobbt, they are coloured blue by iodine, we have no collateral proof in support of the assumption. Again, I have been unable to ascertain why the cellular matter apparent under the microscope is written down as cellulose. It cer- tainly gives a blue coloration with Schultz’s reagent, but I can find no further evidence.Bassovin is R very imperfectly defined substance, and it would indeed be difficult t o identify it from the general description given of it (Gmelin-Kraut, 1862, 4, 650).0 SULLIVAN : GUM TRAGACANTH. 1165 Of the 8olubh gunz, nothing further appears to be recorded than that it resembles arabin. The nitrogenous matter is simply indicated by the presence of nitrogen, and concerning the sugar there is noinforma- tion ; but the knowledge of the ash conetituents is of course satisfactory. 111. Object of t7he Work. With these facts in my mind, and looking upon gum tragacanth as of the nature of plant sap containing reseswe matter in an advanced stage of elaboration, it appeared to me-with the light thrown upon the constitution of the gums in general by my work on the gums of the arabin group-that some clearer idea of the constitution of this gum could easily be obtained, and thence of the more imme- diate materials of reserve matter. From the general description of bassorin, I was led to believe I should have to deal with some gum acid or acids related to those of the arabin or geddin series in a polymerised state; this, however, was found not to be the case; when the gum was treated with water, the solution was only slightly acid, the amount of potash required to render the mucilage neutral being very small, and the solubility thereof being practically unaltered.IV. Mode of Pvocedure. When the gum is treated with water, it swells up very considerably, and at least one hundred times its weight of water must be added before any differentiation occurs.The addition of alkalis does not materially alter this, consequently, the usual method of fractionating such materials was not available. When diluted as t,hus described, however, the so-called cellulose, with jelly-like matter and starch, begins to separate (Kutzing, Ph&sophimha Botanik, i, 203; H. von Mohl, Bot. Zeit., 1857, 32), and by this means a moderately clear solution, difficult to filter, is obtained. The cellulosic shreds occupy a large space in the liquid, but all attempts to get them to settle satis- factorily met with but indifferent success. Consequently, I deter- mined a t the outset to act on the gum with dilute sulphuric acid, to examine the products of the reaction, and then, if possible, to trace each product back to the original material whence it was derived.The investigation was accordingly carried out in this direction, and yielded interesting results, but as I believe I can describe the nature of the gum without utilising this series of experiments, I shall leave them unrecorded. I may, however, say that it was the knowledge derived from a study of the products of the action of dilute sulphuric acid on the gum-in which I was most materially assisted by Dr. Stern, my assistant at that time-and especially of the products other VOL. LXXIX. 4 K1166 O’SULLIVAN : GUM TRAGACANTH. than sugars, which enabled me to deal with the problem in the manner described in this paper. The results, so far as the sugars are con- cerned, will be found in the work of Widtsoe and Tollens (loc.cit.), although no indication is there given as t o the immediate source of these substances. I proceeded as follows : When water is added to the gum, the pieces swell up t o a more or less transparent jelly, some of them being very nearly transparent, others quite white and opaque, whilst every intermediate stage is also observed. Examined microscopically, the transparent jelly masses are found to consist of irregular, blown out ” cells, the thin walls being filled with glairy matter; the opaque pieces consist of the same kind of irregular cells, but containing, in addition to the glairy matter in the interior, many granules, apparently starch, which are coloured blue by iodine, whilst the intermediate varieties are distinguished by the varying numbers of granules they contain, this number apparently increasing with the opacity, V.The Granules. Although the granules are distinctive, and are coloured blue by iodine, T thought it desirable to isolate them, and t o determine abso- lutely their identity with starch. When the gum is submitted to the action of freshly prepared copper- ammonium solution, obtained by treating copper turnings with ammonia in the usual way, nothing is left undissolved but bright, almost transparent granules, which are deposited on the addition of a large bulk of water. On decanting off the copper solution and washing the residue with water, the aqueous filtrate was coloured brown by iodine solution, whilst on treatment with dilute hydrochloric acid, a solution was obtained which mas coloured dark purple by that reagent.On treating the remaining granules with iodine, a coloration not as intensely blue as that yielded by starch with the same reagent was observed. It was difficult to obtain the granules in sufficient quantity t o admit of a determination whether or no they yielded maltose on treatment with malt extract, but in order to procure some evidence on this point, about 150 grams of the gum were treated with 10 litres of water, after soaking, and allowed to stand for some days. The insoluble portion was then strained out, by means of a linen filter bag, washed with cold water so long as anything dissolved, and then pressed. Under the microscope, the mass was seen to consist of unaltered granules and shrivelled, cell-wall de‘bris. It was treated with boil- ing water, cooled down to 60-65’, and digested for several hoursO'SULLIVAK : GUM TRAGACBRTH.1167 with malt extract. A microscopical examination showed that the granules had been-dissolved, and on evaporating the filtered solu- tion to a syrup, and digesting this with alcohol of sp. gr. 0,830, a clear solution and an insoluble residue were obtained. The residue was neglected, but the solution, freed from alcohol by distillation and evaporated again to a thick syrup, mas taken up by alcohol of sp, gr. 0-820, in which it was completely soluble. After freeing this solution from alcohol, the residue was examined. The figures yielded by the aqueous solution were as follows: sp. gr. 1.029, ago* + 8.0'; whence [aID + 53.1'. This number approaches that of dextrose so closely as to leave little