Cenni di Pepo, called Cimabue

Cimabue is the first name in the written history of European art and one of the least documented figures in it. Two things are recorded of him with certainty. On 18 June 1272 a notary in Rome listed a witness as Cimabove pictore de Florencia. Between 2 September 1301 and 19 February 1302 the fabric accounts of Pisa Cathedral paid him ten soldi a day to set the figure of Saint John the Evangelist into the apse mosaic. That mosaic is the only surviving work documented as his while he was alive, and it is not a painting. Everything else attributed to him is attributed by argument.

The consequence is exact and worth stating plainly: an attribution to Cimabue is an attribution to a reconstructed hand, not to a documented artist. Nothing is signed. The accepted body of work on panel runs to fewer than ten pictures. Two of those were unknown to scholarship in 1999. One of them was found in 2019 above a hotplate in a kitchen in Compiegne and sold for the highest price ever paid for a pre-1500 work of art.

The name means ox-head, or, from the verb cimare, one who crushes the views of others. Both readings are late. What survives of the man is a nickname, two documents, a wall of blackened fresco at Assisi, a crucifix that the Arno destroyed in a morning, and about a dozen panels that a century and a half of connoisseurship has agreed to call his.

There is no Cimabue market. There is one sale. So far as is known, the Mocking of Christ at Acteon in Senlis on 27 October 2019 was the first work by Cimabue ever offered at public auction, and every other accepted picture is in an institution or a church, most of them immovable. The Louvre panel arrived in Paris in 1812 as spoil of war; the National Gallery panel arrived in 2000 as a tax settlement rather than a purchase; the Frick panel was bought from a dealer in 1950 for a sum the Frick does not publish.

The single result is therefore both a record and a sample of one. It set the highest price ever paid for a work of art made before 1500, and it did so on an object that had been estimated at 300,000 to 400,000 euros before anyone had looked at it properly. The lesson a collector should take from it is not about Cimabue's price level, which does not exist, but about the gap between a regional estimate and a technically examined attribution: the multiple was more than sixty.

The second lesson is that a record price does not confer title. The French state classified the panel a national treasure within weeks and refused it an export licence, and the buyers never took possession.

Two documents and a lifetime

The documentary base is this. In 1272 Cimabue appears in Rome as a witness to a notarial act, described as a painter from Florence: the National Gallery in London therefore gives his dates as documented 1272, died 1302, which is the honest formula. In 1301 he was commissioned to supply an altarpiece for the Ospedale of Santa Chiara in Pisa; the picture is lost, and Dillian Gordon notes it may never have been painted at all. From 2 September 1301 to 19 February 1302 the Pisa Cathedral accounts record daily payments of ten soldi for work on the apse mosaic, ninety-four days in all, and specify that he finished the figure of Saint John.

He did not finish the mosaic. Francesco da Pisa had already set the Christ; the Virgin was completed by Vincino da Pistoia only in 1321, nineteen years after Cimabue was last paid. The mosaic survived the cathedral fire of 1595 and four subsequent campaigns of renovation.

That is the whole of it. There is no will, no inventory, no letter, no drawing that anyone accepts, no record of a workshop, no list of pupils that is not Vasari's. The training is a blank: Vasari's story that he learned from Greek painters brought to Florence is not credited by modern scholars, and the alternatives offered - an apprenticeship under Coppo di Marcovaldo, the influence of Giunta Pisano - rest on style alone.

The number, and the number that is not the number

Writing in Apollo a fortnight before the 2019 sale, Donal Cooper of the University of Cambridge placed the new panel precisely: it joins fewer than ten accepted works on panel by the artist. That is the figure to quote, and it is the reason a Cimabue coming to market is not a rare event but a first one.

Count every medium and the number rises. Reporting on the Louvre's 2025 exhibition put it at twelve works securely attributed, which takes in the Assisi frescoes and the Pisa mosaic alongside the panels. A stricter reckoning goes the other way: of the fourteen paintings Vasari listed as Cimabue's, five are lost and five are now given to other hands, leaving four that are generally accepted without argument - the Santa Trinita Maesta in the Uffizi, the Santa Croce Crucifix, the frescoed Maesta in the Lower Church at Assisi and the Maesta in the Louvre.

