Leonardo da Vinci

Around 1490, in Milan, Leonardo drew a machine for making needles and calculated that it would produce twenty thousand an hour. In the margin he worked out the profit: sixty thousand ducats a year, at a hundred hours a day of operation.

There are a hundred hours in a day only for someone who is not really building the machine. The calculation is the point, not the manufacturing, and the pleasure lies in seeing how a thing would work. He did this for forty years, on paper, about everything, and finished almost nothing.

The illegitimate son

He was born on 15 April 1452, near the hill town of Vinci in Florentine territory. His father Ser Piero da Vinci was a notary; his mother Caterina was a local woman of no standing, possibly a household servant, possibly - on a recently argued and contested reading of an archival document - an enslaved woman from the Caucasus or the Balkans. They did not marry.

Illegitimacy shaped everything twice over. He could not follow his father into the notarial profession, which was closed to those born outside marriage, and he could not attend a university. So he was apprenticed instead, and he never received the Latin education that defined an educated man of his century. He taught himself Latin in his forties, from a grammar, with limited success.

He referred to himself, with a mixture of defiance and irritation, as omo sanza lettere - a man without letters - and turned it into a method:

They will say that I, having no literary skill, cannot properly express that which I desire to treat of; but they do not know that my subjects are to be dealt with by experience rather than by words. Codex Atlanticus, folio 327v

The same page compares the scholars who quote authorities to men got up in the fruits of other people’s labour, and adds that experience has been the mistress of everyone who ever wrote well. He returns to that argument in a dozen places across thirty years and never once concedes it.

Around fourteen he entered the workshop of Andrea del Verrocchio in Florence - a bottega that produced paintings, sculpture, metalwork and engineering, and which trained Botticelli, Perugino and Ghirlandaio. The story that Verrocchio put down his brush on seeing Leonardo’s angel in the Baptism of Christ is Vasari’s and is probably an invention, but the angel is visibly by a different and better hand.

In 1476 he was anonymously denounced in Florence for sodomy along with three other young men. The case was dismissed for lack of corroboration, twice. He was twenty-four. The episode leaves almost no trace in his writing, and something in his conduct afterwards - the guardedness, the mirror writing, the way he lived - has been read against it, though the reading is speculation.

Leonardo’s sexuality: the accusation of 1476, Salaì and Melzi

On 9 April 1476 someone dropped an unsigned note into the tamburo, the denunciation box set into the wall of the Palazzo della Signoria. It went to the Ufficiali di Notte, the Officers of the Night, a magistracy the republic had created in 1432 for the single purpose of policing sodomy. The note named a seventeen-year-old goldsmith’s apprentice, Jacopo Saltarelli, and four men said to have used him: a tailor called Baccino, one Bartolomeo di Pasquino, Leonardo Tornabuoni of the banking family, and Leonardo di Ser Piero da Vinci, aged twenty-four, living in Verrocchio’s workshop.

Florentine procedure required an accusation to be signed. This one was not. The case was held over on condition that nothing further arrived; a second denunciation came in on 7 June and was dismissed on the same technicality. There was no trial and no verdict.

The scale of the thing is worth stating, because it changes what the document means. Michael Rocke’s work in the Florentine archives established that in the seventy years to 1502 the Office of the Night implicated some seventeen thousand men in a city of around forty thousand people, and convicted roughly three thousand. Sodomy in Florence was a capital offence on the statute books and an ordinary feature of male life in the streets, prosecuted at a volume that only a normal practice can generate. Leonardo was not a scandal. He was one name in a very long ledger.

What survives from him about any of it is nothing. Two months of his twenty-fifth year are documented by the state and by no one else, and he never refers to them.

The rest of the evidence is household evidence. In 1490 a ten-year-old boy from a family of tenants on Leonardo’s vineyard land entered the workshop: Gian Giacomo Caprotti, whom Leonardo renamed almost at once Salaì, little devil, after a demon in Pulci’s Morgante. Within days he had written the boy down in the accounts as a thief, a liar, obstinate and greedy, and he then went on itemising the cost of his cloaks, his hose and his shoes for the next thirty years. Salaì stayed until Leonardo’s death, was left half the vineyard in the will, and was recorded at his own death in 1524 as the owner of a picture that his papers call la Gioconda.

Francesco Melzi was the opposite in every respect: a young Milanese nobleman, educated, literate, who joined the household around 1507 and became the executor. He inherited the manuscripts, the instruments and the drawings, wrote to Leonardo’s half-brothers to tell them of the death in terms no legal formula required, and then spent decades keeping the papers in order.

Around 1910 Sigmund Freud built an entire essay on a single childhood memory in the Codex Atlanticus, in which a bird descends on the infant Leonardo in his cradle and opens his mouth with its tail. The memory is genuine and the bird is a nibbio, a kite. Freud read it in a German translation that rendered the word as vulture, and the vulture, through Egyptian iconography, carried the maternal symbolism his argument needed. Meyer Schapiro took the essay apart in 1956. It survives as a document in the history of psychoanalysis and not as a document about Leonardo.

Modern biography treats him as homosexual and treats the sources as insufficient for anything more precise. That is the honest position. The denunciation is real, the two households are real, and the interior life is not recoverable.

Milan

Around 1482 he wrote to Ludovico Sforza, ruler of Milan, applying for a position. The letter survives and is remarkable for what it emphasises. Ten numbered paragraphs describe portable bridges, methods of draining moats, cannon, covered chariots, catapults, and ways of undermining fortifications. In the eleventh, almost as an afterthought, he mentions that in time of peace he can do architecture and sculpture, and that in painting he can do whatever anyone else can.

The surviving copy, on folio 1082r of the Codex Atlanticus, is not in his hand. It is a fair copy made by someone else, which is what an application actually sent to a duke would require, and it closes with an offer to begin the bronze horse.

He got the job, and stayed seventeen years - the most productive period of his life.

He worked as a military engineer, a designer of court pageants, an architect. He spent sixteen years on a colossal bronze equestrian monument to Ludovico’s father, produced a full-size clay model that astonished the city, and never cast it: in 1494 the bronze was requisitioned to make cannon for the war against France, and in 1499 French crossbowmen used the clay horse for target practice.

His first Milanese painting commission set the pattern for everything afterwards. On 25 April 1483 he and the brothers de Predis contracted with the Confraternity of the Immaculate Conception to supply an altarpiece for their chapel in San Francesco Grande. What the Confraternity received was The Virgin of the Rocks: no haloes to speak of, no gold brocade, the Virgin in a cave among rock formations drawn with a geologist’s attention. The argument about payment ran, on and off, for roughly a quarter of a century, was settled only in 1506, and produced two versions of the painting. One is in the Louvre and one in the National Gallery in London, and which came first is still contested.

And he painted The Last Supper, on the refectory wall of Santa Maria delle Grazie, between about 1495 and 1498.

True fresco requires the painter to work fast, on wet plaster, section by section, with no revision possible. Leonardo could not work that way. So he invented a technique - oil and tempera on a dry, sealed wall - that let him take his time and change his mind. A contemporary described him standing before it for hours without lifting the brush, then arriving suddenly, making two strokes, and leaving.

