I. The Snow of 20 January 1494

A life-size snow figure being cut with a wooden blade in the arcaded courtyard of a Florentine palace, two onlookers in fur-lined cloaks standing out in the trodden snow beside it
The courtyard of the Medici palace on the Via Larga, January 1494. He was known in the house already — Lorenzo had taken him in around 1490 to work in the sculpture garden at San Marco under Bertoldo di Giovanni — and the snow was Piero's way of calling him back.

Nothing in the surviving record puts Michelangelo on a beach. It does put him, once, in a courtyard full of snow. Heavy snow fell on Florence on 20 January 1494, and Piero di Lorenzo de’ Medici, who had inherited the management of the city twenty-one months earlier and was making an unusually poor job of it, sent for the sculptor and asked him to make a statue out of it. Vasari gives the episode a single clause: he caused him to make in his courtyard “a statue of snow, which was very beautiful.”

That is the whole of it. The sculptor was six weeks short of nineteen. Vasari does not say what the figure was, and neither does Condivi, whose 1553 life was dictated in all but name by Michelangelo himself and corrects Vasari wherever Michelangelo felt correction was owed. Between the two of them the commission was preserved and the subject was lost. By the first week of February there was a wet patch on the pavement; by November Piero had surrendered Pisa, Livorno and four fortresses to Charles VIII, been run out of Florence for it, and Michelangelo was in Venice and then Bologna.

So there is a precedent, and it is a real one. He once made a thing everybody knew would be gone by Candlemas, for a patron who asked for it, and made it well enough that Vasari thought it worth a line fifty-six years later. Every claim about the sandcastles rests on that fortnight of Florentine weather, and none of them says so.

II. What Sand Will Actually Hold

Two bare hands pressing a mound of dark wet sand flat on a beach, a wooden pail and an iron blade beside them
Packing, not building. Grain shape decides as much as water does: a mix of fine and coarse angular grains locks together, while grains rounded off by long transport slide past one another, which is why a little silt in the sand helps and beautifully smooth sand builds badly.

Sand is not a soft marble. It is a granular solid that borrows every scrap of its strength from water, and it borrows far less than seems reasonable. Pakpour, Habibi, Møller and Bonn put numbers on it in Scientific Reports in 2012 and found the optimum at a liquid volume fraction of about one per cent — one hundredth, and no more. What holds a wet column up is a mesh of capillary bridges between grains, and what brings it down is not crumbling but elastic buckling under its own weight, the same failure a slender steel section suffers. Maximum height scales as the two-thirds power of the base radius; the measured exponent came out at 0.7. A column twenty centimetres in radius can stand about two and a half metres, against an older rule of thumb that put the ceiling near twenty centimetres whatever the base.

The method is right, too, which is the part that ought to trouble anyone inclined to laugh. Competition sand is packed first and cut afterwards: it goes into bottomless timber forms in layers, is rammed down wet until the block will stand without them, and is then worked from the top with flat blades until a figure comes out of it. That is exactly the sequence a marble carver follows, and exactly the opposite of what a child with a bucket and a shovel does. On the two questions that matter to a sculptor — a compacted mass, and a subtractive tool — a modern beach competition and a sixteenth-century Florentine workshop agree completely.

They disagree about the calendar. A marble figure is a bet on the next five hundred years. A sand figure is a bet on the next six hours, and the odds are set by the moon.

III. Governor and Procurator General of the Fortifications

Red chalk and ink study seen from above: a four-armed earthwork like an X, its arms curving out to four horned points above a dry ditch
The shape of the argument. His bastions curve where everyone else's were ruled straight, and the horns reach far enough forward to flank each other. He is the man who put the type to work, in the earthworks of Florence; the refining is credited to Baldassare Peruzzi and Vincenzo Scamozzi later in the same century, and Vauban made a system of it a hundred and fifty years after that.

In the first months of 1529 word reached Florence that Clement VII intended to bring his family back and that the Emperor’s army would do the carrying. The Medici had been expelled on 17 May 1527, eleven days after the sack of Rome began. The republic that replaced them set up a committee of nine, the Nove della Milizia, to finish the defences begun under the Medici in 1526 and never completed, put Michelangelo on it, and then gave him the office in full: governor and procurator general of the fortifications.

Sixteen sheets of what he drew survive, in the Casa Buonarroti on the Via Ghibellina, and they are the strangest things in his hand. Inventory 13 A — pen and ink, watercolour and red pencil, 410 by 568 millimetres, drawn about 1529 for the Porta al Prato of Ognissanti — sets a city gate behind a pair of horned, curving salients that reach out from the wall like a claw. Nothing else in the period looks remotely like it. The bastion fort as the rest of Europe was drawing it in the 1520s is a ruled polygon, all straight lines and dead angles killed by arithmetic. Michelangelo’s arms bend.

