Making the Cut: Has a Lost Caravaggio Been Discovered in Toulouse?

The art world was split, a few days back, when it was announced that an unknown Caravaggio painting had been discovered in an attic near Toulouse, France (above).  The theme, Judith Beheading Holofernes, is the subject of one of Caravaggio’s most compelling works (below).  A second version would be a major addition to his oeuvre.  Skeptics believe the newly discovered panel to be a fake—perhaps of recent vintage or, more likely, a 17th century work by one of Caravaggio’s many followers.  At best, they say, it is a second-rate copy of a lost Caravaggio.  Some go further, saying the work lacks the drama we would expect from a Caravaggio composition, whether an original or a copy.  True believers are describing it it as the kind of discovery that occurs just once in a lifetime.



The stakes are high.  Caravaggio has displaced Michelangelo on some lists of the most recognized and admired old masters.  I was in Tokyo when I read about the new discovery, and I'd been to a Caravaggio exhibition there, packed with locals eager to have an up-close encounter with original works.  One of the big draws of the show was Caravaggio's Mary Magdalene in Ecstasy, which was discovered in a private collection just two years ago (below).  A new Judith painting means new blockbuster exhibitions and untold ticket sales at the museum lucky enough to own it.  Of course, ownership will come at a price.  The painting has been given an estimated value in excess of 130 million dollars.  Many would like to see the French government purchase the work from those who found it, as it is currently barred from travel and undergoing testing at the Louvre.  Could this really be a lost masterpiece?  I will side with the True Believers, though there are some features of the work that give me pause.  I’ll also ask why and whether it really matters who painted it.

In establishing the authenticity of a work, nothing is more important than provenance.  Here, things are sketchy, but there a little bits of evidence that make it plausible, at least, that this work could be an original.  The crucial piece of evidence is that a painter named Louis Finson, from Bruges, was living in Naples during Caravaggio’s years there, and evidently owned a Caravaggio painting of Judith and Holofernes.  A visitor mentions (but does not describe) the work in a document dated 1617.   We know the painting mentioned is not the more famous depiction of Judith that is universally attributed to Caravaggio, since that one was commissioned by the Roman banker, Ottavio Costa, and would have still been in his possession at this time. What’s more, Louis Finson created a Judith painting, and his composition duplicates the one that was discovered in Toulouse.  Finson’s execution is less polished than the Toulouse canvas, suggesting that this was his source, not the other way around.

This still leaves the question of how the painting ended up in Toulouse.  One clue, mentioned on an art history blog, is that a Caravaggio painting of Judith was documented in Antwerp at the end of the 17th century in the collection of a well-known engraver.  The painting might have made its way from Italy into a Flemish collection via Finson, who was also an art dealer and retained strong ties to his homeland.  Finson owned another Caravaggio, Madonna of the Rosary (below right), which was bought by a Flemish consortium and gifted to a church in Antwerp. Belgium came under French control during the reign of Napoleon, and was part of the French empire between 1793 and 1815.  During occupation, Napoleon’s troops are presumed to have acquired many paintings, by either looting or by other coercive means, such as levying high taxes.  The home in Toulouse where this panel was found once belonged to an officer in Napoleon’s army.  Therefore, it is easy to imagine that he might have acquired it on a campaign in the low countries.

This is an incomplete provenance, but it is certainly suggestive.  Moreover, there are aspects of style and content that link the painting to Caravaggio.  The composition resembled the known version in various respects.  It contains Judith, her slave, and Holofernes at the very moment of his decapitation.  Caravaggio liked to depict moments of peak action, rather than, say, placing the head of Holofernes in a sack, which had been a more popular episode during this period.  The figures are set against a sparse, black background, consistent with Caravaggio’s extreme tenebrism.  Hanging above the figures is a rich red cloth, similar to the one in the known painting.  

The compositions of the two works are also similar in other respects.  Holofernes is on the left lying in bed with his right arm holding up his body weight and his left writhing in pain.  His head is twisting back towards his assassin.  Judith’s arms are stretched out with her sword, hacking into Holofernes’s neck, which spouts jets of red blood.  Her maidservant awaits with a sack for the head.  In the known picture, the maidservant is off to the right, marking the edge of the composition.  In the newly discovered canvas, she stands between Holofernes and Judith.  That is the most obvious difference in arrangement, but in both the two woman merge together as a single form—accomplices in this act of violence.  Both compositions seem to be inspired by a 1540 print by Giulio Romano.  The posture of Judith in the known painting closely parallels that print, as does the position of the maidservant to the left of Judith in the newly discovered painting.