Within twenty years it was flaking. The most reproduced painting in Western art has been a ruin for five centuries, and its ruin is a direct product of the way its maker thought.

The painting itself is a piece of psychology. Every earlier Last Supper is a static, symmetrical row with Judas isolated on the near side of the table. Leonardo paints the instant after one of you will betray me and shows twelve simultaneous reactions in four groups - recoil, protest, confusion, a hand reaching - with Judas among them, leaning back into shadow, clutching a purse and knocking over the salt.

Leonardo’s Sforza horse, and the bronze that became cannon

The monument was Ludovico’s idea before it was Leonardo’s job: an equestrian statue of his father Francesco Sforza, the mercenary captain who had taken Milan in 1450 and founded the dynasty. Leonardo took it on around 1482 and worked at it, with interruptions, for more than a decade.

The problem was scale. Equestrian bronzes existed, and Leonardo had grown up around the best of them; Verrocchio was working on the Colleoni monument in Venice. But those were roughly life size. This was to stand around seven and a half metres high, which is three times life size and beyond anything cast since antiquity, and would take something in the order of seventy tons of bronze.

He solved it, on paper, twice over. The Madrid codices contain the casting notes: a single pour, with the mould buried and the metal entering from below, and an armature of iron bars, and a system of furnaces feeding one channel. He also worked out how to cast the horse in one piece rather than in sections, which was the whole point and the whole risk.

In November 1493 the full-size clay model was set up under an arch in the courtyard of the Castello Sforzesco for the wedding celebrations of his niece Bianca Maria Sforza to Maximilian I, then King of the Romans. It was, by every contemporary account, the most impressive object in Italy, and no equestrian figure that size existed anywhere else.

Thirteen months later the bronze was gone. In November 1494 Charles VIII of France came over the Alps, and Ludovico sent the metal to his father-in-law Ercole d’Este in Ferrara to be made into artillery. The material for the largest sculpture of the Renaissance became field guns for a war Milan then lost anyway.

The clay stayed in the courtyard. When the French took Milan in 1499, their crossbowmen used it for target practice, and what they left the winters finished. Nothing of it survives.

Leonardo’s own comment on the loss, in a draft letter to Ludovico, is as flat as a receipt: of the horse he will say nothing, because he knows the times. In 1999 two bronzes cast from a new model by Nina Akamu, built on the initiative of the American collector Charles Dent, were installed at San Siro in Milan on 10 September and at Grand Rapids in Michigan on 7 October. They are twenty-four feet high, they are handsome, and they are not his horse. They are what a modern foundry makes when it works from his drawings, which is a different object from the one he could not cast.

Why Leonardo’s Last Supper began to fail within twenty years

The wall is in the refectory of Santa Maria delle Grazie in Milan, and the painting is not a fresco. That single fact accounts for almost all of its history.

In true fresco, buon fresco, pigment in water is laid into wet lime plaster and becomes part of the wall as the lime carbonates. It is extremely durable and it is unforgiving: the painter must complete a patch of plaster the same day, cannot revise, cannot glaze, cannot build up depth in layers. It is a technique for people who know exactly what they are going to do.

Leonardo did not know what he was going to do until he had done it. So he sealed the masonry with a ground of pitch, gesso and mastic, laid a white preparatory layer over it, and painted on the dry surface in tempera with oil, the way he would paint a panel. It gave him everything he wanted: slow work, correction, glazing, the deep shadow behind Judas, the modelling of twelve faces.

The ground began to fail almost at once. Gesso contracts as it dries, and a paint film bound in oil and egg on a contracting ground lifts away from the wall in flakes. Antonio de Beatis, visiting Milan in 1517 in the same journey that took him to Leonardo at Cloux, recorded that the picture was beginning to spoil. Vasari saw it in the middle of the century and described a blur. By the end of the sixteenth century writers were already discussing it as a wreck.

Everything else that happened to it happened to something already fragile. The refectory sits low and takes damp, and the painted wall is an outside wall. Kitchen steam and candle smoke worked on it for two hundred years. In 1652 the friars cut a doorway through the bottom of the composition and removed the feet of Christ. Napoleon’s soldiers used the room as a stable in 1796. On the night of 15 August 1943 an Allied bomb brought down the vault and the east wall of the refectory; the painting survived because it had been sandbagged and scaffolded, and then stood in the open air under a tarpaulin until the room was rebuilt in 1947.

Between those events came the restorers, and they did more damage than the bomb. Repainting campaigns in 1726, 1770, 1821, 1854 and 1908 each covered the previous one. The long final restoration under Pinin Brambilla Barcilon ran from 1978 to 1999 and consisted mostly of removing what earlier hands had added, using solvents and a scalpel under a microscope, square centimetre by square centimetre.

What was uncovered is disputed in the way such things always are. The restoration team’s own figures put roughly 42.5 per cent of the surface as Leonardo’s paint, about 17.5 per cent as loss, and the remaining 40 per cent as earlier restorers’ additions retained or toned. Pietro Marani, who co-directed the work, has said no more than half the surface can be called his. James Beck of Columbia, the most prominent critic of the project, put the original at 18 to 20 per cent and argued that the cleaning had gone too far.

What the ruin costs is specific. The composition is intact, because composition is drawing and drawing survives. The colour is largely gone. So is the modelling of the faces, which was the reason he took years over it, and so is the detail of the far wall, and so is the light. The most influential painting in Europe is legible mainly through the copies its own fame produced, and the best of those, the full-size version by Giampietrino now at Magdalen College, Oxford, was used as a reference during the restoration.

Fifteen years of no fixed employer

The French took Milan in 1499 and Ludovico’s rule ended. Leonardo, at forty-seven, went looking.

In 1502 he took service with Cesare Borgia as military engineer, touring the fortifications of the Romagna. Machiavelli was attached to the same court that winter as the Florentine envoy. The two men were in the same rooms, at the same table, in the same small entourage, at the moment Borgia had his rebellious captains strangled at Senigallia. They almost certainly knew each other; the documentary evidence for a friendship is thin and the circumstantial evidence is unavoidable. Two seats at this table were occupied by men who had shared a winter.

Back in Florence he produced the Mona Lisa - begun around 1503, worked on for over a decade, never delivered - and was commissioned to paint the Battle of Anghiari in the Palazzo Vecchio, on a wall opposite one assigned to Michelangelo, who was twenty-three years younger and disliked him. Leonardo experimented again, this time with an encaustic technique, and the paint ran off the wall. It was abandoned and later painted over.

He returned to Milan, then Rome, where he was outclassed for commissions by Raphael and Michelangelo and spent his time on mathematics, optics and dissections until a complaint to the pope about his dissecting stopped that.

In 1516 Francis I of France offered him a manor house at Clos Lucé near Amboise, a pension, and no duties, with the title first painter, engineer and architect to the King. He went, taking three paintings with him - including the Mona Lisa. He died there on 2 May 1519, aged sixty-seven, and Vasari’s account of him dying in the king’s arms is almost certainly false and very durable.