The figure that is not a corpus is forty. The Louvre's 2025 show, Revoir Cimabue. Aux origines de la peinture italienne, assembled une quarantaine d'oeuvres, and the number has been repeated as though it were a count of his work. It is not. The checklist includes Duccio, whose Maesta was shown as a precedent, and closes with Giotto's Saint Francis Receiving the Stigmata, painted for the same church in Pisa as the Louvre Maesta. It is an exhibition about where Cimabue sits, built from comparanda. Anyone citing forty as the size of the corpus has counted the room, not the artist.

What Vasari did to him

Vasari opens the Lives with Cimabue, and the whole shape of Western art history follows from that choice. He also gets a great deal wrong, and because he is first, the errors propagate.

He gives the birth as Florence, 1240, and the death as 1300 at the age of sixty. The Pisa accounts show him alive and working in February 1302. He says Cimabue was buried in Santa Maria del Fiore under an epitaph by one of the Nini: Credidit ut Cimabos picturae castra tenere, sic tenuit, vivens: nunc tenet astra poli. He then says that Dante, in Purgatorio XI, was alluding to that inscription. The dependency runs the other way if it runs at all: Dante was writing within a decade of Cimabue's death, and the Latin couplet reads like a translation of Dante rather than a source for him.

The largest error has a name. Vasari describes a great panel of the Virgin made for Santa Maria Novella, set high between the Rucellai and Bardi chapels, larger than any figure painted up to that time, carried to the church in solemn procession with trumpets. That picture is the Rucellai Madonna, and a contract of 15 April 1285 commissions it from Duccio di Buoninsegna for the Laudesi confraternity, for 150 lire, with the stipulation that he paint it himself and use ultramarine and gold leaf. Vasari's attribution stood until 1889, when Franz Wickhoff argued it back to Duccio on the evidence of Duccio's own Maesta.

The anecdotes have had the longest life and the least support. Cimabue finding the boy Giotto drawing a sheep on a slab of rock at Vespignano; Giotto painting a fly on the nose of a figure of his master's so convincingly that Cimabue tried more than once to brush it off; the painter so proud that he would destroy a finished work rather than let a fault stand in it. The last of these Vasari takes not from life but from a commentator on Dante writing around 1334, whose subject was the vice of pride, not the workshop.

What Dante actually says

The three lines of Purgatorio XI are the most quoted sentence about any medieval painter and are almost always quoted for the wrong reason. They are not Dante's verdict on painting. They are spoken by Oderisi da Gubbio, a manuscript illuminator, on the terrace of the proud, where the penitents walk bent double under stones. Oderisi's whole speech is an argument that earthly fame is a breath of wind that changes direction: Cimabue eclipsed by Giotto is his first example, and Guido Guinizelli eclipsed by Guido Cavalcanti is his second, and both are there to tell the pilgrim that his own name will not last either.

Read as a moral illustration, the passage says that Cimabue was famous enough by around 1310 that a Florentine reader would recognise the point without a gloss. Read as art history, which is how Vasari read it and how it has been read since, it becomes a league table with two entries. The first reading is what the poem is doing. The second is what the discipline made of it.

The kind of object he was making

A Duecento panel is not a picture in the modern sense. It is a cult object with a function, and its licence to exist at all had been settled by argument three centuries before Cimabue was born. The Second Council of Nicaea in 787 permitted images of Christ, the Virgin, the angels and the saints, and drew the distinction the whole tradition rests on: images receive the veneration of honour, not the worship owed to the divine nature alone, and the honour paid to an image passes to the person it represents rather than staying in the wood and pigment. After a second iconoclast period the decision was made final in 843 and is commemorated in the Byzantine rite as the Triumph of Orthodoxy.

That settlement is what a Florentine painter in the 1270s inherited along with the Byzantine formulas themselves. It explains why the compositions are so stable: a panel that works is a panel that carries the honour correctly, and there is no premium on originality of type. Cimabue's Virgin and Child with Two Angels in London depends on a Byzantine prototype of a kind still preserved at Saint Catherine's on Sinai and in the Archaeological Museum at Veroia in northern Greece, down to the angels' fingertips curling over the back of the throne. The angels wear the loros, the pearl-studded ceremonial band of Byzantine emperors and empresses. The court dress is not decoration; it is the argument, made in the visual language of earthly power, that the two figures on the throne are the King and Queen of heaven.

