The pond with its trees laid flat

There is a wall painting from Thebes of a garden with a rectangular pond in the middle. The pond is drawn as seen from above. The trees around it are drawn as seen from the side, lying flat on the page, their trunks pointing outward from each bank like the teeth of a comb.

The same construction is in a medieval church plan with its elevations hinged up around it, in a schoolroom drawing of an opened cardboard box, and in every sheet-metal development the parallel field measures. One construction, four names, and the geometry is the same in all four: a rotation about a hinge line.

What the fold is #

Take a wall standing on a floor. It meets the floor along a line; rotate the wall about that line until it lies in the floor’s plane. Do that for every wall, outward, and what is left on the page is the floor with four rectangles attached to its edges.

The rotation is the whole of the construction, and it is worth saying that it is not a projection. Every other convention in this field maps a three-dimensional world onto a plane by throwing a dimension away. This one moves the world before drawing it, and then draws it orthographically — so the drawing is a true picture of a rearranged garden rather than a distorted picture of the real one.

Every length is true, exactly #

A rotation is an isometry: it preserves distances. So the length of any edge in the drawing is the length of that edge in the building, and the error is zero rather than small.

The measurement bears that out — the worst discrepancy over all eight edges checked is 0.0 × 10⁰ metres, which is exact for the same reason the parallel lines in the previous essay are exactly parallel: there is no arithmetic in the statement to leave a floor behind.

Angles within a face survive too, and for the same reason. A rectangle folded flat is still a rectangle, a right angle in the wall is a right angle on the page, and a circle painted on the wall is a circle in the drawing.

And every angle between two faces is gone #

What is not preserved is the angle between two faces, and it is not preserved in the strongest possible sense: the drawing does not record it anywhere.

Across a hinge, the plan and the wall attached to it are coplanar, so the dihedral angle in the drawing is a hundred and eighty degrees. In the garden it is ninety. The ninety was recorded only in the fold, and the fold has been undone.

That is a much stronger statement than “the angle is distorted”. A distorted quantity can be recovered if the distortion is known; an absent one cannot. And it is why the same drawing serves a garden with vertical banks and a garden with sloping ones: the sheet is consistent with any dihedral at all.

Sixteen assemblies, and four bits #

Four hinges, each of which can fold up or down, gives sixteen ways of reassembling the sheet — and the sheet is equally consistent with all of them. A reader supplies four bits from outside the marks to pick one.

That is the same accounting the same person, twice on one panel makes for a narrative panel’s reading order, and the same accounting the one thing a single view cannot give makes for a photograph’s missing length. In each case the picture is complete, consistent and short of something specific, and the something is countable.

Four bits is in fact an undercount, and the undercount is the essay’s own earlier point read carefully. The sheet is consistent with any dihedral at each hinge, not merely with up or down — so each hinge carries a real number rather than a bit, and a four-hinge sheet determines the garden only up to a four-parameter family. The sixteen assemblies are the discrete part of that freedom, the choice of side; the continuous part is the angle, and it is the larger loss.

Stated that way the fold-out sits with the rest of this collection’s ambiguities rather than apart from them. A single view leaves one real parameter free; a two-centre picture leaves three, which is its shear; a fold-out with $h$ hinges leaves $h$. In every case the picture is complete and consistent and short of a stated number of real quantities, and in every case the reader supplies them from outside the marks.

In practice a reader supplies more than four bits and does it instantly, because a garden’s trees grow upward and a box’s flaps fold inward. The convention leans on that entirely: it is legible because the world it depicts has only one plausible assembly, and it fails exactly where the assembly is genuinely ambiguous.

The same drawing as a cutting pattern #

Read from the workshop rather than from the wall, the fold-out has an obvious purpose: it is a cutting pattern, and a cutting pattern needs exactly the lengths and none of the angles.

That is why the parallel field already has the construction under another name. The drawing and the development is about laying a curved sheet flat so it can be cut from stock, and a picture that can be printed is about the surfaces for which that is possible at all. A box’s development and an Egyptian pond are the same object, made by people with entirely different intentions, because the geometry serves both.

There is a limit worth naming here, because it separates the flat case from the curved one. A box unfolds exactly because its faces are planes and a plane is developable. A dome does not: laying it flat requires stretching, and the amount of stretching is fixed by its curvature rather than by anybody’s skill. A fold-out of a domed room is therefore not a fold-out at all — it is a projection with a choice of distortion, and the whole curved field is about which choice.

Why this is aspective, and where it goes further #

Assembled from several views establishes the general Egyptian construction: a figure built from the views that identify each part best, shoulders from the front and head in profile, glued at the joints. The fold-out is that principle applied to a scene rather than to a body, and it is the case where the gluing has an exact geometric description.

The difference is worth stating. An aspective figure’s parts are drawn from different directions and joined at seams that are not rotations of anything — the shoulders and the head are two orthographic views, and no rigid motion carries one into the other. A fold-out’s parts are related by rotations about lines that are actually in the drawing, so the assembly is a definite operation rather than a convention of joining.

That makes the fold-out the measurable member of the family, and it is why the sixteen assemblies can be counted where an aspective figure’s cannot.

Where the hinge is the only record #

The strongest version of the essay’s claim is worth isolating, because it is the one that generalises past this convention.

A quantity survives a drawing if it is recorded somewhere in the marks, and the test for whether it is recorded is whether two scenes differing only in that quantity would produce different marks. That is the same test this collection applies to every ambiguity it measures, and it is the reason a dihedral counts as absent rather than as distorted: two gardens whose banks meet the water at different angles fold flat to the identical sheet, so no reading of the sheet, however careful, separates them.

