Venice and Paris at the dawn of the Enlightenment; or Rosalba in Paris

A new book by Valentine Toutain-Quittelier, Le Carnaval, la Fortune et la Folie, with a subtitle which translates roughly as the first part of the title of this post, has just arrived. It represents the fruit of many years’ work on this theme (the author’s doctorate was awarded in 2011, and several published articles present material revisited here). And it comes equipped with a preface from Pierre Rosenberg who knows more than anyone about the artistic relations between Paris and Venice. Thus for so many reasons the volume is to be welcomed. Much of it will delight and inform, and I shan’t attempt to summarise the book since you would be far better advised to read it for yourself.

At the centre of the work however is the theme in my subtitle: the seminal visit which Rosalba Carriera made to Paris in 1720–21, and the records (notably in her diaries) which provide crucial evidence of the artistic milieu of the time. Indeed her role could not be better described than by Louis Réau, who called her “le trait d’union entre l’art ascendant de la France et l’art déclinant de l’Italie.” The Paris journal in particular has been studied many times since its publication in 1793, notably by Alfred Sensier in 1865 (with notes expanding a text of a few thousand words into nearly 600 pages) and again in the critical edition of the artist’s writings published by Bernardina Sani in 1985 (not to mention the two editions of her catalogue raisonné – although Sani didn’t include lost “œuvres mentionnées” in the catalogues). Nevertheless mysteries remain – not least arising from Rosalba’s bizarre, but phonetic, spellings of proper names which requires a knowledge of Venetian orthography and orthoepy to disentangle. Valérie Toutain-Quittelier (“TQ” in what follows) brings exactly the right linguistic skills to this (to take an example, Rosalba writes “di Tre” for “d’Estrées” – although as this is preceded by maréchale, the problem is not so difficult as some others).

Sorting out these confusions is worthwhile, as I have tried to do in my annotated English translation of the journal (the current version is available here – I am always grateful for additions or corrections). And it is illuminating to do this for all her contacts, not just the sitters in her portraits that she lists. A couple of examples, not in TQ: Sani and Sensier leave us to understand that the Rollands belonged to the magistracy; but proper analysis shows that they were a different family, of bankers and agents de change. A similar case is the “Dervest” family who appear repeatedly in the diaries: Sani makes no attempt to identify them. Sensier did however connect them to Du Revest, contrôleur of Law’s bank – his name appeared on the bills. (Although TQ has a chapter on the bank I can find no reference to the contrôleur.) But the precise genealogy in fact reveals that he was Scipion de Vétéris du Revest and his wife, Mitilde Priuli, of a noble Venetian family. No wonder Rosalba was so keen to talk to her in a language she understood.

The modes of Venetian–Parisian connection were not merely artistic. But even within the arts the connections were not merely visual.  Anyone familiar with Watteau’s art is aware of the importance of the Comédie-Italienne, reintroduced to Paris in 1716 by Luigi Riccoboni. Yet neither Riccoboni nor his wife’s family, the Ballettis, are even mentioned as far as I could see – although they are connected in so many ways to the book’s theme. It is not widely known that Rosalba did a miniature of Riccoboni (it belonged to his daughter-in-law in 1773). Riccoboni’s niece Manon would in a later generation take us into the worlds of Casanova and Nattier, and would marry the great architect Blondel; La Tour would exhibit a portrait of Zanetta Balletti in the salon of 1751. At the period where TQ’s book is focused, Riccoboni’s sister-in-law, Margherita Balletti, had written to Rosalba seeking advice on painting in miniature, while her husband, the celebrated composer Giovanni Bononcini, met Rosalba several times in Paris in September 1720; she records going with him to see Law.

