S L V . - T h e Dete!rmination of the. Molecular Weights of the Carbo-hydrates. By HORACE T. BROWN and G. HARRTS MORRIS Ph.D. BUT very little is known with certainty at the present time about the true molecular composition of the majority of the carbohydrates. The vaponr-density method is obviously inapplicable to them and chemists have had for the most part to be satisfied with a determi-nation of the niizzimum size of their molecules from a study of their somewhat indefinite metallic derivatives and of their acetyl-compounds. Musculus and Meyer (BUZZ. Soc. Chim. [2] 35 370) have attempted to determine the relative size of the molecules by observing the rate of diffusion of the different members of the group but, although the results are of inierest they do not go far in deter-mining the particular point we have before us.In the case of dextrose and levdose there is a considerable amount of positive evidence that their molecular composition is expressed by the formula C6H120s. This is shown by their intimate relation to the hexahydric alcohol mawnitok and by the recent researches of Kiliani on their cyanhydrins which yield hydroxy-acids containing seven atoms of carbon. With regard to the higher carbohydrates the knowledge we possess as to their molecular weight is entirely indirect. In the case of starch and the dextrim we are quite sure from the way in which they break up under the action of diastase and dilute acid that their molecular structure must be very complex and we are able to learn something about the relative size of the molecules but that is all.Within the last few weeks Victor Meyer (Ber. 1888 556) and Auwers (Ber. 1888 860) hare directed attention to a new method for determining molecular weights which is applicable to all organic substances and is of special value where a determination of vapour-density is impossible. This method was devised by Raoult (Ann. 4 401 1885; id. [6] 8 289 and 317 1886; Cornpi. reizd. 94,1517, 1882; 101 1056 1885; 102 1307 1886) and is the outcome of his elahorate investigation into the laws governing the freezing point of dilute solutions and it is certainly a remarkable fact that these important researches should up to recently have received so small a share of attention from chemists. Cl'im. Phys. [ 5 ] 28 133 1883 ; id.1161 2 66-124 1884; id. [6] MOLECULAR WEIGHTS OF THE CARBOHYDRATES. 611 Blagden as early as 1788 (Phil. Trans. 78 277) established the fact with regard to inorganic salts that the lowering of the freezing point of their aqueous solutions is proportional to the weight of substance dissolved in a constant weight of water. De Coppet ( A n n . Chirn. Plzys. [4] 23 25-26) in 1871-72 clearly pointed out that when this lowering of the freezing point is worked out for a deter. minate quantity of the substance dissolved in 100 grams of water, the result which he terms the “ coeficient of depression,” is constant for the same substance and that the coefficients for different sub-stances bear a simple relation to their molecular weights. It was left for Raoult to show clearly what those relatioils are and to extend the investigation to organic substances and to other solvents besides water.Briefly stated Raoult’s generalisations are as follow :-When certain quantities of the same substance are successively dissolved in a solvent on which it has no chemical action there is a progressive lowering of the point of congelation of the solution and this lowering is proportional to the weight of t h e substance dissolved in a constant weight of the soZvent ; in other words ihe lowering of congelation is dependent solely on the respective masses of the substance and solvent and is in dependent of the temperature. If the observed depression of the point of congelation of a solution be taken as C and the weight in grams of the anhydrous substance in 100 grams of the solvent as P then when the substance exists in solution in the anhydrous state the quotient - which we will repre-sent by A and which Raoult terms “ gross coefficient of depression ’’ (coeficient d’abnissement b r u t ) is the lowering of congelation produced by 1 gram of substance in 100 grams of solvent.If we multiply this C coefficient - = A by the molecular weight of the body dissolved, P we obtain the depression which would be produced if 1 mo1.s of the substance were dissolved in 100 grams of the solvent. This is the “ true molecular depression ” (abaissement rnole’culaire vrai) which is represented by T. We have then-C P f l L T = x M. P This formula Raoult finds to be sensibly correct even for substances which do not exist in solction in the anhydrous form provided we take solutions so dilute that the observed lowering of congelation C is about 1” C.