Proust at the Drawing-Board

/ /

Devotees of Proust have had much to celebrate recently. 2021 and 2022 were significant years, the first marking the 150th anniversary of his birth, the second the centenary of his death. His publishers, Gallimard, rose to both occasions. In 2021 they issued a fully-annotated edition, by Proust’s great-grand-niece, Nathalie Mauriac Dyer, of the previously unpublished Les soixante-quinze feuillets et autres manuscrits inédits. The “seventy-five folios” (in fact seventy-six), dating from 1908 and long believed to be lost, were found in 2018 among the papers of the recently deceased Bernard de Fallois, who had referred to them in his edition of Contre Sainte-Beuve (1954) (hereafter CSB) and had published two extracts. Their relation to À la recherche du temps perdu (hereafter RTP) is discussed below. Then, in 2022, a new edition of Proust’s Essais appeared in the prestigious Pléiade series, including Pastiches et mélanges and CSB, replacing and greatly augmenting the previous Pléiade edition of 1971. Finally, in 2023, we had not only a translation of Les soixante-quinze feuillets by Sam Taylor (The Seventy-five Folios and other Unpublished Manuscripts, Belknap Press, Harvard, $29.95; hereafter 75F) but a new book on Proust by Michael Wood in Oxford University Press’s “My Reading” series of personal reflections on great writers ($24.95), which I shall come to later.

Sam Taylor’s translation is fluent and idiomatic. Inevitably, it employs some American usages which sound odd to British ears, but I have noticed only two things which I would call mistakes. On page 73, the narrator’s mother, speaking to her chambermaid, refers to him as “Monsieur Marcel”, where the corresponding place in the French has “Monsieur Marcel”.  Given the perennial debate about the narrator’s first name, Proust’s deletion should surely be reflected in English. Also, on page 92, I would prefer “in the Underworld” to “in hell” as a translation of “aux Enfers,” particularly as the allusion in context is to Aeneas. More generally, it seems extraordinary that the translation used for quotations from RTP should still be that of Scott Moncrieff as revised by Terence Kilmartin and D. J. Enright, rather than the completely new, and more reliable, version published in 2002 by Penguin, under the general editorship of Christopher Prendergast. Perhaps there were copyright issues here?


The reappearance of 75F further complicates an already Byzantine chronology, of which the following is a summary:

1895—1902: Proust works on Jean Santeuil (hereafter JS), which he then abandons. It is eventually published in 1952, in an edition by Bernard de Fallois, who rearranges, cuts, and in places rewrites the text, with the aim of producing a usable reading version. A different text, following Proust’s manuscript, appears in the Pléiade edition of Pierre Clarac in 1971.

1908:  75F composed, perhaps using lost earlier drafts.

1908—1910: Proust works on CSB, which he also abandons. It is first published in 1954, again in an edition by de Fallois, then in 1971, in a radically different version, in the Pléiade series, edited by Pierre Clarac and Yves Sandre. The editors had conflicting conceptions of a project which Proust himself never brought to final form. De Fallois, judging that CSB was neither an essay nor a novel but simply, unclassifiably, “a work”, printed both the narrative prelude, the conversation between Proust and his mother, and the sections on Sainte-Beuve and other writers. Clarac and Sandre printed the latter only. The 2022 Pléiade Essais adopts a different policy (see below).

1909—1912: most of the “other manuscripts” included in 75F composed.

All the above works contained material which fed into RTP, whose own complicated textual transmission will be outlined later.


Proust’s method of composition resembled that of a jigsaw, in which a huge number of tiny pieces, differing only fractionally, must be moved around until each falls into a place which then seems to have been waiting for it all along. And just as with a jigsaw, sometimes bits which could be made, with an effort, to go together had to be prised apart and tried with other bits before they fitted perfectly. Although 75F was described by the jacket copy of the French edition as the “earliest version” of RTP, this has been sensibly toned down by the English publishers into “early versions of six episodes included in the novel.” To be precise, apart from three pages of Albertine disparue, 75F has contributed exclusively to Du côté de chez Swann and À l’ombre des jeunes filles en fleurs.

