In recent decades, researchers have developed enormous interest in the role of Denis Diderot in Guillaume-Thomas Raynal’s major historiographical work, the Histoire philosophique et politique du commerce et des établissements des Européens dans les deux Indes, known as the Histoire des deux Indes (hereafter HDI), which appeared in four different versions: HDI70 (1770; actually probably released in 1772), HDI74 (1774), HDI80 (1780) and HDI20 (1820), corresponding to their (official) publication dates.¹ This historiographical work is widely regarded as a foundational text of anti-colonialism and abolitionism, cited in an almost unmanageable number of monographs and essays. Scholars have especially emphasised the importance of Diderot’s role as a ghostwriter, claiming that his contributions radicalised the HDI (Wolpe 1956; Benot 1970; Strugnell 1973; Israel 2011, pp.413–25; Stenger 2013). According to these scholars, Diderot’s input was already notable in the first edition and grew even more substantial in the second, and especially the third, version of the HDI.

But did Diderot really write these passages? This question might seem bizarre today. These attributions are so firmly established that many scholars rely on them without even realising their uncertainty. The recent scholarly edition of the HDI includes a list of Diderot’s presumed contributions that is widely accepted (Raynal 2010–2023). A hundred years ago, the question of authorship was a mystery; but today, thanks to new sources discovered in the 1950s and 2010s, this problem is widely considered settled – at least, this is the impression one gets from reading studies by scholars of Raynal and Diderot (see below section 2).

This article presents the results of a study that began as a search for unknown contributions by Diderot to the HDI, but ultimately called into question Diderot’s authorship for the vast majority of the passages attributed to him. The original plan was to use stylometry – especially rolling stylometry – to identify as yet unknown passages by Diderot. When the stylometric analysis not only yielded no new fragments, but also raised doubts about the established attributions, we initially suspected methodological mistakes. We reviewed the corpus and the methods and found some aspects that needed improvement, but nothing that significantly altered the results. It was only when we became acquainted with the history of these attributions that we realised the problem lay neither with our corpus nor with our stylometric methods, but rather with the assumptions underlying previous scholarship.

This study demonstrates that scholarship must reject almost all previous attributions of HDI fragments to Diderot. It reaches this conclusion through two distinct methods: qualitative source criticism (the most classical method in the historical sciences) and stylometric analysis, an ensemble of quantitative and computational techniques that have recently been significantly refined. This study not only raises serious doubts about established attributions but also offers new hypotheses for further exploration of the HDI. Rather than a book that was created primarily by assembling fragments by Raynal and Diderot, the picture that emerges is of a collective work involving many contributors. The philosophical dimension of the HDI cannot be traced to a single author; instead, it was probably the product of a collective – Paul Thiry d’Holbach’s coterie – which was itself not without hierarchies. It is likely that Diderot played a strongly subordinate role in the genesis of the HDI: he probably wrote only a few passages of the HDI, spending much more time on language copyediting. By contrast, d’Holbach may have written a substantial share of the fragments typically attributed to Diderot.

The first section reviews the evidence for the collaboration of different authors on the HDI and critically revisits the history of attributions to Diderot. The result is that, even without any computational distant reading, there are strong reasons to distrust the common attributions to him. Rather, the available evidence suggests that, in quantitative terms, Diderot was probably a second- or third-rank contributor to the HDI. We then explain the stylometric methods used and present the corpus studied. The next sections summarise and discuss the results, while the conclusion formulates hypotheses for further investigations. These new hypotheses better reflect the findings from both source criticism and stylometry and should be tested with different methods.

There is no doubt that Diderot contributed to the HDI. The real question is whether only a few or many passages can be traced back to his pen. The earliest reference to Diderot’s contribution to the HDI appears in a letter from him to Friedrich Melchior Grimm, dated 15 August 1772. It shows that Raynal had asked Diderot to improve the text for a future edition, but Diderot had not yet read the book and was not eager to do the work. Nonetheless, at Raynal’s insistent request, in August and October of that year he wrote some fragments or at least some sentences for the forthcoming edition.² Furthermore, during the 1781 controversy that erupted surrounding the appearance of Raynal’s name and portrait in HDI80, Diderot wrote an unusual piece in which he admitted authorship of some – unspecified – parts of the work. This manuscript, known as the ‘Lettre apologétique de l’abbé Raynal’, was a reply to Grimm’s criticism of Raynal. It did not, it seems, circulate during Diderot’s lifetime, and was only discovered in the 1950s among the papers of Diderot’s son-in-law. In the letter, Diderot makes clear that Grimm’s critique of the HDI’s most provocative passages should also be taken as an attack on him, since he had contributed to them.³ At the same time, he made no secret of the fact that he generally held the HDI in low esteem: of its thousands of pages, he found only about fifty which, he said, drew inspiration from a zeal for virtue and hatred of vice. At no point in the ‘Lettre apologétique’ does Diderot suggest he wrote substantial parts of the HDI, at least not in quantitative terms. On the contrary, the letter implies that his contributions were among those fifty virtue-promoting pages (Diderot 1970, pp.213–14).