It is only by working through a similar level of detail in TQ that questions surface. I’m afraid most of the rest of this post is for specialists – or those who consult the Dictionary in future and wonder why I haven’t followed TQ. Some of the puzzles have more than one solution, and a discussion of recent literature would help identify these even when the author’s proposal is better than those already offered. It can sometimes be hard to tell when the information presented is already known and accepted, or new; and if new, whether it is certain, probable, possible or (as I suspect in a few places) wrong. Unfortunately (despite copious notes) this is not assisted by the often inadequate references that make it hard to follow which picture TQ is discussing: one of my criticisms of Sani was her decision not to include details of auction sales etc., but TQ routinely omits museum inventory numbers, catalogue raisonné references and dimensions (the discussion about Rosalba’s size system on p. 174 is confused by misplaced endnote indicators). Of one pastel she discloses merely that it “appeared once on the art market”. A concordance with Sani would help: at least for the pastels she discusses you can now find the equivalence by searching “Toutain-Quittelier 2017b, fig. x” in the online Dictionary where I give both the Sani reference (if there is one) and my J number (these unique, Googleable identifiers by far the easiest and shortest way to cite the dimensions, location, provenance, exhibition history and literature of any pastel).

But within the visual arts the book explores widely, including less well known figures such as Nicolas Vleughels who had travelled to Venice in 1707. It is perhaps worth emphasizing that his use of pastel (as studies for his paintings) predated Rosalba’s Paris trip by some years: a case in point is the study of a female arm (J.771.127 in the Dictionary, Fig. 201 in TQ; left below): Hercenberg, in his 1975 monograph on the artist, no. 310, regarded it as a study for Loth’s daughter in the 1718 painting Loth et ses filles, 1718; TQ instead captions it as a study for the arm of Campaspe’s servant in the 1716 Louvre painting of Apelle peignant Campaspe, 1716 (detail, middle below). To my eye however the correspondence is much closer with the servant on the far left of another painting, L’Enlèvement d’Hélène, c.1710–12, which is actually reproduced in TQ (fig. 183: detail, right below):

Vleughels also made a pastel copy of a lost work, quite possibly by Rosalba Carriera. One puzzle concerns another copy of it by an anonymous hand (J.9.6063): to this TQ offers an intelligent suggestion, with the name of Madeleine Basseporte. I think it’s an interesting idea, although I would hesitate about making an attribution to an artist whose accepted œuvre consists of a single work unless the technique were absolutely similar (it is not: the treatment of the shadows is especially telling; detail from Basseporte in Rijksmuseum, J.1304.11, left; from J.9.6063 right below):

TQ does not mention two further copies which can be found in the Dictionary, nor does she discuss the relationship with work by Boucher suggested by Alastair Laing, apparently later than the period to which Basseporte’s pastels belong (see Nathalie Strasser’s catalogue of the Collection Jean Bonna, Dessins français du xvi^e au xviii^e siècle, 2016, which is not referenced). But TQ’s argument is given less authority by imprecision: the Rijksmuseum pastel is captioned “vers 1730” (p. 291), but “vers 1727” (p. 289): there is no mention of the fact that it is actually signed and dated on an old label “Peinte par Madeleine Basseporte 1727”. You have to believe the label (there is no other reason to assign the work to Basseporte). But it’s not at all obvious why TQ infers that the portrait is a self-portrait. There was a self-portrait (unlocated) in her posthumous inventory, to which TQ makes no reference, although it also includes two copies expressly after Rosalba “avec mains” – evidence which surely supports TQ’s proposed attribution more firmly than the somewhat hackneyed remark “Elle peignoit le pastel et fut bientôt connue par des portraits qu’on mit à côté de ceux de Rosalba” from the 1780 obituary I also cite. (These two pastels are surely the ones by “Mlle Belleporte” in the Mesnard de Clesle sale, 1804, evidently misreading their labels; they again are not mentioned by TQ. However neither their dimensions nor aspect ratio fit J.9.6063, which is significantly longer also than the Rijksmuseum ratio.) The Rijksmuseum pastel (of which incidentally there is a second version, J.1304.112, once attributed to Rosalba) is not described as a self-portrait in any reliable source I know (Ann Sutherland Harris’s uncontentious remark in 1976 that “it seems possible that it is a self-portrait” has conspicuously not been taken up): if TQ makes the claim on objective evidence, it would have been helpful to cite this; if merely because it “looks like a self-portrait”, we can retain an open mind.