The publisher’s replacement of “earliest version” by “early versions” is also prudent because, as noted above, 75F, and RTP itself, incorporate revisions of material which had already been tried out in JS. These include—although with different place and character names—the episode of Jean’s mother’s goodnight kiss, the meeting with his first little sweetheart in the Champs-Elysées, the description of the village where he spends his childhood, his initiation into the aristocratic milieu, his seaside holidays, the military life of the garrison, the insane jealousy of the lover, and the meditations on the depredations of time, the inexorable processes of ageing and death. We even have the petit phrase of Vinteuil’s sonata, here attributed to Saint-Saëns; and although we do not have the madeleine, there is a sustained passage on the associative power of memory to transcend Time, and an instance of involuntary memory. In addition, there are early sketches in JS for several characters who will reappear, with different names, in RTP, including Norpois (Duroc), Saint-Loup (Betrand de Réveillon), Bergotte (Traves), and Charlus (Lamperolles).

A crucial new feature of 75F is Proust’s adoption of first-person narration in place of the third person used for JS. (“Un Amour de Swann” in RTP, being an historical flashback, is a third-person narrative with occasional interpolations from the “I” who tells the rest.) From the beginning he was uncertain about the form and genre of JS; a rejected preface fragment begins, “Can I call this book a novel?” Employing a creaking metafictional device, he opened with an introduction by an anonymous first-person narrator, who informs us that we are about to read a manuscript by the celebrated author C., now dead, which has fallen into his hands, a work which interweaves passages of fiction with thinly-disguised autobiography and essayistic digressions.  Proust is evidently covering himself in anticipation of criticisms. The supposed manuscript of C. follows, telling Jean’s story. Reversing the common procedure, in which a first novel tends to be a first-person narrative because it draws heavily on its author’s own life, Proust attempted to objectify his experiences by creating the character of Jean as his surrogate and seeing him through someone else’s eyes. This seems to have felt wrong, and the project was abandoned in favour of translating Ruskin. At some point Proust must have realised that the whole point of the project was that it was subjective, not objective, enacting rather than describing the growth of a mind, so he began again with 75F. Furthermore, in eventually making the narrator a novelist he is departing significantly from JS whose hero dreamed of being a poet. Proust saw that the kind of philosophical and psychological exploration he wished to make was only possible if fiction was the subject of a fiction.


75F contains six blocks of text. The first, “An evening in the countryside,” is the longest continuous item; probably incorporating earlier work, it provides a humorous description of the narrator’s family (given actual Proust family names, which will later disappear), stressing its eccentric characters and their tendency to bicker. The episode of the mother’s goodnight kiss reappears from JS, and there is a description of her deathbed which, in RTP, will be transferred in modified form to the narrator’s grandmother. This is followed by a wonderfully comic scene in which the narrator’s brother, about to leave the countryside to return to Paris, throws a histrionic fit of despair at being parted from a goat to which he has become very attached and which he immediately forgets; this found its way into CSB in the section “Retour à Guermantes”, and was adapted in Du côté de chez Swann where it is the narrator who bids a more sentimental farewell to his beloved hawthorns.

The second episode in 75F, “The Villebon Way and the Meséglise Way,” exists in three versions, which evoke scenery, atmosphere, and sensations in striking detail and include a first version of the vision of the three trees, whose significance remains hidden from the narrator, which will reappear in CSB and in “Noms de pays: le nom” in À l’ombre des jeunes filles en fleurs. “A visit to the seaside” opens with a sketch of the narrator’s grandmother’s wilful pursuit of healthy air to the discomfort of other, unnamed hotel guests, some of whom are described and will later play more definite roles in RTP. An extended pen-portrait of the narrator’s philandering uncle contains many ideas that would later contribute to the character of Swann. “Young girls” consists of two versions introducing what would become the petite bande, but without Albertine, who emerged as a character only in 1913; the narrator’s desperate attempts to get to know them, exploiting his acquaintance with Monsieur T. (in a later version a painter, and ultimately Elstir), are detailed with rueful self-mockery. The reflections on the poetry of names in “Noble names” became part of “Noms de personnes” in CSB and, of course, form an obsessive topic in RTP. Finally, the brief section “Venice” refers back to Proust’s study of Ruskin; essentially a lively piece of travel writing, it gives no hint of the significance which Venice will assume in Albertine disparue or Le Temps retrouvé.

