Twenty-Four Boxes Flat To Sixty Microns. Two Of The Lids Are Swapped.

SAQQARA, Egypt — The inner faces of the twenty-four granite boxes in the Saqqara galleries are flat to about six hundredths of a millimetre.

We paid to check, over eleven nights, with an instrument that measures turbine jigs. The number held. Mean deviation across the inner side faces of all twenty-four: 0.061mm. Best 0.041mm, worst 0.11mm. Internal corners square to a mean of 0.043 degrees.

That is the figure everybody quotes and it survived measurement.

Then we measured the lids.

Nine tenths of a millimetre

The undersides of the lids deviate by 0.9mm on average — range 0.41mm to 2.1mm. Fifteen times worse than the boxes they sit on, made of the same rock, in the same room, presumably by the same people.

A workshop that can hold sixty microns does not lose it on the next slab. Something else is going on.

The shape of the error

The something else is visible the moment the two error maps are laid on top of each other.

Each lid is high exactly where its box rim is low, and low exactly where the rim is high. Across the twenty-two pairs we could measure fully, the cross-correlation between a lid's underside and its own box's rim is minus 0.86.

That is not a machining tolerance. That is a fitting tolerance.

What a mason saw in ninety seconds

We showed the error maps to Fenella Achterberg-Osei, who has cut granite for forty-one years, without telling her what the objects were or where they came from.

She looked at two of them side by side and asked whether we wanted the answer about the lapping or about something else.

"You don't grind a lid flat," she said. "You grind it down. You put your abrasive on the rim, you lower the lid, you drag it, you lift it, you look at where the paste has been rubbed off, and you go again. When it sits all the way round, you stop. Nobody in that yard cared whether the lid was flat. They cared whether it sealed."

Two surfaces that are individually mediocre and jointly excellent is the signature of hand fitting. It is also, as Mr Saito-Duval points out at the foot of this page and at some length, an extraordinary thing to have done to a twenty-seven tonne slab.

The marks

The pairing is not only in the geometry. It is written on the stone.

Twenty-one of the twenty-four boxes carry a mark in red ochre on the rim, and a matching mark on the underside of a lid. They are working marks — a stroke count, a wedge, in two cases what appears to be a name. They are the ordinary bookkeeping of a yard that had twenty-four heavy things to put together in the right order.

We photographed all of them. They are published with this story.

The two that do not match

Boxes 14 and 20 are wearing each other's lids.

The geometry says so: each lid's error map mirrors the other box's rim, at minus 0.81 and minus 0.77, while the correlation with its own box is a flat nothing. The ochre marks say so too, crosswise.

And those two boxes are precisely the two with the visible corner gaps — 12mm at the north-west corner of 14, 9mm at the south-east of 20 — which have been shown in books and lectures since the mid-1980s as evidence of damage by some unexplained force.

The clearance journal for November 1852 records lids being raised with jacks and levers and set back down selon la commodité: as was convenient.

It does not number the boxes. It does not have to.

What survived and what did not

The precision claim survived, and this desk expected it to fall. Sixty microns is real, we paid to measure it, and anybody who wants to argue about the Serapeum should start from there.

What did not survive is the inference. The best-fitting surfaces in that gallery are the ones where two stones were rubbed together until they agreed, which is the oldest trick in masonry and needs no machine at all.

And the most cited anomaly in the room turns out to be two men in 1852 putting the wrong lid back on, twice.