Function is equally concrete on the large panels. The Louvre Maesta, four and a quarter metres high, stood on an altar in the Franciscan church at Pisa. Its original gabled frame carries twenty-six painted medallions: the Redeemer and four archangels at the apex, prophets at the corners, the twelve apostles down the sides, and along the lower crossbar Clare, Francis of Assisi, Elizabeth of Hungary and Anthony of Padua. That lower register is a roll-call of the Order that paid for it, placed where a friar kneeling at the altar would meet it at eye level.

Where he leaves the Greek manner

The departure is easiest to see in the smallest works. In the London panel the throne is set at an oblique angle rather than flat to the picture plane; the cushion curls at its edges under the Virgin's weight; and the Child, seated on her raised left knee and turned away from the viewer, steadies himself by grabbing her wrist and finger. Nothing in the Byzantine model does that. The National Gallery's own reading of the change is that Cimabue was adapting formal images for Western Christians for whom a personal relation to God was the point of the exercise, and that his experiments with three-dimensional settings and with gesture were what the next generation took up.

The same movement runs through the large panels in sequence. Most scholars place the Pisan Maesta now in the Louvre before Duccio's Rucellai Madonna of 1285, and before Cimabue's own Santa Trinita Maesta; Hayden Maginnis dissents, and Mariagiulia Burresi and Antonino Caleca have argued the Pisan panel back into the late 1260s or 1270s with a terminus ante quem of 1284. Gordon places the London panel stylistically between the two large Maestas, which is one reason the small pictures matter out of all proportion to their size.

The Santa Trinita panel makes the doctrinal claim explicit in its architecture. Under the throne, in four arched niches, sit four half-length prophets holding inscribed scrolls: Jeremiah, Abraham, David and Isaiah, each carrying an Old Testament verse read as a prophecy of the Incarnation. The Virgin enthroned above them is being presented as the fulfilment of what the four men below are holding in their hands. The panel is an argument with a foundation course.

Gold, blue, and what they actually cost

The technical facts are known in unusual detail for the London panel because the National Gallery examined it after acquisition. The gold is water gilding laid over red bole - the clay underlay that both deepens the colour and stops gold leaf on a white ground reading green. There is no mordant gilding anywhere. The gold was then worked with tools: a single dot punch round the border and the haloes, a finer instrument punching rather than incising a foliate pattern into the border and a pseudo-kufic design into the haloes. Pseudo-kufic is Arabic script copied as ornament by a painter who could not read it, and its appearance in a halo is a straightforward index of how far Mediterranean goods and Byzantine models had travelled.

What gold ground was understood to mean is a separate question from what it cost, and the two are usually run together. Otto Pacht's formulation, that medieval gold ground was always interpreted as a symbol of transcendental light, is the standard statement of the first. The second is counterintuitive: gold leaf is beaten so thin that the bullion in a panel is not the expensive part, and lapis lazuli blue was at least as costly to use. Duccio's 1285 contract is the check on this. It stipulates ultramarine and gold by name, which tells you what a confraternity thought worth writing into a legal instrument.

The palette Cimabue used on the London panel is narrow and deliberate. Flesh is laid over an underlayer of green earth, in the Byzantine manner, and modelled with white, red lake, vermilion and a little black. The only blue found anywhere in the picture is ultramarine, used with varying white for the Virgin's cloak, her cushion and parts of the angels' wings. The cloth of honour is vermilion glazed with red lake. Red lake has faded unevenly across the panel, so the pinks are now paler than he left them.

Assisi, and what the Order wanted the building to say

Cimabue's largest surviving undertaking is the decoration of the transepts and apse of the Upper Church of San Francesco at Assisi, datable to about 1277-80. He had the crossing vault with the four Evangelists, each set against a cityscape labelled for the region he was held to have evangelised - Asia, Greece, Judea and Italy, the last inscribed in the archaic form Ytalia. He had the apse with the death, assumption and coronation of the Virgin, the right transept with scenes from the Apocalypse and the left with a vast Crucifixion, in which Saint Francis kneels at the foot of the cross.

The nave below was given to other hands, Roman and Tuscan, and its programme is typological: thirty-two scenes in two registers, the Old Testament above and the New below, with the life of Saint Francis running beneath them at eye level. The arrangement makes the argument that the Order existed to make. Francis is not illustrated after scripture; he is set in the same column as it, so that the reader of the wall moves from type to fulfilment to friar without a break. Cimabue's Crucifixion, with the founder at the foot of the cross among the eyewitnesses, does the same work in one image.

