The collection

Every essay — page 2

One idea per essay, ordered so that the earlier ones set up the later ones — but nothing here depends on being read in sequence.

Axioms and construction · Flat-folding · Designing a base · Tessellations · Rigid folding · Curves and material · What it costs to know · Who found it, and when · Folding nobody designed

Tessellations

One vertex repeated until the sheet stops being a sheet and becomes a material.

at every vertexthree of one, one of the other15 interior vertices, all identicalwhat the sheet gainsone degree of freedom, not manyit opens and closes in both directions at oncea negative Poisson's ratio22 mountain and 16 valley creases · 6.2 sheet-widths of foldingmountainvalleyraw edge

One vertex, repeated

Take a single flat-foldable vertex and tile the plane with it. The sheet stops being a sheet and becomes a material — with a stiffness, a packing behaviour and a Poisson's ratio that the paper never had.

6 figures
nearly flatwidth ×0.91 length ×0.98ν = -0.22half closedwidth ×0.66 length ×0.88ν = -0.52nearly packedwidth ×0.45 length ×0.55ν = -2.93both dimensions shrink together — pulling it open in one direction opens it in the other

A sheet with one freedom

A Miura-folded sheet can move in exactly one way. Pull it open in one direction and it opens in the other — a negative Poisson's ratio, arriving entirely from the crease pattern and not at all from the paper.

7 figures
4 corners, all alikesectors 90°, 90°, 90°, 90°two equal pairs, so no sectoris strictly the smallestthe assignment256 of 4096 fold6 mountain, 6 valleythe ring takes two lettersthe panels can be orderedwhat is checked4 interior verticesand not the tilinga twist of radius 0.17 sheet-widths12 creases, 4.70 sheet-widths of foldingmountainvalleyraw edge

A square that turns

A twist is a small polygon that rotates as the sheet closes around it. The geometry is forced rather than designed — Kawasaki fixes one sector, the big-little-big lemma forbids a strictly smallest one, and what is left is the pattern Ron Resch was drawing in the 1960s.

6 figures
00.20.40.60.80123how far the sheet is closed−ν, so every curve shown is a negative ratioMiura, slant 0.25Miura, slant 0.42Miura, slant 0.6accordionexactly zerothe same paper,three behaviours,chosen by the creasepattern alone

A material made of creases

A folded sheet has a Poisson's ratio, a stiffness and a packing behaviour that the flat sheet did not. None of them belong to the paper — they belong to the pattern, and changing the pattern changes them without changing the material at all.

6 figures
found rather than designedcrushed drink cans, tree bark,deployable boomsat every interior vertexsectors 60°, 60°, 60°, 60°, 60°, 60°two courses and four diagonalstwo of one letter, four of the otherwhy the height is not freea steeper diagonal makes the topsector strictly smallest, flankedby two of the same letter20 interior vertices · 23 mountain and 56 valley creasesmountainvalleyraw edge

Patterns nobody designed

Crush a thin cylinder and it folds into a diamond lattice. Nobody chose the pattern — it is the buckling mode with the lowest energy, and it satisfies the flat-folding theorems because it just folded.

6 figures
-0.02-0.02-0.03-0.04-0.06-0.10-0.19-0.49-0.05-0.06-0.07-0.10-0.15-0.23-0.43-1.01-0.10-0.13-0.16-0.21-0.30-0.45-0.81-1.79-0.20-0.24-0.30-0.39-0.53-0.79-1.36-2.93-0.34-0.41-0.50-0.64-0.85-1.24-2.10-4.41-0.54-0.65-0.79-0.99-1.31-1.88-3.13-6.48-0.80-0.94-1.14-1.42-1.87-2.64-4.36-8.950.120.200.300.420.550.700.850.080.200.320.440.560.680.800.90panel slanthow far the sheet is closedwhat the map saysevery cell negativefrom -0.02 to -8.95a factor of 535.6a material has one value;this has a working point,and it is chosendarker means more negative

A property you can dial

Steel has one Poisson's ratio. A Miura-folded sheet has a surface of them, and where on that surface it sits is set by the panel shape and by how far it happens to be folded — which is why it is a mechanism rather than a material.