Leonardo against Michelangelo, and two battle frescoes neither finished

In 1503 the restored Florentine republic gave Leonardo the east wall of the Sala del Gran Consiglio in the Palazzo Vecchio, the vast new council hall built under Savonarola, and commissioned a scene from the Battle of Anghiari, the victory over Milan in 1440. Some months later it gave the opposite wall to Michelangelo, then twenty-nine, for the Battle of Cascina.

It is the only occasion in the history of art on which the two of them were set to work in the same room, and it produced nothing except a legend and a great many copies.

They did not like each other. The Anonimo Gaddiano, an anonymous Florentine manuscript of artists’ lives compiled in the 1540s, records the one exchange that survives: Leonardo passing the benches outside Santa Trinita, a group discussing a passage of Dante, someone asking him to explain it, Leonardo seeing Michelangelo approach and deferring the question to him, and Michelangelo answering, in front of everybody, that Leonardo could explain it himself, being the man who designed a bronze horse and abandoned it out of shame at not being able to cast it. The account exists in that manuscript and nowhere else, which is a reason to hold it loosely. It is also the only piece of dialogue between them anyone has.

Leonardo chose as his subject the fight for the standard: four horsemen and two foot soldiers locked around a pole, the horses biting each other, the faces indistinguishable from the animals’. His own notes for it are among the most disturbing pages he wrote, a description of how to paint a battle that dwells on dust, on the blood in the mud, on the expressions of the dying.

Then he chose his technique again. He sealed the wall with gypsum and rosin, laid an undercoat mixed with heated beeswax, and painted in oil over that, apparently on the model of a passage in Pliny about encaustic painting. To force the drying of the upper part of the mural he had braziers brought in. The lower section set; the upper section, too far from the heat, never did, and the colours ran down the wall.

He stopped in 1506 and went back to Milan. Michelangelo never began painting at all: he finished the cartoon of a company of Florentine soldiers surprised while bathing, which artists came to copy for years as a school of the nude, and in 1505 was summoned to Rome by Julius II. Both cartoons were eventually cut up and dispersed.

Between 1555 and 1572 Vasari rebuilt and repainted the hall. Whether Leonardo’s ruined wall survives behind Vasari’s masonry is a live question: Maurizio Seracini spent decades arguing that it does, drilled through the Vasari fresco in 2011 and 2012, and found pigment traces he considered consistent with Leonardo before the project was stopped amid objections from art historians and conservators.

What is certain is a chain of copies. A partial copy of the central group in the Uffizi, the panel known as the Tavola Doria, and above all a drawing Rubens made around 1603 after an engraving after a copy: three removes from the original, and still the most violent image of cavalry in European art.

The Mona Lisa: who Leonardo painted, and what the 1911 theft did

The picture is oil on a single poplar panel, seventy-seven centimetres by fifty-three, unvarnished by Leonardo and heavily varnished since, and it is not signed, dated, documented by contract or mentioned by him anywhere in seven thousand pages of notes.

The sitter is almost certainly Lisa Gherardini, wife of the Florentine silk merchant Francesco del Giocondo, which is where both La Gioconda and Mona Lisa, from monna, madam, come from. The best evidence is a marginal note found in 2005 in a copy of Cicero’s letters at Heidelberg, written in October 1503 by Agostino Vespucci, a chancery colleague of Machiavelli’s. Comparing Leonardo to the Greek painter Apelles, Vespucci notes in passing that he is at that moment working on a head of Lisa del Giocondo. That single line dates the picture and names the sitter from outside the tradition.

He never delivered it. He was still working on it a decade later, carried it to France, and had it with him at Cloux when Antonio de Beatis saw it on 10 October 1517, described in the visitor’s journal as a Florentine lady done from life at the request of the late Giuliano de’ Medici. It entered the French royal collection and has been in France ever since. That is the whole reason it is in Paris.

Its fame is a separate question, and the honest answer is that most of it is modern.

The painting was admired from the start by people who saw paintings. Vasari, who never saw it, wrote a page about the eyebrows. It hung at Fontainebleau, then Versailles, then in Napoleon’s bedroom, then the Louvre. Through the nineteenth century it accumulated a literary cult: Théophile Gautier and Walter Pater wrote about the smile in terms that now read as fever, and Pater’s description of a woman older than the rocks among which she sits did more for the painting’s mystique than any critic before or since. It was known. It was not the most famous picture in the world.

On the morning of Monday 21 August 1911 the Louvre was closed to the public. Vincenzo Peruggia, an Italian glazier who had worked at the museum fitting protective glass, came in wearing a workman’s smock, lifted the panel off its four hooks in the Salon Carré, took it out of its frame in a service stairwell and walked out with it under his coat. Nobody noticed for more than twenty-four hours; the first person to raise the alarm was a painter who had come to copy it and found four pegs and an empty wall.

The theft was the news story of the year in Europe. The Louvre closed for a week, and when it reopened people queued to look at the empty wall. Guillaume Apollinaire was arrested and jailed; Picasso was questioned. Photographs of the missing painting went round the world in newspapers that had never reproduced it before, and it stayed in the papers for two years.

Peruggia kept it in a trunk in his room in Paris and eventually took it home to Italy. In November 1913 he offered it to an antique dealer in Florence, Alfredo Geri, who informed the Uffizi. It was exhibited in Florence, Rome and Milan to enormous crowds and returned to the Louvre on 4 January 1914. Peruggia served a few months and was widely treated in Italy as a patriot repatriating stolen property, which is a story with a hole in it: Leonardo took the picture to France himself, and Francis I bought it.

Since then the fame has been self-sustaining, and it compounds. Duchamp drew a moustache on a reproduction in 1919. It travelled to Washington and New York in 1963 and to Tokyo and Moscow in 1974, and it has not left Paris since. Around ten million people a year file past it. A former director of the Louvre has estimated that eighty per cent of the museum’s visitors have come for that one panel, which they see for a median of a few seconds, at a distance, behind glass, over the heads of a crowd.

How many paintings Leonardo actually finished

Fewer than twenty paintings survive with an attribution most specialists accept, and the exact figure depends entirely on what is being counted.

The pictures nobody argues about are few: The Last Supper, the Mona Lisa, the two versions of The Virgin of the Rocks, The Virgin and Child with Saint Anne, the Saint John the Baptist, the Lady with an Ermine, the Ginevra de’ Benci, the Annunciation, the Portrait of a Musician, La Belle Ferronnière, the Benois Madonna, the Madonna of the Carnation, the Litta Madonna. Add the angel and part of the landscape in Verrocchio’s Baptism of Christ, and two large works abandoned in the underdrawing: the Adoration of the Magi of 1481, left as a brown monochrome of extraordinary complexity when he went to Milan, and the Saint Jerome, never taken beyond a first lay-in.

Several of those are contested in whole or in part. The Litta Madonna is given by many scholars to Boltraffio. The Portrait of a Musician moves in and out of the canon. The Zöllner and Nathan catalogue, which is the standard reference, runs to a few more than twenty entries only by including works with substantial workshop participation.

