Handwritten or typed drafts by modern and contemporary writers are composed of various kinds of paper, ranging from large double sheets of watermarked laid paper to small fragments out of pocket notepads, standard wove « EXTRA STRONG » A4 of typewriter paper or low quality cross‑ruled school notebooks.... Since genetic criticism developed in France as an approach to the study of handwritten drafts in order to explore the writers' process of composition, literary scholars turned to the material analysis of modern manuscripts (from the eighteenth century to our day) with renewed interest1. Among other tools enabling us to grasp the complex interaction between the various phases of writing and editing, paper analysis has in fact played a significant role, but one often performed backstage. It is carried out, with minimal means, in a codicological perspective2, which means that information obtained from the paper cannot be interpreted separately from other aspects such as the handwriting, the visual layout or a linguistic analysis of the manuscript.

The common category of « writing paper » has developed into a wider and wider range of products, undergoing a spectacular growth in the mid‑19th century, as literacy and education improved. Parallel to the market of printing and newspapers, stationers have invented all kinds of specific types of paper answering the needs of schools and administrations, as well as private demand3. Yet, our knowledge of handmade paper from earlier centuries is far wider than that of the recent industrial production : is paper too familiar an object in our daily practice of writing to deserve systematic examination ?

Searching for evidence

Modern codicology has to come to terms with both handmade and industrial papers, the latter bearing less visible clues of its origin. Considering the variety of material available to a modern writer, and the complexity of the working process that takes shape on the surface of paper and leaves material traces in the mass of several hundreds of leaves, what does « looking at paper » mean ? One may even wonder whether the information provided by the analysis of paper is still useful for dealing with the problems raised by the reconstruction of the writing process in modern literary works, as it has proved to be for the study of drawings, engravings and musical drafts of the past, as well as for the history of the book.

Providing a technical context to the literary artefact

As an object made of paper used for writing, a manuscript bears witness to a hidden part of the poetic accomplishment : it shows the hard work, in its material heaviness. When they decide to keep their drafts, writers are aware of their ambiguous testimonial value. Some of them even make a point of enhancing their active and intimate involvement with the material aspect of the creative activity, especially as concerns paper4. André Gide writes in his diary on the 4th of June 1949 :

« There are days when it seems that if I only had a good pen, good ink and good paper, I could easily write a masterpiece ».

Stendhal simply notes down as a landmark in a margin of f° 272 in the Vie de Henry Brulard : « March 6, 1836. New paper, bought in Civita Vecchia », while Hugo turns a similar note into a monument :

« April 29, 1865, I write the last page of this book on the last leaf of the 'Charles 1846' paper. This paper will have started and ended with this book ».

Marcel Proust, provocatively, claimed to be a « manual worker », and the amazing aspect of his notebooks, where most pages grow out into as many as four additional scrolls folded on the sides, gives us an idea of what he meant. Another advocate of the pleasure of writing's craftsmanship, Roland Barthes, also frequently indulged in collage, long before the notion of « cut and paste » was promoted by electronic text processing. In order to describe such drastic material transformations, it is necessary to look at the object without any prejudice about the process of textual elaboration.

The very notion of a manuscript as a single object ‑ a result of its conservation in a library or museum ‑ often betrays the scattered and mobile state of the written material on the writer's desk. Nevertheless, some unexpected characteristics of a specific process may come to light as one takes into account the technical context (including writing material and tools) where singular gestures of the writer have taken place.

As professionnal consumers of paper, writers also respond to the growing offer of the market. In the 19th century, for instance, they tend to behave according to two major trends : a « sedentary », stable consumption, such as that of Flaubert or Zola, attached to one or two major kinds of paper, and faithful to a couple of well known papermakers (e.g. Blanchet Frères & Kleber, Lacroix Frères) during their whole lifetime. Conversely, there is what could be described as a « wandering » consumption, such as shown by the heterogeneous manuscripts of Balzac or Hugo : all kinds of paper qualities and sizes are tested, adopted for a while, used simultaneously or successively out of some material constraint (for instance visits abroad or exile), as one indulges in a whim of fashion or casually picks up some envelope lying on a desk to scribble a note.