Jacques-Henri Meister, editor and co-author of the Correspondance littéraire, was the first to claim that Diderot ranked among the most important authors of the HDI. In a eulogy for Diderot – delivered in 1786, two years after his death – Meister invoked Diderot’s contributions to the HDI to praise his utter selflessness, courage and tireless zeal for the truth. In a somewhat vague formulation, Meister asserted that many ‘belles pages’ from Helvétius’ De l’esprit or d’Holbach’s Système de la nature could only have been penned by Diderot. He even claimed that Diderot had spent two years working almost exclusively on the HDI, so that almost a third of the work could be traced back to him.⁴ Meister initially ‘published’ this panegyric discourse in the Correspondance littéraire, a manuscript journal sent exclusively to foreign princes and princesses (primarily of the Holy Roman Empire) via diplomatic channels. The eulogy served to celebrate one of the most important authors and editors of the Correspondance littéraire - and thus, probably, also to preserve a spark of this glamour for the journal itself. At the same time, Meister had every reason to belittle Raynal’s merit. He was hostile to the abbé, whom he had accused in the Correspondance littéraire in 1781 of damaging the reputation of his patron, Jacques Necker, through his ill-considered behaviour – particularly by printing HDI80 under his own name and featuring his own portrait (Ohji 2012, pp.63–64; Topazio 1958, p.107). However, in the printed version of the eulogy (1788), Meister no longer dared to attribute such a leading role in the development of radical philosophy to Diderot. He left Diderot’s contributions to the HDI unmentioned, and in the following years his claim remained known only to a circle of insiders (Meister 1788).

This changed abruptly in 1791, when Raynal wrote a letter to the French National Assembly denouncing its infringements of the king’s authority and the creation of a schismatic state church. The reaction was tremendous. In the early years of the Revolution, the HDI enjoyed great prestige. It was considered one of the works that had most inspired the Revolution. With his letter, Raynal instantly turned – at least in the eyes of the Assembly’s supporters – from an intellectual pioneer into an opponent of the Revolution. A battle now raged over the cultural legacy and collective memory of the Enlightenment. In this context, polemicists began spreading the claim that Raynal was not the author of the philosophical chapters in the HDI. However, no one singled out Diderot as a major contributor to the HDI; his name was mentioned only alongside several others. The slanderous newspaper article with which Anacharsis Cloots opened the polemic against Raynal listed Jean de Pechméja, Jean-Baptiste-Léon Dubreuil, Diderot, Jacques-André Naigeon and d’Holbach as the authors of the HDI’s philosophical sections. Camille Desmoulins mentioned ‘Pechméja, Diderot, Dubreuil, Naigeon, d’Holbach’. Jacques-Pierre Brissot emphasised Pechman’s contribution. The anonymous pamphlet Raynal démasqué claimed that Diderot’s pen could be recognised ‘en vingt endroits de l’histoire philosophique’. An anonymous article about Raynal’s letter to the Assembly stated that Diderot had written almost all of the pages that made the HDI a success – but not that large parts of this work were attributable back to him.⁵

For these polemicists, Diderot, who had died in 1784, was just one author among others. In contrast, Diderot’s daughter, Marie-Angélique de Vandeul, and her husband understandably sought to support Meister’s thesis that Diderot had made an enormous contribution to the HDI. To make their point, they did not rely on Diderot’s manuscripts but on printed copies of HDI80, from which they selected certain passages and attributed them to him. Although in the sources known to us (see below), they drew exclusively on passages from editions of HDI80, many of the fragments they chose originally came from HDI70 (published 1772) or HDI74. If we accept these attributions, we must then conclude that Diderot played a significant role in all three versions of the text published during his lifetime.