And all of it was to be built out of dirt. The angled bastion had come in because fire cannot be bent round a curve: a round tower left a skirt of dead ground at its own foot that nothing else in the defences could sweep, and pulling the tower out into a diamond point put every inch of that ground under flanking fire from somewhere along the wall. By the 1530s the whole continent was raising them in packed earth revetted with brick rather than stone, for the plain reason that brick does not shatter when a ball strikes it and ashlar does. The state of the art in siege engineering, in the year Michelangelo held the office, was a damp heaped material rammed hard and cut to sharp angled faces, standing up because it was compacted and not because it was strong. The man had already designed a sandcastle. He had simply designed it for a city with no beach.

Very little went up: the projects were complex, the republic short of months, and what was built has gone. The one piece anyone can still stand beside is at San Miniato al Monte, where he turned the hilltop campanile into an artillery position and hung it about with mattresses to soak up shot. The siege closed on 24 October 1529 and lasted until 10 August 1530, roughly eight thousand men holding the circuit. Michelangelo missed the opening. He had left on 10 September after telling the government to its face that it was being sold, and came back in the middle of November to finish the work he had walked out on.

IV. The Argument From Ashes

An old man in a dark gown feeding loose buff sheets into an iron brazier on the floor of a bare workshop
Rome, February 1564. He had done it before and at a distance: in 1518 he instructed Leonardo Sellaio to burn certain cartoons stored at his house in Rome, and on 5 February 1519 Sellaio wrote back that almost all of them were gone, and that he had grieved over doing it.

Everything above is on paper. The case for the sandcastles rests on paper that is not, and that argument runs better here than for almost any other artist of the century, because Michelangelo genuinely did burn his own work, repeatedly and on purpose. Pietro Aretino wrote to him more than once to beg for one of the sheets that normally went into the fire. In the days before he died in Rome on 18 February 1564 he burned drawings, sketches and cartoons by the armful, and Vasari’s explanation is that nobody was to see the labour and think him less than perfect.

The Getty puts the survivors at about six hundred sheets against the twenty-eight thousand he would have made at one a day. Rather more than two hundred of the six hundred never left the family, and Casa Buonarroti’s own account of how they came through is a list of near-misses. So the absence of a sandcastle drawing proves nothing whatever. That is true, and it is where every argument of this shape stops, and it is the wrong place to stop.

Six hundred out of twenty-eight thousand is a two per cent sample, and a two per cent sample is very good evidence of what was routine and none at all of what was rare. What it contains is instructive: fortification studies, architectural elevations, studies for the vault, and a household shopping list on which he drew the herring, the anchovies and the two dishes of fennel beside the words, because the servant sent to fetch them could not read. A man who illustrated his own grocery order would have left something behind if he had spent his summers on a beach. He left nothing.

V. Carrara, and the Figure Already in the Stone

A squared white marble block roped to a timber sledge on a broad quarry road cut level along the mountain, six oxen in the traces and men bracing the runners with levers
Coming down, which is the expensive part. The block that became the David left the mountain by ox cart, went to the sea, and came up the Arno on a barge; the marble bill for a colossal figure was always as much haulage as stone.

The second objection is temperamental rather than documentary. Carrara marble did not turn up at his door. Leo X handed him the design of a marble front for San Lorenzo in the autumn of 1516, the contract was drawn on 19 January 1518, and on 10 March 1520 Michelangelo himself went and registered its rescission. He spent the better part of those three years in the mountains above Carrara and at Pietrasanta, choosing blocks, specifying the dimensions to which they were to be cut, and arguing about the roads that would bring them down. The church still wears the bare brick face it had when he started.

Stone repaid him unevenly, and it mattered to him in a way it could not matter to a man working in sand. The David is 5.17 metres of second-grade bianco ordinario out of the old Roman Fantiscritti quarry, a block two other sculptors had already given up on. Four unfinished Prisoners stand in the Accademia still half inside their blocks, in the condition art history calls non finito and which is only legible as a condition because the surplus stone is still attached and can be seen not to have come off.

That is the difference no missing evidence can close. Sand keeps no record of its own failures: a collapsed sand figure is a mound, and a mound is what you started with. A failed marble figure stands in a Florentine gallery with a label on it while a million and a half people a year file past, looking at the part that was never removed.

VI. Sonnet XV

Michelangelo wrote more than three hundred sonnets and madrigals, about sixty of them addressed to men, and one settles the question more firmly than any drawing could. John Addington Symonds put it into rhymed English in 1878 under the title The Lover and the Sculptor:

The best of artists hath no thought to show
Which the rough stone in its superfluous shell
Doth not include: to break the marble spell
Is all the hand that serves the brain can do.

There is a complete theory of sculpture in those four lines, and it is a theory of removal. The figure is inside the block before the sculptor arrives; the work is getting the rest of the block off it, and the virtue lies in knowing what is surplus. Everything else follows — the months in the quarries choosing which block held which figure, the Prisoners abandoned when the answer stopped being obvious, the lifelong preference for cutting over modelling.