Different kinds of paper may be used at various periods or for specific purposes by a writer, as the manuscripts of Stendhal illustrate quite clearly, since they combine stable options of good quality paper used in full in‑folio format for his major works (mainly when in Italy on diplomatic service, but also when working with a secretary in Paris) and haphazard sets of laid or wove paper folded or cut in smaller dimensions for short notes and undeveloped drafts. A writer is liable to develop meaningful habits as well as to react to incidental events (which may remain for the most part unknown to us) not only in choosing the paper, but in using the writing surface, as well as folding, cutting or gluing it.

Another element should be taken into account before interpreting any information based on the material aspect of manuscripts : however well ordered, neatly bound and fully documented they may be, the documents we encounter today in the library or museum collections are but a partial amount of the total mass of papers actually used by a writer to compose his works. Moreover, their present state may possibly reflect a selective disposition adopted by the author in consideration of the posterity of his or her work, or even a posthumous arrangement resulting from someone else's appreciation of the archive. Such limitations should not stop our investigation, but remind us that all the clues do not lie in paper analysis alone.

Collecting datas in modern manuscripts

Some remarkable research has been done and published in the last years on the use of paper by artists. The material object had first undergone a close scrutiny inasmuch as it led to identifying or dating drawings, watercolors or prints5. Now the results of such detailed and systematic analysis have led to further conclusions concerning the choice of paper by artists (Turner, Rembrandt), and their experimenting on it various technical devices6. In other words paper is not to be considered any more as a passive surface receiving the creative work, but as a tool in the creative process, as P. Bower writes about Turner's work :

« The paper is always an integral part of the work (...). It is never merely a ground to carry an image ».

Concurrently, in the field of textual studies, material description of manuscripts has been published either as a part of the most thorough scholarly editions of complete works, such as Heine's Säkularausgabe (Akademie Verlag, CNRS), or as a separate codicological catalogue of the manuscripts, such as Der Handschriftliche Nachlass G. W. F. Hegels from the Staatsbibliothek zu Berlin presented by E. Ziesche (Harrassowitz Verlag, 1995). In both cases, the search for evidence appears fruitful : yet it remains uncertain how much of such minute and precise data will be actually put to use in the interpretation of the work, especially when the codicologist's propositions happen to question some well established scholarly interpretations.

As a first step, the limited observation that can be carried out by a researcher in the situation of a library reading room may provide enough data to identify the various kinds of paper. As long as we limit ourselves to data accessible to the naked eye, the specific guidelines a codicological description should follow in order to gather all the details needed for identification, classification or dating purposes are quite simple7. In case the paper was not bought and used as single loose sheets, the kind of object (such as a notebook or notepad) the sheet belonged to originally should be mentioned before describing its physical aspect, and it may be useful to determine whether it was separated from the binding before or after writing.

Besides the description (and reproduction) of watermarks, some physical characteristics allowing us to sort out different types of paper should be registered : for example dimensions, manufacturing technique (hand/ mechanic) and type (laid/wove) of paper, colour, smoothness, rigidity and thickness ; when laid, the average rhythm of chain and wire lines ; when printed, the color and rhythm of ruling or cross‑ruling lines, as well as the headband. Alterations such as folding, cutting, tearing or creasing should also qualify each leaf or fragment individually.

Because of such alterations, modern manuscripts present a number of obstacles to identification and dating : full sheets are seldom encountered ; when cut and folded, wove paper gives less information about the initial format than laid paper ; often partially hidden by heavy ink erasures, watermarks are not easily accessible on in‑4° or smaller folios (especially when they are bound), etc. Yet the cross‑checking of the various criteria mentioned above, when they are entered in a database, allows us to reduce the amount of unidentified papers, and may even suggest hypotheses to gather together some isolated pieces.