Unfortunately, it is not possible to determine exactly when Diderot’s daughter and son-in-law began attributing passages of the HDI to him. It is, however, very unlikely that this occurred before the Revolution. In fact, the immediate reason the Vandeuls selected passages from the HDI was to prepare a projected edition of Diderot’s Œuvres complètes, in which extracts from HDI80 (included in volume 38) would appear alongside bold, still-unreleased writings by Diderot in other volumes – texts they could not have had printed before 1789 without considerable risk to themselves due to heavy censorship and effective policing (Dieckmann 1951, pp.92–95). The most likely scenario is that the Vandeuls drew on the HDI during or after the polemic against Raynal in order to portray Diderot as an intellectual father of the Revolution.

The Vandeuls produced at least three different, overlapping selections of passages from HDI80. Vandeul tore some pages from a printed edition of HDI80 and first marked passages with a pencil, which he then copied under the title ‘Pensées détachées’. He later interrupted this work and made a second selection, which differed significantly from the first. He called this second set ‘Mélanges’ and integrated it into the manuscripts of Diderot’s Œuvres complètes, which he clearly intended to have printed (Duchet 1978, pp.13–24). His wife Marie-Angélique took a similar approach, but probably with a different aim. She also picked up at least one printed copy of HDI80 that belonged to a friend and marked passages she wished to attribute to her father – though unlike her husband, she probably did not copy them. Her marking of passages was already known in early 19th-century bibliographic circles (Goggi 2018, p.247). However, it was only in the 21st century that a copy of the HDI was identified in which Diderot’s daughter had apparently marked passages allegedly penned by her father: the so-called ‘exemplaire Hornoy’, to which we will return. The selection contained therein shows considerable overlap with Vandeul’s two selections, but also notable deviations (Goggi 2018).

If the Vandeuls are to be believed, Diderot may have written nearly a third of the HDI – over 200 000 words out of approximately 635 000, as Meister also claimed.⁶ However, several men of letters who knew Diderot and Raynal well painted a very different picture of the HDI from that of either Meister or the Vandeuls. Diderot’s close friend Jacques-André Naigeon asserted that Diderot had taken no part in preparing the second edition (published in 1774), or at least not in the chapters on slavery and the slave trade – chapters that the Vandeuls ascribed to him. Rather, claimed Naigeon, the chapters critical of slavery were the work of Jean de Pechméja (Dieckmann 1951, p.93). During the Restoration, Charles de Pougens, a close acquaintance of Raynal and Pechméja, named Jacques Antoine Hippolyte de Guibert, Pechméja, Antoine-Léonard Thomas, Jean-Baptiste Dubucq (or Du Buc) ‘et bien d’autres encore’ as having played ‘la plus grande part’ in the HDI (Pougens 1826, p.146). He also claimed that Pechméja wrote the section discussing the slave trade (Pougens 1826, p.89). These witnesses did not suggest that Diderot stood out among the contributors to the HDI.

As a result of the polemics against Raynal and the desire to emphasise the contributions of different authors, there emerged a certain confusion surrounding the authorship of the HDI in the first half of the 19th century. The most prominent bibliographer of the period, Antoine-Alexandre Barbier, offered vague and contradictory information on the subject, shaped in part by his hostility towards Raynal and his work. Without citing his sources, he initially repeated rumours, without commenting on them, in a note on HDI70: ‘On assure que, pour la partie philosophique, Raynal a été aidé par Diderot, Pechméja et d’Holbach’ (Barbier 1852, p.45). He quoted Meister’s claim that Diderot had written a third of the HDI, without comment. At the same time, he asserted – again without naming his sources – that the person who had written the most for the HDI was a certain abbé Martin. More specifically, he wrote that Pechméja privately claimed authorship of the ‘pages éloquentes sur la traite des noirs’; a remark Barbier had probably heard from Naigeon or Pougens. He was explicit only in attributing Book 19 (the Tableau de l’Europe) of HDI74 to Alexandre Deleyre. It remains unclear to what extent Barbier accepted Meister’s claim with regard to Diderot. He was evidently unsure about the details. Still, Barbier consistently emphasised the role of authors other than Diderot. In his view, the HDI was not chiefly the product of collaboration between Raynal and Diderot, but rather an ‘ouvrage fait par tant de mains’ – a patchwork of fragments from numerous authors (Barbier 1852, p.45).

