Chapter 8

with reference to the highest nitrate, and the decomposition of the nitrate by alkalis with formation of hydroxypyruvic acid. While these reactions afford no very sure ground for deductions as to constitutional relationships, it certainly appears that, if the aldose view of the unit group is to be retained, this form of the anhydride contains suggestions of the general tendency of the celluloses on treatment with condensing acids to split off formic acid in relatively large quantity [Ber. 1895, 1940]; the condensation of the oxycelluloses to furfural; the non-formation of the normal hydroxy-dicarboxylic acids by nitric acid oxidations. Indirectly we may point out that any hypothesis which retains the polyaldose view of cellulose, and so fails to differentiate its constitution from that of starch, has little promise of progress. The above formula, moreover, concerns the assumed unit group, with no suggestion as to the mode of aggregation in the cellulose complex. Also there is no suggestion as to how far the formula is applicable to the celluloses considered as a group. In extending this view to the oxycelluloses, Vignon introduces the derived oxidised group

CHO.(CHOH)_{3}.CH . CO |_O__|

--of which one is apportioned to three or four groups of the cellulose previously formulated: these groups in condensed union together constitute an oxycellulose.

These views are in agreement with the experimental results obtained by Faber and Tollens (p. 71). They regard the oxycelluloses as compounds of 'celloxin' C_{6}H_8{O}_{6} with 1-4 mols. unaltered cellulose; and the former they particularly refer to as a lactone of glycuronic acid. But on boiling with lime they obtain dioxybutyric and isosaccharinic acids; both of which are not very obviously related to the compounds formulated by Vignon. We revert with preference to a definitely ketonic formula, for which, moreover, some farther grounds remain to be mentioned. In the systematic investigation of the nitric esters of the carbohydrates (p. 41) Will and Lenze have definitely differentiated the ketoses from the aldoses, as showing an internal condensation accompanying the ester reaction. Not only are the OH groups taking part in the latter consequently less by two than in the corresponding aldoses, but the nitrates show a much increased stability. This would give a simple explanation of the well-known facts obtaining in the corresponding esters of the normal cellulose. We may note here that an important item in the quantitative factors of the cellulose nitric ester reaction has been overlooked: that is, the yield calculated to the NO_{3} groups fixed. The theoretical yields for the higher nitrates are

Yield p.ct. N p.ct. of cellulose of nitrate Pentanitrate 169 12.7 Hexanitrate 183 14.1

From such statistics as are recorded the yields are not in accordance with the above. There is a sensible deficiency. Thus Will and Lenze record a yield of 170 p.ct. for a product with 13.8 p.ct. N, indicating a deficiency of about 10 p.ct. As the by-products soluble in the acid mixture are extremely small, the deficiency represents approximately the water split off by an internal reaction. In this important point the celluloses behave as ketoses.

In the lignocelluloses the condensed constituents of the complex are of well-marked ketonic, i.e. quinonic, type. In 'nitrating' the lignocelluloses this phenomenon of internal condensation is much more pronounced (see p. 131). As the reaction is mainly confined to the cellulose of the fibre, we have this additional evidence that the typical carbonyl is of ketonic function. It is still an open question whether the cellulose constituents of the lignocelluloses are progressively condensed--with progress of 'lignification'--to the unsaturated or lignone groups. There is much in favour of this view, the evidence being dealt with in the first edition, p. 180. The transition from a cellulose-ketone to the lignone-ketone involves a simple condensation without rearrangement; from which we may argue back to the greater probability of the ketonic structure of the cellulose. We must note, however, that the celluloses of the lignocelluloses are obtained as residues of various reactions, and are not homogeneous. They yield on boiling with condensing acids from 6 to 9 p.ct. furfural. It is usual to regard furfural as invariably produced from a pentose residue. But this interpretation ignores a number of other probable sources of the aldehyde. It must be particularly remembered that lævulose is readily condensed (a) to a methylhydroxyfurfural

C_{6}H_{1}O_{6} - 3H_{2}O = C_{6}H_{6}O_{3} = C_{5}(OH).H_{2}.(CH_{3})O_{2}

and (b) by HBr, with further loss of OH, as under:

C_{6}H_{12}O_{6} - 4H_{2}O + HBr = C_{5}H_{3}(CH_{2}Br)O

and generally the ketoses are distinguished from the aldoses by their susceptibility to condensation. Such condensation of lævulose has been effected by two methods: (a) by heating the concentrated aqueous solution with a small proportion of oxalic acid at 3 atm. pressure [Kiermayer, Chem. Ztg. 19, 100]; (b) by the action of hydrobromic acid (gas) in presence of anhydrous ether; the actual compound obtained being the omega-brommethyl derivative [Fenton, J. Chem. Soc. 1899, 423].

This latter method is being extended to the investigation of typical celluloses, and the results appear to confirm the view that cellulose may be of ketonic constitution.

The evidence which is obtainable from the synthetical side of the question rests of course mainly upon the physiological basis. There are two points which may be noted. Since the researches of Brown and Morris (J. Chem. Soc. 1893, 604) have altered our views of the relationships of starch and cane sugar to the assimilation process, and have placed the latter in the position of a primary product with starch as a species of overflow and reserve product, it appears that lævulose must play an important part in the elaboration of cellulose. Moreover, A. J. Brown, in studying the cellulosic cell-collecting envelope produced by the _Bacterium xylinum_, found that the proportion of this product to the carbohydrate disappearing under the action of the ferment was highest in the case of lævulose. These facts being also taken into consideration there is a concurrence of suggestion that the typical CO group in the celluloses is of ketonic character. That the typical cotton cellulose breaks down finally under the action of sulphuric acid to dextrose cannot be held to prove the aldehydic position of the carbonyls in the unit groups of the actual cellulose molecule or aggregate.

We again are confronted with the problem of the aggregate and as to how far it may affect the constitution of the unit groups. That it modifies the functions or reactivity of the ultimate constituent groups we have seen from the study of the esters. Thus with the direct ester reactions the normal fibrous cellulose (C_{6}H_{16}O_{5}) yields a monoacetate, dibenzoate, and a trinitrate respectively under conditions which determine, with the simple hexoses and anhydrides, the maximum esterification, i.e. all the OH groups reacting. If the OH groups are of variable function, we should expect the CO groups _a fortiori_ to be susceptible of change of function, i.e. of position within the unit groups.

But as to how far this is a problem of the constitution or phases of constitution of the unit groups or of the aggregate under reaction we have as yet no grounds to determine.

The subjoined communication, appearing after the completion of the MS. of the book, and belonging to a date subsequent to the period intended to be covered, is nevertheless included by reason of its exceptional importance and special bearing on the constitutional problem above discussed.

~THE ACTION OF HYDROGEN BROMINE ON CARBOHYDRATES.~[4]

H. J. H. FENTON and MILDRED GOSTLING (J. Chem. Soc., 1901, 361).

The authors have shown in a previous communication (Trans., 1898, 73, 554) that certain classes of carbohydrates when acted upon at the ordinary temperature with dry hydrogen bromide in ethereal solution give an intense and beautiful purple colour.[5] It was further shown (Trans., 1899, 75, 423) that this purple substance, when neutralised with sodium carbonate and extracted with ether, yields golden-yellow prisms of omega-brommethylfurfural,

CH:C.CH_{2}Br | | | O | | CH:C.CHO.

This reaction is produced by lævulose, sorbose, cane sugar, and inulin, an intense colour being given within an hour or two. Dextrose, maltose, milk sugar, galactose, and the polyhydric alcohols give, if anything, only insignificant colours, and these only after long standing. The authors therefore suggested that the reaction might be employed as a means of distinguishing these classes of carbohydrates, the rapid production of the purple colour being indicative of _ketohexoses_, or of substances which produce these by hydrolysis.

By relying only on the production of the purple colour, however, a mistake might possibly arise, owing to the fact that _xylose_ gives a somewhat similar colour after standing for a few hours. Hence, the observations should be confirmed by isolation of the crystals of brommethylfurfural. No trace of this substance is obtained from the xylose product.