Theoretically, on a higher level of investigation, measuring exact thickness, degree of smoothness or colour with more sophisticated instruments or comparing the orientation of fibers by means of optic and laser analysis would facilitate comparison and checking as well as further identification of unmarked wove sheets or smaller fragments, which are very common for instance in the manuscripts of Balzac or Stendhal. Practically, such an examination has to be undertaken under the responsability of the conservation departments, which are not always open to research projects aiming at literary analysis rather than preservation8.

However precise the level of description, the next stage of research consists of comparing the papers found in a single work to other papers used by the same writer or some of his contemporaries, with a preference for complete sheets and dated documents. Identifying the papers in the author's letters may prove very useful for dating purposes, inasmuch as the same sorts might appear in some of the undated manuscripts. When Diderot, Montesquieu9 or Stendhal worked with several secretaries, in various places or at different periods, the distinction of the paper sorts gives solid ground to confirm the identification of handwritings. The high number of items and criteria thus used for comparison requires a heavily structured database system. Furthermore, cross‑checking results issued by the interrogation of the codicological database with the results of linguistic and stylistic analysis, as well as with available biographical or autobiographical datas, often sheds a new light on the perception of the genetic process.

In order to interpret properly some of the information which comes out of the codicological description, another stage of the work involves research in historical sources. Concerning recent history, available paper history essays and watermark catalogues are far from sufficient, and. existing watermark databases must improve before they become workable tools on the net, so we must still rely most of the time on sources such as commercial or industrial exhibitions catalogues10. Gathering scattered and often partial data in such a context is time consuming and often disappointing. Nevertheless some significant results have been reached through a systematic study of papers in the manuscripts of writers such as Stendhal, Hugo or Roussel, with a relatively low input from the bibliography on paper history : more discoveries are to be expected when consistant documentation on 19th and 20th century papers becomes available.

Of course, paper by itself does not hold all the clues, but looking at it closely may reveal unexpected twists and turns of the creative process, and thus provide decisive milestones to find one's way in the written labyrinth of the drafts. The fact that a piece of paper may have been used by the writer on more than one occasion, as shown by differences in inks and writing tools or by the shape of handwriting itself, makes it in most cases irrelevant to attempt a linear sequential ordering of the sheets. When scholars in genetic criticism establish virtual sequences based on the results of the codicological analysis correlating the various stages of the « work‑in‑progress », they are often more attached to the relative chronology than to dating strictly speaking.

Bearing in mind the limits of such an investigation, the question one follows up in a systematic research on the paper found in literary manuscripts should not merely be : « Which kind of paper was used ? » — leading to an identification process based mainly on datas related to the history of paper production —, or even : « When was it used ? » — a question which requires cross‑checking historical data about production and biographical data concerning the writer, and to compare as many autograph documents as possible — but also : « How was it used ? » — a question opening up a wider perspective on the history of writing in its material and cultural dimensions, at the cost of an even more cautious, or skeptical, approach to what historians would call « material evidence ».

Case studies

Dating and chronology based on paper evidence

Following in Briquet's steps, art historians, bibliographers and philologists have often focused their interest in paper on watermarks. Yet, as Pr Irigoin underscored11, dating a single item, or a group of works, with the help of watermarked papers implies a much wider knowledge of the history of paper than is provided by a glance at the major watermark catalogues available. His point is even more convincing when applied to modern documents, considering the scarce proportion of 19th and 20th century watermarks in the published catalogues. Moreover, as I hinted at earlier, the lack of documentation is not the only obstacle : the supposedly « individual » shape of the hand‑made watermark on the mould used to produce paper sheet by sheet has little to do with the stereotyped metal cast fixed on the dandyroll whose imprint is repeated at a regular rhythm on the surface of the endless real of machine‑made paper12. The latter is not always strictly connected to a single sort of paper or even to a determined papermaker. Hence the difficulty to get any useful information from such standard watermarks as the famous « J WHATMAN » found in a great number of wove sheets under the quill of Balzac in the 1830's, for instance. To what degree is variation in dimensions or position of the mark significant ?