In the early 20th century, the literary historian Anatole Feugère offered a significantly different assessment in the first academic study on this topic. Like Barbier before him, Feugère emphasised the collective nature of the HDI. Without subjecting the sources to criticism or even quoting them precisely, he assembled all available information on authorship throughout the HDI and produced a list that reflected the sum of earlier claims: Thomas, Jean-François de Saint-Lambert, Guibert, the baron de Knyphausen, d’Holbach, Nicolas de Lagrange, Naigeon, a certain La Roque, the abbé Martin, Dubreuil, Pechméja, Valadier (a poet from Raynal’s hometown), the abbé Pestre, the abbé Bonnaterre, Deleyre and Diderot. According to Feugère’s account, the contributions of Pechméja, Deleyre and Diderot were particularly extensive (Feugère 1913). Relying on a letter of December 1772 from the abbé Ferdinando Galiani, Feugère showed that Pechméja worked intensively for Raynal. In this letter, Galiani referred to the young Pechméja as a ‘coadjuteur’ to Raynal, who had received 100 louis (i.e. 1150 livres) for his work, and hinted at a conflict between the two, since Raynal had had an affair with Pechméja’s wife (Galiani 1881, vol.1, p.403). Feugère claimed that Pechméja was primarily responsible for language revision; Deleyre, by contrast, had a ‘mission plus haute: il doit maintenir la pureté de la doctrine’ (Feugère 1913, p.348). Feugère offered no sources to support these assertions. His information on Diderot’s contribution to the HDI was equally unsubstantiated. Somewhat contradictorily, he claimed that Diderot had intervened twice – once for HDI74 and once for HDI80 – and then suggested that Diderot had already contributed to the first edition (p.350). He quoted a letter written by Diderot in 1765, in which the latter admitted having kept the abbé Raynal waiting six months for an answer about an ‘ouvrage politique’, which Feugère believed must refer to the HDI. Feugère supposed that Raynal later instructed Diderot not only to correct the HDI, but also ‘d’augmenter considérablement l’ouvrage, en y intercalant des réflexions et des dissertations’ with new fragments (pp.350–51). He uncritically accepted Meister’s claim that Diderot had written a third of the HDI, though he offered no new sources to support it.

Feugère’s greatest achievement remains the recognition that some of Diderot’s 1772 contributions to the Correspondance littéraire (known as ‘Fragments politiques’) were later incorporated, albeit in heavily revised form, into HDI74. This marks the first occasion – and, as we shall see, the last – on which Diderot’s contribution to the HDI was actually documented (Feugère 1913, pp.345–70). Nevertheless, it should be noted that alongside this important insight, Feugère’s essay also included a number of dubious and unsubstantiated claims.

In the 1950s, researchers undertook for the first time a source-critical examination of all the evidence concerning Diderot’s involvement in the HDI and reached conclusions that differed significantly from those of Feugère. Herbert Dieckmann presented the first scholarly description of the Fonds Vandeul, i.e. the papers left by Diderot’s son-in-law, Caroillon de Vandeul, which contain an important part of Diderot’s legacy. Dieckmann made two important discoveries: first, the manuscripts of Diderot’s contributions to the Correspondance littéraire (‘Fragments politiques’), which appeared in a heavily revised version in HDI74 (seven fragments, 14 folios, in an unknown handwriting); second, the selections of passages from the HDI that Vandeul had extracted from a printed edition of HDI80. Dieckmann proceeded with scholarly thoroughness and concluded that the selections made by Vandeul could not be equated with Diderot’s contributions to the HDI. In fact, Vandeul altered the selection over the course of his work, suggesting either that he had no special knowledge of Diderot’s authorship or that he deliberately spread falsehoods about it. Vandeul obviously had neither instructions nor manuscripts from Diderot when selecting parts of HDI80 to attribute to him. There was therefore no compelling reason to trust him, Dieckmann concluded (Dieckmann 1951, pp.92–155).

Virgil Topazio’s 1958 essay was another decisive contribution to the scholarly debate over Diderot’s role in the HDI. Topazio subjected the claims about Diderot’s involvement to critical scrutiny based on all available sources. He proved that the existing evidence strongly argues against Diderot’s involvement in HDI70. Only Topazio’s most compelling arguments will be mentioned here. The political work referred to by Feugère, which Diderot mentioned in 1765, was very likely not the HDI but Raynal’s Histoire des guerres, an unpublished work that the chevalier de Chastellux commented upon in a letter to Raynal on 3 September 1765 (Topazio 1958, p.104). Above all, Topazio analysed Diderot’s correspondence, showing that, according to his letters to Grimm between August and October 1772, Diderot had obviously not read the HDI, expressed very little interest in it, had still not written any fragments for Raynal and was very reluctant to do it (Diderot 1965, pp.101–102, 146–47, 149). Diderot’s correspondence thus speaks clearly against his collaboration before October 1772. This finding further reinforces the view that the selections made by Diderot’s daughter and son-in-law – many of which derive from the first edition of 1770 – cannot be regarded as reliable.

