In the period 1895-1900, which has elapsed since the original publication of our work on 'Cellulose,' there have appeared a large number of publications dealing with special points in the chemistry of cellulose. So large has been the contribution of matter that it has been considered opportune to pass it under review; and the present volume, taking the form of a supplement to the original work, is designed to incorporate this new matter and bring the subject as a whole to the level to which it is thereby to be raised. Some of our critics in reviewing the original work have pronounced it 'inchoate.' For this there are some explanations inherent in the matter itself. It must be remembered that every special province of the science has its systematic beginning, and in that stage of evolution makes a temporary 'law unto itself.' In the absence of a dominating theory or generalisation which, when adopted, gives it an organic connection with the general advance of the science, there is no other course than to classify the subject-matter. Thus 'the carbohydrates' may be said to have been in the inchoate condition, qualified by a certain classification, prior to the pioneering investigations of Fischer. In attacking the already accumulated and so far classified material from the point of view of a dominating theory, he found not only that the material fell into systematic order and grew rapidly under the stimulus of fruitful investigation, but in turn contributed to the firmer establishment of the theoretical views to which the subject owed its systematic new birth. On the other hand, every chemist knows that it is only the simpler of the carbohydrates which are so individualised as to be connoted by a particular formula in the stereoisomeric system. Leaving the monoses, there is even a doubt as to the constitution of cane sugar; and the elements of uncertainty thicken as we approach the question of the chemical structure of starch. This unique product of plant life has a literature of its own, and how little of this is fully known to what we may term the 'average chemist' is seen by the methods he will employ for its quantitative estimation. In one particular review of our work where we are taken to task for producing 'an aggravating book, inchoate in the highest degree ... disfigured by an obscurity of diction which must materially diminish its usefulness' ['Nature,' 1897, p. 241], the author, who is a well-known and competent critic, makes use of the short expression in regard to the more complex carbohydrates, 'Above cane sugar, higher in the series, all is chaos,' and in reference to starch, 'the subject is still enshrouded in mystery.' This 'material' complexity is at its maximum with the most complex members of the series, which are the celluloses, and we think accounts in part for the impatience of our critic. 'Obscurity of diction' is a personal quantity, and we must leave that criticism to the fates. We find also that many workers whose publications we notice in this present volume quite ignore the plan of the work, though they make use of its matter. We think it necessary to restate this plan, which, we are satisfied, is systematic, and, in fact, inevitable. Cellulose is in the first instance a structure, and the anatomical relationships supply a certain basis of classification. Next, it is known to us and is defined by the negative characteristics of resistance to hydrolytic actions and oxidations. These are dealt with in the order of their intensity. Next we have the more positive definition by ultimate products of hydrolysis, so far as they are known, which discloses more particularly the presence of a greater or less proportion of furfural-yielding groups. Putting all these together as criteria of function and composition we find they supply common or general dividing lines, within which groups of these products are contained. The classification is natural, and in that sense inevitable; and it not only groups the physiological and chemical facts, but the industrial also. We do not propose to argue the question whether the latter adds any cogency to a scientific scheme. We are satisfied that it does, and we do not find any necessity to exclude a particular set of phenomena from consideration, because they involve 'commercial' factors. We have dealt with this classification in the original work (p. 78), and we discuss its essential basis in the present volume (p. 28) in connection with the definition of a 'normal' cellulose. But the 'normal' cellulose is not the only cellulose, any more than a primary alcohol or an aliphatic alcohol are the only alcohols. This point is confused or ignored in several of the recent contributions of investigators. It will suffice to cite one of these in illustration. On p. 16 we give an account of an investigation of the several methods of estimating cellulose, which is full of valuable and interesting matter. The purpose of the author's elaborate comparative study is to decide which has the strongest claims to be regarded as the 'standard' method. They appear to have a preference for the method of Lange—viz. that of heating at high temperatures (180°) with alkaline hydrates, but the investigation shows that (as we had definitely stated in our original work, p. 214) this is subject to large and variable errors. The adverse judgment of the authors, we may point out, is entirely determined on the question of aggregate weight or yield, and without reference to the ultimate composition or constitution of the final product. None of the available criteria are applied to the product to determine whether it is a cellulose (anhydride) or a hydrate or a hydrolysed product. After these alkali-fusion processes the method of chlorination is experimentally reviewed and dismissed for the reason that the product retains furfural-yielding groups, which is, from our point of view, a particular recommendation, i.e. is evidence of the selective action of the chlorine and subsequent hydrolysis upon the lignone group. As a matter of fact it is the only method yet available for isolating the cellulose from a lignocellulose by a treatment which is quantitatively to be accounted for in every detail of the reactions. It does not yield a 'normal' cellulose, and this is the expression which, in our opinion, the authors should have used. It should have been pointed out, moreover, that, as the cellulose is separated from actual condensed combination with the lignone groups, it may be expected to be obtained in a hydrated form, and also not as a homogeneous substance like the normal cotton cellulose. The product is a cellulose of the second group of the classification. Another point in this investigation which we must criticise is the ultimate selection of the Schulze method of prolonged maceration with nitric acid and a chlorate, followed by suitable hydrolysis of the non-cellulose derivatives to soluble products. Apart from its exceptional inconvenience, rendering it quite impracticable in laboratories which are concerned with the valuation of cellulosic raw materials for industrial purposes, the attack of the reagent is complex and ill-defined. This criticism we would make general by pointing out that such processes quite ignore the specific characteristics of the non-cellulose components of the compound celluloses. The second division of the plan of our work was to define these constituents by bringing together all that had been established concerning them. These groups are widely divergent in chemical character, as are the compound celluloses in function in the plant. Consequently there is for each a special method of attack, and it is a reversion to pure empiricism to expect any one treatment to act equally on the pectocelluloses, lignocelluloses, and cutocelluloses. Processes of isolating cellulose are really more strictly defined as methods of selective and regulated attack of the groups with which they occur, combined or mixed. A chemist familiar with such types as rhea or ramie (pectocellulose), jute (lignocellulose), and raffia (cutocellulose) knows exactly the specific treatment to apply to each for isolating the cellulose, and must view with some surprise the appearance at this date of such 'universal prescriptions' as the process in question.