It is also possible that the newly discovered painting inspired Peter Paul Rubens.  Rubens was a fan of Caravaggio, and often produced works that elaborated on Caravaggio's compositions.  (For example, he has an Entombment of Christ that exaggerates the composition of Caravaggio's version, as can be seen below.)  

Rubens painted Judith Beheading Holofernes, and though the painting is now lost, we know it from copies.  He painted the maidservant to the left of Judith, and he also painted Judith's right arm below her left, consistent with the Toulouse composition.  Rubens may have seen Caravaggio's painting when it came to Antwerp.

There is another reason for thinking that newly discovered work owes to Caravaggio: it seems to include Caravaggio's models.  Judith in the newly discovered work resembles Judith in the known work.  The resemblance isn't perfect but the faces align quite exactly making it plausible that they were depictions of the same model.  Below, you can see a gradual transition from one to the other.

Notice that the lips match, the eyebrows taper in the same way (though, in the new version they are arched), the hairlines coincide, the cheek bones protrude at the same angle, and the noses are nearly identical.  They are dressed differently, but they both have wavy hair and nearly identical earrings.

The model in Caravaggio's famous Judith painting also sat for several of his other masterpieces including the Conversion of Mary Magdalene and Saint Catherine (both below).  She is believed to be a woman named  Fillide Melandroni.  The resemblance between the woman in the newly discovered painting and Melandroni supports attribution to Caravaggio, since hers was a face he clearly liked to paint.  

In principle, of course, the resemblance between the Judith in the Toulouse painting and the Judith in the known painting could be a consequence of copying.  Caravaggio had many imitators, after all.  But copies of Caravaggios tend to introduce distortions.  More often then not, faces are altered and resemblance to the original models is lost.  Some copies of Caravaggio's Conversion of Mary Magdalene, featuring Fillide Melandroni, are reproduced below (source).  Even Finson's copy of the Judith painting, which is unusually faithful, introduces distortions in Melandroni face.  He has elongated her eyes, reddened her cheeks, and fattened her neck.



Some scholars believe that Caravaggio painted Melandroni  only between 1597 and 1599.  This would present a puzzle since, as we will see, the newly discovered work is likely to have been created much later.  But I think there is at least one other clear case of a Melandroni portrait well after 1599: the Mary Magdelene figure in his Entombment, dated around 1604 (the figure on the right in the detail below).  During the early years of the 17th century, Caravaggio was getting some large Church commissions, which may have made it difficult to use Melandroni as a model.  She was a notorious person in Rome (for some biographical information, see Peter Rabb's M: The Caravaggio Enigma).

Melandroni was born in Siena and came to Rome at an early age looking for opportunities to help support her family.  By the age of 13, police records indicate that she had gotten involved in sex work.  By her late teens, she seems to have become one of the most highly paid courtesans in Rome, known to members of elite society. Caravaggio probably met her when he was commissioned to paint her portrait by Marchese Vincenzo Giustiniani, a wealthy banker (below, right).  (That painting was destroyed during the bombing of Berlin in the Second World War. ) It is conceivable that Melandroni helped Caravaggio network with other powerful figures, advancing his career.  It is also presumed that they became lovers.

Melandoni, like Caravaggio, was born into modest means and came to Rome looking for success.  Both had a number of run-ins with the law.   Caravaggio had fled his hometown of Milan because of a violent altercation, and would later be brought up on a wide range of charges including two stabbings, beating a man with a stick, libel, weapons possession, assault on a Vatican notary, chucking artichokes at a waiter, throwing stones at a policeman, and throwing stones at his landlady’s window.  Melandroni seems to have had a violent temper as well.  One police report indicates that she cut the hand of another courtesan, Prudenza Zacchia, while attempting to slash her face.  Melandroni was enraged because she found Zacchia in the bed of her lover, Ranuccio Tomassoni, a local pimp and neighborhood boss.  