7 figures
polygonsectors at the twist vertexKawasakitiles the plane3-gon60.0 · 60.0 · 120.0 · 120.0180.0° = 180.0°yes — 6 round a point4-gon90.0 · 90.0 · 90.0 · 90.0180.0° = 180.0°yes — 4 round a point5-gon108.0 · 108.0 · 72.0 · 72.0180.0° = 180.0°no6-gon120.0 · 120.0 · 60.0 · 60.0180.0° = 180.0°yes — 3 round a point7-gon128.6 · 128.6 · 51.4 · 51.4180.0° = 180.0°noevery one of these twists satisfies the local theorems and folds flat on its ownthe interior angle has to divide 360° for the twists to meet, which only 3, 4 and 6 dobeyond 7 sides the assignment search runs out — 21 free creases, and the enumerator refuses above 22

Which polygons twist

Twist tessellations come in three kinds — triangle, square, hexagon — and it is natural to read that as a fact about twists. It is not. A twist can be built around any regular polygon and every one of them folds; what stops at three is the tiling, and the tiling is a fact about the plane.

6 figures
two kinds of vertex, both forced16 of degree 490°, 90°, 90°, 90°9 of degree 690°, 45°, 45°, 90°, 45°, 45°40 mountain and 36 valley creases14.3 sheet-widths of foldingmountainvalleyraw edge

The base that tiles

The waterbomb base is the first thing most people fold and the last thing they think about. Repeat it across a sheet and it becomes a tessellation with two kinds of vertex, an assignment that has to be searched for rather than remembered, and a folded state thirty-two times smaller than the paper.

9 figures
562 × 24 cells323 × 39 cells324 × 416 cells325 × 525 cellsrepeating rules that pass every condition, out of 51224 rules pass on the smallest patch and on none of the others

A unit that folds is not a tessellation

Of the 512 repeating rules for the waterbomb tessellation, 56 pass every condition on a two-by-two patch and 32 pass on every larger one. The twenty-four that die were never foldable — the small patch simply contained one of the four kinds of vertex the pattern makes, and the failures were at the other three.

9 figures
patternhow much smaller it foldscreasingper sheet-widthpreliminary8.0 layers, 8 at the deepest8.0×4.81.65×footprint × depth = 1.004 of the sheetmiura8.1 layers, 16 at the deepest8.1×6.21.31×footprint × depth = 1.001 of the sheetyoshimura32.0 layers, 36 at the deepest32.0×11.82.71×footprint × depth = 1.000 of the sheetwaterbomb31.6 layers, 32 at the deepest31.8×14.32.22×footprint × depth = 0.992 of the sheettwist3.0 layers, 9 at the deepest3.0×4.70.64×footprint × depth = 0.995 of the sheetthe shrinkage is the pattern's, not the paper's — nothing here knows what the sheet is made of

What a corrugation costs

Every tessellation this repository can fold, measured the same way: how much smaller it gets, how deep the stack becomes, and how much creasing was needed to buy it. The last column is the one nobody quotes and the one a folder feels.

9 figures
the flat pattern6 columns · 5 rowsthe course, closedradius 0.16676 sides of 0.1667 close into a ring of radius 0.1667, in the sheet's own unitsthe circle through the corners is 4.72% longer than the ring, which is what a 6-sided polygonowes its circle

The cylinder the pattern chooses

A Yoshimura pattern folds into a tube, and the tube's diameter is not a property of the paper. The course of diamonds has to go round exactly once, so the sheet's width is spent on the circumference the moment the columns are drawn — and what a larger sheet buys is a longer tube, never a fatter one.

9 figures
24682468times shorter across the sheettimes shorter down itaccordionMiuraleaf corrugationYoshimurasquare twistwhat each one doesaccordion8.00× across, 1.00× downMiura2.73× across, 1.75× downleaf corrugation3.57× across, 1.22× downYoshimura4.00× across, 4.00× downsquare twist1.52× across, 1.52× downthe dashed line is a directionthe fold never touchesone direction untouched: 1 of 5 · equal both ways: 2 · neither: 2the single number a corrugation is usually quoted by is these two multiplied together

A shrink is two numbers

How much smaller a folded sheet gets is quoted as a single factor, and that factor is a product. Measured along each axis separately, an accordion turns out to leave one direction of the paper exactly alone, a twist draws in equally both ways, and the Miura does neither — which is the whole of what makes it a Miura.