Then there is Salvator Mundi, which is a case of its own. A panel of Christ blessing, badly damaged and heavily restored, resurfaced in 2005, was cleaned, was attributed to Leonardo by a group of specialists convened in 2008, hung in the National Gallery’s Leonardo exhibition in 2011, and on 15 November 2017 sold at Christie’s in New York for $450.3 million, the highest price ever paid for a work of art at auction. It was bought for the Louvre Abu Dhabi and has not been publicly exhibited since.

The attribution is not settled. Matthew Landrus of Oxford has argued that most of the execution is by Bernardino Luini and that Leonardo’s own hand accounts for something between five and twenty per cent. The Louvre’s 2019 exhibition, by most accounts, declined to show it as an autograph Leonardo. Christie’s maintained that there was a broad consensus. What is not in dispute is that the panel was substantially repainted in the modern conservation, so that a share of what a viewer looks at was applied in the twenty-first century.

The count matters for a reason beyond bookkeeping. Titian left several hundred paintings, Rubens something over a thousand with his workshop, Rembrandt around three hundred. Leonardo produced in a working life of some forty-five years a body of finished pictures that a diligent painter of the period might complete in three or four. That is the datum every account of him has to explain.

What sfumato is, and how Leonardo made it

Sfumato comes from fumo, smoke, and Leonardo’s own description of the effect is that a painter should work without lines and without borders, in the manner of smoke. It means the abolition of the contour. In fifteenth-century Florentine practice a form was drawn and then filled: the boundary between a cheek and the air behind it was a decision, and it was visible. Leonardo removed the decision. In his mature pictures there is no edge anywhere, only a rate of change.

He arrived at it through optics rather than through taste. The notebooks work steadily at how light behaves in air, at why distant hills go blue and lose their detail, at how the eye handles a boundary between two tones, at the fact that a shadow has no line where it begins. His theory of vision is stated at the opening of the material Melzi later assembled as the treatise on painting:

The eye, which is called the window of the soul, is the principal means by which the central sense can most completely and abundantly appreciate the infinite works of nature. Notebooks, Richter §653; from the Codex Urbinas Latinus 1270

The technique that follows from it is glazing, and it is slow past the point of reason. Instead of mixing a mid-tone and applying it, he applied a great many transparent films, each one shifting the tone a little, each one allowed to dry, so that the transition is built out of accumulated depth rather than out of blending on the surface.

How thin those films are was not known until 2010. Philippe Walter’s team at the Louvre and the European Synchrotron used X-ray fluorescence spectrometry on seven Leonardos, including the Mona Lisa, without taking a sample from any of them, by measuring how the emission from the underlying layers is attenuated by what lies on top. The glaze layers came out at one to two micrometres, with total build-ups in the shadows of a few tens of micrometres. A micrometre is a thousandth of a millimetre. In the darkest passages he had laid up to thirty layers, none of them thicker than a fifth of a red blood cell, over an area the size of a hand.

The consequence is the thing people notice and cannot name. Because the corners of the mouth in the Mona Lisa are indeterminate, with no edge to fix them, the expression is genuinely unstable: peripheral vision, which is coarse and sensitive to low spatial frequencies, reads a smile there, while foveal vision, looking straight at the mouth, does not. The face changes as attention moves across it, and it changes because the painter removed the information that would have settled it.

Leonardo’s Vitruvian Man, and the problem it was solving

The drawing in pen, ink and metalpoint at the Gallerie dell’Accademia in Venice, made around 1490, is the most reproduced image in the history of science and is almost never described accurately. It is not a celebration of the human form. It is a solution to a specific geometrical problem that had been open for fifteen hundred years.

Vitruvius, the Roman architect, wrote in Book III, chapter 1 of De architectura that a well-made man’s outstretched arms and height give a square, and that a man lying on his back with arms and legs extended, with a compass centred on his navel, touches a circle. He meant the two figures as evidence that the human body is proportioned, and that a temple should be too.

The trouble is that the two statements are not compatible with one body. If the square and the circle share a centre, the man will not fit both. Renaissance draughtsmen had tried it repeatedly, and their figures come out either with absurd limbs or with a circle that does not touch.

Leonardo’s move is to give the two figures different centres. The square is centred at the genitals and the circle at the navel, and the same man occupies both by changing posture: feet together and arms out for the square, legs apart and arms raised for the circle. The two positions are drawn superimposed, which is why there are four arms and four legs. The problem dissolves, and the geometry is exact.

Around the figure, in mirror writing, he sets out the proportional system: the span of the arms equals the height, four fingers make a palm, four palms a foot, six palms a cubit, four cubits a man, and so on. These are Vitruvius’s ratios, checked against measurements Leonardo took from live models, and where the two disagreed he followed the models.

He was not the only person working on it. In 1986 the architectural historian Claudio Sgarbi found another Vitruvian figure, in an illustrated manuscript of Vitruvius in Ferrara, drawn by Giacomo Andrea da Ferrara: a friend of Leonardo’s, a Vitruvius specialist, a man Leonardo names in his notes as a dinner companion. Sgarbi dates it to about the same year and argues it came first. Other scholars read the two drawings as the product of a conversation between them rather than a copy. Either way, Leonardo’s version is the one where the line does not waver and the geometry closes, and Giacomo Andrea’s is full of corrections. Luca Pacioli, who knew them both, described the two men as almost brothers. On 12 May 1500 Giacomo Andrea was beheaded in Milan for conspiring against the French occupation, and his body was quartered and displayed on four of the city gates. Leonardo, in the same city and the same year, took French service.

The notebooks

He left around 7,200 surviving pages, which scholars estimate to be perhaps a quarter or a third of what he wrote. They are not organised. A page may carry a study of a horse’s leg, a shopping list, a joke, a geometrical proof and a note about the flight of birds.

Nearly all of it is in mirror writing, right to left. The romantic explanation is secrecy. The likely explanation is that he was left-handed and this kept his hand out of the wet ink.

Anatomy. He dissected perhaps thirty human bodies, working at night in hospital mortuaries in Florence and Rome, in an era before preservation - which means the smell and the decay were part of the work. He produced around 240 drawings whose accuracy was not matched for centuries. He was the first to draw the human spine with its correct curvature, and he left what is usually taken to be the earliest description of arterial thickening with age, having dissected a centenarian who had died peacefully and found his arteries dried and constricted. He made a glass model of the aortic valve, pumped a suspension of grass seeds through it, and observed the vortices that close the valve - a mechanism not confirmed until fluid-dynamics studies in the 1960s.

He also drew a foetus in the womb with a precision nobody would match for three hundred years, and made errors where he was reasoning from Galen instead of from the body in front of him.

Water. He was obsessed with it for forty years and it is the connecting thread of everything else. He studied vortices, wave interference, erosion, the way currents behave around obstacles - and then used the same forms for hair, for drapery, for the deluge drawings at the end of his life. He proposed diverting the Arno to cut Pisa off from the sea, and surveyed the route.