Melandroni’s involvement with Ranuccio Tomassoni helps to explain one of the most momentous events in Caravaggio's life.  In 1606, the artist and some friends got into a fight with Tomassoni and several members of his family.  The most widely accepted reason for the fight is that Caravaggio had refused to pay Tomassoni money lost in a bet over a tennis match. But police reports indicate that Caravaggio attempted to castrate Tomassoni during their fight.  This suggests that the motive was not just monetary.  Perhaps Caravaggio’s contempt for Tomassoni derived from his affection for Melandroni.  Whatever the motive, the outcome was dire.  Tomassoni died from his wounds and Caravaggio was forced into hiding.  Sentenced to death in absentia, Caravaggio fled from Rome never to return.  He spent his final four years on the run, and ultimately died of unknown causes—possibilities include malaria, typhoid, vengeance (we was wanted by papal police, Roman street thugs, and perhaps the Knights of Malta), or lead poisoning from paint.

Caravaggio painted Melandroni as a sex worker (Magdalene), a martyr (Catherine), and a heroine who saved her people (Judith).  In each role, she is presented with strength and dignity.   The figure in the newly discovered Judith painting follows this pattern.  She is portrayed not as a violent killer or sultry temptress, but as a fearless and resolute heroine.  Her maidservant is also familiar from Caravaggio’s work.  She appears in his Crucifixion of St. Andrew, easily identified by her enlarged goiter (below, with detail).  The cause is unknown, but the model probably suffered from Hashimoto's disease or some other thyroid condition.  The male victim has no known identity, though he resembles the Holofernes in Caravaggio’s other treatment of this theme.  He also bears a slight resemblance to Caravaggio himself.  While on the run, Caravaggio painted himself as the decapitated head of Goliath (above), but that choice was probably prompted by the fact that there was a price on his head—perhaps a manifestation of anxiety or a plea for clemency.  

The newly discovered painting is not dated, but there is reason to believe it was painted well after the more familiar treatment of Judith.  The known version was painted around 1598-1599, when Caravaggio was still in Rome.  If the newly discovered canvas is the one that Finson owned, then it probably wasn’t painted until around 1607, when Caravaggio was in Naples.  This would explain why it includes the older woman who also appeared in the Crucifixion of St. Andrew, painted around that time.  A 1607 painting of Salome also includes an older women who may be drawn from the same model (below).  In the Salome painting, the woman's goiter is concealed by clasped hands.  Apparently, Caravaggio has originally planned to conceal here goiter in the same way in the St. Andrew scene.  An infrared spectrograph of that work reveals that the older woman had clasped hands in the underpainting (above, source).  Scholars believe that Caravaggio chose to reveal her enlarged goiter there because St. Andrew was said to cure throat ailments.  In the Judith painting, there is no evidence that he considered hiding her ailment.  Tthe enlarged throat amplifies the violence of the decapitation.




A further bit of evidence for timing concerns the sword in the new Judith painting.  It closely resembles a sword in another painting of David, which was also executed in 1607.  Caravaggio was no stranger to weapons, and we see different swords appearing in his work.  These two pictures are also stylistically similar.  They have an illustrational quality.  Many earlier works are richer in detail, and have more angular elements, and many of the later works have more painterly brushwork.

If the newly discovered Judith was painted in 1607, Melandroni would not have been able to sit for it.  She was back in Rome.  But that could explain why the likeness to Melandroni is imperfect—Caravaggio would have been painting her from memory.  This may seem like an extravagant supposition for an artist who was so famous for painting models from life (Caravaggio did not do preparatory drawings--he painted directly from model to canvas), but other paintings from the period indicate that Caravaggio was working from both models and memory or imagination.  For example, the lead figure in the aforementioned Salome painting looks like an idealized type rather than a portrait from life.  Caravaggio would have had greater accuracy drawing his friend, Melandroni from memory, given that he had painted her many times.  He also would have had some reason to do so, since she appears in the original work.  It is noteworthy that both the new Judith and the Salome are painted with little anatomical detail in a heavy black dress with frilled white sleeves.  The lack of carefully rendered limbs and the visual simplicity of the clothing provide further evidence that Caravaggio was working without a live model in these works.  He was in dire financial straits when he arrived, and it might have been hard for him to find a female model without social connections.