doubt that we have that substance present.From this observation, I do not, however, think it can be said that we have not starch to deal with, still I feel I am justified in recording the observation that these granules, apparently consisting of starch, yield on treatment with diastase, not maltose, but dextrose. There is still one more point t o which attention should be drawn here, namely, that the cell-walls are dissolved by copper-ammonium solution. Numerous attempts were made to isolate definite entities from this solution, but the results were not satisfactory; hence the use of that reagent in the investigation was discontinued. VI. 2% Cellulosic Xuktunce. Since the cell-walls described above did not appear identical in every respect with ordinary cellulose, an examination of the substance seemed desirable.Accordingly, a portion of the original gum was treated with three hundred times its weight of boiling water, digested for some time, and then submitted t o filtration through linen bags. The residue left in the bags was repeatedly treated with boiling water, and was finally collected on filter paper, a hot water funnel being employed. The filtered shreds which comprise the whole of the cellular matter are completely soluble in freshly prepared copper-ammonium solution, in which only the granules remained insoluble when the whole gum was treated with the reagent. Examined microscopically, the cellular mass was found to consist of shreds and d%bris, a portion of which gave a blue colour with zinc chloride solution of iodine, the remainder being coloured reddish-brown; in fact, i t was quite obvious that the shreddy mass, although showing in its parts no material differences under the microscope, consist of two or more distinct substances.Besides the cellular shreds, a few hard, well-defined crystals, which are neither sulphate nor carbonate of calcium, are also present, but as they occur in extremely small quantity, I have not examined them further. * The number gives the length of the tube in which the observation was made, in mm. 4 ~ 21168 O'SULLIVAN : GUM TRAGACANTH. The amount of cellular matter present in the gum varied betwsen 1 and 3 per cent. A portion of it wae repeatedly extracted with boil- 'ing water, dehydrated with strong alcohol, and dried, h s t in a vacuum and then in a current of dry air at 100' On combustion : 0.322 gave 0.498 CO,, 0.175 H20, and 0.030 ash.C=46*51 ; H=6.65 per cent. (in substance free from ash). Ash = 10.3 per cent. These results appear to indicate that the cell-walls do not consist of cellulose proper, however closely they may resemble the oxycelluloses. The ash contained silica, lime, phosphoric acid, ferric oxide, &c. Action of Xulphuric Acid on t h Cellular Substance.-Some of the cellular matter, purified as described above, was washed, first with a dilute solution of potasb, then with dilute hydrochloric acid, and finally with water, till free from chlorine. The jelly thus obtained consisted of shreds, and was digested at 98-99' with a 5 per cent.solution of sulphuric acid for an hour. During the digestion, a marked odour of furfuraldehyde was observed. At the end of the time stated, the solution was filtered, and the insoluble portion washed with boiling water till free from sulphuric acid. The filtrate was neutralised with baryta water, the precipitated barium sulphate filtered off, and the filtrate concentrated in a vacuum to a syrup; this was then taken up with alcohol of sp. gr. 0.820, in which it was almost entirely soluble. After removal of the alcohol by distillation, the syrup which remained behind yielded, on standing, a crop of microscopic, skiff-shaped crystals of arabinose, such as this sugar yields at times on crystallisation from slighty impure solutions. The crystals wepe washed with dry methyl alcohol, and dried in a current of dry air at 100'.1.820 grams, dissolved in water and made up to 100 c.c., gave ago + 3-5', whence [a], + 96.2'. This figure is low for arabinose, but no lower than I have obtained before under similar conditions. The airdried crystals lost no weight at 100'. On recrystallising the preparation from water, crystals with the characteristic terminations of arabinose, as yielded by arabic acid, were obtained. These were washed with methyl alcohol and dried, first in air, and then at 100'. 1.054 grams, dissolved-in water and made up to 100 c.c., gave UY + 2-15', whence [aJD + 102*0°. As the low opticity of the sugar at the outset might have been due t o the presence of galactose, a portion of the original syrup was treatedO'SULLIVAN : GUM TRAGACANTH.1169 with nitric acid of the proper strengh, but no mucic acid was produced ; dextrose it might have been, but sufficient material was not at my disposal to determine the point. I consider, however, that most prob- ably the disturbing cause is the Eubstance of low rotatory power which is always prc;duced when arabinose is subjected to the action of sulphuric acid for any length of time. It is evident, therefore, that the cellular matter, on treatment with sulphuric acid as described, yields practically nothing but arabinose. The residue from this treatment, insolubIe in sulphuric acid, is a jelly-like mass which, under the microscope, appears to consist of a number of shreds with no definite structure, in fact it differs very little in appearance from the original cellular matter, Iodine solution colours some of.the shreds blue, others brown, whilst the remainder are unchanged. More of the shreds are coloured blue by zinc chloride solution of iodine than by iodine alone, but some are still only coloured brown; a solution of iddine and sulphuric acid also acts in much the same way, except that a larger proportion of the shreds appear to be coloured blue in this case. A few acicular crystals are still observable. It is clear from these experiments, therefore, that even after digestion with 5 per cent. sulphuric acid, the residue is not homogeneous, never- theless a portion of it was dried in a current of dry air a t looo, and subjected to combustion, to enable me to form an idea of the change, if any, which had taken place.C = 45.75 ; H = 6.52 per'cent. (in substance free from ash). Ash = 6 -15 Hence there is a clear indication o€ a diminution of carbon and hydro- gen, which cannot be explained by the elimination of an arabinose group alone ; the time at my disposal, however, did not permit of my going more deeply into the question. It may be pointed out, however, that the sugar solution contained a considerable quantity of uncrystal- Zisable matter, a careful examination of which would, very possibly, throw some light on the point. With the object, however, of ascer- taining whether or not the above cellulose-like substance contained a fixed cellulose residue, it was repeatedly treated with bromine water and then with ammonia.Under this treatment, it soon became evident that the substance was dissolving gradually, and after about one-third of the original substance had disappeared, the residue was collected, thorouglg washed with boiling water, dehydrated with strong alcohol, and finally dried in a current of dry air at looo. On combustion of a portion of it, 0.336 gave 0.531 GO,, 0.186 H,O, and 0.0195 ash. per cent.1170 O’SULLIVAN : GUM TRAGACANTH. 