There are other examples where TQ may well have a good point but has not always presented the best evidence with the clearest argument to support it. There is a temptation to employ Holmesian deduction (when you eliminate the impossible, whatever is left, however improbable, etc.): but this is rarely a safe approach in art history, where the mysteries and confusions are not of the closed-box, detective story character, which is all the more reason why we should work together to pool thinking on these questions. The logic may not be mathematical, but common sense can help.

Another case is illustrative: the pastel last seen in 1926 when it belonged to the duc de Doudeauville and was described simply as of the duchesse de Chevreuse (J.21.05021; fig. 129) is named in my Dictionary as probably Louise-Léontine-Jacqueline de Bourbon-Soissons-Neuchâtel (1696–1721), as the right duchesse de Chevreuse of that generation. TQ, examining the 1926 owner’s genealogy, finds it impossible to see any connection between the duc de Doudeauville and this duchesse de Chevreuse, and so goes on to propose a different identity – that of the maréchale d’Estrées whose name is so garbled in the Diari. But seven generations of the owner’s pedigree would provide 64 ladies of the highest nobility to choose from, a warning in itself about the probabilities. However it is TQ’s premise that is wrong: the owner in 1926, Armand de La Rochefoucauld, 5^e duc de Doudeauville (1870–1963) (not his son, Sosthène, as TQ states) was connected with the ducs de Chevreuse: he was in fact the great-great-great-great-grandson of the very same duchesse de Chevreuse (his father’s father’s mother was Pauline-Hortense d’Albert de Luynes, and you can make the connection by consulting just two files in my iconographic genealogies, Albert de Luynes and Montmorency).

TQ does cite my work in connection with my reidentification of the Charolais/Clermont sisters (J.21.0382 and J.21.0411 had been switched).

The princess in the Chantilly pastel wears a white muslin dress; her sister is in brown. Curiously TQ mentions twice, on p. 150 with its note 16 on p. 186 and on p. 165 with note 70 on p. 188 (do these repetitions reflect the genesis of the book as a thesis?), a phrase in Rosalba’s diary, referring to Mlle de Charolais, “vestita di ganzo d’argento”. If you think silver might mean white this would indeed reinforce the identification. But the entry is for 10 March 1721, after the pastel was finished, and so doesn’t imply that this was the same costume as in the pastel. Moreover the words immediately following “d’argento”, “con gli ornamenti di fiori da Vicenza” (omitted in the first discussion, although given in the note to the second discussion) surely confirm that this was a heavily decorated court dress of the kind the princesse would have worn at that time, rather than the diaphanous, quasi-allegorical confection in the pastel. The point is worth considering at least as it suggests that the princesses may have worn quite different costumes from those shown in their portraits. Dresses in this type of fabric abound in contemporary portraits, for example by Largillierre: that Rosalba substituted something simpler (perhaps using the 2½ ells of “Mussolino” she records asking her sister to buy on 21 December 1720?) tells perhaps something more about her working methods (and pressure of work), as the patience required to depict such woven patterns is vastly greater that the broad sweeps with which Rosalba often enveloped her women.