The Franciscan reading extends into the smallest details, and here the evidence is argued rather than documented. Anne Derbes has argued that the transparent loincloth worn by Christ in Cimabue's Santa Croce Crucifix, which has Byzantine antecedents, emphasises his nakedness and therefore his poverty, a reading with particular weight for an order founded on it. Holly Flora's monograph, which took the 2020 Premio San Francesco, makes the broader case that Cimabue worked with his Franciscan patrons on ideas about materials, memory and beauty, and that the results were designed to enable new kinds of devotion rather than merely to look new. He was not exclusively a Franciscan painter: the Arezzo crucifix was made for the Dominicans at San Domenico.

The negative on the wall

The Assisi frescoes are ruins, and they are ruins for a chemical reason that is now precisely characterised. Cimabue used lead white heavily in his highlights, in preference to bianco di San Giovanni, which covers less well. The lead white has blackened. Where he laid his lights the wall now reads dark and where he laid his darks it reads pale, so that the Crucifixion in the left transept has the appearance of a photographic negative of the picture he painted.

A team led by Manuela Vagnini published the analysis in Vibrational Spectroscopy in 2018, using Raman spectroscopy and X-ray diffraction on fragments of the work. They found complete oxidation of lead white to lead dioxide, the mineral plattnerite, throughout an external layer twenty to thirty microns thick, both where the pigment was used alone and where it was mixed with azurite - the azurite itself partly converted to copper chlorine compounds. The finding that matters is negative: the samples are painted a secco, which rules out the standard explanation that the alkaline fresco mortar drove the oxidation. The authors point instead to a possible role for chlorine compounds. In old master drawings the same pigment blackens by a different route, forming lead sulphide from hydrogen sulphide in the air; the mural mechanism is not that one, and is still not fully explained.

The frescoed Maesta in the Lower Church, showing the Virgin and Child with four angels and Saint Francis, was cleaned in a campaign completed in February 2024 by Sergio Fusetti, the basilica's chief restorer, with a team from Tecnireco of Spoleto, at a cost of 300,000 euros paid by Ferrari. The findings there were mundane and instructive: the surface had been dulled by dirt deposited by millions of visitors and by a protective coating applied in the 1970s that had gone reflective and yellow.

26 September 1997

Two earthquakes of magnitude 5.7 and 6.0 struck Umbria in quick succession on 26 September 1997. Four people inside the Upper Church were killed by falling masonry. Two sections of vault came down. One of them was Cimabue's.

Of the four Evangelists in the crossing vault, the sections showing John, Luke and Mark broke into relatively large pieces and were reassembled. The Saint Matthew section collapsed completely into roughly 120,000 fragments, which were collected, scanned and catalogued into more than a thousand labelled drawers. A neutral curved panel now fills the hole in the ceiling. About thirty fragments have been put back.

The fragments have had one scientific use already: the samples in the 2018 lead white study came from the Saint Matthew scene, taken from material that could not be replaced on the wall. In December 2025 the Galleria Nazionale dell'Umbria in Perugia, the engineering department of Perugia University and the culture ministry's conservation office began a feasibility study, running at least six months, into whether artificial intelligence can match the fragments by shape and colour against a high-resolution photograph of the vault taken shortly before the earthquake. In the meantime a projection made by the Florence imaging firm Ikare, at a cost of 5,000 euros, throws a digital reconstruction of the missing section onto the blank panel.

4 November 1966

The Crucifix from Santa Croce is the single most consequential loss in Cimabue's corpus and the reason his name is known to people who could not name a picture by him. When the Arno came into the refectory on 4 November 1966 the water rose above the height of Christ's halo, and took the paint with it when it went down. Roughly sixty per cent of the painted surface was destroyed. Oil, mud and naphtha soaked the panel, which swelled, expanded and cracked what was left. Volunteers recovered specks of pigment from the water by hand.

The restoration ran for ten years under Umberto Baldini and Ornella Casazza at the Florence restoration laboratory, and it is a landmark in conservation quite apart from the object. Baldini rejected Brandi's tratteggio as it was then being applied and developed two alternatives, selezione cromatica for losses that can be reintegrated without invention and astrazione cromatica for losses that cannot. The Cimabue Crucifix is where the second was first used. Because the gaps interrupt the face and the body of Christ, reconstructing them was judged unjustifiable; instead the hatched retouching does not reconstruct any form at all, its colour derived from the overall chromatic character of the picture, to make what Baldini called a neutral link between the surviving passages. His stated aim was to give the object back as a work of art while ensuring that the restoration itself had, in his phrase, no independent action on it.