Not all the material in 75F was used subsequently, and large parts of the novel we know were yet to come. Moreover, those parts which are drawn upon are dispersed in different parts of the final work, which ballooned from the original project of three volumes as its publication was delayed by World War I. For example, two pages in “Combray” use folios 41, 27, 40 and 51 in that order, while “Noms de pays: le pays” jumps within a few pages from using Folios 12 and 13 to using folio 53. In addition, some passages in 75F appear, revised, in the drafts for CSB. Mauriac Dyer has done a marvellous job in disentangling Proust’s wayward procedures.

Of the other manuscripts included in 75F, some of the material looks back to JS, some overlaps with CSB, but most of it represents early, or the earliest, versions of episodes that would find their final form in RTP, especially that of the goodnight kiss (no. I) and the madeleine (in no. IV, a distinctly unpoetical piece of stale bread, in no. VII a biscotte). The most important extract, no. XII, comes from Notebook 4, which, as Mauriac Dyer says, has been considered the place “where In Search of Lost Time was born” on account of its fusion of strands of CSB with reworked passages from 75F concerning the evening walk, the “two ways” and the seaside holiday. These are now more closely integrated into a narrative framework, and enriched with descriptive detail and character sketches. Notebook 4 also contains the first mention of Swann, whose character takes over from the earlier M. de Breteveille (as the evening visitor to the narrator’s parents) and the narrator’s uncle (as an inveterate womaniser) and affords an opportunity to develop the theme of Jewish identity. The new volume of Essais also contains extensive extracts from Notebook 4 among its materials on CSB, seen as marking the swerve from essay into novel.

No. VII of the additional manuscripts was printed, with some textual variations, as the authorial preface to CSB in de Fallois’ edition. It contains the key episodes of involuntary memory: the biscotte (later the madeleine), the line of trees, the uneven paving-stone in Venice, and the spoon dropped on the dinner plate, all of which will reappear in Le Temps retrouvé as part of the “Adoration perpétuelle,” the exposition of aesthetic theory. CSB thus not only forms the centrepiece of the new volume of Essais but also acts as the bridge between 75F and RTP.   


What kind of text did Proust think he was producing in CSB? In de Fallois’ version, Proust’s preface, introducing the idea of associative memory and relegating intelligence to a secondary role, is followed by four chapters of reminiscence before the conversation with Maman, which leads into the critical sections of the work and back, via Balzac and M. de Guermantes, to fiction (including the dissertation on homosexuality) and a concluding meditation on the nature of literary creation. The original readers, in 1954, would have had the impression of a baffling mélange of genres, in which, nonetheless, they would have recognised much that was familiar to them from RTP. Although de Fallois interfered with the manuscript, he followed Proust’s intentions with regard to the form of the whole. Antoine Compagnon, in his general introduction to Essais, establishes that Proust distinguished the article, written for ephemeral newspaper publication, from the essai, a more substantial piece intended for a heavyweight review, and the étude, the product of scholarly research aimed at an academic readership. The distinctions were not hard and fast, however, and in the correspondence about CSB printed in Essais, Proust uses all three terms to describe his work in progress, and also récit for the narrative of the conversation with his mother. Sometimes he simply calls what he is writing un livre or un long ouvrage. In one letter, making an unsuccessful sales pitch to the editor Alfred Vallette, he even calls it un véritable roman.

