Ceramics · iron redox · show the check

The Black That Was Never Paint

When the painter set the brush down, nothing on the vase was black. The gloss carries no black colourant at all: it is a clay slip, and its colour is decided later by how much oxygen is in the kiln, and it can be undecided again. Fire one below. The oxide boundary the pot crosses is computed from NIST-JANAF thermochemistry, and the vase changes colour at the instant the firing path crosses it. Then the parts the field is still arguing about, which are more of them than the museum labels suggest, including the firing temperature everyone repeats and nobody appears to have measured.

An Athenian amphora is among the most recognisable objects in the ancient world, and the first thing most people assume about it is wrong. The black is not a pigment. Nothing in it is black when it goes into the kiln, nothing black is added, and if you fire the same pot a second time in the wrong air you can take the black off again without touching it. What the painter laid down was a suspension of very fine clay. What made it black was the removal of oxygen from the room.

The consequence is stranger than the fact. A painter working on a leather-hard pot in the Kerameikos around 520 BC was brushing on a slip that looked, wet, much like the pot beneath it. The design existed as a difference of sheen and a faint difference of tone. Everything that makes the object famous arrived hours later, in the dark of a closed kiln, and arrived for the whole vase at once. The craft is not in making the black. It is in making it stay on only half.

Fire one

Three decisions, which are the three a potter actually makes. How hot the kiln gets while the vents are shut, how rich the fire runs while they are shut, and how hot the pot still is when the vents open again. The dot on the right is the kiln, moving through temperature and oxygen pressure. The line it crosses is the boundary between the red iron oxide and the black one, computed rather than drawn.

Iron has two colours and the kiln picks one

Almost all common clay carries iron. Fired in air it finishes as haematite, Fe₂O₃, the red of every flowerpot and roof tile you have seen. Take the oxygen away and the same iron finishes as a black spinel. Nothing is added and nothing is removed but oxygen, and the oxygen does not have to be taken out of the clay so much as out of the room.

Where the switch happens is not a matter of opinion. Six moles of haematite give up one mole of oxygen to become four of magnetite, and thermodynamics fixes exactly what oxygen pressure that reaction sits at for any temperature. Integrating the NIST-JANAF heat capacities and entropies for haematite, magnetite and oxygen puts the boundary at at an oxygen partial pressure of bar. Open air is bar. The kiln's whole job, in one sentence, is to travel orders of magnitude in oxygen pressure and then come back.

That is the number worth sitting with. Not a change of pigment, not a change of clay, not even principally a change of temperature: a change in how much oxygen is present by a factor of more than a million, produced by blocking a hole and throwing green wood on a fire.

Two independent routes to the same line

One computed curve is a claim, so the page carries a second that shares no arithmetic with the first. The line above comes from integrating tabulated heat capacities and entropies. The check comes from an experimental fit to the magnetite-haematite equilibrium used in geochemistry (Fegley 2013, as implemented in the Australian National University's oxygen fugacity calculator), which was calibrated against furnace measurements rather than against thermochemical tables.

Across the range a kiln can reach they differ by at most in log₁₀ of the oxygen pressure, a factor of in the pressure itself. That is a real disagreement between two good sources and this page will not hide it. It is also of no consequence whatever to a potter, which is the point: the uncertainty in where the line sits is about times smaller than the swing the kiln performs across it. A target that wide cannot be missed by an error that narrow.

Why heat alone will not do it

Haematite does come apart on its own if you get it hot enough in ordinary air, with no reducing gas at all. The same computation says where: . Greek kilns did not approach it and most modern ones do not either. So the blackening cannot be an accident of heat. It has to be supplied, deliberately, by making the fire burn dirty.

The window a potter can actually hit

Once a wood fire has eaten the free oxygen in a closed kiln, what sets the oxygen pressure is how much of the carbon leaves as carbon monoxide rather than carbon dioxide. Both gases stand in equilibrium with oxygen, so a CO:CO₂ ratio names an oxygen pressure exactly. Running that combustion equilibrium at from the same tables gives the two ratios that bound the craft.

One part of carbon monoxide in of carbon dioxide is already enough to take the iron black. One part in is where magnetite itself starts to give way and the pot begins to bloat. Between them lies a working window wide, and it stays between across the whole 850 to 950 °C band the measurement literature supports, so it is a property of the craft rather than of any one temperature.

This is the quiet explanation for something that ought to be puzzling: how a technique that sounds impossibly delicate was in fact routine production for two centuries, turned out by the shipload, by people with no thermometers and no gas analysis. The target is not narrow. It is more than three orders of magnitude wide, and its lower edge sits so close to a clean flame that any wood fire in a shut kiln falls past it almost at once. Getting the black was never the hard part. Getting the red back was.

Why one layer keeps the black and the other gives it up

Here is the whole difficulty. During the reduction the pot goes black all over, gloss and ground alike, because the atmosphere reaches everything. Then the vents open, air returns, and the iron in the body goes back to red. If the gloss followed it there would be no picture. It does not follow, and the accepted reason is that by then it is shut: the fine layer has vitrified, its pores have pinched off, and oxygen has no route in.

Falling through water is a sieve no sieve can match

The gloss is made from the very fine end of a clay, separated by suspending it in water and waiting. Large particles fall fast, small ones fall slowly, and if you pour off the top after a while you keep only what is still up there. This is not folklore, it is Stokes' law, and it gives a cut size you can compute exactly. It is also the field's own instrument: when archaeometrists report a gloss particle size they report a Stokes equivalent spherical diameter.

What the published recipes were actually selecting

Modern attempts to reproduce Attic gloss report a settling time, and the times are wildly different from one another. Nobody, as far as this page could find, has put them on one axis. The same law the field uses to report particle size will say what each protocol selects, so here they are, side by side.