The third division of our plan of arrangement comprised the synthetical derivatives of the celluloses, the sulphocarbonates first, as peculiarly characteristic, and then the esters, chiefly the acetates, benzoates, and nitrates. To these, investigators appear to have devoted but little attention, and the contribution of new matter in the present volume is mainly the result of our own researches. It will appear from this work that an exhaustive study of the cellulose esters promises to assist very definitely in the study of constitutional problems.

This brings us to the fourth and, to the theoretical chemist, the most important aspect of the subject, the problem of the actual molecular structure of the celluloses and compound celluloses. It is herein we are of opinion that the subject makes a 'law unto itself.' If the constitution of starch is shrouded in mystery and can only be vaguely expressed by generalising a complex mass of statistics of its successive hydrolyses, we can only still more vaguely guess at the distance which separates us from a mental picture of the cellulose unit. We endeavour to show by our later investigations that this problem merges into that of the actual structure of cellulose in the mass. It is definitely ascertained that a change in the molecule, or reacting unit, of a cellulose, proportionately affects the structural properties of the derived compounds, both sulphocarbonates and esters. This is at least an indication that the properties of the visible aggregates are directly related to the actual configuration of the chemical units. But it appears that we are barred from the present discussion of such a problem in absence of any theory of the solid state generally, but more particularly of those forms of matter which are grouped together as 'colloids.'

Cellulose is distinguished by its inherent constructive functions, and these functions take effect in the plastic or colloidal condition of the substance. These properties are equally conspicuous in the synthetical derivatives of the compound. Without reference, therefore, to further speculations, and not deterred by any apparent hopelessness of solving so large a problem, it is clear that we have to exhaust this field by exact measurements of all the constants which can be reduced to numerical expression. It is most likely that the issue may conflict with some of our current views of the molecular state which are largely drawn from a study of the relatively dissociated forms of matter. But such conflicts are only those of enlargement, and we anticipate that all chemists look for an enlargement of the molecular horizon precisely in those regions where the forces of cell-life manifest themselves.