Moreover, Topazio explained that not only is there no evidence of Diderot’s extensive collaboration on HDI74, but also that his correspondence (including a letter from Diderot to Raynal) and Naigeon’s testimony make such intensive involvement unlikely (Topazio 1958, pp.105–107). We therefore have no reason to assume that Diderot wrote more for the second edition than the seven fragments that first appeared in the Correspondance littéraire (‘Fragments politiques’) and then in heavily revised form in the HDI. Furthermore, Topazio considered it unlikely that Diderot penned a large portion of HDI80. Here, too, he demonstrated that the available evidence argues against Diderot being one of the principal authors of the work. Topazio pointed to Diderot’s ‘Lettre apologétique de l’abbé Raynal,’ which suggests that the sum of Diderot’s contributions probably amount to fewer than fifty pages in HDI80. He showed that all other assumptions rested on dubious sources and unsound methods. As Topazio noted, Diderot may have worked primarily as a stylistic copyeditor for HDI74 and HDI80; while we have very little evidence of his authorship in the HDI, his correspondence shows that, under pressure from Raynal, he did perform such tasks – just as he famously did for d’Holbach (Topazio 1958, pp.103–16; Sciuto 2023).

Dieckmann and Topazio thus mounted a compelling challenge against the claims advanced by Meister and the Vandeuls. So how did scholars come to believe that Diderot wrote large parts of HDI70, HDI74 and HDI80? Hans Wolpe had taken the first steps in this direction in his monograph Raynal et sa machine de guerre (1956). In it, Wolpe scrutinised two sets of documents from the Fonds Vandeul: the printed pages of HDI80 marked in pencil and ink, and the collection of fragments titled ‘Pensées détachées’. He remained cautious about the former, emphasising that one could not be sure the compiler intended to document Diderot. Above all, Wolpe was alarmed that some of the selected passages originally came from HDI70 – an edition to which, in his view, Diderot had not contributed. In contrast, he mistook the ‘Pensées détachées’ for a collection of fragments that Diderot had sent to Raynal, believing them to derive from Diderot’s original manuscripts – even though they too contained many HDI70 passages. Wolpe did not realise that the ‘Pensées détachées’ themselves were derivatives of selections from printed fragments. Finally, Wolpe argued that the vigour and philosophical depth of these fragments signalled they came from the pen of a great writer – in his view, this meant that they must be by Diderot rather than Raynal (Wolpe 1956, pp.186–252).

While Topazio rightly dismissed Wolpe’s methods as untenable, it was Michèle Duchet who, from the 1960s onwards, equated the fragments of the Fonds Vandeul with Diderot’s contributions to the HDI. This work resulted in a list of Diderot’s alleged fragments, which Duchet eventually printed in her Diderot et l’Histoire des deux Indes (1978). As this monograph is the original source of the misattributions discussed in this article, it is worth examining Duchet’s argument closely. As we shall see, this study presents serious shortcomings, all the more astonishing given that it is based on an admirably detailed examination of the Fonds Vandeul, which has enabled better reconstruction of the stages of Vandeul’s work. For example, Duchet corrects Wolpe by demonstrating how Vandeul selected passages from a printed edition of HDI80 to integrate successively not only into the two collections of fragments described by Dieckmann (‘Pensées détachées’ and ‘Mélanges’), but also into manuscripts by Diderot such as ‘Observations sur le nakaz’.⁷ Logically, these meticulous observations should have led her to question Vandeul’s reliability.

But, strangely enough, Duchet chose the exact opposite interpretation. She considered the documents produced by Vandeul to be ‘documents de première main’, a phrase suggesting that they were Diderot’s own, even though Duchet was well aware that she was dealing with collections compiled by Vandeul. In her view, the passages from the HDI in question thus represent ‘fragments’ of Diderot that Raynal incorporated into his work. Without citing any supporting sources, she asserted that Vandeul probably followed instructions from the late Diderot that have since been lost (Duchet 1978, p.134; Dulac 2013, pp.118–19). In her account, the equation between Vandeul’s selections and Diderot’s œuvre takes on the status of an axiom to be defended at all costs. At no point does Duchet address the context in which Vandeul chose the passages from HDI80, or the interests that may have influenced him in this work. To attribute these fragments to Diderot, she relies partly on an argument from authority: if no one excludes the passages from the HDI added by Vandeul to ‘Observations sur le nakaz’, then we must also admit that Vandeul’s texts are authoritative, and so ‘Pensées détachées’ and ‘Mélanges’ (i.e. Vandeul’s two selections) must be incorporated into Diderot’s complete works (Duchet 1978, pp.45–48). This appeal to authority no longer holds today because, of course, researchers who have studied the informal notes we now call ‘Observations sur le nakaz’ reject Vandeul’s version (Dulac 2013, pp.107–51).