In order to identify the substance, the ether extract, after neutralisation, is allowed to evaporate to a syrup, and crystallisation promoted either by rubbing with a glass rod, or by the more certain and highly characteristic method of 'sowing' with the most minute trace of omega-brommethylfurfural, when crystals are almost instantly formed. These are recrystallised from ether, or a mixture of ether and light petroleum, and further identified by the melting-point (59.5-60.5°), and, if considered desirable, by estimation of the bromine.

It is now found, so reactive is the bromine atom in this compound, that the estimation may be accurately made by titration with silver nitrate according to Volhard's process, the crystals for this purpose being dissolved in dilute alcohol:

0.1970 gram required 10.5 c.c. _N_/10 AgNO_{3}. Br = 42.63 p.ct., calculated 42.32 p.ct.

This method of applying hydrogen bromide in ethereal solution is, of course, unsuitable for investigations where a higher temperature has to be employed, or where long standing is necessary, since, under such circumstances, the ether itself is attacked. Wishing to make investigations under these conditions, the authors have tried several solvents, and, at present, find that chloroform is best suited to the purpose. In each of the following experiments, 10 grms. of the substance were covered with 250 c.c. of chloroform which had been saturated at 0° with dry hydrogen bromide. The mixture was contained in an accurately stoppered bottle, firmly secured with an iron clamp, and heated in a water-bath to about the boiling temperature for two hours. After standing for several hours, the mixture was treated with sodium carbonate (first anhydrous solid, and afterwards a few drops of strong solution), filtered, and the solution dried over calcium chloride. Most of the chloroform was then distilled off, and the remaining solution allowed to evaporate to a thick syrup in a weighed dish.

The product was then tested for omega-brommethylfurfural by 'sowing' with the most minute trace of the substance, as described above. It was then warmed on a water-oven, kept in a vacuum desiccator over solid paraffin, and the weight estimated. When necessary, the product was recrystallised from ether, and further identified by the tests mentioned. The following results were obtained:

Weight of crude residue. Swedish filter paper 3.0 crystallised at once by 'sowing.' Ordinary cotton 3.3 " " Mercerised cotton 2.1 " " Straw cellulose[6] 2.3 " " Lævulose 2.2 " " Inulin 1.3 " " Potato starch 0.37 " " Cane sugar 0.85 " " Dextrose 0.33 uncrystallisable. Milk sugar 0.37 " Glycogen 0.34 " Galactose 0.34 "

The products from _dextrose_, _milk sugar_, and _galactose_ absolutely refused to crystallise even when extracted with ether and again evaporated, or by 'sowing,' stirring, &c.

The _glycogen_ product deposited a very small amount of crystalline matter on standing, but the quantity was too minute for examination; moreover, it refused altogether to crystallise in contact with the aldehyde. It may fairly be stated, therefore, that these last four substances give absolutely negative results as regards the formation of omega-brommethylfurfural; if any is formed, its quantity is altogether too small to be detected.

The specimen of _starch_ examined was freshly prepared from potato, and purified by digestion for twenty-four hours each with _N_/10 KOH, _N_/4 HCl, and strong alcohol; it was then washed with water and allowed to dry in the air. It will be seen that this substance gave a positive result, but that the yield was extremely small, and might yet be due to impurity. Considering the importance of the behaviour of starch, for the purpose of drawing general conclusions from these observations, it was thought advisable to make further experiments with specimens which could be relied upon, and also to investigate the behaviour of dextrin. This the authors have been enabled to do upon a series of specimens specially prepared by C. O'Sullivan, and thus described by him:

1. Rice starch, specially purified by the permanganate method.

2. Wheat starch " " "

3. Oat starch, contains traces of oil, washed with dilute KOH and dilute HCl.

4. Pea starch, first crop, washed with alkali, acid (HCl), and strong alcohol.

5. Natural dextrin, D = 3.87, alpha_{D} = 194.7; K = 0.95, (c 2.628).

6. alpha-Dextrin, C equation purified without fermentation, 30 precipitations with alcohol (Trans., 1879, 35, 772).