Nevertheless, in cases where the mark does bear either a mill's name, or several names (or initials) of associated partners referring to a papermaking company whose official existence can be precisely  determined, identifying the paper's watermark can still provide at least a terminus a quo to date a work by. Among the ink drawings of Victor Hugo, some happen to be completely abstract, depriving art historians of any thematic or stylistic landmark, which entails two opposite interpretations : some experts date them as part of the earliest graphic experiments carried out by the poet, others on the contrary as a freehand capriccio by a mature artist. Fortunately, one of the wash‑drawings was bearing a watermark « DAMBRICOURT FRERES/ Hallines ». After exploring a large sample of early and late manuscripts of Hugo, we found out it appeared only in later works, such as La Légende des siècles (1875‑1878) — which did not come as a surprise considering the style of the watermark as well as other technical aspects of the paper. A further enquiry gave us the exact location and dates of activity of the mill, confirming the hypothesis of a late date, after 187513.

When a watermark includes a date of manufacture, the problem of dating seems quite easily solved. Paper historians have even determined an average delay between date of manufacture and date of use, although such a delay is bound to undergo important variations according to historical and geographical circumstances14. One also has to take into account a greater degree of freedom among private users, as compared to the normal professional (publishers, secretaries) and administrative consumption of paper, which is compelled to have a quick turnover of the stock. Furthermore, a number of examples bring to our attention the deliberate choice of writers or painters to buy some old paper, or to keep some in a drawer for a long period of time. A note in Victor Hugo's diary gives us some details about the paper he calls « Charles 1846 » (whereas the watermark actually reads « C. Harris/ 1846 »), which he bought in January of1864, during his exile in Guernsey15. This information is valuable, considering his note at the opening of the manuscript of Les Travailleurs de la mer : « June 4, 1864. I start using today this Charles 1846 paper which Bichard sold me as unalterable » as well as the final remark, quoted earlier. In the present case, it would obviously be a mistake to question the autograph date of the manuscript (although they are often misleading in Hugo's manuscripts) in favor of the watermark evidence.

In most cases literary scholars are not so much confronted with a problem of dating as they are concerned with establishing a relative chronology, enabling them to situate one folio « before » or « after » another one. For that purpose the connexion between the paper used for drafts and for letters may provide a precise chronological reference. Such a connexion is most effective with writers of the 18th century, who did not have at their disposal a specific kind of paper for letters, as would be the case in the next century. In a thorough study of Diderot's papers, M. Bockelkamp was able to assign to each kind of paper found in the manuscripts a limited span of time, according to the kinds identified in the philosopher's almost lifelong daily correspondence with Sophie Volland16. Some undated autograph pages kept in the Russian national archives among the late copies of his manuscripts ordered by Catherine II could be approximately dated thanks to this chronological classification of paper.

Similar research has been recently undertaken on the « Notes » of Marcel Duchamp, for the most part undated. This corpus combines half French and half American papers, kept in major museums and collections on both sides of the Atlantic. Here the question of dating each fragment becomes crucial, since the artist kept coming back to some specific projects on several occasions after as much as ten or twenty years. Yet the methodology adopted with Diderot's papers proves to be not as effective for several reasons, depending both on the history of paper and the individual modalities of use by the artist. On the one hand, industrially produced paper does not bear as many hints betraying its origins as handmade papers do : in particular, watermarks in the 20th century are less frequent and less informative, and many sorts have become a standard product, available in a great number of slightly different versions, such as the famous American « legal pad » yellow paper17.