Whether it succeeded is disputed in print. Kim Muir's survey of the field records that Alessandro Conti, Roger Marijnissen and Leon Kockaert, M. Kirby Talley and Sarah Walden have all questioned whether the treatment met its aims. The crucifix went back on public display in 1976, travelled in Europe and America, and in 2013 was hung in the sacristy at Santa Croce, high enough to survive the next flood.

Even its basic data are contested. The Opera di Santa Croce gives 433 by 390 cm and dates it before 1288; the Web Gallery of Art and Wikipedia give 448 by 390 cm, the first dating it 1287-88 and the second about 1265. A twenty-year spread on the date of the largest surviving object in the corpus is the kind of thing that does not appear in general accounts of him.

The diptych: eight scenes, three survivors

The most important development in Cimabue studies in a century has been the recovery of the small devotional work he made around 1280, and it happened in two instalments nineteen years apart.

The Frick Collection bought a panel of the Flagellation from Knoedler in Paris in 1950. Its authorship was argued over for fifty years: Roberto Longhi proposed Cimabue, Millard Meiss insisted on Duccio, Meyer Schapiro read it as evidence of a flagellant confraternity, and Meiss eventually conceded he might have been wrong. Neither John White nor James Stubblebine would put it in Duccio's work; White called the Duccio attribution a flight of scholarly fancy.

In 2000 a small panel of the Virgin and Child with Two Angels emerged from the Gooch collection at Benacre Hall near Lowestoft in Suffolk, where it had been catalogued since at least 1933 as Old Italian School. It was entered for sale at Sotheby's on 6 July 2000 as lot 76, withdrawn, and accepted by the British government in lieu of inheritance tax and allocated to the National Gallery. Its attribution to Cimabue, endorsed by Miklos Boskovits and by Luciano Bellosi in a letter of 26 May 2000 now in the Gallery archive, settled the Frick panel too: the two are the same poplar, the same thickness, the same underdrawing, the same craquelure, the same punched border.

In 2019 the third panel appeared and fixed the geometry. Acteon in Senlis established that the wood grain of the Mocking of Christ ran continuously into that of the London panel, and that its barbs - the torn edges left where an engaged frame was prised off - matched. The Virgin and Child was the top left scene, the Mocking directly beneath it, the Flagellation to its right. That in turn confirmed Boskovits's reconstruction of a diptych by the Master of San Martino alla Palma of about 1320, eight scenes across two wings, which Dillian Gordon had already argued was copied from a Cimabue. Five scenes of Cimabue's are therefore missing and can be named: a Betrayal from the left wing, and an Ascent to Calvary, Crucifixion, Entombment and Last Judgement from the right.

None of that is stylistic. Grain, barbs, worm channels and craquelure are carpentry and physics, and they are the strongest evidence in the whole Cimabue file.

What the small panels were for

A folding object twenty-five centimetres high, showing the Passion scene by scene in gold-ground compartments divided by painted red bands, is not an altarpiece and not a picture to be looked at across a room. It is an instrument for a practice, and the practice is documented.

In the copy made a generation later by the Master of San Martino alla Palma, Saints Clare and Francis kneel at the foot of the cross. Boskovits and Gordon read that as evidence of a Franciscan and probably Clarissan patron, and Gordon takes the correspondence as strengthening the case for a Clarissan patron for the Cimabue original. Holly Flora had already proposed, on the strength of the Frick and London panels alone, that Cimabue's series is the earliest identifiable example of a tradition of Clarissan devotional image; Cooper judges that the third panel significantly strengthens her reading. The literature these communities used runs in parallel: the Meditations on the Life of Christ, which Cooper describes as composed for a Clarissan nun in Tuscany around 1300, instructs the reader to place herself inside each episode of Christ's life and dwell there. Its authorship is disputed - pseudo-Bonaventure, John of Caulibus, and a Poor Clare have all been proposed - but its method is not in doubt, and the panel is built for it. Cooper's collaborators describe exactly what such an object gives a reader: Andrea De Marchi points to choral narration, the crowding of figures known from the Assisi frescoes brought down to the scale of the hand; Joanna Cannon to interweaving gestures and exchanged looks offering multiple points of focus that converge on the standing Christ.