The examination of these specimens was conducted on a smaller scale, but under the same conditions as before, _one gram_ of the substance being treated with 12.5 c.c. of the saturated chloroform solution and heated in sealed tubes for two hours as above. The results were as follows:

Weight of crude residue. 1. Rice starch 0.046 crystallised at once by 'sowing.' 2. Wheat starch 0.044 " " 3. Oat starch 0.049 " " 4. Pea starch 0.064 " " 5. Natural dextrin 0.088 " " 6. alpha-Dextrin 0.055 " "

The results may therefore be summarised as follows:--Treated under these particular conditions all forms of cellulose give large yields of omega-brommethylfurfural, some varieties giving as much as 33 per cent. Lævulose, inulin, and cane sugar give yields varying from 22 to 8.5 per cent.; various starches give small yields (average about 4.5 per cent.); and dextrins 5 to 8 per cent., whereas dextrose, milk sugar, and galactose give, apparently, none at all.

The yields represent the solid crystalline residue; this when purified by recrystallisation gives, probably, about three-quarters of its weight of pure crystals. (In the case of dextrose, &c., the yields represent the weight of syrup.)

These numbers, however, by no means represent the maximum yields obtainable, owing to the comparatively slight solubility of hydrogen bromide in chloroform. The process was conducted in the above manner only for the sake of uniform comparison. The ether method previously described gives much larger yields; for example, 12 grms. of inulin treated with only 60 c.c. of the saturated ether gave 2.5 grms. of substance. For the purpose of obtaining larger yields, other methods are being investigated.

The facts recorded above, taken in conjunction with those given in our previous communications, appear to point definitely to the following general conclusions. First, that the various forms of _cellulose_ contain one or more groups or nuclei identical with that contained in _lævulose_, and that such groups constitute the main or essential part of the molecule. Secondly, that similar groupings are contained in _starches_ and _dextrins_, but that the proportion of such groupings represents a relatively small part of the whole structure.

The nature of this grouping is, according to the generally accepted constitution of _lævulose_, the six-carbon chain with a ketonic group:

C·C·C·C·C·C || . O

But the results might, on the other hand, be considered indicative of the anhydride or 'lacton' grouping, which Tollens suggested for lævulose:

C·C·C·C·C·C \ / \ / . O

The latter very simply represents the formation of omega-brommethylfurfural from lævulose,[7]

O----- | H H | | | | | OH·C-----C---C---C--C-----CH_{2}·OH H_{2} OH OH OH H

giving

H H HC·C:C·C:C·CH_{2}Br || \ / , O \ / O

although by a little further 'manipulation' of the symbols the change could, of course, be represented by reference to the ketonic formula.

~The Ketonic Constitution of Cellulose.~

C. F. CROSS and E. J. BEVAN (J. Chem. Soc., 1901, 366).

In this paper the authors discuss more fully the theoretical bearings of the observations of Fenton and Gostling, the two papers being simultaneously communicated. The paper is mainly devoted to a review of the antecedent evidence, chemical and physiological, and to a general summing up in favour of the view that cellulose is a polyketose (anhydride).

* * * * *

(p. 79) ~Composition of the Seed Hair of Eriodendron~ (~Anf.~)--Some interest attaches to the results of an analytical investigation which we have made of this silky floss. There is little doubt that cotton is entirely exceptional in its characteristics: both in structure and chemical composition it fails to show any adaptation to what we may regard as the _more obvious_ functions of a seed hair--which certainly do not demand either structural strength or chemical resistance. The following numbers determined for the kapok differentiate it widely from the cottons:

Ash, 1.3; moisture, 9.3; alkaline hydrolysis (loss) (a) 16.7, (b) 21.8. Cellulose, by chlorination, &c., 71.1.

In reacting with chloride it shows the presence of unsaturated groups, similar to the lignone of the woods. This was confirmed by a well-marked reaction with ferric ferricyanide with increase of weight due to the fixation of the blue cyanide.