0.284 gave 0.457 CO,, 0.1595 H,O, and 0.008 ash. C=45*15 ; H=6*42 per cent. (in substance free from ash). Ash= 2.8’1 per cent. Here, again, we have a diminution in the carbon and hydrogen, and ako, as might be expected, in the ash, with.a gradual approach, so far as the carbon and hydrogen are concerned, t o the normal composition of cellulose, but since the substance gradually went into solution under the influence of bromine and ammonia, the absence of an unalterable residue of normal cellulose was conclusively proved.Hence it is perfectly clear that the cell walls, seen under the microscope, although they yield many of the reactions of normal cellulose, have so far no ascertainable relationship to it, Whether the substance or substances are converted in the ordinary process of anabolism into true cellulose or not, it is impossible to say, and the series of changes by which such a transformation could be effected is not easy to imagine. I can only repeat that the cellular matter of gum tragacanth yields many of the reactions of true cellulose and arabinose on treatment with sulphuric acid, but leaves no permanent residue when acted on with bromine and ammonia.VII. The Portion of Gum Tragacanth soluble in Water. Sample I. On evaporating the aqueous filtrate from the cellulosic shreds, a gelatinous scum forms on the surface of the liquid, and on cooling the solution, after some concentration, a considerable quantity of jelly-like matter falls out ; in fact, the solution becomes a jelly if the concentra- tion is carried far enough. An attempt at fractionation was made by skimming off the scum and pressing out the jelly, but no satisfactory results were obtained ; hence the solution was concentrated to about one-sixth of its original bulk, and alcohol added gradually in small portions at a time till a ‘‘ break ” was effected, when further addition of alcohol produced a bulky, curdy precipitate, ‘‘ fraction a.” This could be readily separated from the clear, alcoholic mother liquor, which we shall call “fraction p.” VIII.Fraction p. The alcohol was distilled off, and the aqueous solution concentrated in a vacuum to a syrup. On carefully adding alcohol of sp. gr. 0.840 t o this syrup, a flocculent precipitate ‘‘ fraction 7’’ was thrown out, which was allowed t o settle, and on decanting off the clear, supernatant solution and adding more alcohol, a waxy precipitate, similar in appearance to that yielded by gerlda gum, was thrown out. An examination of fraction p was first made.O'SULLIVAN : GUM TRAGACANTH. 1171 This precipitate was dissolved in a little water, roprecipitated by alcohol of sp. gr. 0*830 in presence of hydrochloric acid, and handled in precisely the same way as described by me in connection with the gedda gums (Trans., 1891, 59, 1032). It was divided into three portions, fractions I, 11, 111, fraction I being the least soluble.These fractions were freed from hydrochloric acid by repeated pre- cipitation with alcohol, dehydrated with strong alcohol, and finally dried in a current of dry air at 100' under 250 mm. pressure. The optical activity of each fraction was then determined with the follow- ing results : Fraction I. [.ID - 78*0°. ,, IT. [alD -58.8 . ,, 111. [aID -88.1 . From these it is evident that we have not a homogeneous body to deal with, although the product is not such a mixture as was the case with the gedda gum acids. A barium salt of fraction I1 was prepared by neutralising a solution of i t at the boiling temperature with clear baryta water, using neutral Iitmus as indicator.The solution of the barium salt thus prepared mas concentrated, precipitated by alcohol, and dehydrated by alcohol of sp. gr. 0.820. On analysis, 0.783 gram of this salt, dried in a current of dry air at loo', gave 0.037 gram of barium sulphate, BaO =3*23 per cent. This figure indicates a molecular weight corresponding t o that of a fraction of the gum acids of gedda gum C (Trans., 1891, 59, 1073); the rotations, too, of the fractions from the two sources are nearly equal in amount, although opposite in sign. Hence the soluble gum acids of gum tragacanth resemble those of gedda gum, but are lsvorotatory instead of dextrotatory. VIIIA.Action of SuZphuric Acid on Fraction /3 p r i j e d . Fractions I1 and I11 of the soluble gum acids mentioned above were mixed, dissolved in water to a 30 per cent. solution, and this was heated in the water-bath to loo', sufficient diluted sulphuric acid being added to make the mixture contain 3 per cent, of that acid. The digestion was continued for 15 minutes, and, after cooling, the acid was neutralised with clear baryta water. The whole was then con- centrated in a vacuum to a syrup, because the barium sulphate could not be satisfactorily removed by filtration. This syrup was pre- cipitated by excess of alcohol of sp. gr. 0.820, and the clear alcoholic solution containing the sugar or sugars was freed from alcohol by distillation, concentrated in a vacuum to a syrup, which was then1172 O’SULLIVAN : GUM TRAGACANTH.taken up by as little dry methyl alcohol as possible; a small amount of barium salt remained insoluble. The clear methyl alcoholic solution, -on standing, yielded stellate groups of crystals, with the terminations and general characteristics of arabinose. These were collected, washed with methyl alcohol, and their optical activity determined. It was found to be [aID +90*8O. As, however, this number was far too low for arabinose, the crystals were redissolved in methyl alcohol, a little insoluble matter filtered off, and the solution again allowed t o crystallise. Three crops of crystals were collected, namelJr, a, 6 , and c, and the optical activity of each was determined, with the following