The spread is more than an order of magnitude, and the middle row is the interesting one: the single documented recipe that is agreed to have worked, Aloupi-Siotis's Thetis reproduction, specifies a fraction under 330 nm, and a still column would need to deliver that. One laboratory settled for twenty hours, which is nowhere near it. One reconstruction settled for six months, which is far past it.

Two honest cautions, because this is the page's own arithmetic rather than anybody's measurement. Depth is unreported in every one of these papers, and depth enters the answer directly, which is why each row is a band over tub depths from 50 to 200 mm rather than a single figure. And a real tub is not a still column of separate particles. Clay in plain water flocculates and falls as clumps, so settling only sorts by size if the suspension is dispersed to begin with. That is not a footnote to the levigation question, it is the same question: whether the Athenians used a dispersing agent decides whether their tub was a sieve at all, and the field has not settled it.

And a fine powder sinters thousands of times faster

Now the mechanism that shuts the door. Sintering is how a powder, held hot, welds itself into a solid: particles grow necks, the pores between them shrink, and eventually the channels linking one pore to the next pinch off. A body with connected pores lets the kiln atmosphere reach every grain in it. A body whose pores have pinched shut does not.

Conway Herring proved in 1950 that this obeys a scaling law of brutal simplicity. Take two powders identical in every way except that one is k times finer. They reach the same stage of sintering in times related by t₂/t₁ = (d₂/d₁)ⁿ, where n is 1, 2, 3 or 4 according to which transport mechanism carries the material. Herring's law does not say which mechanism runs. It says that whichever one runs, fineness wins, and wins as a power.

A potter who levigates his gloss clay to a few times finer than his body clay has not merely made it smoother. Under lattice diffusion he has made it sinter about a thousand times faster; under grain-boundary diffusion, ten thousand. So there exists a band of temperature and time in which the gloss has vitrified and closed and the body has barely begun. That band is where an Attic vase is fired, and everything else follows. The gloss goes black with the pot, seals while it is black, and then cannot hear the kiln any more. The body, still open, hears everything, and turns back.

Notice how much the potter had to understand for this to work, which is nothing. He had to make the design clay finer than the pot clay, which he wanted to do anyway because it brushes better and dries glossier. The physics did the rest, and it was twenty-four centuries before anyone could have said why.

There is one more requirement, and it is the reason the technique needs a particular clay rather than any clay. The gloss must be low in calcium. Calcite in a clay gives off carbon dioxide as it heats, and that gas blows pores through exactly the layer that was supposed to seal, letting oxygen back to the iron and turning the black red. A gloss too rich in lime does not merely come out wrong, it flakes.

Black-figure and red-figure are one technology

The two great styles of Athenian pottery are usually presented as a succession, as though the second replaced the first the way oil replaced tempera. They are not different techniques. The clay is the same clay, the slip is the same slip, the firing is the same firing. The only thing that changes is which side of the outline the brush is on.

In black-figure the painter puts slip on the figures and leaves the background bare. In red-figure, from around 530 BC, he paints the background and leaves the figures bare. Press the two buttons on the kiln above and fire it again: the same three stages, the same boundary crossing, the same instant of change, and a completely different object.

What red-figure buys is not colour, it is line. On a black-figure vase interior detail has to be cut through the dried slip with a needle, which gives a hard, even, scratched contour. On a red-figure vase the detail is drawn with a brush in slip on bare clay, which gives a line that can swell and taper and lift. The style changed because the drawing changed, and the drawing changed because the painter stopped fighting the material.

The temperature everybody repeats

Look up how hot an Athenian kiln ran and you will be told 945 °C. It is a strikingly precise number for a wood-fired kiln in the sixth century BC, and that precision is what made this page go looking for where it comes from.

It comes, as far as could be traced, from two places. The first is a widely syndicated article on the World History Encyclopedia, credited to the Trustees of the British Museum, which states that The temperature in the kiln continued to rise to around 945º C. That article carries no bibliography, no references and no citations of any kind. The second is the English Wikipedia article on three-phase firing, which says the temperature was held probably at about 945 °C (1,733 °F).

The Wikipedia article attributes its account to Noble 1960 and Schumann 1942, and both citations are defective in ways that are checkable in a minute. It gives Noble as AJA 63 (1960); the American Journal of Archaeology's own table of contents puts The Technique of Attic Vase-Painting in volume 64, issue 4, pages 307 to 318. It gives Schumann's paper as volume 32 of the Berichte der deutschen keramischen Gesellschaft, where the archaeometric literature consistently gives volume 23. This page verified the Noble reference directly against the journal and reports the Schumann conflict without adjudicating it, not having held the volume.

What the measurement literature gives is a band, not a number. Estimates from gloss microstructure put the maximum at 850 to 950 °C, and the best-documented successful modern reproduction fired at 890 to 900 °C. Nothing in the archaeometric papers consulted for this page contains 945 °C at all. The likeliest explanation is that it is a prescriptive target from a mid-century replication recipe that has been laundered, through museum copy, into sounding like a measured property of ancient pots.

Here is the part that makes it worth reporting rather than merely correcting. The physics does not care. Across the entire 850 to 950 °C band the haematite boundary moves by in log₁₀ oxygen pressure, and even at the hot end it still sits orders of magnitude below open air. The CO:CO₂ window stays within a factor of two of itself. Every conclusion on this page survives the whole band, which is exactly why nobody noticed for decades that the number had no source: it never had to be right.

What is genuinely unsettled

Museum labels present this technique as solved. The specialist literature does not. These are live disputes between serious groups, as of the most recent work this page could reach.