But the most characteristic feature is the high yield of furfural on boiling with condensing acids. The following numbers were determined:

Total furfural from original fibre 14.84 In residue from alkali hydrolysis 11.5 In cellulose isolated by Cl method 10.4

Treated with sulphuric acids of concentration, (a) 92.1 grs. H_{2}SO_{4} per 100 c.c., (b) 105.8 grs. per 100 c.c., the fibres dissolve, and diluted immediately after complete solution it was resolved into

(a) (b)

Reprecipitated fraction 68.7 43.7 Soluble fraction yielding furfural 13.2 14.3

By these observations it is established that the furfuroids are of the cellulose type and behave very much as the furfuroids of the cereal celluloses.

This group of seed hairs invites exhaustive investigation. The furfuroid constituents are easily isolated, and as they constitute at least one-third of the fibre substance it is especially from this point of view that they invite study.

RECHERCHES SUR L'OXYCELLULOSE.

L. VIGNON.

~Résumé of investigations (1898-1900) of Oxycellulose, published as a brochure~ (Rey, Lyon, 1900).

(a) A typical oxycellulose prepared from cotton cellulose by the action of HClO_{3} (HCl + KClO_{3}) in dilute solution at 100° for one hour gave the following numbers:

C H O Elementary composition 43.55 6.03 50.42

Oxycellulose Original cellulose Analysis by Lange's method Soluble in KOH (at 180°) 87.6 12.0 Insoluble in KOH (at 180°) 12.4 88.0

Oxycellulose Original cellulose Heat of combustion 4124-4133 4190-4224 Heat evolved in contact with 50 times wt.} normal KOH per 100 grms. } 1.3 cal. 0.74 cal.

Oxycellulose Cellulose Absorption of colouring } Saffranine 0.7 0.0 matters at 100° per 100 grms. } Methylene blue 0.6 0.2

(b) _Yield of furfural from cellulose, oxy- and hydro-cellulose._--From the hydrocelluloses variously prepared the author obtains 0.8 p.ct. furfural; from bleached cotton 1.8 p.ct.; and from the oxycelluloses variously prepared 2.0-3.5 p.ct. The 'furfuroid' is relatively more soluble in alkaline solutions (KOH) in the cold. The insoluble residue is a normal cellulose.

(c) _Nitrates of cellulose, oxy- and hydro-cellulose._--Treated with the usual acid mixture (H_{2}SO_{4} 3 p., HNO_{3} 1 p.) under conditions for maximum action, the resulting esters showed uniformly a fixation of 11.0 NO_{2} groups per unit mol. of C_{24}. The oxycellulose nitrate was treated directly with dilute solution of potassium hydrate in the cold. From the products of decomposition the author obtained the osazone of hydroxypyruvic acid [Will, Ber. 24, 400].

(d) _Osazones of the oxycelluloses._--Oxycelluloses prepared by various methods are found to fix varying proportions of phenylhydrazine (residue), viz. from 3.4-8.5 p.ct. of the cellulose derivative reacting, corresponding with, i.e. calculated from, the nitrogen determined in the products (0.87-2.2 p.ct.). The reaction is assumed to be that of osazone formation.

The author has also established a relation between the phenylhydrazine fixed and the furfural which the substance yields on boiling with condensing acids. This is illustrated by the subjoined series of numbers:

Phenylhydrazine Furfural Fixed p.ct. formed p.ct. Cotton (bleached) 1.73 1.60 Oxycellulose (HClO_{3}) 7.94 2.09 " (HClO) 3.37 1.79 " (CrO_{3}) (1) 7.03 3.00 " (CrO_{3}) (2) 7.71 3.09 " (CrO_{3}) (3) 8.48 3.50

(e) _Constitution of cellulose and oxycellulose._--The results of these investigations are generalised as regards cellulose (C_6) by the constitutional formula

CH--CH_{2} / | | (CHOH)_{3} O | \ | | CH--O .

The oxycelluloses contain the characteristic group

COH / (CHOH)_{3} \ CH--CO \ / O

in union with varying proportions of residual cellulose.

QUANTITATIVE SEPARATION OF CELLULOSE-LIKE CARBOHYDRATES IN VEGETABLE SUBSTANCES.

WILHELM HOFFMEISTER (Landw. Versuchs-Stat., 1897, 48, 401-411).