results : a [a], + 102O.6 [a], +104. c [aj, + l o 6 . These numbers clearly agree with those accepted for arabinose, and although the crystals have not absolutely the same form as that assumed by arabinose obtained from arabic acid, there can be no doubt about their identity with that substance. The amount of arabinose, including the uncrystallisable matter in the mother liquors, yielded by the gum acid varied between 73-0 and 76.7 per cent. of the dry material taken. When the percentage of arabinose thrown out in arriving at this stage is considered, the high molecular weight attributed t o the original soluble gum acid is amply confirmed. The precipitate thrown down by alcohol, from which the sugar solu- tion was separated by decantation, was treated with a little water, and sufficient sulphuric acid added to convert nearly all the barium present into sulphate. Alcohol was then added until the solution ‘‘ broke.” In this way, the barium sulphate easily falls out and can be separated, whilst the partially degraded gum acid is retained in the dilute alcoholic solution.This can be precipitated by the addition of strong alcohol, and may be easily freed from excess of sulphuric acid by redissolving in water and reprecipitating with alcohol several times. After freeing from sulphuric acid, the degraded gum acids were divided into two fractions by means of alcohol. Each fraction was converted into a powder by means of strong alcohol, Barium salts were also made by neutralising a portion of each with clear baryta water, and precipi- tating the concentrated solutions with strong alcohol, and these were dried in the usual way.An examination of the free acid and barium salt of each fraction was made, with the following results : salt thereof BaO = 10 per cent, Free acid. Most soluble portion gave +335O, and the bariumO'SULLIVAN : GUM TRAGACANTH. 1173 Free acid. Least soluble portion gave [.ID + 33*2', and the barium salt thereof BaO = 8.8 per cent. I n each of theso instances, the substance was dried in a current of dry air a t 100'. If these factors are compared with those obtained for tri- and di- galactangeddic acids (Trans., 1891, 59, 156-157), an almost exact agreement is observed, hence these degraded gum acids are a mixture of the tri- and di-galactangeddic acids.I n further confirmation of the galactan nature of these degraded gum acids, a determination was made of the amount of muciq acid formed from them by treatment with nitric acid in the usual way. Seventy to eighty per cent. of mucic acid was thus obtained, and this result confirms the identity of these degraded acids with those of the geddic series. As was naturally to be expected, on further treating these acids with sulphuric acid, galactose and a further degraded acid, corresponding to, if not exactly identical with, geddic acid were obtained. I n order to determine whether or not other samples of gum tragacanth would yield like results, a fresh sample of the gum from a different source was examined. IX. Portion ofthe Gum soZubEe in Water. Sarnph 11. A. Examination of Praction corresponding to p- Fraction. On treating the second sample of gum tragacanth in precisely the same way as described for the first, a product was obtained correspond- ing to /?-fraction, which yielded fractions with opticities [a], - 9 6 O to [a], - 115' ; these gave barium salts which contained from 4 to 5.4 per cent.BaO, the evidence, curiously enough, being that the neutral- ising power for barium increases with rising rotatory power. When these results are compared with those obtained for the similar acids yielded by sample I, it becomes evidebt that we are dealing with members of the same series. Sample I. Free acid [a]D - 88'; barium salt 3-23 per cent. BaO. )) I r e ?? [.ID - 9 6 ; ,, 44 1, , I 9 9 11* 19 [a]D -115; ,? 5.4 99 99 It is clear, therefore, that although the gum acids belong t o the same series, the number of arabinose yielding groups is less in the more optically active acids than in those producing a smaller rotation.IXB. Action of Xulphwic Acid on P-Fraction. SampZe 11. We have now to decide whether or not this is the only difEerence between the gum acids. To determine it, fractions of opticity between1174 O'SULLIVAN : GUM TRAGACANTH. [ a ] , - 95" and [ a ] = - loo", neutralising about 4 per cent. barium, were mixed and digested with 3 per cent, sulphuric acid for 30 minutes. The products were separated as usual, and 72.5 per cent. of sugar of [ a l D +looo was obtained, This was recrystallised from methyl alcohol, and several crops were separated which possessed the charac- teristic terminations of arabinose and yielded the following factors : [ a ] , +101-104".K 101-105, These figures conclusively prove that the sugar is arabinose. Much non-crystallisable matter possessing a low opticity was present in this as in-the first case, but whether this is produced by the action of sul- phuric acid on arabinose, or represents a sugar which, like dextrose, crystallises with difficulty in presence of contaminating matter, I am unable to say. Amongst some of the products of the react.ion, galactose is present without doubt, for on treating this uncrystallisable matter with nitric acid, mncic acid is produced, X. The Degvaded Gum Acids-from Sample 11. Using the method of purification and fractionation already men- tioned, products were obtained having opticities varying between [ .ID + 34' and [ a 3, + 40.6", and yielding neutral barium salts con- taining 10-11.8 per cent.of BaO. These figures agree almost abso- lutely with those obtained when the first sample of the gum employed was subjected to similar treatment, notwithstanding the fact that the original acids acted on by sulphuric acid varied considerably in optical activity. I may here indicate that these degraded acids are necessarily mix- tures, the more highly dextrorotatory portions being those from which a portion of the constituent yielding galactose has been removed, whilst those of lower opticity have either lost less of this constituent, or only part or the whole of that yielding arabinose. XI. Purlher Bydrolysis of the Degraded Acids. The fractions having [ aID + 34" to + 4 0 * 6 O just mentioned were mixed, dissolved in water to a 30 per cent.solution, and digested at about 98' for 2 hours with 4 per cent. sulphuric acid. Clear baryta water was added t o neutralise the cooled liquid, and then alcohol of sp. gr. 0.850 so long as a precipitate was formed. This consisted of the bariumLsalt of the degraded acids, some sugar, and barium sulphate. The clear alcoholic solution containing the greater part of the sugar or sugars was decanted off, and the precipitate further freed from sugar by digesting with alcohol of sp. gr. 0.870, redissolving in a littleO'SULLIVAN : GUM TRAGACANTH. 