9 figures
what the construction produced9 twists, 36 interior verticesturned 24.1° from the tiling's edgespleats 0.118 to 0.118 wide2.20× smaller once the pleats are taken upevery vertex passes all four conditionsmountainvalleyraw edge

Any tiling makes a twist

A twist tessellation is usually drawn, admired and copied. It can be derived instead: hand the construction any tiling of the plane and it returns a crease pattern that folds flat, with the twist polygons' shapes forced by the tiling's own angles and nothing left to choose but how large and how turned.

8 figures
what the construction produced23 twists, 122 interior verticesturned 24.1° from the tiling's edgespleats 0.068 to 0.068 wide1.58× smaller once the pleats are taken upevery vertex passes all four conditionsmountainvalleyraw edge

Where two twists share a pleat

Every twist tessellation the tradition draws has one size of twist, because every tiling it is drawn on has one kind of vertex. Hand the construction a tiling with two, and the pleat between a large twist and a small one turns out to fix their sizes exactly — three to one, and nothing else folds.

8 figures
30°60°90°0.250.400.550.700.85how much of the room between two vertices the twists takeno paper leftno assignment existstwist angleboth curves are measured rather than plotted from a formula

Fenced at both ends

The twist angle of a tessellation looks like a free dial, and it is fenced twice. Turn too far and the pleats have no paper left. Turn too little and something stranger happens: every angle condition in the subject goes on holding and the pattern loses its mountain-valley assignment entirely.

8 figures
parallelstraight creases spread 0.0°flat, they spread 0.0°fan 5.2°straight creases spread 25.5°flat, they spread 30.9°fan 9.2°straight creases spread 46.6°flat, they spread 55.0°fan 13.8°straight creases spread 72.4°flat, they spread 82.5°

The corrugation that curves

A Miura is a flat sheet that becomes a flat slab. Open its straight creases into a fan and the same construction gives a corrugation that wraps a cone — exactly a cone, with every straight crease passing through one point to fifteen decimal places, at every moment of the fold, with the apex travelling as the sheet closes.

8 figures
the equation on an edge, at both of its endsas drawnevery vertex moved 12 per centratio − 1round a loopratio − 1round a loopthe square grid000.200.90the triangular grid000.170.75the honeycomb000.351.73the rhombille tiling2.0002.992.14the elongated triangular tiling000.171.27the rhombille's tiles are not regular, its ratios are three and a third, and they still multiply to one

The propagation that never had to work

The twist construction carries one equation per edge of its tiling and propagates the twist sizes outward from a seed. On every tiling anybody has drawn a twist on, every one of those equations is satisfied trivially — both ends of an edge read the same two numbers, because a regular polygon has one interior angle. The construction has been running and doing nothing, and the one tiling where it did something is the one whose tiles are not regular.

8 figures
what the construction produced7 twists, 60 interior verticesturned 17.2° from the tiling's edgespleats 0.123 to 0.123 wide1.76× smaller once the pleats are taken upevery vertex passes all four conditionsmountainvalleyraw edge

The dial that decides nothing

Turn a twist tessellation's angle from one fence to the other and every measurable thing about it changes: the smallest sector goes from 88 degrees to under one, the pleats swallow a quarter of the sheet and then almost none of it, the folded footprint changes by a third. The number of ways it can be creased does not change at all — sixteen, at every angle tested — because the lemma reads which sector is smallest and never how small.

8 figures
10 mountains, 14 valleys3.95 wide, 2.10 deep13 mountains, 11 valleys1.98 wide, 1.79 deep

The Miura folds two ways

One vertex repeated is what makes the Miura buildable: identical panels, identical creases, one degree of freedom. It is also what makes it ambiguous. At one fold angle on one crease the sheet has two folded states, differing in three letters and in half its width, and both of them close exactly — while a mesh with no two vertices alike has one.

9 figures
-0.15-0.00-0.12-3.53-0.46-2.69-0.09-0.300.44the sheet as a whole reports -0.175spread 3.97, 23× the sheet's

Nothing to average over

A folded corrugation is reported with a Poisson's ratio, and both of this site's measurements of one were made on a sheet that repeats a single cell. On such a sheet every cell behaves the same way and the cell's number is the sheet's number. On a sheet with no repeating cell the cells run from −3.5 to +0.4 — some widening while others narrow — and the sheet's own figure describes none of them.