There’s a short paragraph on p. 167 which raises another issue. In it TQ considers two of the names of English (or more precisely British) sitters. She notes that the Duchess of Richmond commissioned her portrait in miniature on 1 October 1720, and wonders what happened to that of the Duke; an endnote describes his expression in a Kneller portrait of him in the NPG. I think a far longer discussion would be required to deal with the issues this cryptic entry throws up: to take just one, might not Rosalba (who had the greatest difficulty with names of any kind, let alone the titles of English aristocracy) have confused the Duke of Richmond with his son, recte the Earl of March, if the Duchess (whom TQ correctly identifies as born Anne Brudenell) were travelling with Lord March (for it was the son who was on the grand tour at that date, and of whom Rosalba made several further images not mentioned by TQ, perhaps from the miniature she made in Paris then)? But there is an even simpler explanation when one examines the manuscript: what appears to be a second entry for “D di Richend” looks as though it is just the marginal summary referring to the portrait of the duchess mentioned in the main text rather than a second portrait.

In the same paragraph, TQ addresses the entry for “Molgneux” (commenced 7 October 1720, finished and paid for within the week), whom she identifies as Richard, 5^th Viscount Molyneux. This apparently is on the strength of a passport for a three-week journey issued to the “vicomtesse de Molineux” with her femme de chambre – on 7 January 1721. The identification sounds reasonable (and would have been reinforced, as also would the logic of combining this discussion within the same paragraph as the Richmonds, had TQ revealed that this Mary Brudenell was Anne’s sister) – except that it is far from certain. If Lady Molyneux was travelling with her maid, that surely suggests she was not travelling with her husband, and so offers no evidence that he was in Paris some months earlier. In fact the reference is at least as likely to refer to her mother-in-law, the young and recently widowed dowager Viscountess Molyneux (who continued to use that title until her death, despite remarriage). But more to the point, TQ makes no mention of the Dictionary identification (first suggested by Francis Russell in his 1989 Burlington Magazine review of Sani and found also in Ingamells), with a (at most distantly related) Pooley Molyneux (1696–1772) who was in Padua in March 1721 and probably passed through Paris at exactly the right time. Until the portrait is located such questions cannot be definitively resolved (unlikely: litigation about Pooley’s will reached the House of Lords half a century after it was written): but a modern in-depth study needs to refer to the identification previously published in three serious sources, if only to present reasons why they might be wrong.

A rather different problem arises with the lost self-portrait (J.21.0101) known from an engraving by Lépicié (left, below). Sani 2007 reproduces a weak copy (J.21.0104) as the original; TQ correctly ignores that, and analyses the Crozat correspondence to show that the engraving was made at Crozat’s behest in 1736, from a self-portrait in his collection (incidentally not one which found its way to the Hermitage). But TQ then goes on to argue that this pastel is the problematic work in the Bowes Museum (J.21.0107; below, right) which is related to the Uffizi portrait of the artist with a drawing of her sister (J.21.0106); she argues that the Bowes pastel is the self-portrait which she believes belonged successively to Crozat (my J.21.0101), the Erzbischof von Köln and Jullienne (my J.21.0108) and then Mariette (I think she refers to J.21.0117, the second part of lot 7 in the Mariette sale: this however was sold to Lempereur, not Paillet, who bought the other part of lot 7; further it is a small head which has nothing to do with the Bowes pastel; Saint-Aubin’s sketch will be found in the Dictionary). Leaving aside the conflations of what I think are different works (and which would require proper detail, such as lot numbers, to follow what TQ is suggesting: the only note refers to Isabelle Tillerot’s thesis where the Washington Allégorie de Peinture, J.21.2, is invoked), I’m afraid I don’t follow this at all. Engravers don’t make the changes of orientation TQ requires for there to be any connection between the Lépicié engraving and the Bowes pastel (and if they were following the portrait of Rosalba’s sister for the three-quarters angle, could this be regarded as a self-portrait?). But the discussion also ignores the obvious questions about the status of the Bowes pastel, for which no provenance is established before the late nineteenth century: it is an almost exact reversal of the Uffizi pastel, suggesting that it was a pastiche derived from an unknown engraving (with some minor adjustments, such as the implausible braided hair at the back of the sister’s head which the pasticheur has had to invent): this however renders the conspicuously right handed artist left handed, and also uniquely among Rosalba’s œuvre (setting aside the half-dozen colour reproductions in Sani which were printed back to front) has lighting from the right. There seems to be another confusion in these records: the Uffizi pastel is dated 1709, when the artist was 36 (if anything the subject of the Lépicié engraving is younger), while the archbishop’s pastel was “dans un âge avancé”. The Bowes pastel won’t illuminate the source of the Lépicié engraving.