Caravaggio's financial needs may explain another feature of the newly discovered painting.  In trying to establish its authenticity, French expert, Stéphane Pinta, has used x-rays to search for pentimenti--signs of the artist's thought process as the work was executed.  Judging from press photos (right), the x-rays show remarkably little difference between the underpainting and the final product.  This could be taken as evidence against the authenticity of the work, since copies rarely show pentimenti.  X-rays of the original Judith painting reveal that Caravaggio changed the positions of Holofernes's head to emphasize its separation from the body.  No such changes are visible here.  Experts allege that the fingers of Holofernes were widened during its execution, but I am not convinced.  Does this mean the painting is a fake?  Another explanation is possible: Caravaggio could have been quickly churning out reworked versions of an old themes in order to make money.  It is even possible the Toulouse painting is an exact copy that Caravaggio made of another version, which is now lost.  It is known that he made copies of some of his works.  For example, in 1602, he made two nearly identical John the Baptist patintings (below).  He also seems to have made duplicates of his celebrated Supper at Emmaus and the Incredulity of Thomas.  Ten years ago copies of both these compositions were identified in a Church in Loches, France.  Studies of the canvas and pigments that were used indicate that they were created by Caravaggio.

Still, many are unconvinced that the Toulouse painting was created by Caravaggio.  Jonathan Jones, an art critic at the Guardian calls it a fake.  His evidence?  It lacks the intensity of the known Judith painting.  Here's Jones at length:

Compared with this, the Toulouse painting is hopeless. It certainly has the shine and colour of a Caravaggio, the cinematic light effects he is so famous for. But it has none of his psychological intensity. The Barberini painting is the visual equivalent of a Shakespearean tragedy: a profound imaginative insight into killing and conscience. What’s the Judith of Toulouse playing at? She seems bizarrely disengaged from what she’s doing, having drifted off right in the middle of hacking off a man’s head. She’s looking towards us – why? Caravaggio never posed people as vacantly as this. It is the precision of his emotional dramas that makes them electrifying. The Toulouse painting is dramatically incompetent.

I beg to differ.  First off, there are some vacant stares in Caravaggio's oeuvres, like the Salome of 1607, which I mentioned above.  I would concede that the newly found Judith does not convey conscience or ambivalence, like the earlier painting.  There is a slight wrinkle in Judith's brow in the Toulouse painting,  but it is barely perceptible, unlike in the original where the act of killing seems to fill Judith with disgust.  Even so, Judith's expression in the newer work is, if anything, more effective.  Judith is saving the Jews from a murderous general.  She will soon be proudly delivering his head to her people, and she will display it on a stake for all to see.  This same general, drunk with wine and lust, wanted to bed Judith and learn secrets that would allow him to annihilate the remaining Jewish resistance.  Judith is a fierce warrior at her moment of victory, and her expression shows her triumphant resolve.

Caravaggio uses outward glances sparingly in his art, but he does rely on the the device in some works--and to great effect.  He uses it in his homoerotic paintings of Bacchus, Amor, and the prepubescent St. John paintings reproduced earlier (here the outward state entices the viewer).  He also uses in his Ecco Homo, above, and Entombment Paintings (here the viewer is implicated in the crimes against Christ).  Most relevantly, he uses it in a 1609 version of Salome with the head of John the Baptist.  The latter example comes closest to the Toulouse Judith, both in time, and conceptually.  The Salome painting is more introspective (one might even say she is staring inward, not out), but, in both cases, the female executioner seems to be delivering a subtle warning to the viewer.  

 Still, there are some features of the Toulouse Judith that chip away at my confidence about the attribution.  My concerns involve some stylistic elements and the handling of paint.  Stylistically, I am troubled by the fact there is no figure in complete profile.  Almost all of Carravagio's paintings with three or more figures have one in profile.  It is almost a formula in his work.  I've always liked the formula because it imparts a kind of archaic formalism on works that otherwise aspire to naturalism and drama.   Another missing ingredient: dark eyes.  In many Caravaggio paintings, the figures eyes are shadowy--the sclera look dim rather than white.  Less so here.  In terms of brushwork, there are places where the rendering is poorly executed.  One issue is that Judith's cleavage shows little dimensionally (compare the Conversion of Mary Magdalene, above, for example).  This is odd given that Caravaggio is one of art history's most carnal painters.  