Duchet’s claim that the Fonds Vandeul presents ‘documents de première main’ raises several problems. First, Vandeul had made not one but two separate selections – thus contradicting himself. Even if his work were trustworthy, which of the two selections should one follow? It is striking that Duchet never asked this question and simply aggregated both of Vandeul’s sets. Second, Vandeul included passages that contemporaries attributed to authors other than Diderot. Pougens (1826) and Barbier (1852) suggest that Pechméja wrote the chapters on slavery and the slave trade in HDI70 and HDI74. Yet Vandeul’s selections include five fragments from those HDI70 chapters and one from HDI74. Of these, three were not altered for HDI74 or HDI80 (see Duchet 1978, pp.85–86). Vandeul’s choices therefore contradict Naigeon’s claim that Diderot wrote nothing on these subjects in the first two editions of the HDI. Likewise, Vandeul’s two selections include twenty-one fragments from Book 19 in HDI74 – fourteen of which survived intact in HDI80 (see Duchet 1978, pp.97–103). Since Barbier suggests that Deleyre wrote the 1774 version of that book, this decision further undermines Vandeul’s credibility.

Even more importantly, there is a third major problem with Duchet’s claim: Diderot’s letters show, as Topazio revealed, that he did not take part in drafting HDI70. Duchet addresses this only briefly – even though she suggests that the unaltered HDI70 fragments make up about half of Vandeul’s selections (Duchet 1978, p.51). But Duchet’s rebuttal of Topazio’s argument is surprisingly perfunctory. She argues that the proximity of certain HDI70 passages to the ‘Observations sur le nakaz’ proves Diderot’s authorship, even though the ‘Observations’ were written after Diderot had read HDI70 and even though the version she cites is the one into which Vandeul had already incorporated HDI80 passages, a fact she herself uncovered. Duchet also ignores the evidence in Diderot’s correspondence, which rules out any contribution to HDI70. Topazio had identified the 1765 political work for which Raynal asked Diderot’s help as his Histoire des guerres. Curiously enough, Duchet contends that a history of armed conflicts between states could not be described as a political work, so that this term must refer to HDI70. In a circular, apodictic move, she reverts to her fundamental premise – that the Fonds Vandeul as a whole is a first-hand document: arguing that, as fragments penned by Diderot would appear in HDI70, then this must be the political work of 1765.⁸ Finally, Duchet dismisses the value of the correspondence by contrasting ‘le témoignage des textes’ (Vandeul’s selections) with the ‘silence’ of the correspondence – a silence which, in her view, proves nothing. Yet the correspondence is not silent: it clearly shows that Diderot did not contribute to the HDI before October 1772, as Diderot himself states. Moreover, since no source supports a contribution by Diderot to HDI70, the burden of proof lies with Duchet, not Topazio.

Adding up the fragments from HDI70, Book 11 and Book 19 that, according to Naigeon, Pougens and Barbier, are not by Diderot, we arrive at 107 out of 271 whose attribution by Vandeul is problematic. Even setting aside Vandeul’s contradictions and his cavalier treatment of Diderot’s manuscripts – and even ignoring his personal interest in inflating his father-in-law’s political legacy – this should have sufficed to confirm Dieckmann and Topazio’s judgement: Vandeul’s selections cannot be regarded as a faithful record of Diderot’s contributions.

Regarding HDI74, Duchet’s observations are less problematic. She closely examines the seven fragments that Diderot sent to Grimm for the Correspondance littéraire (‘Fragments politiques’) and demonstrates how liberally Raynal adapted them. The passages published in HDI74 were substantially rewritten. However, this erudite analysis is followed by puzzling conclusions. For example, Duchet asserts that: ‘Dans l’état actuel des choses, il faut admettre que les “Pensées détachées” et les “Mélanges”, préparés pour l’impression par M. de Vandeul à partir de pages détachées de l’Histoire, sont une base solide pour l’étude des contributions de Diderot à l’ouvrage de Raynal’ (Duchet 1978, p.44). Yet nothing from her examination of the seven 1772 fragments incorporated by Raynal into HDI74 supports this claim. Equally problematic is her suggestion that Diderot was genuinely enthusiastic about contributing to the HDI in 1772 (Duchet 1978, pp.17–18). His correspondence proves the contrary, and Duchet provides no sources to support her interpretation.