9 figures
counting rules and counting objects are different measurementsrepeating rulesnine binary choices, one per crease of the repeating unit512pass on a small patchevery vertex of a two-by-two patch satisfies every condition56pass on a larger oneand on a patch that contains all four kinds of vertex32folded objectscounted by comparing the folded panels, not the letters1

Thirty-two rules, one object

Five hundred and twelve repeating rules for the waterbomb tessellation, fifty-six that pass on a small patch, thirty-two that pass on one containing every kind of vertex. Fold all thirty-two and compare their panels: the same panels, in the same places, with the same areas, every time. The rules are thirty-two labels on one object, and a count of them has counted the labels.

8 figures
a 3 × 3 patch: 42 creases, 18 of them buriedthe bar is on a log scale, because the two numbers differ by four orders of magnitudepieces the 3×3 patch has262,144pieces containing a repeating rule32, one eachevery one of the 32 rules is in a piece no other rule is in

Thirty-two rules, thirty-two pieces

The waterbomb tessellation's surviving repeating rules fold to one object — same panels, same places, same areas. Put them in the space of letterings the patch admits and they occupy thirty-two different pieces of a quarter of a million, so no two of them can be reached from one another without unfolding the sheet.

8 figures
the bar is the draws whose letters do not contradict themselvesa loop of panels is a proof that no flat folded state exists, and it costs one passone square twist39 of 409 panels · 12 creasesone hexagon twist40 of 4013 panels · 18 creasesa small square tiling24 of 4049 panels · 72 creasesthe square tiling7 of 4049 panels · 84 creasesthe patch a propagation returns first is not a draw and has no reason to be among these

The tiling the unit could not promise

Every twist on this site carries the same caveat: the unit is verified and the plane is not, because deciding a whole pattern is intractable. There is one thing about a whole pattern that costs a single pass over its crease list, and it says no. The square twist tiling was drawn with a lettering that contains a loop of twenty-eight panels, so the patch on this site had no flat folded state at all — and only seven of forty independent redraws avoid one.

8 figures
the same tessellation on the same square, cut out of the plane two waysassembled from whole unitsclipped from the plane12 crossings · panels 1.73 apart0 crossings · panels closemountainvalleyraw edge

Cutting a patch out of a plane

A tessellation is infinite and a sheet is not, so every picture of one is a decision about where the paper stops. Assembling whole twist units on a square and running the outstanding pleats to the rim puts 12, 18, 12 and 5 creases across other creases on four of five tilings; generating the pattern over a larger region and clipping it puts none. The panels then place exactly — and what is waiting behind the repair is a different refusal that could not be asked about before.

8 figures
the bar is the average number of layers over the folded footprint0.5 of the sheet3.891 whole of 9 · 89% cut by the rim0.42 of the sheet3.641 whole of 9 · 89% cut by the rim0.34 of the sheet4.049 whole of 9 · 0% cut by the rim0.28 of the sheet4.309 whole of 21 · 57% cut by the rim0.22 of the sheet4.349 whole of 25 · 64% cut by the rim0.18 of the sheet4.7925 whole of 45 · 44% cut by the rima bulk property arrives as the boundary leaves, and neither a page nor a sheet of paper reaches the end of it

The property a patch does not have

A folded corrugation is described as a material — a packing ratio, a stiffness, a Poisson's ratio — and every one of those is a statement about an unbounded medium. Fold the same tiling at six sizes on the same square and the compaction climbs from 3.89 layers to 4.79 as the share of units the rim cuts falls from nine tenths to four, and it has not settled at the fine end. The number a patch gives is the material's number minus its own boundary.

6 figures
the bar is the proportion of the sheet the pattern asks for2 columns, 2 rows1.183(2 + tan 20°) / 2 = 1.1833 columns, 3 rows1.122(3 + tan 20°) / 3 = 1.1224 columns, 3 rows1.455(4 + tan 20°) / 3 = 1.4556 columns, 4 rows1.591(6 + tan 20°) / 4 = 1.5916 columns, 6 rows1.061(6 + tan 20°) / 6 = 1.0618 columns, 6 rows1.394(8 + tan 20°) / 6 = 1.394a square sheet needs a proportion of exactly one, which the counts and the slant have to be chosen for

The paper a pattern asks for

A Miura of c columns and r rows at a slant α wants a sheet whose proportion is (c + tan α) / r — one equation tying the two counts, the angle and the shape of the paper. A square is the case where it comes to one, which needs the tangent of the slant to be a whole number: 45° for a pattern one row taller than it is wide, 63.43° for two, and nothing at all for the slants anybody draws.