Much more troubling in Holofernes's face (above left).  Caravaggio was a master of sfumato.   Like Leonardo, he could create contours that blend so well that outlines entirely disappear.   The face of Holofernes in the Toulouse painting shows no evidence of this gift.  The lines are both visible and amateurish.  For example, the line extending from Holofernes's nostril to his mouth is sharply drawn, and the flesh on either side of it looks more like two disjointed, flat planes rather than a continuous, undulating surface.  The lines under his eyes and the dark bars on the bridge of his nose are also unlike anything in Caravaggio's work.  Other details in the face are equally unconvincing.  Caravaggio normally shows a great attention to detail.  In the original Judith (above right), he includes untamed hairs in the mustache which overlap with a gaping mouth.  In the Toulouse painting, the moustache is a cartoonish paste-on.  Worse still, the teeth look like a picket fence; there are no gums and no impression that they curve into the mouth.  The mouth cavity is little more than a black pit.  The face in the original Judith is masterful.  The new one could be a comic book villain.  Admittedly, these flaws do not diminish the overall impact of the composition, but they are glaringly obvious when one attends to it in detail.  Moreover the Finson copy (below), though graceless, is somewhat more gently rendered.  Was Finson improving on Caravaggio here?  Or was he copying from another version?

I think there is good reason to think that Finson was copying the Caravaggio that was found in Toulouse.  The best evidence involves a small detail on the arm of Holofernes.  If you notice, in the head detail above, there is a curious line extended up from the armpit of Holofernes.  This is painted as if it were a wrinkle or crease in the skin.  That is an anatomical error.  The Toulouse canvas has no wrinkle painted there, and is, as such, more anatomically accurate.  But happens to be a crease in the canvas at just that spot.  It seems likely that the crease was there when Finson owned the work and he mistook it for part of the painting!  (See comparisons below.)

Even after noticing this, I was haunted by the hackish brushwork in the face of the Toulouse Holofernes.  Could Caravaggio really have sold a painting with a face so poorly rendered?  Perhaps.  There is one other example in Caravaggio's oeuvre of a face rendered with clumsy outlines.  It is in Caravaggio's aforementioned copy of his Incredulity of Thomas, which was recently discovered in Loches (below).  Here, the faces are like masks.  The faces in this copy could have been painted by a 20th century expressionist.  They are so stylized that they actually work, giving the painting a modern feel.  This would cast doubt of the authenticity of the painting were it not for the chemical analyses of its pigments and canvas.  For me, the Loche Thomas work helps confirm the authenticity of the Toulouse Judith.  It suggests that the Toulouse painting may be a copy made by Caravaggio of his own work.  Further testing will settle whether the right materials were used, and I would bet on a positive verdict.  Of course, there is always a chance that a follower of Caravaggio managed to use materials purchased from the same source.  Does this possibility mean we should discount the significance of such forensic tests?  I don't think so.  We can't rule out that Caravaggio's had a follower who used the exact same materials, but in an era before mass production, we should not take that mere possibility too seriously.  Moreover, the overall quality of the work is high enough (again, compare Finson) to believe that it was painted by Caravaggio himself.

My overall assessment is that the Toulouse painting is a Caravaggio.  Indeed, it is an extremely fine work.  It has subtle features that invite reflection, commentary, and discussion: Judith's dark attire, the juxtaposition of her alabaster skin with the weathered face of her maidservant, the way Holofernes looks up at the maidservant, who looks at Judith, who then looks at us.  To me, this picture is as compelling as the known Judith painting, and comparisons between the two will be fascinating to explore.

I've occupied a lot of space arguing for authenticity, and the experts may be testing and debating this painting for years. Why?  Why bother with this debate about attribution?  Obviously attribution matters for market value, but should it?  And should we care whether this canvas was painted by Caravaggio or one of his followers.