Diderot et l’Histoire des deux Indes contains further examples of Duchet drawing conclusions that do not follow from the data she presents. For example, in showing that Vandeul collaborated with Raynal on the 1820 edition, she suggests that this discovery makes it difficult to doubt that ‘Pensées détachées’ and ‘Mélanges’ belong to Diderot’s œuvre, comparable, as she puts it, to the articles in the Encyclopédie (Duchet 1978, p.45). The logic of this claim is hard to discern. Moreover, from the general coherence of the HDI text and the difficulty of distinguishing Diderot’s passages within it, Duchet concludes that all these passages should, in their entirety, be considered works by Diderot (Duchet 1978, pp.55–56). A far more logical conclusion, of course, would have been that Raynal and his team skilfully merged contributions from multiple authors into a single text. But for Duchet, the fact that Raynal and his team rewrote Diderot’s supposed passages does not alter her view that these fragments are wholly by Diderot: she maintains that the distinction between first- and second-hand passages should not be overemphasised, since all ultimately amount to contributions by Diderot (Duchet 1978, p.63).

Of course, a great historian like Duchet was well aware that her method was problematic, yet she refused to draw the necessary conclusions. She admitted that, unlike the case with the articles in the Encyclopédie, there is no proof that Diderot is the author of the text of the HDI, but chose to disregard this fact because, in her opinion, ‘il est impossible de ne pas reconnaître les thèmes majeurs de son œuvre entre 1770 et 1780’ (Duchet 1978, pp.45–47). She maintained that the selections made by Vandeul from a printed version of HDI80 should be published as isolated fragments ‘pour voir surgir un Diderot inédit’ (Duchet 1978, pp.62–63). This was the goal relentlessly pursued throughout – the driving force behind the entire study. To reach it, as she herself admitted, it was necessary not to give undue weight to the ‘les réserves qui s’imposent’ – that is, to scholarly caution – but instead to deploy ‘toute l’audace nécessaire […], puisqu’il s’agit de rendre à Diderot quelque 270 morceaux’ (Duchet 1978, p.48). This wording is telling: it reveals how, in this case, the cult of the great man and the desire to find unknown texts by Diderot overrode caution and scientific rigour.

Duchet’s work is therefore a surprising mixture of admirably erudite analyses and illogical conclusions. Astonishingly, this remains the most recent discussion of the selections made by Vandeul. Whether out of deference to Duchet, whose monumental Anthropologie et histoire au siècle des Lumières (1971) is, justifiably, widely admired – or out of fascination with the brilliant Diderot, the circle of HDI specialists emerging in the 1970s uncritically adopted Duchet’s axioms. Thus, when Anthony Strugnell published a now classic work on Diderot’s late political thought, he relied on Duchet’s attributions without expressing the slightest doubt (Strugnell 1973, p.vii). In France, Yves Benot also fully accepted the new attributions (Benot 1970). Strugnell and Benot believed that Diderot was the true precursor of the French Revolution, and this thesis increasingly captivated subsequent scholarship. From one study to the next, the attribution of Vandeul’s selected passages to Diderot was repeated so often that it eventually came to seem self-evident. Diderot came to be viewed as a proto-revolutionary hero – staunchly democratic, anti-colonialist and abolitionist.

It was against this intellectual backdrop that a copy of HDI80, known as the Hornoy copy (‘exemplaire Hornoy’), was discovered in the 2010s, bearing marks claimed in a manuscript note to be by Madame de Vandeul. Without any real discussion, it was immediately regarded as another faithful documentation of Diderot’s contributions to the HDI. Gianluigi Goggi, who had already developed great erudition in analysing Diderot’s works in the 1970s, undertook a study of these passages, which enriched the list of fragments attributed to Diderot (Goggi 2018). Just as Duchet had incorporated two divergent sets of passages selected by Vandeul, Goggi now added those selected by Madame de Vandeul, without questioning whether any one of these three sets should carry greater authority. Here again, the context in which Diderot’s daughter undertook this work – and her probable motivations – was not taken into account. Finally, on the basis of thematic similarities between passages in the HDI and works by Diderot, Goggi identified other fragments that he thought were probably by Diderot. The result of this effort – based on questionable premises and yet showing great erudition – is the list of Diderot’s contributions at the end of each volume of the critical edition of HDI80 (Raynal 2010–2023).