Geology. In the Codex Leicester he addresses the fossils found in mountain rock. The standard explanation was Noah’s flood. Leonardo rejected it on evidence: the shells appear in distinct layers, which one flood could not produce; many are intact and paired, meaning the animals were living where they lay rather than washed there; and some show boreholes made by other creatures. He concluded that these had been sea beds, raised over enormous time. That is stratigraphy, arrived at around 1508, and it did not enter geology until the eighteenth century - because he told nobody.

Optics and light. He understood that the eye receives light rather than emitting it, described the camera obscura, worked out atmospheric perspective - that distance shifts colour toward blue and softens edges - and built sfumato out of it: transitions with no outline, in the manner of smoke. The corners of the Mona Lisa’s mouth are deliberately left indeterminate, and the shift in her expression as the gaze moves is a consequence of how peripheral and foveal vision handle low-contrast gradients.

Flight. Hundreds of pages on birds - the Codex on the Flight of Birds is a single tight study of how a wing works in air. He designed ornithopters that could not have worked, because human muscle cannot generate the power-to-weight ratio required, and the pages contain the observations that would eventually matter: that a bird’s wing is a curved surface and that the air moving over it does the lifting.

And none of it was published. Francesco Melzi, his heir, kept the papers carefully; after Melzi’s death in 1570 the collection was broken up, pages were cut out and sold, whole codices vanished, and two turned up in Madrid in 1965 after being misfiled for a century and a half. By the time the anatomy was engraved and circulated, Vesalius had published De humani corporis fabrica in 1543 and the discoveries had been made again by someone else.

This is the central fact about Leonardo as a thinker. Science is not what one person finds out; it is what a community can build on. He did the work and withheld the transmission, and so his effect on the history of knowledge is almost nil, while his effect on the history of art is total.

Where Leonardo’s notebooks are now, codex by codex

Nothing that Leonardo bound survives as he bound it. What exists is a set of collections assembled, dismembered and reassembled by other people over five centuries, and the shape of the surviving material is largely the work of one Spanish sculptor.

Melzi took the papers to his family villa at Vaprio d’Adda and kept them for fifty years. After his death in 1570 his son Orazio put them in an attic, and they began to leave the house. Pompeo Leoni, sculptor to Philip II of Spain, acquired the bulk of them in the 1580s and did the thing that determined everything afterwards: he took the notebooks apart and re-sorted the sheets by subject, pasting them into large albums, cutting some pages and mounting others so that one side is hidden. The Codex Atlanticus and the Windsor collection are both Leoni’s arrangements, not Leonardo’s.

The Codex Atlanticus. Biblioteca Ambrosiana, Milan. The largest single collection: 1,119 leaves, 2,238 pages, covering 1478 to 1519, on mechanics, hydraulics, mathematics, weapons, flight, architecture and everything else. Leoni made it; the Ambrosiana has held it since 1637, apart from a spell in Paris after Napoleon took it in 1796; between 1968 and 1972 it was taken out of Leoni’s mounts, restored and rebound as twelve volumes.

The Windsor sheets. Royal Collection, Windsor Castle. Around 600 catalogued sheets, including the entire surviving anatomical corpus, the deluge drawings, the horse studies, the maps and most of the portrait drawings. They were in England by the seventeenth century and in the royal collection by the reign of Charles II. The 153 anatomical sheets were bound in three volumes: Anatomical Manuscript A, of eighteen sheets, B, of forty-two, and C, of ninety-three.

The Paris manuscripts, A to M. Institut de France. Twelve small notebooks, the only ones that survive in something like their original form as notebooks, taken from the Ambrosiana by the French in 1796 and never returned. Manuscript B, which contains the aerial screw and the earliest flying machines, dates from the first Milan years. In the 1840s Count Guglielmo Libri, a mathematician entrusted with inspecting French libraries, cut leaves out of manuscripts A and B and sold them to Lord Ashburnham; they were recovered in 1890 and are known as the Ashburnham codices.

The Codex Arundel. British Library, Arundel MS 263. Some 283 leaves of different sizes, mostly 1480 to 1518, on mechanics, geometry, optics and hydraulics, bought in Spain by the Earl of Arundel in the 1630s.

The Madrid codices. Biblioteca Nacional de España. Madrid I, 192 folios, on mechanics and statics, and the single best source on how he intended to cast the Sforza horse; Madrid II, 158 folios, on fortification, geometry and the Arno diversion. They reached the Spanish royal library by 1712, were miscatalogued in the course of the library’s moves, and were effectively lost until Jules Piccus, an American professor of Romance languages, identified them in 1965. They hold something like a seventh of the surviving notes, and nobody had read them for something like a century and a half.

The Codex Leicester. Privately owned. Eighteen sheets folded to make seventy-two pages, written between about 1508 and 1510, on water, geology, fossils and the light of the moon. Bought by Thomas Coke in 1719, sold by the Leicester estate to Armand Hammer in 1980 for $5.1 million and briefly renamed the Codex Hammer, and sold again at Christie’s on 11 November 1994 to Bill Gates for $30,802,500. It is the only Leonardo manuscript in private hands.

The rest. The Codex Trivulzianus at the Castello Sforzesco in Milan, 55 of an original 62 folios, much of it lists of Latin words he was teaching himself. Three small Forster notebooks at the Victoria and Albert Museum in London. The Codex on the Flight of Birds, eighteen folios measuring 21 by 15 centimetres, at the Biblioteca Reale in Turin. And the Codex Urbinas Latinus 1270 in the Vatican Library, which is not Leonardo’s manuscript at all but Melzi’s compilation from it, and the source of the printed treatise on painting.

Set against that inventory, the losses are the larger figure. Perhaps three quarters of what he wrote is gone, cut out by collectors, sold by the sheet, or simply thrown away in an attic at Vaprio. The surviving notebooks contain their own summary of the method, in six words:

Wisdom is the daughter of experience. Notebooks, Richter §1150; from the Forster codices

Why Leonardo wrote backwards

Nearly every private page he wrote runs right to left in mirror script, with the individual letters reversed. Held to a mirror it reads as ordinary Italian.

The interesting explanation, and the one that has attached itself to him, is secrecy. It is almost certainly wrong, for two reasons that require no scholarship. The script is trivial to read: any flat reflective surface decodes it, and any reader willing to spend an afternoon learns to read it directly. And he did not use it when it mattered. Letters intended for other people, presentation drawings, the fair copy of the application to Ludovico Sforza: all conventional, left to right.

The dull explanation is mechanical. He was left-handed. A left-hander writing left to right with a quill pushes the nib into the paper rather than pulling it, catches the fibres, spatters, and then drags the heel of the hand back across ink that is still wet. Writing right to left with the left hand does none of that. The nib is pulled, the hand moves away from what has just been written, and the page stays clean. It is the same solution a left-handed child arrives at spontaneously, which is a documented phenomenon; the difference is that Leonardo was never made to unlearn it, because he was never in a school.

There is one complication worth adding. In 2019 conservators at the Uffizi examined his earliest dated work, the pen drawing of the Arno valley of 1473, and concluded that the mirror-writing inscription on the front is by the left hand and a second inscription on the back is by the right. On that reading he could use either hand, and the mirror script was a working default rather than a necessity.