6 figures
the bar is the share of draws that agree with themselvesthe rows are ordered by panel count, which is the only thing changing along them49 panels26 of 200square · 84 creases · 26 of 20062 panels5 of 200elongated · 106 creases · 5 of 20077 panels2 of 200hexagonal · 142 creases · 2 of 20083 panels0 of 200triangular · 142 creases · 0 of 200157 panels0 of 200rhombille · 282 creases · 0 of 200a zero is a zero of the draws taken and not a proof that no consistent lettering exists

Letters that agree get rarer

Two hundred letterings drawn independently from a square twist tessellation patch, and twenty-six of them have letters that do not contradict themselves. On the next patch up it is five, then two, then none, then none. What the share falls with is not the size of the patch and not the angle of its twist: it is the number of independent closed chains its panels form, which is Euler's relation on the drawing and is fixed before a single letter is chosen.

8 figures
each arrow points from the lower panel to the higher one9 panels · 12 creases · 12 arcsa loop of 8 panels — no order existsthe arrows are the whole of the test — nothing here asks which panels lie over which

The ring is the loop

The square twist's central polygon is four creases enclosing one panel, and a lettering that gives all four the same letter has no folded state. That was established by enumerating the orderings of nine panels. It can now be read off the crease list in one pass, because the eight panels the letters send round in a circle are exactly the ring — the twist's own defining feature, contradicting itself.

9 figures
every curve is one construction grown, and the axis is the same for all threethe Miura, grownthe Yoshimura, growntwist patches00.2500.5000.7501255075100125independent closed chains of panelsshare of letterings that agree with themselvesthe horizontal axis is read off the drawing before any letter is chosen, and it is the number of interior vertices

A corrugation agrees with itself

A Miura fold of forty-eight panels and a twist tessellation patch of forty-nine have almost exactly the same number of independent closed chains for their letters to contradict themselves round — thirty-five against thirty-six. Sixty-four per cent of the Miura's drawn letterings are consistent and thirteen per cent of the patch's. A Yoshimura at thirty-three chains manages ninety-three. The room to fail sets the scale; the construction decides where in it a pattern lands.

9 figures
two kinds of vertex, both forced16 of degree 490°, 90°, 90°, 90°9 of degree 690°, 45°, 45°, 90°, 45°, 45°40 mountain and 36 valley creases14.3 sheet-widths of foldingmountainvalleyraw edge

The rule that breaks the count

The waterbomb tessellation has five hundred and twelve repeating rules for its letters and thirty-two of them fold. A hundred and twenty of the other four hundred and eighty send four panels round in a circle — the shortest circle a crease pattern can have — and every single one of those hundred and twenty has broken Maekawa's count at the very vertex the circle goes round. The theorem that closes the shortest circle, caught doing it, a hundred and twenty times.

8 figures
the 16 repeating rules that fold, written outrows first, then the two column classes — and every one of them alternates down the columnrows · columns above|below20VV · MV|MV21MV · MV|MV22VM · MV|MV23MM · MV|MV24VV · VM|MV25MV · VM|MV26VM · VM|MV27MM · VM|MV36VV · MV|VM37MV · MV|VM38VM · MV|VM39MM · MV|VM40VV · VM|VM41MV · VM|VM42VM · VM|VM43MM · VM|VMfour ways of writing the rows times four ways of alternating the columns is sixteen, and there is nothing else

Sixty-four rules, sixteen fold

The Miura fold's letters are usually given as a recipe: rows one way, columns changing at every row. Write down every rule of that shape — the letter on a crease depending only on which row and which column it is in — and there are sixty-four. Sixteen fold flat. They are exactly the ones whose columns change at every row, the row letters do not matter at all, and every one of the forty-eight refusals is the counting theorem's alone.

7 figures
the bar is how many rules the two tests agree aboutone reads three bits of the rule; the other folds the sheet and walks the arcsthe Miura fold64 of 6438 rules predicted to close a loop · 0 disagreementsthe tapered leaf64 of 6438 rules predicted to close a loop · 0 disagreementsthe closed form says a loop is available exactly where the columns fail to change letter and the row disagrees with them

The loop is in the rule

Of the forty-eight repeating rules that do not fold a grid corrugation, thirty-eight send four panels round in a circle and ten merely fail the count. Which is which can be read off three of the rule's six bits, without building the pattern, folding it or walking a single arrow — and the closed form agrees with the arrows on all sixty-four rules of both grid families.