* By “ 1 mol. of substance ” dissolved in 100 grams of water Raoult means of course a weight of the substance in grams equal to its molecular weight. 2 T 612 BROWN AND MORRIS THE DETERMINATION OF T is a quantity varying with the nature of the solvent but with the same solvent remaining constant f o r numerous groups of compounds, and it may consequently be considered as a known quantity. On the other hand the coejicient of depression - or A can be obtained ex-perimentally so that if the molecular weight M of the substance is unknown it can be calculated by the formula-C I? and from amongst the possible molecular weights of the substance T under experiment we take that which approaches nearest to - A' In carrying out this method any liquid may be used as a solvent, provided it is capable of solidifying a t a definite temperatme.It suffices merely to know the value of T for that particular solvent and for certain groups of bodies analogous to the one under experiment. The solvents which Raoult recommends in his latest paper ( A n n . Chim. Phys. [6] 8 317 1886) are water acetic acid and benzene. For inorganic saZts the values of T for water have been found by Raoult to be six in number corresponding to certain well-defined groups of salts with benzene the values of T are reduced to two, and for acetic acid T has a constant value for all inorganic compounds.With oi-ganic compounds with but very few exceptions the respective values of T for the solvents mentioned above remain constant ; they are as follows :-T . Water 19 Acetic acid 39 Benzene . 49 Raoult has examined a large number of organic substances the molecular weights of which have been put beyond doubt by determi-nations of their vapour-densities and the results illustrate in a remarkable manner the accuracy and general application of his method. I n determining the value of T €or water Raoult has recorded in his tables numbers which he obtained f o r cane-sugar, invert-sugar milk-sugar and mannitol and which indicate that the commonly received forrnuh for these substances express their molecular composition but he does not call any special attention to these experiments nor does he appear to have examined any other of the carbohydrates.It seemed therefore a matter of interest in the present state of our knowledge of this important group of substances, to submit its various members to a method which promises in th THE MOLECULAR WEIGHTS OF THE CARBOHYDRATES. 613 near future to throw much light on the molecular size of non-volatile organic compounds. Owing to the comparative ease with which most of the carbo-hydrates with the exception of dextrose and levulose are hydro-lised the employment of acetic acid as a solvent was impracticable, and our choice was necessarily limited to that of water. With the exception of oxalic acid and the amines all the bodies examined by Kaoult in aqueous solution give normal results.The method of experiment we have adopted is extremely simple and is essentially tbe same as that described by Auwers (Ber. 1888 712), with the exception that when working with aqueous solutions it is not necessary t o take any precautions to exclude the air. A solution of the carbohydrate is prepared containing a known weight of the substance in 100 C.C. of the liquid. About 120 C.C. of this are introduced into a thin beaker of about 400 C.C. capacity closed with an india-rubber plug with three perforations through one of which a small glass stirrer passes and through the second a thermometer graduated to .l,th of a degree C. This is viewed through a tele-scope and since the scale is an open one there is no difficulty in taking readings to &th of a degree The beaker is immersed in a, mixture of ice and brine a t a temperature from 2" to 3" below the freezing point of the solution which is allowed to fall in temperature from 0.5" to 1" below its point of congelation; this i t