The systematic comparison of Duchamp's notes with his letters, although it does allow much valuable cross‑checking, is less useful in view of the fact that numerous letters are not dated and many of his notes are scribbled on second‑hand material, such as pieces of printed material (margins of a page torn form a telephone book, back of label for a Camembert box) which are seldom used for letterwriting18. On the other hand, some of the printed papers provide useful clues for dating or localizing the original document, which might give us a terminus a quo : it is the case with dated electricity bills, invitations to weddings or exhibit openings, telegram forms printed in Spanish issued by a Buenos Aires company, as well as with printed letter‑head from a hotel in Copenhagen or a transatlantic travel company. Some envelopes of letters addressed to the artist also fortunately show a dated postal stamp, when they have not been torn in smaller fragments.

As a matter of fact, Duchamp's habit of dividing the page into smaller pieces appears as one of the major obstacles to our investigation. The most frequent treatment of a large enough sheet of paper would be for him to fold it in two, write on the first, fourth (exterior) and then second and third (interior) pages, and eventually separate the two halves. When the two halves have both been preserved and if we found them among hundreds of fragments, the irregularly torn side allows us to put them back together as two matching pieces of a jig‑saw puzzle. Another frequent manipulation consists in tearing a single corner out of a fully written page, the smaller piece being usually preserved whereas the biggest one has been either lost or destroyed, thus depriving the remnent piece of its original (genetic) context.

In other words, it is not by chance this large collection of scattered fragments keeps resisting all attempts at ordering, numbering or classifying. Our hope is that the material analysis may reveal some of the deliberately hidden relationships linking the wandering pictorial or verbal messages that Duchamp himself published in fac‑simile, the Green box, as loose unbound fragments, so that the reader would make up his own random combination as he went. For that reason his notes never became a « text », but remained as a living unfinished form perpetuously showing its own genesis in statu nascendi.

Paper against textual evidence : Deconstructing the manuscript as an object

Taking for granted the object as it is preserved nowadays in a library binding has led some philologists astray for several years before a more precise codicological analysis of the material leads to criticism of previous interpretations. This remark does not mean the classification adopted by the librarian is pointless : as a matter of fact, the purpose to give access to a « readable » manuscript often entails a material paradox when the draft gives several parallel layers and conflicting versions intertwined on the same sheet of paper, so that they are materially inseparable although they belong to different phases of the genesis19. Yet, some early parts of a work, erased and deleted or scattered by later interventions of the author, may well pass unnoticed in the process of « ordering a manuscript » according to the final and best known (i.e. in most cases published) version. Even before this ultimate step, we are aware of the risks undergone by the mass of papers as they are handled by the family, friends, heirs or collectors before they find a material stability. Moreover, working on drafts teaches us how mobile a « work in progress » can be throughout its career in the writer's hands ! Thus, there are many reasons to question the object in its present state and search for safer grounds for interpretation...

Among the recently discovered drafts of Raymond Roussel (1877‑1933), the Bibliothèque nationale de France has bound in three volumes some 915 autograph leaves related to the novel Impressions d'Afrique20. Two volumes gather the chapters of the first and second part respectively, whereas all the fragments which did not appear in the published version — either because they represent early summary versions of some episodes, or because they have been rejected at some point from later stages of the work — are found in the third volume. At first sight, describing the papers seemed an easy task, since everything was written on the same kind of material : the 915 leaves came from standard school notebooks (approximately 222 x 172 mm) composed of low quality ruled paper that Roussel apparently used as loose sheets. Yet, a closer examination revealed the presence of seven different sorts of laid paper, of which four happened to bear a watermark, and two sorts of wove paper without watermarks but bearing different ruling, one of them significantly smaller in size than all the other kinds (209 x 166 mm.).