The practical consequence for everyone since is considerable. Reversed handwriting, in fifteenth-century Tuscan, in an idiosyncratic private orthography, with no punctuation and no word division in places, on sheets rearranged by a sixteenth-century collector, is why the notebooks took four hundred years to publish and why the standard editions still disagree with each other.

Leonardo’s anatomy: thirty dissections, and a valve nobody could confirm for four centuries

The anatomy is the strongest claim he has to being something other than a painter with interests, and it is the part of the work least known outside the specialist literature.

It began as a painter’s problem. A figure painter needs surface anatomy, and Verrocchio’s workshop taught it. It ended somewhere else entirely: with a projected treatise in 120 books, an organ-by-organ programme, and a working method that anatomy did not otherwise acquire for another century.

The scale is on record from his own mouth. On 10 October 1517 Cardinal Luigi of Aragon called on him at Cloux, and the cardinal’s secretary Antonio de Beatis wrote the visit up. Leonardo showed them volumes of anatomical drawings and told them he had dissected more than thirty bodies, men and women, of all ages. De Beatis, who had no reason to invent it, records the claim flatly, and it is the only figure that comes from Leonardo himself.

There was no preservation. A body kept for a week in a Roman summer is a hazard and an ordeal, and he worked at night, in the mortuaries of Santa Maria Nuova in Florence and later in Rome, by candlelight, alone. He wrote out the discouragements himself in a note to any reader tempted to follow: that the reader will be turned back by his stomach, and if not by that then by the fear of spending the night hours in the company of quartered and flayed corpses, and if not by that then by lacking the draughtsmanship, or the perspective, or the geometry, or the arithmetic.

What he did with the dissections was a method, and the method is the achievement. He drew each structure from several fixed points of view, rotating around it the way a sculptor turns a piece, so that the reader could reconstruct the three-dimensional object from the flat page. He drew exploded views. He drew a limb layer by layer, skin, fat, muscle, tendon, bone, in sequence. He cut a structure in cross-section and drew the section. Every anatomical atlas since Vesalius uses those conventions, and he arrived at all of them before 1513 in a manuscript nobody read.

The winter of 1510. For one season he had a collaborator. Marcantonio della Torre held the chair of anatomy at Pavia, was about thirty years old, and was among the best of the new anatomists working directly from bodies rather than from Galen’s text. The two of them appear to have worked together at Pavia through the winter of 1510 and into 1511, and Leonardo may have dissected as many as twenty bodies in that period.

That winter produced Anatomical Manuscript A: eighteen sheets, densely worked on both sides, carrying more than 240 individual drawings of bones and muscles and some thirteen thousand words of annotation. It is the finest surviving anatomical work of the Renaissance, better than anything printed for another generation, and it was made in perhaps four months. Della Torre died of plague in 1511, in his early thirties. The joint treatise stopped there and nothing of it was ever printed.

The centenarian. Around 1508, in the hospital of Santa Maria Nuova, Leonardo talked to an old man who told him he was more than a hundred years old and who, sitting on his bed a few hours later, died without any movement or sign of distress. Leonardo dissected him, in his own words, to see the cause of so sweet a death.

He found the arteries dry, thin and withered, thickened in their walls, and lengthened and twisted like a snake, and the vessel feeding the heart narrowed. He attributed the death to a failure of blood supply to the heart and the lower parts of the body. And he did the second half of the experiment: in the same period he dissected a two-year-old child and recorded that everything was the reverse. That is a description of arterial disease, an account of ageing as a vascular process, and a controlled comparison, three hundred and fifty years before the pathology had a name.

The heart. He established, against Galen, that the heart is a muscle nourished by its own vessels like any other. He established that it has four chambers rather than two. Around 1509 he poured molten wax into the ventricles of an ox heart and let it set, taking a cast of the interior shape, which is a technique still in use. He identified and drew the arched band of muscle crossing the right ventricle that later anatomists would call the moderator band, and called it the chain of the right ventricle.

He also drew, carefully, the pores in the wall between the ventricles that Galen said carried blood from one side to the other and that do not exist. It is the single most instructive error in the whole corpus. In the one place where the body offered him nothing to see, he drew what the authority told him was there.

The aortic valve. The best thing he did in any field is on a few sheets from about 1512 and 1513.

The question was mechanical. The aortic valve has three cusps that must close against the whole column of blood in the aorta trying to fall back into the heart, and they must close fast, completely, and millions of times without tearing. Nothing in the visible anatomy obviously does the shutting.

He noticed that the aorta, just above the valve, bulges into three pouches, one behind each cusp: the sinuses of Valsalva, named later for someone else. He suspected they were not decorative. So he built an apparatus. He took a mould from the aortic root of an ox and cast it in glass, producing a transparent model of the real geometry, and pumped water through it carrying a suspension of grass seeds, so that the movement of the water became visible.

The seeds showed him what he wanted. Blood leaving the heart does not simply pass the sinuses; some of it curls into them, decelerates, reverses along the sinus wall and forms a standing vortex behind each cusp. That vortex pushes the cusp inwards, away from the wall, while flow is still going forwards, so that the valve is already most of the way shut before any backflow arrives. He drew the vortices, and he drew the cusps closing under them.

He was right, and there was no way to prove it for four hundred and fifty years. The mechanism was confirmed in 1968, when Brian and Francis Bellhouse published in Nature the results of a rigid model of the aortic root fitted with a flexible valve, perfused with pulsatile flow, traced with dye and filmed at speed. They found the trapped vortices, and they found that the vortices also scour the sinuses and keep clots from forming behind the cusps. Confirmation inside a living body waited for phase-contrast magnetic resonance imaging in the decades after that.

Leonardo had the question, the answer, the apparatus and the technique. What he was doing with the grass seeds is flow visualisation with tracer particles, which is a standard method in fluid dynamics and had no name until the twentieth century.

The foetus. The drawing of a foetus in the womb at Windsor, made around 1511, is the most reproduced anatomical image he made and is accurate about the posture, the single-chambered human uterus, the umbilical cord and the vessels of the cervix. It also shows a cotyledonary placenta, with the tufted attachments found in cows and not in humans, because he had dissected a cow and had not dissected a pregnant woman. The drawing is exactly as good as his access to material, which is the honest description of the whole enterprise.

And then nothing. The projected treatise was never written. The sheets went into Melzi’s boxes and then into Leoni’s albums and then to Windsor, and were not systematically published until the end of the nineteenth century. Vesalius brought out De humani corporis fabrica in 1543, twenty-four years after Leonardo died, and became the founder of modern anatomy on work that in several respects was less good.

This is the clearest case in his life of the difference between finding something out and making it known, and the two are not the same activity.

Whether Leonardo invented the helicopter, the tank and the parachute

The question is usually put as if a single answer covered all three, and it does not. The three drawings are of three different kinds and they get three different verdicts.

The aerial screw. Paris Manuscript B, folio 83 verso, drawn in Milan around 1487 to 1490. A helical sail wound around a vertical shaft, a radius of some five metres, made of linen stiffened with starch over a reed frame, with men walking the shaft round to spin it. His note says that if the instrument is turned quickly it will screw itself upward into the air.