will readily do without the formation of ice.Freezing is now brought about by dropping into the beaker through the third aperture in the plug, a very small fragment of ice from a little of the same solution which has been previously frozen in a test-tube. The liquid is stirred briskly and as freezing commences the thermometer rises very rapidily and in a few second8 becomes stationary at the true freezing point of the solution the concentration being always so arranged that the observed depression is never more than 1" to 2" below zero.I f we take C = observed depression of freezing point, x = grams of substance in 100 C.C. of solution, y = grams of water in 100 C.C. of solution, then the " coefficient of depression " is expressed by the equation and the molecular weight M of the substance by 19 &I = -A 614 BROWN AND MORRIS THE DETERIiIINATION OF As the molecular composition of dextrose may be looked upon as fairly well established it will be well in the first place to show how far Raoult's method when applied to this substance bears out the generally accepted formula C6H1206. In all the experiments which follow, Column E gives the observed temperature. , C , the depression of freezing point corrected.* , A , the " coefficient of depression." , M , the molecular weight deduced from experiment.Dextrose C6H,,06. M = 180. Freezing point of water used O"-OOO. Strength of solution 12.616 grams dextrose in 92.25 grams water. E. C. A. M. - 1.450" 1.450" 0.106 179 - 1.450 1.450 0.106 179 - 1.450 1.450 0.106 179 St'rength of solution 8.3704 grams dextrose in 94.86 grams water. E. C. A. M. - 0.945" 0.945" 0.107 177 - 0.940 0.940 0-106 179 - 0.940 0.940 0.106 179 Strength of solution 4.1140 grams dextrose in 97-47 grams water. E. C. A. M. - 0.445" 0.445" 0.103 184 - 0.445 0.445 0-103 184 - 0.440 0.440 0.104 182 Calculated for C S H 2 0 6. Found (Mean). A = 0.106 A = 0.1052 M = 180.0 M = 180.2 These experiments with dextrose show very clearly the concordant nature of the results obtained by this method even when the solutions vary considerably in density.If dextrose had been a previously unknown substance whose relations and derivatives had not been studied and f o r which only * The correction here applied is the difference between the observed freezing point of the solution and the freezing point of water determined in the same apparatus and under exactly similar conditions THE MOLECULAR WEIGHTS OF THE CARBOHYDRATES. 615 the empirical formula CH,O had been determined by combustion, we should have had no hesitation in selecting from among the possible molecular weights 30 60,90 120 150 180 210 &c. the one which corresponds most nearly with the observed value of M = 180, and with the formula CsHlZO6. Dextrose exhibits in a pre-eminent degree the phenomenon of birotation ; the action on polarised light of a freshly-prepared solution of crystallised dextrose being double that of the same solution after standing for some hours.The phenomenon of birotation has never received any physical explanation and as it seemed to us possible that it might be in some way intimately connected with the size or" the molecule in solution we submitted t o Raoult's method freshly prepared solutions in which the amount of birotary carbohydrate was concurrently estimated by the polariscope. Freezing point of water used + 0.025". Strength of solution 10.013 grams dextrose in 93.85 grams water. x. C. A. M. -1.115" 1.1 40" 0.106 180 [ a ] j at time of experiment 105.6" = 72.6 per cent. birotary dextrose. E. C. A. M. -1.115" 1.140" 0.106 180 [ a ] j at time of second experiment 97.2" = 58.6 per cent.birotary .dextrose. Calculated for C6H1206* Found. A = 0.106 A = 0.106 M = 180.0 M = 180.0 It is clear from the above that whatever may be the cause of birotation it is certainly not to be attributed to a condensation of the molecule. Cane-sugar. ClzHz2011. 