What conclusions can we draw from this detailed overview of the attribution history and the evidence on Diderot’s contributions to HDI70, HDI74 and HDI80? There are no fewer than seven reasons why we should not trust the selection of fragments by the Vandeuls as evidence of Diderot’s contributions to the HDI and thus reject the lists given in the recent critical edition of the HDI.

While it is certain that Diderot contributed to HDI74 by submitting a few fragments to Raynal – fragments that were significantly rewritten by Raynal or one of his team to fit the HDI – one conclusion becomes inescapable: unless one places blind faith in Meister’s account and Diderot’s children, there is no reason to believe that the editor of the Encyclopédie was a major contributor to the HDI. His principal function may have been similar to the role he played for d’Holbach: that of a language copyeditor. It seems far more reasonable to trust Diderot’s own account rather than the versions circulated after his death by those who had been close to him and who had every incentive to aggrandise him. Even without invoking computational quantitative methods, we must side with Dieckmann and Topazio and assume that – barring the unlikely discovery of new manuscripts – Diderot wrote nothing for HDI70 and probably only a handful of fragments for HDI74 and perhaps for HDI80.

In addition to qualitative source criticism, a second method may be applied to test the reliability of the attributions to Diderot: stylometry. In modern computational humanities, stylometry refers to a study field that uses statistical methods applied to analyse writing styles, particularly employing quantitative data on linguistic features (e.g. word frequencies) to identify unique stylistic ‘fingerprints’ of authors. Studies in this field have demonstrated that authorial individuality is often found not in meaningful words (such as nouns and verbs) and concepts, but in function and frequently used terms – words used universally within a language and often unconsciously, unlike content words (Pennebaker 2011; Kestemont 2014). Over the past two decades, computational authorship attribution methods have shown that combining function and frequent words with appropriate statistical analysis can accurately determine the authorship of human-written texts (Grieve 2007; Koppel et al. 2009; Stamatatos 2009; Schöch et al. 2023). In literary history, computational stylometry helped to attribute works by major historical authors such as Ovid (Nagy 2023), Shakespeare (Craig and Kinney 2009; Plecháč 2021) and Molière (Cafiero and Camps 2019), while also resolving modern authorship cases like those of Elena Ferrante (Tuzzi and Cortelazzo 2018) and Robert Galbraith (Juola 2015).

In the following, we will analyse the fragments attributed to Diderot on the basis of Vandeul’s selections of passages from HDI80, thereby testing the soundness of common attributions to Diderot. Where Diderot is not identified as a probable author, we will present preliminary results that suggest authors who are closer to the HDI80 than Diderot, though actual attribution to these authors will require further examination not conducted in this paper.

This study aims to contribute to computational authorship attribution methods for 18th-century non-fictional texts in French. This is an ambitious goal for two reasons. First, French political writings have not been extensively studied in computational stylometry, unlike 18th-century French fiction and drama (see Schöch and Riddell 2014; Schöch 2017; Cafiero and Camps 2019; Gabay 2021; Röttgermann 2024). Non-fictional, ‘impersonal’ texts are likely to employ linguistic features that differ from those found in literary genres that rely, for example, on first-person narration and dialogue; these differences can significantly affect verb forms, pronoun use and overall textual style. Second, there is no well-established methodology for analysing non-fiction historical texts that feature text reuse, intertextuality and multiple authorship. The object of the study, fragments from the HDI, require solutions to these challenges, and this study provides preliminary results on how standard stylometric methods perform with such data. We expect our results to be refined in future research.

Through both qualitative source criticism and stylometric analysis, this article has demonstrated that the prevailing method of attributing passages in the HDI to Diderot is fundamentally flawed. This leads to several conclusions and proposals.

An exploration of the HDI’s authorship shows that intellectual history has often placed too much emphasis on individual geniuses. It is hard to resist the temptation to find more texts by a star author, or to interpret them as expressions of exceptional convictions and unique visions. Yet, as the sociology of knowledge demonstrated already a century ago, thinking is better understood as an incremental and collective enterprise shaped by organisations and asymmetrical interdependencies (Mannheim 1936). Searching for originality may introduce important biases in our narratives of Enlightenment history.

Martynenko A., Csapó J., Dostert C. and Tricoire D. 2025. ‘Supplementary code and data for the paper ‘Ghostwriting and collective authorship in the Enlightenment: Denis Diderot, d’Holbach’s coterie, and the problem of authorship in Raynal’s Histoire des deux Indes’ https://zenodo.org/records/17227546. Referred to throughout as ‘Supplementary materials’.