1175 water without separating the barium sulphate, and reprecipitating several times with alcohol of the strength mentioned.The alcoholic solutions thus obtained were mixed, the alcohol dis- tilled off, and the residual solution concentrated in a vacuum to a syrup. On examination, the solid matter in this was found to have an optical activity [ aID + 60--65O, and to contain barium salts of gum acids. Consequently the whole of the syrup was digested f o r some time with alcohol of sp. gr. 0.520 in a flask with a reflux con- denser. Much gummy matter remained insoluble, and, on cooling, more was precipitated, leaving the alcoholic solution clear. When this solution was again freed from alcohol and concentrated in a vacuum, a syrup was obtained which crystallised on standing. After a few recrystallieations, the substance had a crystalline form similar to that exhibited by galactose under certain conditions.[ aID + 80-83O. On examination, the crystals yielded the following results ; K = 93-95', and on treating them with nitric acid under the most favourable conditions, 72-75 per cent. of mucic acid was obtained, Hence the sugar is undoubtedly galactose. Returning now to the barium salt, partially freed from the sugar as just described. It was treated with the least possible quantity of water, and an attempt unsuccessfully made to separate the barium sulphate by filtration; it was not possible to obtain a clear filtrate, consequently sulphuric acid was carefully added until the barium present was completely precipitated, and then alcohol until a '' break " was effected, when the barium sulphate with a little gum acid separated out, leaving the alcoholic solution clear.This was decanted off; it contained the greater part of the degraded gum acid. Treatment with alcohol of sp. gr. 0.820 enabled me to separate this into two fractions. a, a less soluble one, and b, one more soluble; these were dried in the usual way and the optical activity of each determined : a gave [aID +93*So. 6 ,, [aID +94*1°. A barium salt of each was prepared, that of: a yielded BaO= 18.1 per cent. b ?t 17-6 I, In another series of experiments in which the acids with opticities between [ a]D + 34" and 40° were acted upon with 4 per cent. sulphuric acid for 4 hours, degraded acids were obtained in precisely the same way as described, of which the optical activity was somewhat higher, namely, [aID + 109.7-109*5', but the barium salts of these yielded practically the same numbers, namely, BaO= 18.6-1 8.03 per cent.1176 O'SULLIVAN : GUM TRAGACANTH.To enable me to form some notion of the composition of these sub- stances, one of the apparently purest barium salts, dried in dry air at 100' under 275 mm. pressure, was burned with the fcllowing results : C=34*41; 0.474 gave 0-573 CO,, 0-184 H20, and 0.113 BaCO,. C2,H,,0,1,Ba0 requires C = 34.45 ; H = 4.49 ; BaO = 19.10 per cent, I have attributed (Trans., 1891, 59, 1054) the formula C23H38022 to geddic acid of [ a ] D + 71' and C2,H,,02,Ba0 to its barium salt. The latter formula requires C = 33-70 ; H = 4.64 ; BaO = 18.58 per cent. On comparing these results, we observe a broad agreement between them, but one not sufficiently close as to enable us to state that the degraded acids from both sources are identical, although it can hardly be doubted that there is a close relationship.The composition of the two acids is no doubt the same, namely, C23H38022, but geddic acid has uD + 71', whilst the acid under consideration has aD + looo to logo, and the barium salts are different, as is shown by the analyses. I n dealing with non-crystallisable substances of this kind, and taking cognisance of the fact that they are exceedingly difficult to dry, I do not wish to draw more inferences than the analytical numbers warrant. But, I may say these have not been based on one set of experiments, but on many. Hence, I feel justified in concluding that the degraded acid of tragacanth, if not absolutely identical with geddic acid, very closely resembles it, and, on comparing the products obtained from both the original undegraded acids after being acted upon by sul- phuric acid for 10 to 15 minutes, I feel convinced that the partially degraded acids obtained in each case are very closely related, if not identical.They both have the game optical activity, and their barium salts yield the same amount of baryta, as has been shown above. It is sufficiently clear that the primary acids of gum tragacanth are not identical with those obtained from gedda gum, although they have many properties in common, the chief differences being that the former acids are laevorotatory, whilst the latter are dextrorotatory, and, unlike the gedda gum acids, do not yield ayabinon as a pro- duct of their partial hydrolysis with sulphuric acid.Possibly, this occurrence of arabinon-y ielding acids may have something to do with the differences in the direction of the rotation, especially as the products after the elimination of the arabinon groups agree so closely in optical properties and neutralising power. It is thus evident that the- acid8 of gum tragacanth, soluble in dilute alcohol are polyarabinon-trigalactan-geddic acids, namely, XC oHlsO,, 30, 2H20010,C23H38022, with the differences indicated above and by this formula. H = 4.32 ; BaO= 18.51.O’SULLIVAN : GUM TRAGACANTH. 