31 = 342. Freezing point of water used O*OOO. Strength of solution 13.052 grams in 91.98 grams water. E. C. A. M. - 0.825" 0.825" 0.059 322 -0.835 0.835 0.058 328 - 0.835 0.835 0-058 32 616 BROWN AND MORRIS THE DETERMlNhTION OF Strength of solution 10.1410 grams sugar in 93.77 grams water. E. C. A. M. - 0*600" 0.600" 0.055 345 - 0.600 0.600 0.055 345 - 0.600 0.600 0.055 345 Strength of solution 8.2580 grams sugar in 94.93 grams water.E. C. A. M. -0.500" 0.500" 0.05 7 333 - 0.490 0.490 0.056 340 - 0.490 0.490 0.056 340 Strength of solution 6.064 grams sugar in 96.28 grams water. E. C. A. M. -0.355" 0.355" 0.056 340 -0.355 0.355 0.056 340 -0.350 0-350 0.055 345 Calculated for C12H22011- Found. A = 0.0555 A = 0.0562 M = 342.0 M = 337.5 The lowest possible empirical formula C12H22011 evidently repre-sents the molecule of cane-sugar in solution and our results con-sequently do not bear out a suggestion which Winter appears to make in his recent paper on levulose (AnnuZen 244 1888 308) that the molecule of cane-sugar is more complex than this. Ihv er t e d Cane- sugar. Since the molecular weight of cane-sugar is 342 and that of dextrose 180 it seemed almost certain that Raoult's method applied to invert-sugar would yield a value for M of 180 and that the value of A would consequently be approximately doubled during the process of inversion.The experiment was made by determining the depression of the freezing point of the same soZutiorc of cane-sugar both before and after inversion which was brought about by the addition of a little invertase.* f Invert,ase is readily prepared by triturating fresh solid yeast with fine sand, digesting the pasty mass with water a t the ordinary temperature for a few hours, filtering and precipitating the filtrate with alcohol of about 80 per cent. The pre-cipitate is well washed with alcohol dehydrated with absolute alcohol and dried over sulphuric acid in a vacuum. As thus prepared invertnse is a white friable substance completely soluble ;n water and readily inverting several hundred times its own weight of cane-sugar THE MOLECULAR WEIGHTS OF THE CARBOHYDRATES.617 The results were as follows :-Cane-sugar Xolution before Inversion. Freezing point of water +0.025". Strength oE solution 4.9818 grams sugar in 96.94 grams water, E. C. A. M. - 0.27O" 0.295" 0.058 328 -0.270 0.295 0.058 328 - 0.2 75 0.300 0.058 328 Calculated for C,zHz,O 1' Found (mean). A = 0.0555 A = 0.058 M = 342.0 M = 328.0 The above solution was completely inverted with 0.030 gram of invertase the volume being maintained constant ; it gave After Inversiom. Freezing point of water +0.025". Strength of solution 5.344 grams sugars in 96.72 grams water.E. C. A. M. -0.585O 0*6?0" 0.110 173 -0,580 0.605 0.109 175 - 0.580 0.605 0.109 1'75 Calculated for CBHlZOIS. Found (mean). A = 0,106 A = 0.1093 M = 180.0 M = 174.3 This experiment may be taken as a proof that the value of M f o r levulose like that for dextrose is 180. MaZtose. C12H2,011. M = 342. That this is a saccharose having the same elementary percentage composition as cane-sugar there can be but little doubt but whether these two compounds are mebameric or polymeric is still open to question. The metallic derivatives and acelyl compounds oE maltose tend t o show that the simplesb possible foiamula C12H22011 also expresses its molecular composition but Herzfeld has recently questioned this (Annalen 220 1883 220) and is inclined from hi 61s BROWN AND XORRIS THE DETERXINATION O F experiments on the behaviour of maltose towards Fehling's solution, to assign to it a molecular formula a t least three times as large.To put this matter to the test of direct experiment the following deter-minations were made by Raoult's method :-Freezing point of water +0.030. Strength of solution 15.785 grams maltose in 90.40 grams water. E. C. A. M. - 1.010" 1.040" 0-059 322 - 1.000 1.030 0.059 322 - 1.005 1.035 0.059 322 Strength of solution 10.499 grams maltose in 93.68 grams water. E. C. A. M. - 0.635" 0.665" 0.059 322 - O.635 0.665 0.059 322 -0.635 0.665 0.059 322 Strength of solution 5.124 grams maltose in 96.89 grams water. E. C. A. M. -Q2S5" 0.315" 0.059 322 - 0.28.3 0.315 0.059 32.2 -0.285 0.315 0.059 322 Calculated for C12H,,Oll* Found (mean).A = 0.0555 A = 0.059 111 = 342.0 M = 322.0 These experiments prove beyond doubt that the molecules of cane-sugar and maltose when in solution are of equal size and that these substances are therefore metameric not polymeric. We must con-sequently attribute the difference in their properties to the different arrangement of atoms in the molecule. Milk-sugar C,2H,,0,,. M = 342. Freezing point of water + 0.030". Strength of solution 10.263 grams in 93.59 grams water. E. C. A. M. - 0.580" 0-610" 0.055 345 -0,585 0.615 0.055 345 - 0.585 0.615 0.035 345 Calculated for C,,HZ,Oll. Found. A = 0.0555 A = 0.055 14 = 342.0 X = 345. THE MOLECULAR WEIGHTS O F THE CARBOHYDRATES. 