A quantity survives a drawing if it is recorded somewhere in the marks. A length survives the fold because it is drawn as a length. An angle within a face survives because it is drawn as an angle. The dihedral does not survive because there is no mark in the drawing whose measurement returns it — not a distorted version of it, not a version needing a known correction, nothing.

That distinction runs through the whole collection under a different name. What a projection destroys lists length, angle, area and the ratio of lengths among the casualties of a perspective, and the cross-ratio among the survivors — and the reason the cross-ratio survives is that it is recorded in the marks, as a ratio of ratios that the projection carries through. The fold-out’s ledger is the same kind of list for a different operation, and reading the two side by side is what makes the field a comparison rather than a catalogue.

There is a practical corollary. Given a drawing and a question, the useful first move is to ask which mark would answer it — and if no mark would, the answer has to come from outside. That is the same move the metrology field makes when it asks which single fact closes a measurement, and it is the reason both fields end up counting bits.

The reader’s own hinge #

A reader who wants to check all of this needs a cereal box and a pair of scissors, and the checking takes a minute.

Cut along the vertical edges and lay the box flat. Measure any edge on the flattened sheet and the same edge on a second, unopened box: they agree, and they agree exactly rather than nearly, because nothing was stretched. Then measure the angle between a side panel and the base on the flat sheet — a hundred and eighty degrees — and on the unopened box — ninety.

That is the whole essay, performed. And the last part is worth doing too: hand the flat sheet to somebody who has not seen the box and ask them to fold it. They will get it right, and they will get it right by using the printing rather than the geometry, which is the four bits arriving from outside the marks.

Two hinges meeting at a corner #

There is a detail the simple account glosses over, and it is where the convention gets genuinely awkward.

Two adjacent walls meet each other along a vertical edge as well as meeting the floor along their own hinges. When both are folded outward, that shared vertical edge is torn: it appears twice in the drawing, once on each wall, and the two copies are in different places. Every real net has this property and every real box shows the seam.

So a fold-out sheet contains a set of edges that are the same edge and are drawn apart, and nothing in the sheet says which pairs they are. A reader identifies them by length and by adjacency, and a cardboard box helps by putting a glue tab on one of them.

That is an extra thing supplied from outside the marks, and it is not in the four bits above — the four bits are the fold directions. The pairing of torn edges is a separate count, and for four walls it is four more pairs to identify. The convention survives because the identification is nearly always forced, and the case where it is not forced is exactly the case where a net is hard to fold: many faces, similar lengths, no printing.

What a fold-out is not #

Three things it is often taken for, and it is none of them.

It is not a perspective, and the absence is total rather than partial: there is no station point, no horizon, no diminution, and no vanishing point, because the drawing is orthographic after the rotation. A picture with no eye is where the collection prices that absence generally.

It is not an exploded view, which separates parts along the directions they assemble in and keeps their orientations. An exploded view preserves the dihedral angles and spends the contact; a fold-out preserves the contact and spends the dihedral. Opposite trades, easy to conflate, and the ledger tells them apart.

And it is not an elevation set. A plan with four separate elevations printed beside it carries the same lengths, and it carries the assembly too, because each elevation is labelled with the wall it belongs to. The fold-out throws that label away and replaces it with adjacency, which is why it needs the four bits and the elevation set does not.

The boundary, stated #

Plane faces, straight hinges, and a fold that does not overlap.

Curved faces cannot be laid flat without stretching, as above. Hinges that are not straight lines are not rotation axes and the construction has nothing to rotate about. And a fold-out of a shape whose faces are large relative to the plan runs its walls into each other on the page — which is a real constraint on the convention, and it is why Egyptian ponds are drawn with short trees and wide banks rather than tall trees and narrow ones.

There is a fourth, quieter one. The construction assumes the walls fold outward, and nothing in the sheet says so. A wall folded inward would lie over the plan and be hidden, so the convention picks outward for legibility rather than for geometry — which is one of the four bits being supplied by the drawing after all.

The convention in a modern drawing office #

The construction did not survive as a curiosity. It is the standard first sheet of any sheet-metal job, and the vocabulary a drawing office uses for it is worth setting beside the Egyptian one, because it says which properties the trade cares about.

A development is drawn to give the true lengths of every edge and the true shape of every face, and those are exactly the two things the fold preserves. Nothing on a development sheet is a dihedral angle; bend angles are written as notes beside the fold lines, in words, because the drawing has no way to carry them. That note is the four bits, made explicit and printed.

A ruler on an isometric drawing is where this collection prices the same distinction for the other parallel systems — which directions a ruler may be laid along and which it may not. A development is the extreme member of that family: every direction on the sheet is true, because every face is drawn in its own plane, and the price is paid entirely in the assembly.

So the Egyptian painter and the sheet-metal draughtsman made the same trade for the same reason, two and a half thousand years apart, and neither of them needed the geometry stated to make it. What the geometry adds is the count.

What is measured here #

Three numbers, and two of them are exact.

Every edge in a fold-out is the length it is in the solid, to 0.0 × 10⁰ metres over all eight edges checked. The dihedral angle across every hinge reads a hundred and eighty degrees in the drawing and is ninety in the solid, so ninety degrees is spent. And four hinges admit sixteen assemblies, which is four bits a reader supplies to fold the sheet back up — a count rather than an impression.

The short version #

A fold-out rotates every face into the plane of the plan about the line where it meets it. A rotation is an isometry, so every length and every angle within a face survives exactly — the residual is zero rather than small. What does not survive is the angle between two faces: it was recorded only in the fold, and the drawing reads a hundred and eighty degrees across every hinge.

Four hinges give sixteen assemblies and the sheet is consistent with all of them, so a reader supplies four bits from outside the marks. In practice the world supplies them, which is why a cereal box is easy to reassemble and why an Egyptian pond is legible three thousand years later.