1177 One of the fractions of the soluble gum acid under consideration, namely, that with [all) - 8 8 O and containing 3-23 per cent. BaO in the barium salt, is probably represented by the formula : 1 1 oH,cO,, 3Ci 2H2 00,o ,c, 3H31302,~ BaO.This requires 3.29 per cent. BaO, and should yield on partial hydro- lysis 71.7 per cent. of arabinose. It has been shown above that between 72 and 76 per cent. of moderately pure arabinose was obtained from the mixed gum acids. From these facts, it appears clear to me that the constitution of the soluble gum acid of gum tragacanth has, broadly, been sufficiently well established and indicated. The Nitrogenous Matter. Of this, I cannot a t present say more than that it is less soluble in alcohol than the gum acids just described, in fact, it is fraction y, mentioned on page 1170 ; that, when once precipitated and dried, it does not again give a clear solution in water, whilst it is soluble, with a slight brown colour, in potash, and that it does not give the proteid reactions given by the nitrogenous substance from gedda gum (Trans., 1891,5Q, 1061).XII. Fraction a. Flit? Jelly-like Portion. I ‘ Bassorin.)) It has been pointed out in the earlier portion of this paper that the clear filtrate, obtained in the process of removing the cellulose-like shreds from the product of the digestion of gum tragacanth with a large bulk of water, yielded on evaporation a gelatinous scum and a jelly-like deposit, and that the solution, when brought to this state of concentration, yields the bulky precipitate with dilute alcohol which was mentioned when the elimination of the soluble gum acids, fraction p, was described, This precipitate was repeatedly washed with dilute alcohol, dissolved in the least possible quantity of water, acidified with hydrochloric acid, and placed upon a dialyser to free it from salts of calcium, magnesium, &c.When the solution on the dialyser was free from hydrochloric acid, dilute alcohol of sp. gr. 0.870 was added to i t ; a white, curdy precipitate was thrown out, which, on decanting off the supernatant liquid and adding stronger alcohol, became shred-like. This was well washed with alcohol of sp. gr. 0.850, and then de- hydrated by alcohol of sp. gr. 0.S25, when it could readily be rubbed down to a powder. It is undoubtedly the substance known as 6ussorin.1178 O'SULLIVAN : GUM TRAGACANTH. This preparation, dried until constant in a current of dry air at loo', gave [ + 98O. I do not give the analytical numbers, because up to this point I have no criterion of the purity of the substance.It may, however, be here pointed out that the opticity of the gum acids soluble in alcohol of sp. gr. 0.870 is almost identical with the above number, although opposite in sign. A barium salt of the preparation was made by neutralising an aqueous solution thereof with clear baryta water. While doing this, it was noticed that at the point where the baryta water dropped into the solution, and, consequently, where the reagent was in excess, a lemon-yellow colour was developed. By cautious addition of the reagent, however, a neutral solution was obtained without the pro- duction of much colour, and this, on addition of alcohol, yielded a curdy precipitate, which, after being mashed with dilute alcohol and dried in a current of dry air at looo, coutained BaO = 9.2 per cent.XIIA. Conversion of Bassorin into a- and P-Tragacanthan-xyZan- bcmsoric Acids. When an attempt was made to dissolve the above barium salt for the estimation of the barium, a portion was found to be insoluble; hence, it seemed as if some change had taken place, presumably under the influence of the alkali employed. Numerous attempts were made to purify the original bassorin by partial precipitation with alcohol, but with no great success, conse- quently the whole preparation was converted into a barium salt by treatment with baryta water. After this salt had been dried in a vacuum over sulphuric acid, and finally in a current of dry air at looo, it was observed that a part was insoluble in water, as already mentioned.The insoluble portion admitted of being easily filtered off. The filtrate was submitted to dialysis in presence of hydro- chloric acid until free from barium. To tbe solution, alcohol was added, and the precipitate, after redissolving and reprecipitating with alcohol several times until all the hydrochloric acid had been removed, was strained out, pressed, dehydrated, dried in a vacuum over sul- phuric acid, and finally in a current of dry air at 100'. 0.971 gram, dissolved in water and made up to 50 c.c., gave a solu- tion of sp. gr. 1.0063 and an opticity a:' +2'5', whence [a], + 128'. D = 4.27. It will be noticed that the [a], is much higher than that of the original substance, and the divisor much greater than that of any known carbohydrate.Further, it is obvious that we have in no way to deal with the original material.O'SULLIVAN : GUN TRAGACANTH. 1179 On attempting to prepare a barium salt of this substance by neutralising its aqueous solution with clear baryta water, the addition being made drop by drop, a gelatinous precipitate was immediately thrown out. This precipitate, formed in as neutral a solution as pos- sible, was collected and mashed, first with water, then with dilute alcohol, and finally dehydrated and dried as usual. 0.448 gave BaSO,=0*116. BaO= 17.0 per cent., a result which confirms the conclusions arrived a t from the [a],, and D. As has been mentioned previously, the neutralisation of the original solution of bassorin with clear baryta water gave rise t o no pre- cipitate of any kind, whereas the neutralised solution yielded on pre- cipitation with alcohol a substance which, after drying, was not again entirely soluble in water.The portion, of the neutralised precipitate insoluble in water was also insoluble in hydrochloric acid, but mas soluble in potash or ammonia. From these facts, it is evident that the substance known as bassorin is not a stable compound, but undergoes some change under the influ- ence of the alkali employed in the preparation of one of its salts. What the nature of this change is, I do not propose to discuss a t present ; it is sufficient to say that when an excess of milk of lime or baryta water is added to a strong solution of bassorin in water, a dense, curdy, light lemon-yellow precipitate is thrown down, which can be collected, washed with alcohol, dehydrated, and dried at 100' without developing a more intense coloration.On treating the dry mass with an excess of dilute hydrochloric acid, much carbon dioxide is evolved, and a large, semi-curdy portion remains insoluble ; this can be easily filtered off, the filtrate being perfectly clear. The insoluble portion admitted of being washed with cold water, and dehydrated and dried as usual by treatment with strong alcohol. This may be called fi*action B. Addition of alcohol to the clear acid filtrate yielded a curdy precipi- tate calledfractiofi A . On analysis : We shall deal with the latter first. Frccction A . The white, curdy precipitate was washed free from hydrochloric acid by dilute alc~hol, and was pressed and,dried.A small portion of this substance was found to be insoluble in water. This was collected, and alcohol added to the filtrate so long as a precipitate was produced. On drying as usual, this yielded the following results on analysis : 0.83 gram, dissolved in