Thanks to the earliest autograph foliotation of the pages, we have been able to reconstruct several long sequences developing a coherent narrative. Written initially on either one of the two kinds of wove paper, this early version is now scattered amongst the three volumes, since some of the leaves have been corrected and kept in the late version, in a different order, while others were abandoned and put aside. A number of pages have been cut and glued onto a different kind of paper bearing new developments, yet many of these fragments can be matched thanks to the irregularly cut edges, allowing the reconstitution of their original state. As opposed to the continuity found in the use of the wove papers, the different kinds of laid paper are generally used in short sets of less than ten leaves, which means they had been inserted later in the genetic process. In most cases though, the early strata plainly dissappeared underneath heavy erasures and additions superposed on the same page. Starting with 89 disconnected leaves, we ended up restauring five early narrative sequences (comprising altogether 232 written pages), whose existence was not even suspected.

Such a spectacular result is all the more striking since the apparent homogeneity of the material components of the manuscript sustained very low initial expectations. Without the guiding thread of paper analysis, the chances for a philological study to find a coherence between so many scattered fragments was rather limited. In fact, the cohesion of the last version, set forth by the finalized organization of the volumes, hid its own genetic background to the eye of the reader : to discover it, one precisely had to give up reading in order to just look at the paper. A similar method was applied to the manuscripts of Stendhal's last novel, Lamiel, which was left unfinished at his death. In that case, a long series of posthumous publications have attempted at organizing the various components of the archive as a whole, accessible to reading. But the critics had to argue that only the author's degrading state of health could justify a great number of mistakes, gaps and discrepancies in the narrative, such as constant alterations of the main characters names.

In his genetic approach to the problem, Serge Linkès followed the methodology which proved successful with Roussel's drafts, and started with the description of the paper21. Using a database to cross‑check the codicological data with the information about the handwriting, the places of work, and the dates of Stendhal's travels between Italy and France at the end of his life, he discovered the chronology generally accepted was not compatible with the material composition of the manuscript, especially concerning the genesis of the very first period of the work.

To make a complex story short, after jotting down a sketchy program of action and characters in Paris in May 1839, the writer is supposed to have developped a first draft of 71 pages while he was back in Civita Vecchia at the beginning of October, followed by 87 more pages at the end of November. He would have resumed work on Lamiel's first chapter during his stay in Rome at the beginning of January 1840, completing the October version up to some 112 pages as he dictated it to a secretary. He would have started correcting this new version in the first days of February, then kept on working sporadically on the second chapter for two years, until a few days before his death, in Paris. From several remarks in his letters, we learn that he highly praised the capacity of his secretary in Paris, named Bonavie, and was not so satisfied with the one he found in Rome (no one was apparently available to work in French in Civita Vecchia) — although this second secretary remained anonymous, Stendhal refers to him in the marginalia of Lamiel by means of an ugly crow's head, probably a pun on the man's name.

The correlation between the types of paper and the identification of the handwritings gives a quite different picture of the beginning of the work. According to Linkès, the 80 pages which were obviously dictated to the French secretary Bonavie on French paper (a wove paper marked by an embossed stamp to the name « Chambellan, Paris », probably provided by Bonavie's agency) must date from Stendhal's stay in Paris, that is in May 1839, shortly following the autograph sketch. In October, back in his office in Civita Vecchia, the consul starts re‑writing the first pages of the novel on a different wove paper, then gives up the autograph copy to work directly on Bonavie's copy. Once the correction is finished, since he does not have a secretary in Civita Vecchia, Stendhal goes on writing a few pages in his own hand, apparently without much conviction. It is only when he goes to Rome in January that he is able to develop a new version, amplifying as he dictates to the Roman secretary the first Parisian version he reviewed during the fall. This amplified second version (112 pages) is written in the « Crow's" hand on Italian wove paper bearing the watermark « Feliciano Innamorati ».

The most important point in this new perception of the chronology is that the short autograph version of the beginning of the novel is not a first draft but an aborted new version written in Italy. Thus the earliest narrative development is done orally, dictating to Bonavie, and precedes the writer's autograph tentative revision : a working scheme that apparently contradicts most critics' preconceptions, since none of them tried to question the model of the autograph first draft followed by a neat copy in a secretary's hand. This new chronology established on codicological grounds solves most of the incongruencies and inner contradictions between the successive versions. It also shows that Stendhal obviously lacks energy to write in his own hand, but that his sickness does not prevent him from mentally composing some 80 pages in a few days, and that he actually keeps control over the narrative and onomastic transformations of his novel through a year of discontinued work in various locations. Linkès's approach also demonstrates that a long‑lasting biographical legend may obscure material evidence for several generations before a new comer simply endeavours to look at the paper...