Mariette refers to “une teste d’une brune qui revient ou qui va au bal” (my J.21.026, described in Mariette’s sale as “Une joli Vénitienne, ayant sur la tête un petit chapeau où sont attachées des fleurs, & tenant de la main droite un masque noir”), and TQ illustrates (fig. 121: below, left), as by Rosalba, the pastel I reproduce as J.21.0254, which again I consider to be a weak, probably non-autograph copy of a pastel of which at least three other versions exist (one is implausibly described as of La Barbarina, which is why they are gathered under that headline in the Dictionary; but the J numbers will take you there). TQ does not mention the other versions, and while she cites the Mariette sale catalogue she doesn’t illustrate or seem to mention the fact that it is sketched by Saint-Aubin in the 1777 sale catalogue (apparently with a little more space around the work, reinforcing my belief that J.21.0254 is a copy of the original: below, right).

As we all know Rosalba studies are dogged by the plethora of copies and pastiches. Sani’s approach is to omit them altogether (except by mistake). I take the opposite view, regarding it as within the scope of the Dictionary to identify works I consider to be copies by other hands, even of later periods (with indicators of my opinion of their status). Much of this classification can only be done by eye. TQ has a curious discussion (on p. 172, inexplicably separated from the discussion of the Dresden version, p. 153) of the terribly bad copy of Louis XV in the Forsyth Wickes collection at Boston (J.21.0702). According to the Dictionary, this is simple a copy (i.e. not autograph), while noting that Sani includes it as autograph. As far as I am aware there is no provenance before Paul Cailleux sold it to Forsyth Wickes in 1937. TQ makes no reference to my description, and seems tentative in her classification, describing it in the caption (fig. 123) as “Rosalba Carriera et atelier”, while in the text hesitates between blaming Rosalba’s sister or subsequent restoration but without considering the possibility of a non-workshop copy. Incidentally, regarding the real pastel of Louis XV, TQ simply follows Sani when she mentions (p. 172) “un certain ‘abbé Peroz’ s’engage avec enthousiasme à payer la bordure” for the king’s portrait: he was in fact (as you will find from my note in my annotated translation of the Paris journal) abbé Robert Perot (1661–1742), lecteur et garde de la bibliothèque du cabinet du roi – a position which suddenly makes sense of his enthusiasm, for he presided over where the pastel was to be hung.

On the front cover of the second edition of Sani (2007) was the pastel of an unknown man (J.21.0433) which had once been tentatively identified as of Pierre Crozat. In an earlier article (2007), TQ justly questioned the basis of the identification, pointing out that “cet exemple plus mercantile que scientifique montre qu’une fois encore, le mythe a pris le pas sur la raison.” But the hunger for identifications is driven by many motives. Unaccountably Sani (also in 2007) proposed that it was of the prince de Conti – despite the fact that he wore no riband for the Saint-Esprit, which is pretty well inconceivable for this type of representation. TQ correctly rejects this (as I had done), noting this as a salutary example of “les limites de l’analyse morphologique”. One could wish that she had stuck to this principle elsewhere.