The cellulose group has been further differentiated by later investigations. The fibrous celluloses of which the typical members receive important industrial applications, graduate by insensible stages into the hemicelluloses which may be regarded as a well-established sub-group. In considering their morphological and functional relationships it is evident that the graduation accords with their structure and the less permanent functions which they fulfil. They are aggregates of monoses of the various types, chiefly mannose, galactose, dextrose, &c., so far as they have been investigated.

Closely connected with this group are the constituents of the tissues of fungi. The recent researches of Winterstein and Gilson, which are noted in this present volume, have established definitely that they contain a nitrogenous group in intimate combination with a carbohydrate complex. This group is closely related to chitin, yielding glucosamin and acetic acid as products of ultimate hydrolysis. Special interest attaches to these residues, as they are in a sense intermediate products between the great groups of the carbohydrates and proteids (E. Fischer, Ber. 19, 1920), and their further investigation by physiological methods may be expected to disclose a genetic connection.

The lignocelluloses have been further investigated. Certain new types have been added, notably a soluble or 'pectic' form isolated from the juice of the white currant (p. 152), and the pith-like wood of the Æschynomene (p. 135).

Further researches on the typical fibrous lignocellulose have given us a basis for correcting some of the conclusions recorded in our original work, and a study of the esters has thrown some light on the constitution of the complex (p. 130).

Of importance also is the identification of the hydroxyfurfurals as constituents of the lignocelluloses generally, and the proof that the characteristic colour-reactions with phenols (phloroglucinol) may be ascribed to the presence of these compounds (p. 116).

The pectocelluloses have not been the subject of systematic chemical investigation, but the researches of Gilson ('La Cristallisation de la Cellulose et la Composition Chimique de la Membrane Cellulaire VÉgÉtale,' 'La Revue,' 'La Cellule,' i. ix.) are an important contribution to the natural history of cellulose, especially in relation to the 'pectic' constituents of the parenchymatous celluloses. Indirectly also the researches of Tollens on the 'pectins' have contributed to the subject in correcting some of the views which have had a text-book currency for a long period. These are dealt with on p. 151. The results establish that the pectins are rather the soluble hydrated form of cellulosic aggregates in which acid groups may be represented; but such groups are not to be regarded as essentially characteristic of this class of compounds.

Furfural-yielding Substances (Furfuroids).—This group of plant products has been, by later investigations, more definitely and exclusively connected with the celluloses—i.e. with the more permanent of plant tissues. From the characteristic property of yielding furfural, which they have in common with the pentoses, they have been assumed to be the anhydrides of these C₅ sugars or pentosanes; but the direct evidence for this assumption has been shown to be wanting. In regard to their origin the indirect evidences which have accumulated all point to their formation in the plant from hexoses. Of special interest, in its bearings on this point, is the direct transformation of levulose into furfural derivatives, which takes place under the action of condensing agents. The most characteristic is that produced by the action of anhydrous hydrobromic acid in presence of ether [Fenton], yielding a brommethyl furfural

C₆H₁₂O₆ - 4H₂O + HBr = C₅H₃.O₂.CH₂Br

with a Br atom in the methyl group. These researches of Fenton's appear to us to have the most obvious and direct bearings upon the genetic relationships of the plant furfuroids and not only per se. To give them their full significance we must recall the later researches of Brown and Morris, which establish that cane sugar is a primary or direct product of assimilation, and that starch, which had been assumed to be a species of universal matiÈre premiÈre, is probably rather a general reserve for the elaborating work of the plant. If now the aldose groups tend to pass over into the starch form, representing a temporary overflow product of the assimilating energy, it would appear that the ketose or levulose groups are preferentially used up in the elaboration of the permanent tissue. We must also take into consideration the researches of Lobry de Bruyn showing the labile functions of the typical CO group in both aldoses and hexoses, whence we may conclude that in the plant-cell the transition from dextrose to levulose is a very simple and often occurring process.

We ourselves have contributed a link in this chain of evidence connecting the furfuroids of the plant with levulose or other keto-hexose. We have shown that the hydroxyfurfurals are constituents of the lignocelluloses. The proportion present in the free state is small, and it is not difficult to show that they are products of breakdown of the lignone groups. If we assume that such groups are derived ultimately from levulose, we have to account for the detachment of the methyl group. This, however, is not difficult, and we need only call to mind that the lignocelluloses are characterised by the presence of methoxy groups and a residue which is directly and easily hydrolysed to acetic acid. Moreover, the condensation need not be assumed to be a simple dehydration with attendant rearrangement; it may very well be accompanied or preceded by fixation of oxygen. Leaving out the hypothetical discussion of minor variations, there is a marked convergence of the evidence as to the main facts which establish the general relationships of the furfuroid group. This group includes both saturated and unsaturated or condensed compounds. The former are constituents of celluloses, the latter of the lignone complex of the lignocelluloses.