It would not have gone up, for reasons that are worth separating rather than lumping together.

The first is power. A fit man sustains a few hundred watts. Hovering a machine of that size and weight is a job of several kilowatts. There is no arrangement of four men, or eight, that closes a gap of that order.

The second is materials. Linen, reed and wood have the wrong strength-to-weight ratio for a rotor. Made stiff enough to hold its shape against air pressure at speed, the screw becomes too heavy for the lift it can generate.

The third is torque, and it is the one Leonardo appears not to have seen at all. Nothing in the drawing counters the reaction to the spin. Turn the screw one way and the platform, with the crew on it, turns the other. Anti-torque, whether by a tail rotor or by contra-rotating blades, is the problem that kept helicopters on the ground until the twentieth century, and there is no sign of it on the folio.

What has changed recently is the assessment of the geometry itself, which used to be dismissed along with the rest. In 2022 a team at the University of Maryland built a working drone with rotors on the aerial screw’s helical form. In 2025 a computational study by Prakhar, Seo and Mittal, at present a preprint rather than a peer-reviewed paper, compared the screw with a conventional two-bladed rotor at matched lift and found that the screw drew about 42 per cent less mechanical power and produced roughly 72 per cent lower acoustic intensity per unit of lift. Its lift per unit of area is much worse; its total area is very large, so it can produce the same lift turning slowly, and rotor noise rises steeply with speed.

That is not a vindication of the machine. It is a finding that the shape was not foolish, and that what the design was missing in 1490 was not insight but an engine.

The parachute. Codex Atlanticus, folio 1058 verso, around 1485. A pyramid of sealed linen over a wooden frame, drawn small in the corner of a crowded sheet, with a man hanging beneath it. The note beside it, in modernised spelling:

Se un uomo ha un padiglione di pannolino intasato, che sia 12 braccia per faccia e alto 12, potrà gettarsi d’ogni grande altezza senza danno di sé. Codex Atlanticus, folio 1058v

In English, translated for this article: if a man has a tent of sealed linen twelve braccia on each side and twelve high, he will be able to throw himself down from any great height without injury. A Florentine braccio is about 58 centimetres, so the canopy is roughly seven metres across and seven high.

This one works. In June 2000 the British skydiver Adrian Nicholas had a canopy built to the drawing in wood and canvas, weighing about 85 kilograms, and jumped with it from a balloon at around 3,000 metres over Mpumalanga in South Africa. It descended stably; Nicholas said afterwards that the ride was smoother than a modern parachute. He cut away at about 600 metres and landed under a conventional canopy, not because the pyramid had failed but because 85 kilograms of timber and cloth arriving on top of a man at ground level is a separate and worse problem.

That last part was solved eight years later. On 26 April 2008 the Swiss jumper Olivier Vietti-Teppa was released from a helicopter at 600 metres over the airfield at Payerne under a 12-kilogram version of the same pyramid, and landed with it, completing the descent Nicholas had not.

So the parachute is a genuine case. The design is sound, the geometry is right, and the only thing wrong with it in 1485 was the weight of the available cloth.

The fighting vehicle. The drawing is in the British Museum, registration 1860,0616.99, from around 1487. A circular wooden shell like a tortoise’s carapace, reinforced with metal plates and sloped so that shot glances off, guns projecting all the way round the rim, a lookout turret on top, and men inside turning cranks geared to four wheels. A note on the sheet says eight men work it.

The gearing on the sheet does not work. As drawn, the drive turns one pair of wheels forward and the other pair backward, and the machine stands still.

Three explanations have been argued for a century. The first is sabotage: that Leonardo put the fault in deliberately, so that a stolen sheet would be useless, in the same spirit as the mirror writing. The second is that the sheet is a presentation drawing and the error belongs to whoever laid out the gears on it. The third is that it is simply a mistake by a man who in 1487 had not yet done the work on gearing that fills Madrid Codex I fifteen years later.

The third is the least flattering and the most likely, and the evidence for it is his own later manuscripts: the mechanics improves enormously, and it improves in exactly the areas where this drawing is weak. Reconstructions built for museums and television have reversed one gear train and driven the vehicle, which settles the engineering question and leaves the historical one where it was. Even corrected, it is a heavy wooden drum pushed by eight men over broken ground, with no visibility and no ventilation. It is a siege curiosity, not the ancestor of the tank.

The rest of the list. The same distinctions apply across the whole catalogue. The diving apparatus, drawn around 1500 when Venice was expecting a Turkish fleet, is a leather suit breathing through tubes to a float on the surface; it would work at the surface and kill the wearer anywhere below about a metre, because a chest cannot expand against the pressure of water through an open tube. The thirty-three-barrelled organ is a real solution to the problem of reload time and was never built. The self-propelled cart on folio 812r of the Codex Atlanticus, driven by two coiled springs with its steering programmed by pegs set in a slotted disc, was reconstructed by Carlo Pedretti and the roboticist Mark Rosheim and demonstrated in Florence on 24 April 2004, where it drove itself along a preset path; it is probably the closest thing to a programmable machine designed before the seventeenth century, and it was designed as a stage effect. The ornithopters would never have flown, for the power reason. The note on the inside cover of the Codex on the Flight of Birds about a great bird taking flight from Monte Ceceri is real; the story that his assistant Tommaso Masini launched from that hill and flew a kilometre appears in no contemporary document at all.

The general answer is that he invented almost nothing in the sense the word normally carries, because a device that is never built and never described in print does not enter the world. What the sheets contain instead is rarer and harder to categorise: a forty-year private record of one man testing, on paper and against physics, what the world would and would not allow.

Whether Leonardo was a vegetarian

Probably, and the direct evidence is a single sentence written by someone who barely knew him.

Andrea Corsali was a Florentine navigator sailing with the Portuguese in the Indian Ocean. In a letter to Giuliano de’ Medici dated January 1516 by the Florentine reckoning, he described the Gujarati of the Indus delta as people so gentle that they eat nothing containing blood and permit no injury to any living thing, and then added the comparison: like our Leonardo da Vinci.

That is the whole of the external testimony, and it is worth reading carefully. Corsali is writing to a patron who knew Leonardo well, in an offhand aside that assumes the reader will recognise the reference, which is the strongest thing about it. It is also ambiguous. It might mean Leonardo did not eat meat, or only that he would not see an animal harmed.

The notebooks support it without settling it. He kept household accounts and shopping lists, and they are almost entirely bread, wine, fruit, vegetables, mushrooms, grain and eggs. In one note he addresses the meat-eater directly and calls him not the king of the animals but the king of the beasts, a man who spares the parents in order to eat the children and has made his own gullet a burial place for every species. In another he states the same thing without the anger:

Man and animals are really the passage and the conduit of food, the sepulchre of animals and resting place of the dead, one causing the death of the other, making themselves the covering for the corruption of other dead bodies. Notebooks, Richter §847

And Vasari records that he bought caged birds in the market in order to open the cages:

taking them with his own hand out of their cages, and having paid to those who sold them the price that was asked, he let them fly away into the air, restoring to them their lost liberty. Vasari, Life of Leonardo da Vinci (trans. de Vere)

Vasari is Vasari, which means the anecdote could be true, embroidered or borrowed. But it is consistent with Corsali, with the shopping lists, and with a man who dissected animals all his life and wrote about them with a tenderness he almost never extended to people.