To take one example, on p. 181, fig. 130, she reproduces a pastel (J.21.0601; below, left) as of Louis-Henri de La Tour d’Auvergne, comte d’Évreux, citing it as dated 1720 (although that I think is a deduction from her identification), and offering as evidence the similarity with a Rigaud portrait “présumé” of Évreux in the Metropolitan Museum of Art. To me they don’t look sufficiently similar to justify the connection. Further the Met portrait is dismissed by Ariane James-Sarrazin (it is no. P.1442 in her Rigaud catalogue) as not of Évreux (James-Sarazin simply doesn’t agree that he looks like him). TQ states that the Met portrait was engraved by “Johann-Georg Friedrich Schmidt” [sic]: this presumably is the engraving (FD 2155) by the father, Georg Friedrich, which is after a different, 1705 portrait by Rigaud (AJS P.917). But while one can dispute such niceties as whether the man in the Met portrait has the right baton for a maréchal (he doesn’t), the question could simply have been resolved by reference to the Dictionary: for, as I noticed a few years ago, Rosalba made a preparatory drawing (detail, below, right) of the sitter in J.21.0601 , including not only the exact composition but costume details down to the unusual braid on his shoulder. In that drawing he is identified as Henry, Lord Cornbury, later Baron Hyde, shown in a double study with Edward Walpole made when they were both on the Grand Tour in 1730/31. There is certainly room to argue that Rosalba used the same composition for two different sitters (that’s why I don’t conflate J.21.0601 fully with the diary entry for the pastel version of Hyde, J.21.0599), but even such a substitution would have happened at the same time, and there is every reason to believe that this portrait was not made in Paris at all, but in Venice ten years later.

The Walpole family takes me to positively the last case I shall discuss in this post: the proposed identification of a splendid pastel in a private collection as the lost portrait of John Law. (I note in passing that the identification of the Louvre’s girl with a monkey, J.21.0575, which TQ considered to be of John Law’s daughter in 2007 is now updated to Mlle Languet de Gergy, later marquise d’Havrincourt; however the girl was born in Regensburg on 6 June 1717, and when the pastel was done she was nearly 8, not 2, as TQ thought.) This is discussed at great length in the book, in much the same terms as in an earlier article which allowed me to draw the author’s attention to my objection back in June: perhaps the book went to print before it could be changed, but in any case the evidence has been visible in the online Dictionary for some years, and it is regrettable that this and the other points raised here were not at least discussed. (I am happy to respond to emails if the discussion in the Dictionary is too compact, and I am always happy to correct it when, as all too often, I’m wrong.)

There are in short three related works. The larger pastel (J.21.0863: detail below, left), in a private collection, has a smaller version, in Dresden (J.21.0867). When a corresponding drawing in the Biblioteca Marciana (fig. 133 in TQ) was discovered some years ago, the identity of this man was resolved to everyone’s satisfaction: he was the Paduan mathematician, marchese Giovanni Poleni (1683–1761), and the drawing was unsurprisingly in the Poleni family papers. Despite this, TQ has devised a theory which appears to rest on the fact that the lost portrait of John Law is known to have had a smaller version, and this is the only example where the Dresden version corresponds to a larger pastel…so it must be of Law. Thus the Dresden pastel appears as fig. 124, “Portrait de John Law”; the larger pastel as fig. 132 (once again the discussion is divided between two places for no obvious reason), also as “Portrait de John Law”, while the Marciana drawing, evidently of the same man, is fig. 133, “Portrait présumé de Giovanni Poleni”. To make TQ’s position even more surprising, she reproduces Schmidt’s print (detail below, right) after the lost Rigaud portrait of Law, which is not disputed (and broadly corresponds to the Schenk print and other Law iconography, all of which show his aquiline nose and fleshier jowls), and merely notes that “la confrontation morphologique entre le pastel et la gravure pose problème.”

Indeed it does: to my eye they cannot possibly be of the same man. But such subjectivities aside, there is a further objection. The portrait of Law was acquired by Horace Walpole, and though it was subsequently lost, it has left a sufficient trace in the watercolour by Thomas Sandby, Paul Sandby and Edward Edwards of the gallery at Strawberry Hill, in 1781, where we can blow up a detail visible in the niche to the left of chimney (right).

Hardly a high resolution reproduction (the perspective has introduced some distortion), but sufficient surely to dispel any idea that the Poleni pastel is the lost Rosalba of John Law.