The actual production of furfural by boiling with condensing acids is a quantitative measure of only a portion, i.e. certain members of the group. The hydroxyfurfurals, not being volatile, are not measured in this way. By secondary reactions they may yield some furfural, but as they are highly reactive compounds, and most readily condensed, they are for the most part converted into complex 'tarry' products. Hence we have no means, as yet, of estimating those tissue constituents which yield hydroxyfurfurals; also we have no measure of the furfurane-rings existing performed in such a condensed complex as lignone. But, chemists having added in the last few years a large number of facts and well-defined probabilities, it is clear that the further investigation of the furfuroid group will take its stand upon a much more adequate basis than heretofore. On the view of 'furfural-yielding' being co-extensive with 'pentose or pentosane,' not only were a number of important facts obscured or misinterpreted, but there was a barrenness of suggestion of genetic relationships. As the group has been widened very much beyond these limits, it is clear that if any group term or designation is to be retained that of 'furfuroid' is 'neutral' in character, and equally applicable to saturated substances of such widely divergent chemical character as pentoses, hexosones, glycuronic acid, and perhaps, most important of all, levulose itself, all of which are susceptible of condensation to furfural or furfurane derivatives, as well as to those unsaturated compounds, constituents of plant tissues which are already furfurane derivatives.

From the chemical point of view such terms are perhaps superfluous. But physiological relationships have a significance of their own; and there is a physiological or functional cohesion marking this group which calls for recognition, at least for the time, and we therefore propose to retain the term furfuroid.[1]

General Experimental Methods.—In the investigation of the cellulose group it is clear that methods of ultimate hydrolysis are of first importance. None are so convenient as those which are based on the action of sulphuric acid, more or less concentrated (H₂SO₄.3H₂O - H₂SO₄H₂O). Such methods have been frequently employed in the investigations noted in this volume. We notice a common deficiency in the interpretation of the results. It appears to be sufficient to isolate and identify a crystalline monose, without reference to the yield or proportion to the parent substance, to establish some main point in connection with its constitution. On the other hand, it is clear that in hydrolysing a given cellulose-complex we ought to aim at complete, i.e. quantitative, statistics. The hydrolytic transformation of starch to dextrins and maltose has been followed in this way, and the methods may serve as a model to which cellulose transformations should be approximated. In fact, what is very much wanted is a systematic re-examination of the typical celluloses in which all the constants of the terms between the original and the ultimate monose groups shall be determined. Such constants are similar to those for the starch-dextrose series, viz. opticity and cupric reduction. Various methods of fractionation are similarly available, chiefly the precipitation of the intermediate 'dextrins' by alcohol.

Where the original celluloses are homogeneous we should thus obtain transformation series, similarly expressed to those of starch. In the case of the celluloses which are mixtures, or of complex constitution, there are various methods of either fractionating the original, or of selectively attacking particular monoses resulting from the transformation. By methods which are approximately quantitative a mixture of groups, such as we have, for instance, in jute cellulose, could be followed through the several stages of their resolution into monoses. To put the matter generally, in these colloidal and complex carbohydrates the ordinary physical criteria of molecular weight are wanting. Therefore, we cannot determine the relationship of a given product of decomposition to the parent molecule save by means of a quantitative mass-proportion. Physical criteria are only of determining value when associated with such constants as cupric reduction, and these, again, must be referred to some arbitrary initial weight, such as, for convenience, 100 parts of the original.

Instead of adopting these methods, without which, as a typical case, the mechanism of starch conversions could not have been followed, we have been content with a purely qualitative study of the analogous series obtainable from the celluloses under the action of sulphuric acid. A very important field of investigation lies open, especially to those who are generally familiar with the methods of studying starch conversions; and we may hope in this direction for a series of valuable contributions to the problem of the actual constitution of the celluloses.