Why Leonardo finished so little

The complaint is as old as the record. Vasari, writing as an admirer, makes it the theme of the whole life, and puts it first:

in learning and in the rudiments of letters he would have made great proficience, if he had not been so variable and unstable, for he set himself to learn many things, and then, after having begun them, abandoned them. Vasari, Life of Leonardo da Vinci (trans. de Vere)

Vasari’s own explanation is that Leonardo’s conceptions outran any hand’s ability to execute them, which is flattering and is also the explanation of an admirer. The record supports several other causes running at once.

The commission opens a question. This is the most characteristic pattern. A job arrives; the job raises a problem; the problem is more interesting than the job. The Adoration of the Magi of 1481 turns into a study of the geometry of a crowd and is abandoned in the underpainting. The Sforza monument turns into a treatise on casting. Novellara’s report to Isabella d’Este in 1501 catches it exactly: geometry has taken him over and he can hardly bear to pick up a brush.

The technique fails. Twice on a scale that destroyed the work. The Last Supper because he would not paint in fresco, and the Battle of Anghiari because he tried to dry an oil-and-wax wall with braziers. Both failures follow directly from the same requirement, that he be able to go back and change things.

The patrons collapse. He is unlucky in a way that has nothing to do with temperament. Ludovico Sforza loses Milan in 1499 and the whole Milanese programme stops mid-sentence. Cesare Borgia’s state evaporates within a year of his taking service with it. The Florentine republic that commissioned the Anghiari mural falls. Giuliano de’ Medici, his patron in Rome, dies in 1516.

Old age and the hand. By October 1517 de Beatis is recording that a paralysis of the right hand means nothing more can be expected of the brush, though Leonardo could still draw and still teach. A 2019 paper by Lazzeri and Rossi argued from a contemporary portrait that the cause was a peripheral nerve injury rather than a stroke, since his cognition and his left hand were unaffected. Whatever it was, the last years of painting were done under a physical limit.

And a reluctance that looks like something else. He kept the Mona Lisa for at least fourteen years and never gave it to the man who commissioned it. He kept the Saint Anne and the Saint John to the end. These are not abandoned works; they are finished works he would not part with, and they suggest a man for whom delivery, rather than completion, was the difficulty.

The phrase he used about it himself recurs in the late notebooks, in a hand that has begun to shake, and Kenneth Clark called it the refrain of his old age: Di mi se mai fu fatta alcuna cosa - tell me if anything was ever done.

Where the Leonardo myth came from: Vasari, thirty years after

Giorgio Vasari published the first edition of the Lives in 1550, thirty-one years after Leonardo died, and a revised and expanded edition in 1568. He was a painter and architect of ability, a Medici courtier, and the inventor of art history as a genre. He is also the origin of a large share of what everyone knows about Leonardo, and a considerable share of that is not reliable.

He had real access. He was a Florentine working in Florence and Rome, he knew people who had known Leonardo, and for the 1568 edition he did fresh research and corrected himself in places. He had also never met his subject, wrote to a thesis about the progress of art culminating in Michelangelo, and made things up when the story needed them. Several of the most repeated Leonardo stories rest on him and on nothing else.

Verrocchio giving up painting. Vasari’s account of the Baptism of Christ is that the pupil’s angel was so much better than the master’s figures that Verrocchio would never touch colour again, out of disdain that a child should know more than he did. Verrocchio in fact ran a busy workshop for years afterwards, and the story is generally treated as invention. The angel really is by Leonardo, and really is better, which is presumably where the story started.

The shield with the monster. A peasant on his father’s estate brings a wooden buckler to be painted; Leonardo fills a room with lizards, crickets, snakes, bats and locusts, composes a chimera from them, paints it in a stench that he does not notice, and presents it in a darkened room so that his father recoils. No such shield survives, and no other source mentions one.

The head of Judas and the prior. Vasari has the prior of Santa Maria delle Grazie complaining to the duke about how little Leonardo paints, and Leonardo explaining that he still needs a face for Judas and will use the prior’s if nothing better presents itself. It is a good anecdote and it has never been corroborated. Vasari also states that the head of Christ was left unfinished, which the painting itself does not clearly support.

Death in the arms of the king. This is the durable one:

his spirit, which was divine, knowing that it could not have any greater honour, expired in the arms of the King, in the seventy-fifth year of his age. Vasari, Life of Leonardo da Vinci (trans. de Vere)

Leonardo was sixty-seven, not seventy-five, which gives some measure of the care taken. And Francis I was not at Amboise. The court was at Saint-Germain-en-Laye for the birth of the king’s second son, and a royal document was signed there on 3 May, the day after Leonardo died. The scene was painted repeatedly in the nineteenth century, most famously by Ingres, and it is a fiction.

None of which makes Vasari useless. He preserves the physical description, the presence, the strength, the beauty, the singing, the charm in company, the clothes. He preserves the bird-buying. He preserves the tone of a reputation as it stood a generation after the death, which is a genuine historical object even where the anecdotes are not. The rule for reading him is the ordinary rule for any single late source: what he says is evidence of what was said, and where he is the only witness, that is what the claim amounts to.

Why he is at this table

Seat six, in the right wing, immediately beside the empty chair. He is the counsellor who answers a question by taking it apart to see how it moves.

His function is method. When an argument turns on what a thing means, Leonardo asks how it works, and the redirection is not a dodge. He is the table’s demonstration that the question why and the question how are different tools, and that a great deal of what passes for the first is the second badly asked. Saper vedere, he called it: knowing how to see.

He is also the only counsellor here who is primarily a maker. Everyone else at the table thought, taught or governed; Leonardo built things and painted things, and his conclusions come with a residue of physical trial. When Plato says the visible world is a shadow, Leonardo has spent the night with a corpse and a candle establishing exactly where the shadow falls.

Across the table, Machiavelli - with whom he shared a winter in the Romagna in 1502 - is his closest methodological relative and his opposite in temperament. Both look at the thing rather than at what is said about the thing. But Machiavelli looked at men and concluded something bleak, while Leonardo looked at everything except men and remained cheerful. He bought caged birds in the market and released them.

Beside him at seat seven is the man he painted more than any other, and whose account of the world he never quite argued with and never quite adopted. Leonardo’s God, insofar as the notebooks have one, is closer to the order discoverable in a river than to a person; he wrote about nature as a system that never breaks its own laws, which is a theology of sorts, arrived at through hydraulics.

And the charge he answers at every episode is one he wrote out himself. Among the last notes, in a shaky hand, is the line: Di mi se mai fu fatta alcuna cosa - tell me if anything was ever done. He had every capacity and forty years and he left the Sforza horse in clay, the Anghiari running down a wall, the Last Supper flaking, the anatomy in a drawer, and a portrait he never handed over.

The reply he might make is the one his notebooks make on every page: that finishing is what is left to do once the interest has gone, and that he never once stopped.