The use of paper in the writing process : beyond evidence ?

Although describing the physical characteristics of modern literary manuscripts and searching through the recent history of papermaking to document one's findings offer to the codicologist a lifetime project, there is a second motivation for our enquiry, which eventually takes us beyond the aim of discovering and gathering evidence. Our purpose in examining manuscripts is not only to help define and understand the writing process by providing « factual » elements : it also opens up a different approach to the creative phenomenon. The question is not only « which » paper has been used, « when » and « by whom » (the major topics for the art historian or expert), nor by whom and how this paper was produced and sold (main focus for the paper historians), but ultimately how it was used by the artist or writer.

The results of such an investigation give a revelating picture of some material aspects of the work that are seldom commented upon by writers themselves. While they may be tempted to modify their manuscript's dates or tell tales about the genesis of their works, it would be surprising to see authors intentionally alter physical details of their drafts a posteriori — unless they decide to destroy them. It is precisely because they are performed in most cases out of the control of the writers' social image that the gestures performed on paper to shape the work materially according to the poetic necessity are so unique, and so meaningful : looking at paper to uncover the indexical meaning of its transformations is a way to approach the creative phenomenon qualitatively, in its dynamic singularity22.

The never‑ending outgrowth of Proust's collage for La Recherche du temps perdu and, in a less spectacular way, the unexpected treasures hidden in Roussel's drafts show that a material as common and basic as a school notebook may be used, under the urge of creation, in so many completely original fashions. As reveals the radical change of perspective on Stendhal's working habits, a minute detail such as an embossed stamp barely showing at the corner of a folio may lead one to reconsider a whole conception of the genetic process. It is true that the scarce available resources in recent paper history do not always provide sufficiently solid ground to new interpretations ; yet, many of those could not be reached without the contribution of paper analysis. Through their common search for evidence, codicologists and literary scholars are beginning to find out to what extent the paper really plays an active part in the genetic process. Looking at paper as it is, without yielding to well‑established preconceptions about what writing means, brings to light some of the writers' most familiar and compulsive gestures, up to now unnoticed. It seems paper still has a lot to teach us about how we write. When Yves Bonnefoy argues that « writers' manuscripts raise more problems than they can solve »23, Duchamp's stimulating answer would be :

« There is no solution, because there is no problem »

1  Cf.A. Grésillon, Eléments de critique génétique, Lire les manuscrits modernes, Paris, PUF, 1994, and L. Hay (ed.) Les Manuscrits des écrivains, Paris, CNRS Edition, Hachette, 1993.

2  Codicology, a recent diciplin devoted to the study of the « codex », was first applied to medieval manuscripts. Since the 1970’s, the Institut des textes et manuscrits modernes (C.N.R.S., Paris) has developped and adapted this approach to literary manuscripts of the modern times : C. Bustarret « L'archéologie du papier et la génétique textuelle, de Blaise Pascal à Marcel Duchamp », Actes du Congrès de Leipzig (1996), Association internationale des Historiens du Papier, IPH Congress Book, vol. 11, 2000, p. 23‑28.

3  Cf. A. Lacroix, Historique de la papeterie d'Angoulême, Paris, Laîné et Havard, 1863.

4  Cf. C. Bustarret, « L'énigme de l'extra‑strong », Cahiers de médiologie, 4, 1997, p. 85‑97.

5  Cf.Essays in Paper Analysis, ed. Spector, Folger Books, 1987.

6  P. Bower, Turner's Papers, A Study in the Manufacture, Selection and Use of his Drawing Papers, (2 vol. 1787‑1820, 1820‑1851), London, Tate Gallery, 1990, 1999.