One detail in the London panel has been read further, and should be labelled as argument rather than fact. Gordon suggests that the transparent cloth with black embroidered borders draped over the Virgin's cushion is not a piece of dress but an altar cloth of the kind seen in Simone Martini's Mass of Saint Martin in the Lower Church at Assisi, and that its presence underlines a reading of the Virgin as an altar holding the body of Christ. That reading is hers, made in a catalogue, and stated here as hers.

The Compiegne panel, and the two prices

In June 2019 an auctioneer clearing a house at Compiegne, north of Paris, found a small gold-ground panel hanging on the kitchen wall above the hotplate. The owner, a woman in her nineties, believed it a Greek icon. The Louvre's own record dates the family's possession of it to at least 1871, which is a longer and better-documented tenure than the coverage of the find suggested.

Acteon's first estimate, before attribution work, was 300,000 to 400,000 euros. After Eric Turquin and his colleagues in Paris had examined it and the infrared reflectogram had shown an underdrawing consistent with Cimabue's, the estimate was raised to four to six million. It sold at Senlis on 27 October 2019. The hammer was 19.5 million euros; with fees the total was about 24.1 million. Both numbers are real and almost every account carries only one of them without saying which. It was, so far as is known, the first work by Cimabue ever offered at public auction, and it set the record for any work of art made before 1500.

The buyers were the Alana Collection, formed by Alvaro Saieh and Ana Guzman, bidding through the London dealer Fabrizio Moretti. They did not get to take it home. The French state classified the panel a national treasure - the Louvre records the classification on 20 November 2019, the press on 23 December - and refused an export licence, freezing it in France for thirty months while the money was raised. The Louvre completed the purchase at about 24.1 million euros, funded partly from the Louvre Abu Dhabi brand licence, with a final million from the Societe des Amis du Louvre and a contribution from Harry and Linda Fath. It entered the inventory on 25 October 2023 as RFML.PE.2023.33.1.

Anyone working from the 2019 reporting alone would conclude the panel is in a private American collection. It has been French state property since 2023.

How the reputation moved

Cimabue's posthumous career is a study in what happens when a name gets attached to the wrong object. Vasari gave the Rucellai Madonna to him; the picture is enormous, famous and in Florence; and for three centuries the most celebrated Cimabue in the world was a Duccio.

The Victorian consolidation of the error is a painting in itself. Frederic Leighton exhibited Cimabue's Celebrated Madonna is carried in Procession through the Streets of Florence at the Royal Academy in 1855, two hundred and twenty by five hundred and twenty-one centimetres, with Giotto, Dante, Arnolfo di Cambio and Charles of Anjou on horseback in the crowd. Queen Victoria bought it on the opening day for 600 guineas. The picture being carried through the streets is Duccio's. Leighton's canvas is now in the National Gallery on long loan from the Royal Collection and is not on display.

Wickhoff's correction in 1889 subtracted the most famous work from the corpus. The twentieth century added almost nothing until 2000, and the National Gallery states the position bluntly: until the London panel surfaced, Cimabue was known only for large-scale works - frescoes, painted crucifixes and monumental enthroned Virgins. Two small pictures and a kitchen in Compiegne have changed what the name means in twenty-five years.

How you would check any of this yourself

There is no authentication board and there never has been. Attribution to Cimabue is settled, so far as it is settled, in three places, and all three are consultable.

The first is the catalogue literature: Luciano Bellosi's Cimabue of 1998, with the critical apparatus and catalogue by Giovanna Ragionieri, which is the reference work and includes the rejected attributions alongside the accepted ones; and Dillian Gordon's The Italian Paintings before 1400 of 2011 for the National Gallery, whose entry on NG6583 is published free online with the technical notes, the full attribution history and a curator's update of 10 April 2024 that takes account of the Compiegne panel. Holly Flora's Cimabue and the Franciscans of 2019 is the standing monograph on the patronage.

The second is the conservation record: the National Gallery's examination of NG6583, the Turquin firm's infrared work on the Louvre panel, the 2018 Vibrational Spectroscopy study of the Assisi fragments, the Louvre's restorations of the Maesta and the Derision completed in 2024.

The third is the wood. Grain continuity, barbs, worm channels and craquelure decided the diptych, and they are the only part of the Cimabue argument that does not depend on anyone's eye.