One might think it matters because Caravaggio was a great innovator, and an original work is better than a derivative work.  But Caravaggio's innovations are often exaggerated (see Mina Gregori's excellent essay in this volume).  He came from Lombardy where both naturalism and tenebrism have long histories.  Both can be found in Leonardo a century before Caravaggio was painting, and lesser known masters, such as Moretto da Brescia, anticipate Caravaggio's work by decades.   The Moretto above (top) may have been an inspiration for Caravaggio's evocative Entombment, and the Moretto below it (depicting the prophet Micah) closely resembles the shadowy figure on the right side of Caravaggio's 1609 Salome painting (reproduced earlier).  Moretto advocated naturalism and emotional directness, and he eschewed the classicism and symbolism associated with High Renaissance.  He also makes generous use of shadows and dark pigments.  Other painters from Brescia produced works that anticipate Caravaggio's famous tenebrism.  There is a Giovanni Savoldo from 1534 depicting a candle-lit St. Matthew against a dark background (below, left).   In 1571, Antonio Campi painted the Beheading of John the Baptist (also below, right).  It lacks the brutality associated with Caravaggio's work, but the lighting certainly anticipates his imitable style.  Caravaggio was familiar with these painters and learned much from them.  He pushed some elements farther then his predecessors in Brescia, but he did not emerge ex nihilo.  

I already noted, as well, that Caravaggio's original Judith painting may have derived from a print source.  There were many outstanding versions by artists such as Botticelli, Cranach, and Michelangelo.  These helped to establish conventions for the genre, and they are often graphically violent, like the Veronese version, below left.  Caravaggio was clearly inspired by this tradition. He may also owe a debt to Fede Galizia, whose version proceeds his first attempt by several  years, and it includes the trademark red fabric (below, right).

Moreover, Caravaggio, like many successful artists, spent years replicating his own style after he hit upon a salable formula.  If we consider originality a prerequisite for greatness, can we really value any but the earliest Caravaggio paintings?  Doesn't he become an imitator of his own earlier self?  Finally, it is both inaccurate and unfair to say that artists who were deeply inspired by Caravaggio cannot be great in their own right.  There are clear counterexamples:  Artemisia Ghentileschi, Georges de la Tour, and Diego Velázquez.  All of these painters are recognized in contemporary canons of Western art.  In addition, there are many other baroque painters who enjoyed fame in their day, but are now less known.  Some of them are exceptionally good in their own right, despite owing debts to Caravaggio: Mario Minetti, Bartolomeo Manfredi, Simon Vouet, Trephine Bigot, and Valentin de Boulogne, to name a few.  

The best of the Caravaggisti are talented enough to deserve widespread recognition.   We don't discount Titian for learning from Giorgione, and, likewise, we should recognize that those who adopted methods from Caravaggio sometimes created paintings that compare favorably with his.   Gentileschi's Judith is better: the violence is more chilling and one can feel the struggle between Holofernes and his attackers much more intensely (above).   There is also a poetic Judith by de Boulogne (below) that is even more naturalistic than Caravaggio's work, and a moody version by Bigot, which takes tenebrism to a new height (bottom).  Caravaggio's popularity stems, in part, from the accessibility of his work.  Though grizzly, his paintings are often easy to digest for modern, cinema-addicted viewers.  Caravaggio manages to convey emotional depth, but his work can be accused of superficiality at times and their mass appeal is an indication that they pander to popular tastes.  That is not always a bad thing, but it should also remind us that Caravaggio's current stardom does not necessarily establish that his was the greatest painter of his time.  Some of his followers produced works that are more subtle, more challenging, and, in various ways, more inventive. 

When assessing the Toulouse Judith, it does matter whether it was painted by Caravaggio, but not because its artistic value depends on that.  It matters because we are interested in understanding Caravaggio's development, and the new painting may shed light on his artistic process.  Our interest in Caravaggio is relatively recent.  For years he was neglected, and current trends should not lead us to think his handiwork is the ultimate measure of worth. We should take an interest in the painting because it is an impressive treatment of the time-honored subject.  It holds its own against  other great Judith paintings.  Art history is a long dialogue between artists working and reworking familiar themes, and this canvas is part of that conversation.  Its rediscovery means it will re-enter the flow of artistic ideas, perhaps influencing future artists or leading to new perspectives on achievements from the past.