water and made up to 50 c.c., gave a solution of sp. gr. 1.00696 and a:' +4.5', whence [aID + 135. D=4.19.1180 O'SULLIVAN : GUM TRAGACANTH, It will be observed that this opticity is still higher than that of the substance with which I started. Is the substance even now homo- geneous ? To ascertain this, an aqueous solution of the acid was treated with sufficient clear lime water to precipitate about half the solid matter present. A bulky precipitate of calcium salt was thrown out, which was easily filtered, and could be readily washed with water and dilute alcohol ; this was pressed and dried as before.We call this portion fruction A,. To the filtrate, hydrochloric acid in excess was added, and then alcohol so long as a precipitate was formed. This mas collected on a filter, washed with dilute alcohol until free from chlorine, and then dried. Both fractions were redissolved in water containing hydrochloric acid, then reprecipitated with alcohol, and washed free from the acid. They were again redissolved and reprecipitated several times, and analyses of each made, with the following results. Fraction A,.-1.288 grams (dried in dry air at 100' under 300 mm. pressure), dissolved in water and made up to 50 c.c., gave a solution of sp.gr. 1.01062 and u'g0 + 7-14', whence [:.ID + 138.6" and D 4.12. Fraction A,.-1.700 grams, dissolved in water and made up to 50 c.c . gave a solution of sp. gr. 1.01404 and u y + 9*4', whence [a]= + 138.2' and D 4-13. These results justified me in considering that I had a pure substance Combustions of fractions A, and A,, dried in ,dry air at looo and Fraction A,.-o*341 gave 0.537 CO,, 0.169 H,O, and 0.003 ash. Fraction A,.-0.338 gave 0.530 CO,, 0.166 H,O, and 0.004 ash. We call it fraction A,. t o deal with, and accordingly I mas able to proceed to its analysis. 300 mm. pressure, were made, with the following results : C = 43.43 ; H = 5.56 ; ash = 0.80 per cent. C = 43.41 ; H = 5.52 ; ash = 1.20 per cent. The ash in both cases was mainly calcium carbonate.and D, namely, that the substance is homogeneous. which requires : These numbers confirm the conclusions arrived at from the opfiicities The empirical formula calculated from these numbers is C24H36021, C = 43.63 ; H = 5.46 per cent. Barium 8ak-A portion of the acid used in the foregoing analysis was dissolved in water, neutralised at the boiling point with ammonia, neutral litmus paper being used as indicator, and barium chloride solu- tion added so long as a precipitate was formed. The precipitate SOO’SULLIVAN : GUM TRAGACANTH. 1181 obtained was collected and washed with cold water until free from chlorine,* The barium salt so prepared can be easily dried in the usual may, and rubbed down to a fine powder. On analysis : 0.388, dried in dry air at looo, gave BaCO, 0.096.BaO= 19.21. C,,H,,O,,,BaO requires BaO = 19 94 per cent. Determinations as BaSO, yielded the same results. behaves in like manner, but its solubility in water is slightly greater. Calcium XaZt.-This salt is prepared similarly to the above, and 0.326, dried in dry air at looo, gave CaCO, 0.048. CaO= 8-24, C,,H,,020,Ca0 requires CaO = 8.02 per cent. Further details concerning these products need not now be given, as I propose to treat the subject more fully in a future paper. h U u e r SuZt.-An attempt was made to prepare the silver salt of the acid, but the task was beset with many difficulties. About 0.5 gram of the acid was dissolved in water, neutralised with ammonia, and the solution made up to about 45 C.C. On adding ordinary silver nitrate solution to this, the whole was converted into a transparent jelly of sufficient density to admit of the beaker being inverted without dis- turbing its contents.One hundred c . ~ . of water were added, and the mixture well stirred and thrown upon a filter, the operation being conducted in a room lighted by ruby light. Much of the liquid passed through, but it was difficult to free the precipitate entirely from its mother liquor. This end, however, was t o some extent achieved by pressing the gelatinous mass between folds of filter paper, but on attempting to wash the fairly dry substance with water, i t again swelled up, and the above process of filtration and drying had to be repeated. On drying in a vacuum over sulphuric acid in the dark, the salt became brown, and when an attempt was made to dry it in a current of dry air at looo, it blackened completely.Under these circumstances, i t seemed hardly worth while to estimate the silver in the preparation, but in the hope that an analysis might lend some support to the conclusions indicated by the previous experi- ments, a portion of the salt, dried in a vacuum over sulphuric acid (not, however, until constant, owing to slow decomposition of the salt) was nnslysed, with the following results : 0,372 gave 0.11 AgC1. Ag=22*2. C,,H,,O,,,Ag,O requires Ag = 24-77 per cent. * The barium salt is sufficiently soluble in cold wntcr t o give a precipitate with silver nitrate, that is, the silver salt of the acid is less soluble than the barium salt, consequently the absence of chlorine must be proved by silver nitrate iu presence of nitric acid.VOL. LXXIX. 4 1,1182 O’SULLIVAN : GUM TRAGACANTR. This result is, considering the circumstances, sufficiently close to support the conclusions based on previous experiments. To the gum acid giving rise to these salts, the name a-tragacanthccn- xyZan-bassoric acid may be given, for reasons which may simply be indicated here. Hydrolysis of a-Tragacanthan-xylan-basswic Acid. When a 20 per cent. solution of the acid is digested for 20 minutes at the temperature of a boiling water-bath with a 5 per cent, solution of sulphuric acid, a sugar is obtained on the one hand which proves to be a pentose having the opticity [a], - 30’ or thereabouts. This is a new sugar which I propose to name “t~agacanthose ” (it may be I-xylose, but I have not sufficient of the substance to enable me to decide the point at present); on the other, a new acid is produced the barium salt of which was found on analysis to contain 22.5 per cent.BaO. ClgH2,01,Ba0 requires BaO = 23.07 per cent., and the acid from this salt on combustion gave numbers corresponding with those required by the formula C19H2801p The action of sulphuric acid upon a-tragacanthan-xylan-bassoric acid-[ a ] D + 135O-is represented by the following equation : C24H36021 + H2° = C5H1005 + C19H2f401? Tragmanthosc Xylan-bassoric acid. - or 1-xylose. The opticity of the new gum acid in neutral solution (potassium salt) was found to be [ a ] , + ZOOo or thereabouts. This acid is only slightly soluble in cold water, and its barium and calcium salts are almost insoluble.On acting upon xykm-bassoric acid still further with 5 per cent. sulphuric acid in the same way as on the original substanee, another sugar, xylose, was eliminated and another new acid, namely, bassoric acid, was produced. The action of sulphuric acid is represented by the following equation : C19H28017 + H2° = c,H1oo, + cl*H,oo,, X ylose. Bassoric acid. The sugar produced was proved to be xylose by its crystalline form, K, optical activity [ + 21’ (it has a high temporary activity), and by the depression of the freezing point of its solutions. Bassoric acid is only very slightly soluble in cold water, it gela- tinises on treatment with boiling water and separates from this jelly, on cooling, in much the same way as inulin does from its solutions.The optical activity of the acid in alkaline solutions was [ a ] D + 255”, and the neutral barium salt contained BaO=28 to 29 per cent.O'SULLI VAN : GUM TRAGACANTH. 1183 Cl,H,801,Ba0 requires BaO=28*S per cent. This salt is almost insoluble in water. I shall not describe this final acid further on this occasion, but I may point out that the composition seems to indicate a combination of two hexans with C,O,. I have suflicient acid to determine this point and other things con- nected therewith definitely, and the work is in hand. Fraction B. The crumb-like substance, left insoluble in hydrochloric acid when the dried precipitate obtained with milk of lime was treated with that reagent (p. 1179)) was washed free from hydrochloric acid with cold water, rubbed down to a powder with alcohol of sp.gr. 0.820, dried in a vacuum over sulphuric acid, and then in dry air a t 100". 0,972 gram so dried, dissolved in the least possible quantity of potash, and the solution made up to 50 c.c., gave aim +6.0", whence Barium XaZt.-A portion of the acid was converted into barium salt by precipitating the neutral potassium salt solution with barium chloride, filtering off the precipitate, mashing free from chlorine, and drying it with alcohol, then in a vacuum over sulphuric acid, and finally in dry air at 100'. 0.378 gram of the salt thus dried was treated with excess of dilute hydrochloric acid in the cold for 24 hours, and the insoluble residue filtered off and washed free from hydrochloric acid.The barium in the filtrate was precipitated in the usual way as sulphate. It amounted to 0-107 gram BaSO,, hence BaO = 18.6 per cent. The percentage of BaO agrees sufficiently closely with that yielded by the barium salt of fraction A, to indicate that we have a substance of the same composition to deal with, but there is no evidence, so far, that the material is homogeneous. I converted the whole of the portion of [a], + 154O into barium salt. It may be pointed out that unless the solution of the ammonium or potassium salt is not moderately concentrated, the barium salt is not thrown out in an easily filterable condition on the addition of the barium chloride. The barium from the salt was extracted with hydrochloric acid as already described, and the free insoluble acid was divided into bwo fractions by dissolving i t in the least possible quantity of ammonia and partially precipitating the solution with barium chloride.The first precipitate, which may be called B,, after continued stirring was [ u ] D + 154'.1184 O'SULLIVAN : GUM TRAGACANTH. filtered off, and a second, B,, precipitate was obtained from the filtrate by adding more barium chloride solution, Both fractions were dried and freed from barium by treatment with hydrochloric acid as described already. Fraction B,, 0.766 gram (dried in a vacuum over sulphuric acid and in dry air at loo'), dissolved in the least possible quantity of dilute potasb, and the solution made up to 50 C.C. gave 2:O +5.0°, whence [.ID + 163.2'. Fraction B,. The fraction from the second precipitate with barium chloride mas treated and examined in the same way. 0.684 gram in 50 c.c dilute potash solution, gave +4*5', whence [.ID + 1645'. These numbers are in sufficient agreement to warrant the conclusion that the substance dealt with is practically homogeneous. On com- bustion of a portion of each of the fractions, the following results were obtained. Fraction B,. 0.3525 gave 0.5565 CO,, 0.1795 H,O, and 0.0020 ash. Fraction B,. 0.3715 gave 0.5580 CO,, 0.1870 H,O, and 0.0005 ash. BP B,. C.. . . . 43-30 per cent. H.... 5-68 ,, 1 5.59 7, } free from ash. Asb.. 0.57 ,, 0.17 ,, 43.24 per cent. Calculated on substance These numbers agree sufficiently well among themselves, and if they be compared with those obtained for a-tragacanthan-xylan-bassoiic acid, the agreement is so close as to warrant the conclusion that the two acids are isomeric. 1 shall call this acid P-t~~gacant~ccn-~~~a?a- bassoric mid. This conclusion was confirmed by an examination of the products of the sulphuric acid, hydrolysis of the /I-acid, the laevorotatory sugar, xylan, the intermediate acid [ aID + 200' (barium salt of which contains 22.46 per cent, BaO), and the final acid with [a]= + 255'and a barium salt containing 28.8 per cent. BaO being obtained, So far, these two acids are described with sufficient distinctiveness to enable anyone t o recognise either of them, and in this position I leave them for the present. I hope in the immediate future to describe bassorin itself more accurately, to indicate the reactions by which it is connected with a- and P-tragancanthan-xylan-bassoric acids, and to describe more closely the laevorotatory pentose, which is the first product of their hydrolysisCONDENSATION OF PHENOLS WITH ESTERS. 1185 by sulphuric acid. There can be no doubt that xylose is the second sugar eliminated. Of this, however, I may have something further t o record later on. Both xylan-bassoric and bassoric acids were eliminated from the products of the action of sulphuric acid on theundifferentiatedgum, biitit was difficult to prepare pure substances, as mixtures of tragacanthan- xylan-bassoric acids with these acids existed amongst the products. I: have to thank my assistants, Dr. A. L. Stern and Mr. J. A Walker, M.A. (Oxon.), for the helpful assistance they have given me in carrying out the work of which, I in justice must say, this paper is only a brief summary, but which I hope to amplify in the immediate future, more especially in dealing with the cellulosic residue, the nitro- genous constituent, and bassorin.