Given that there can be no definitive text of CSB, the new Essais prints what the editors call a “dossier” of materials, which runs to five hundred pages, and a further selection of appendices—authorial notes, fragments, and relevant letters from 1908 and 1909 about the project from Proust to his friends. This is what de Fallois, in 1954, rejected as unwieldy, “a scholarly edition aimed exclusively at specialists, which would present all the variants [and] respect the incoherence of the episodes.” While one can look askance at the previous editors’ taking it upon themselves to make decisions on Proust’s behalf, the undigested mass of drafts, revisions, repetitions, and discarded versions with which the reader is confronted is daunting. The excellent critical discussion by the editor of CSB for Essais, Matthieu Vernet, shows that the tidy separation between “critical” and “creative,” which has hitherto shaped the editorial tradition, misconceives the evolution of the work and that, in fact, it was precisely the tension between the two modes which catalysed the emergence of RTP. The dossier contains passages—abandoned and redrafted time and again with obsessive concentration—on Combray, Swann, the Guermantes, encounters with the young girls, Charlus (here called Guercy) and the race des tantes, among other subjects.

Proust approached the reading of Sainte-Beuve, both for the pastiche published in Pastiches et mélanges (1919) and CSB, in a characteristically unacademic fashion. Although he read widely, he could hardly be expected to work systematically through the critic’s total output of more than six hundred essays, but he tended to skimp the middle of the essays he did read, homing in on the beginning and the conclusion rather than following the argument closely. His notes for the project, often little more than a few quoted words or an enigmatic allusion, occupy twenty-two pages in Essais, and require almost five hundred footnotes from the indefatigable Vernet.  Proust’s objection to Sainte-Beuve’s biographical method of interpretation anticipated T. S. Eliot’s separation of “the man [sic] who suffers” from “the mind which creates.” For Proust these two entities were different kinds of self, the social moi extérieur and the moi profond who is the source of creativity. In RTP, the je of the narrator and the je of the character blur the boundary between the work and its creator, acting as what Roger Shattuck has neatly described as “an eternal pivot chord”. (Bernard de Fallois, in his preface to the 1954 CSB, quoted Proust’s “the character who says ‘I’” and neatly qualified this with “it’s not so much a character as a tone”.) Moreover, to write a pastiche of an author is also, implicitly, to produce a critique of that author: the creative and the critical are not at odds. The imaginative parts of CSB, such as the passages about waking from sleep (which alternate between third and first person in the drafts) or the ray of sunlight on the balcony, are themselves a demonstration of the case against Sainte-Beuve, because they seek to render impressions, to capture the fleeting moment or sensation, rather than to conduct a detached, rational analysis. Vernet comments that “if the distinction between the moi social and the moi profond came to Proust in the course of working on Contre Sainte-Beuve, it may also be the case that the autobiographical portion of The Seventy-Five Folios was still uppermost [in his mind] and that the identification of author and narrator could have come about quite spontaneously.”


Faced with such an abundance of material, the critic has to choose between a potentially superficial general survey or a potentially infinite scrutiny of details. At only 120 pages of text, Michael Wood’s Marcel Proust opts for the broad sweep but, even so, it is difficult to get through. It is too compressed, gnomic and oracular in places; the meaning of whole sentences at a time, and the logic of their arrangement, is unclear (at least to me). Wood rehearses the story of 75F, remarking how odd it is that these pages give us “so much of the thematic material” of RTP yet “so little of the style and mood.” The reason is, as we saw above, that it was in working on CSB, which gave him the opportunity to write, and obsessively rewrite, passages which would feed into the novel, that Proust found his own voice, as distinct from the ventriloquial voice he used in JS.

What emerges (eventually) from Wood’s book is that Proust is a kind of epistemologist.   Recovering, or losing, a memory is a retrieval or a disappearance of a piece of knowledge about ourselves, and also about other people insofar as they figure in our memories. History, which is the communal memory of society, is equally the repository of knowledge, which hinges on matters of evidence, proof, testimony, probability, and judgement, all of them connected with what we know, or claim to know, or set aside as unknowable. The Dreyfus affair is an object lesson in the difficulty, if not the impossibility, of establishing the truth about historical events. In RTP, the Affair is alluded to more obliquely and quizzically than in the more documentary presentation in JS; it becomes an indicator of people’s prejudices and assumptions about ethical and legal systems, rather than a “case” to be judicially determined. (An analogy may be made with the presence of World War I in Le Temps retrouvé; Proust is not writing an historical novel.)