619 These results indicate that the usually accepted formula for milk-sugar expresses its molecular composition and that the suggestion made by Herzfeld (AnnuZen 220,1883,222) that the true formula is a multiple of C12H22011 is not correct but that the three sugars of the saccharose group cane-sugar maltose and milk-sugar have the same molecular weight.Arabinose C,H,,05. M = 150. This sugar a product of the action of dilute acid on gum-arabic, was formerly considered t o have the formula c,H:,,06. Receat researches of Kiliani ( B e r . 20 1887 343 and 2710) have however, established the fact that it is a compound with only 5 atoms of carbon, and that it is represented by the formula C5Hlo05. We have submitted this substance t o Raoult's method with the following results having first convinced ourselves of its purity by a determination of its optical activity :-Freezing point of water +0*030".Strength of solution 4.1355 grams in 97.45 grams water. E. C. A. M. -0.510" 0.540" 0.127 149% -0.505 0.535 0.126 150-7 - 0.505 0.535 0.126 150.7 Calculated for C5H1005. Found (mean). A = 0.126 A = 0.1263 M = 150.0 M = 150.3 These results fully confirm Kiliani's conclusions. Raflnose. This sugar which occurs in the molasses of beet-sugar and in the seeds of various plants has been the subject of considerable investi-gation. It crystallises in well-defined needles or prisms and from analyses tlie formula C18H32016,5H20 or a multiple of this has been deduced preference being given to a molecule represented by double this formula.We are indebted to Dr. Griess for a specimen of pure crystalline raffinose as well as that of the pure arabinose mentioned above. The substance had an optical activity of [a]j 116.6' or [a]= 104.6, which agrees exactly with the numbers given by Tollens for hydrated raffinose. Preezing point of water + 0.030" (5.30 THE MOLECULAR WEIGHTS O F THE CAABORYDRATES. Strength of solution 8,2225 grams hydrated substance in 94.51 grams water. E. C. A. M. - 0.280" 0.310" 0.0356 533 - 0.880 0.310 0.0356 533 - 0.290 0.320 0.0367 518 Calculated for C,,H320,6,5H,O. Found (mean). A = 0,032 A = 0.036 M = 594.0 M = 528.0 Our results indicate that the molecule of raffinose approximates to the above formula and that it is not a multiple of this. Manizitol C6HBla06.M = 182. Although this substance is not strictly included in the ordinarily accepted definition of a carbohydrate yet its relations to the group are of so intimate a nature and the size and constitution of its mole-cule have been so accurately determined by its i*eactions that we have considered it well as a further proof of the accuracy of Raoult's method when applied to compounds of this class to include the results of its examination. Freezing point of water + 0*030". Strength of solution 7.5382 grams in 95.09 grams water. E. C. A. M. - 0.8O5" 0.835" 0.105 181 -0*803 0-835 0.105 181 - 0.805 0.835 0.105 181 Calculated for C6H1406. Found. A = 0.104 A = 0-105 M = 182.0 M = 181.0 The application of this new method to starch and t o the non-crystallisable products of its transformation soluble starch the dextrins and malto-dextrin seemed full of promise since chemists are still divided in their opinions as t o the true nature of these compounds, and as to whether the differences in the properties of the dextrins are such as to justify the view that they are polymeric or on the other hand compounds ha,ving the same molecular weight but differing in constitution RAMSAY NITROGEN TRIOXIDE AND NITRIC PEROXIDE. 621 Certain difficulties have however arisen at this stage of our inquiry owing to the very high molecular weight which these sub-stances evidently possess. As a result of this the freezing point of even very strong solutions is depressed t o such a small extent as to render it necessary before we can assign any approximately accurate numerical value to our resultr to determine the limits of error of the method which manifestly increase with the molecular weight of the substance. We have however convinced ourselves that the mole-cular complexity of these compounds is very great indeed and me hope to lay certain results before. the Society at an early date