The agreement conceded earlier — a compacted mass, a subtractive tool — turns out to be an agreement about method only, and method is the smaller half of what he is describing. A heap of sand has no superfluous shell. Nothing is inside it that was not shovelled there an hour ago, and no cut can be wrong because nothing in it is surplus. Whatever wet sand is good for — and by the arithmetic it is good for two and a half metres — it cannot be the thing he described. He could have raised a castle on any beach between Livorno and Viareggio and, by his own account of what a sculptor is for, he would not have been making a sculpture.

Which leaves the snow, and the snow is the better story anyway. It stood in a courtyard on the Via Larga for as long as the weather held, nobody troubled to write down what it was, and it remains the only work of his made in the certain knowledge that it was going. It has the further advantage of having happened.

Plates: the snow, the stone and the earthworks

The stone that stayed, the earth that did not, and one tide coming in.

A hilltop stone bell tower hung all round with lashed woollen mattresses, gunners at a small cannon on its top stage, the city below
The tower he padded was itself nearly new: the original campanile of San Miniato fell down in 1499 and its replacement went up in 1523 and was never finished. The eagle on the church front below belongs to the Arte di Calimala, the cloth merchants' guild, which had paid for the building's upkeep since 1288.
Labourers digging a long angled earth bank outside a city wall, barrows and baskets of spoil moving up the slope
Clearing the ground cost Florence a ring of its own buildings — the church of San Gallo and a number of outlying convents and monasteries were pulled down so the approaches could be seen and swept. The army coming for them was Imperial and Spanish, under Philibert of Chalon, Prince of Orange.
A long weathered marble block on its side in a walled cathedral workyard, a shallow figure roughed out along it, weeds at its base
The David block as it stood for most of a working lifetime. Agostino di Duccio began cutting about 1464 and stopped when Donatello died in 1466; Antonio Rossellino was given it in 1476 and the contract was rescinded. An inventory of 1500 describes a figure of marble called David, badly blocked out and supine.
A white marble group of a seated woman holding a dead man across her lap, lit by one diagonal band of daylight in a bare chapel
The Pieta of 1498–99, 174 by 195 cm, cut for the funeral chapel of Cardinal Jean de Bilheres. It is the only work Michelangelo ever signed, and Vasari says he cut his name across the sash after overhearing the piece credited to Cristoforo Solari, then regretted the vanity and never signed another thing.
A man standing upright on a timber scaffold under a barrel-vaulted chapel ceiling, head tilted back, brush raised
He painted the ceiling standing with his head back, not lying on a plank. Bramante's rope-hung scaffold was refused because it would have left holes in the finished vault, so Piero Roselli built a freestanding one. Over three hundred figures, unveiled 31 October 1512, and a verse letter to Giovanni da Pistoia complaining of the goitre it gave him.
Four half-carved marble figures still joined to their rough blocks along a workshop wall, chippings on the floor, mallet and chisels on a bench
Four Prisoners are in the Accademia — the Young, the Bearded, the Atlas and the Awakening Slave — and two more, the Dying and the Rebellious, are in the Louvre. All were cut for the tomb of Julius II, a commission that ran from 1505 to 1545 and ended up in San Pietro in Vincoli rather than St Peter's.
An open walnut chest of loose buff sheets in a shuttered room, a hand lifting the topmost one clear, a brass candlestick beside it
What the family kept. Casa Buonarroti holds over two hundred sheets and says, without apparent strain, that this was the largest such collection in the world then and still is. His heir Leonardo gave the drawings to Cosimo I about 1566 and Cosimo II handed part of them back; at the end of the eighteenth century Filippo Buonarroti sold some to the painter Jean-Baptiste Wicar, and in 1859 Cavalier Michelangelo Buonarroti sold more to the British Museum; the survivors went to the Uffizi in 1960 for restoration and came home in 1975.
A vast stepped white quarry face cut into a grey mountain, small figures working at its foot with bars and wedges, oxen waiting on the track
More marble has come out of the Apuan Alps above Carrara than out of any other place on earth — over six hundred and fifty quarry sites. The Romans knew the stone as marmor lunense and built the Pantheon and Trajan's Column from it. The statuario grade holds the finest detail and takes the highest polish, and the David is not made of it.
A mound of wet sand collapsing at high water, the sea running through a breach in its side, a single figure standing well back
The one test no marble ever has to sit. Too little water and the capillary bridges never form; too much and the mass behaves like a landslide. Modern competition sand is trucked in and packed in timber forms — the 15.1 m tower at Myrtle Beach in 2007 took ten days and three hundred loads, and the 2019 record at Rugen used eleven thousand tons.
A tall bare brick church front above an empty paved square, rough unfaced brickwork and blocked openings across its width
San Lorenzo in Florence, still faced in nothing at all. Leo X gave Michelangelo the design in the autumn of 1516, the contract followed on 19 January 1518, and its rescission was registered on 10 March 1520. A wooden model, 216 by 283 centimetres, is all that three years and a quarry road produced. In February 2007 a computer reconstruction of the design was projected onto the brick so the city could argue about finishing it.