Similarly with sexuality: the agonies of Swann over Odette, or of the narrator over Albertine, warn against the difficulty—even the futility—of trying to know another person. The narrator’s shock at realising the truth about Charlus and Jupien, which hardly diminishes as one after another of his friends and acquaintances turns out to be homosexual or bisexual, emphasises the element of concealment necessary to maintain a façade of social respectability. Social class is predicated on differentiating the people one can “know”—e.g. by acknowledging them in the street, inviting them to one’s house, or considering their eligibility to marry into one’s family— from those one cannot, taking into account their kinship network, religious or racial identity, or moral code: this underlies the exhaustive taxonomy of snobbery, by turns appalling and hilarious, which so fascinates Proust and his narrator. The discovery of a vocation, which is made in Le Temps retrouvé, is the final piece of self-knowledge for which the narrator has been waiting.

Yet, if Proust was an epistemologist, he was not a Cartesian rationalist. Rather, he was a Platonist who held that memory is recollection. The passage printed by de Fallois as the preface to CSB expresses a distrust of “intelligence” which is the product of Proust’s dissatisfaction with Sainte-Beuve’s confidence about the usefulness of biography to literary criticism. No amount of knowledge about the life of Stendhal prevented Sainte-Beuve from judging his novels “frankly detestable,” a verdict which Proust finds absurd. Michael Wood explains that Proust was here “attacking the mind at what almost everyone thought was its reflective and practical best.” He was not advocating mindlessness, or a variety of surrealist automatic writing, but maintaining that books, at any rate if they are any good, are not written by the moi extérieur but by the moi profond. The knowledge the artist needs is more a matter of intuition, impression, instinct, than of logic or rationality, and the same is true of the critic. Proust is close to Pater here (whom we know he read), an aesthete but not a dandy. The illusion that he was a casual or impulsive writer will not survive a glance at his manuscripts. There was always another nuance to capture, if one only tried again. Wood is interested in how Proust shakes the mental kaleidoscope (a favourite image of his) to displace tidy definitions and categories. Knowledge is neither given a priori, nor static.


How much did Proust know about his own novel? This is not a frivolous question. In 1912, he envisaged RTP as a work in two volumes, and he already knew how it was going to end, but the halt on publication caused by the outbreak of war led to the manuscript doubling in length, and there is no reason to think that Proust believed, when he wrote the word “Fin” on the last page, that the work was done. What he did know was that he would live to write no more of it. As Wood says, “Writing the word ‘end’ […] is not the same as finishing.” The mobile textual situation which, as we have seen, is characteristic of CSB applies to RTP also, at least in part, and equally raises questions of editorial principle. This is amply illustrated by the differences between the two Pléiade editions of 1954 (three volumes) and 1987 (four volumes). When the latter, overseen by Jean-Yves Tadié, was published, it drew a severely critical reaction from Roger Shattuck, which can now be found in his Proust’s Way: A Field Guide to In Search of Lost Time (2000).  Tadié and his team took the decision to print—in addition to a continuous reading text with the usual apparatus of introductions, variants and explanatory annotations—appendices giving multiple drafts (esquisses) of major passages. This meant, for example, that the reading text of Du coté de chez Swann occupied 416 pages, which required 171 pages of notes and variants. There were then 86 esquisses in smaller print, running to 360 pages; these had their own notes and variants in turn (a further 67 pages). Pléiade I, which printed notes and selected variants only, contented itself with six pages for the whole volume. That is, admittedly, the most extreme case, but the runners-up are À l’ombre des jeunes filles en fleurs (76 esquisses) and Le Temps retrouvé (71). The obvious value of the esquisses is their proof that much of what Proust decided not to publish is of equal literary merit with the novel we know.