7  Our own descriptive grid almost matches the data mentioned as indispensable (marked with a *) in the International Standard for the Registration of Watermarks published by IPH, 1996.

8  Our institute has recently undertaken a project in collaboration with the Bibliothèque municipale, the « Equipe des Manuscrits de Stendhal » (Université Grenoble III) and the Ecole nationale de papeterie in Grenoble aiming at a systematic description of papers in the manuscripts of Stendhal.

9  Bustarret, C. « Papiers de Montesquieu : une approche codicologique du fonds de La Brède », Revue Montesquieu, 3, 1999, p. 179‑187.

10  P. Bower, « Who made this rubbish? — The historical investigation of particular twentieth‑century papers », IPH 1996, 1, p. 12‑20.

11  J. Irigoin, « La datation par les filigranes du papier », Codicologica, 5, Leiden, 1980, p. 9‑36.

12  M. Bockelkamp, « Wasserzeichenmotive im Maschinenpapier aus literarischen Handschriften », Papiergeschichte(n), Papierhistorische Beiträge W. Schlieder zum 70.Geburtstag, hrsg. von F. Schmidt, Wiesbaden, Harassowitz, 1996.

13  Catalogue Soleil d'encre, Manuscrits et dessins de Victor Hugo, Bibliothèque nationale, Ville de Paris, 1985, n°424, 425, p. 283.

14  R. Jones, « From papermill to scribe : the lapse of time », Papers from the III European Colloquium on Malay and Indonesian Studies, Istituto Universitario Orientale, Napoli, 1988, 153‑169.

15  M.‑L. Prévôt, « Écrit sur une page blanche, Les écrivains et leurs papiers », Cahiers de Médiologie, 4, 1997, p. 171‑178.

16  M. Bockelkamp, « L'analyse bétaradiographique du papier appliquée aux manuscrits de Diderot », Studies on Voltaire and the Eighteenth Century, 254, 1988.

17  As the descriptive part of this project is coming to an end, I would sincerely appreciate any help to localise bibliographical or commercial sources enabling me to identify or date some of the american papers he used between 1915 and the 1960's.

18  At the back of a letter to Katherine Dreier, written on a rough paper torn from an accounting notebook printed with red and blue columns Duchamp hastily wrote in big capital letters « EXCUSE PAPER !!! ».

19  Therefore a genetic classification of manuscripts is necessarily virtual, either through verbal reconstruction, or by means of electronic devices allowing an hypertext (i.e. dematerialized and non linear) presentation of the written material :cf. J.‑L. Lebrave, « Hypertextes‑ Mémoires‑ Ecriture », Genesis, 5, 1994, p. 9‑24 ; « Hypertexte et édition génétique : l'exemple d'Hérodias de Flaubert », Banques de données et hypertextes pour l'étude du roman, N. Ferrand (dir.), Paris, PUF, 1997, p. 137‑154.

20  We are here leaving aside the typed versions as well as the sets of proofs which belong to the final phase of the genetic process. See C. Bustarret, A.‑M. Basset, « Les cahiers d'Impressions d'Afrique : l'apport de la codicologie à l'étude génétique », Genesis, 5, 1994, p.153‑165, and Basset's Doctorate dissertation La genèse d’ Impressions d’Afrique de Raymond Roussel ou le mythe de la création (Université Paris III, 1996).

21  Cf. S. Linkès « Des clés pour les manuscrits de Stendhal : le cas Lamiel », Genèses, Proceedings of the 2nd international congress of genetic criticism, Paris, Sept. 1998 (to be published shortly by ITEM, CNRS).

22  Cf. C. Ginzburg, Mythes, emblèmes, traces, Flammarion, 1989.

23  « Enchevêtrements d'écriture », entretien avec M. Collot, Genesis, 2, 1992, p. 123‑130.