Shattuck, who persists in following Pléiade I for his references, isn’t saying, of course, that the drafts have no interest or that we shouldn’t study them, but he attacked the “genetic” approach of Tadié on the grounds that it relegated the author’s conception of his work to a lower status than that of the textual scholar: “The genetic critics […] were able to do something the deconstructionists never succeeded in accomplishing. They unmade a work of literature.” Shattuck points out that Tadié’s use, in his editorial introduction, of the word oeuvre exploits its ambiguity in French: it can refer to a single work or to the entire corpus of a given author.  Essentially, Shattuck’s objection is that, whereas traditional source-criticism traces the evolution of a work from drafts to final version, involving processes of choice and selection, genetic criticism reverses the procedure and reduces the final text to a primeval chaos of possibilities; the oeuvre (the free-standing work) can only be understood in relation to the oeuvre (all its previous states, and all the author’s other works).

Whole books have been written on the genesis and development of RTP by people much better qualified to comment than I. But the textual transmission is far from providing, in all cases, the final version Shattuck desiderates. Naturally, the text of the volumes published in Proust’s lifetime—Du côté de chez Swann (1913), À l’ombre des jeunes filles en fleurs (1919), Le côté de Guermantes I (1920), Le côté de Guermantes II (1921), Sodome et Gomorrhe I (1921) and Sodome et Gomorrhe II (1922)—carry his final authority, and the drafts and variants for those volumes are clearly subordinate. But each of the three posthumously-published volumes, La Prisonnière (1923), Albertine disparue (La Fugitive) (1925) and Le Temps retrouvé (1927), suffered editorial interference from the outset, which necessitated strategic decisions by those working on the two Pléiade editions.  Specifically:

    1. The published text of La Prisonnière was prepared by Proust’s brother, Robert, and Jacques Rivière, who made cuts, rearrangements and revisions to the manuscript. Pléiade 1 based its text on the manuscript and on typescripts, resorting to the original edition only when unavoidable. Pléiade 2 sought to follow Proust’s last known revisions, which involved moving between different typescripts.
    2. Albertine disparue, based on a typescript, was seen through the press by Rivière, Gabriel Marcel, and Jean Paulhan, again with the advice of Robert Proust. The typescripts were no longer available to the editors of Pléiade 1, who therefore returned to Proust’s manuscript, which was in a chaotic state, making the best sense of it they could and relegating to the foot of the page numerous passages whose intended place in the text was unclear. Their version also ended earlier than the ending of the 1925 text, involving a different division between the last two volumes of RTP. In 1986 a third text appeared, based on a duplicate typescript which had been preserved in the National Library. This had Proust’s handwritten corrections, but he had made them with intended revisions to the last volume in mind, which he did not live to carry out. The Pléiade 2 editors were therefore faced with three versions of the text, none of which represented Proust’s final decisions. They followed the National Library version as far as it went, and the manuscript thereafter.
    3. Le Temps retrouvé was again edited by Paulhan and Robert Proust. The latter had a typescript made of his brother’s manuscript, but this left blanks where the handwriting was difficult or impossible to read; in addition, there were editorial cuts and rearrangements. Both Pléiade editions followed the manuscript, but the order of paragraphs—and even the opening of the volume—differs considerably, according to different judgements about where Proust intended his numerous insertions to be placed.

It can therefore be seen that a “genetic” approach to the latter parts of the novel is not only defensible but inevitable. The case is not as extreme as that of JS or CSB, where no one published text can claim to represent Proust’s final intentions, but we are still faced with a significant degree of instability.


Earlier, I compared Proust’s compositional methods to those of someone assembling a jigsaw. Still other comparisons spring to mind: a mosaic, perhaps, or a patchwork quilt.  Jean-Yves Tadié, in his monumental biography (1996), saw Proust as a chess player who pursues several strategies simultaneously, moving from one part of the text to another, introducing, amplifying, abridging or eliminating, keeping everything in play. Proust himself famously compared his method to the building of a cathedral, with a preliminary ground plan evolving over time as the structure grew and expanded. In the chapter of the published CSB called “Sainte-Beuve et Balzac,” he provides another analogy in noting Balzac’s habit of interweaving different time-schemes in his novels, which come to resemble a piece of ground which is built up from successive deposits of lava. The text of Proust’s major works is similarly geological, preserving fossils of earlier versions. Given his addiction to (literal) cutting and pasting, it is perhaps as well that Proust did not live to have access to the word-processor, or we might never have had a final text of anything.