6 figures
the bar is the middle run of a hundred and twentysame pattern, same code — only the order the letters are tried in differsthe square patch2725 at best · 27 at the middle · 36 at worstthe elongated patch3432 at best · 34 at the middle · 39 at worstthe hexagonal patch4339 at best · 43 at the middle · 51 at worstthe triangular patch4539 at best · 45 at the middle · 53 at worstthe rhombille patch16684 at best · 166 at the middle · 48 of 120 unfinished at 20000an unfinished run is left out of the middle rather than counted as its budget

Four easy patches and one that is not

Run the same search a hundred and twenty times on each of five tessellation patches, changing nothing but the order the letters are tried in. Four of them answer in between twenty-five and fifty-three steps every single time. The fifth answers in eighty-four steps at best, a hundred and sixty-six in the middle, and does not answer at all in forty-eight runs of the hundred and twenty.

6 figures
the bar is how many rules send four panels round in a circleout of the rules that already fail the count at some vertexthe Miura fold3848 refused · every loop four panels · vertices of degree 4the tapered leaf3848 refused · every loop four panels · vertices of degree 4the Yoshimura pattern038 refused · not one closes a loop · vertices of degree 6the waterbomb tessellation120480 refused · every loop four panels · vertices of degree 4 and 6a loop of four needs the letters to alternate round one point, and only a degree-four vertex lets a repeating rule do that

Where a rule can close a loop

Three corrugation families have repeating rules whose letters send four panels round in a circle, and one has none at all. The one that has none is the one whose vertices are all of degree six — and the reason is that a straight line through a point carries a single letter under any repeating rule, while a strict alternation round six creases needs the two halves of that line to differ.

6 figures
each cell is one patch, searched to a verdictgreen: a lettering exists · magenta: none exists, by exhaustion0.150.250.350.50.70.91.11.3turn angle, in radianssquare2626262626262626elongated1515323231313232hexagonal1515394545464545triangular1515393939373737the number in a cell is the nodes the search visited; 6 of 32 patches have no lettering at all

The dial and the tiling that is not alike

Four of the five tilings a twist tessellation can be built on behave identically under every dial the construction has. The fifth has two kinds of vertex, and everything about it is different: it is the only one whose search has a tail, the only one whose shallow patches take minutes to draw, and the only one where a distance has to be solved rather than assumed.

8 figures
each point is one pattern: panels across, nodes up00100100200200one node per panelnodes visitedpanels2 by 2 to 16 by 16, and not one backtrack anywhere in the family

A corrugation never backtracks

As a box-pleating grid goes from two divisions to sixteen, the share of random letterings that agree with themselves falls from a hundred in a hundred to one. The cost of finding one that does stays at exactly one step per panel — four, nine, sixteen, twenty-five, and two hundred and fifty-six — with not a single wrong guess anywhere in the family.

9 figures
the twist patches: nodes against panels050100150one a panel0 panels157every vertex of this family keeps 4 labellings

The most decided vertex here

Sixteen ways to letter four creases; Maekawa allows eight; the big-little-big lemma allows four. A twist polygon's corner is one of the few vertices in this collection where the second cut applies, so it keeps four labellings where a grid, a leaf, a Miura and a crumple all keep eight — and the family the collection long called difficult turns out to be the one whose conditions decide the most.

9 figures
one similarity, five tilingslong: a cell of the flat sheet · short: where it lands foldedthe square grid ×0.41037344the triangular grid ×0.41037344the honeycomb ×0.41037344the elongated triangular tiling ×0.41037344the rhombille tiling ×0.41037344turned 36.62°, the same on every onethe scale is a property of the pleat, and the tiling does not enter it

Folding it flat is one similarity

Where a cell of paper goes when a twist tessellation collapses is a scale and a turn: multiply the plane's lattice by 0.410373441 and rotate it by 36.62°. That is the answer on the square grid, the triangular grid, the honeycomb, the elongated triangular tiling and the rhombille alike, agreeing to eight figures — while the collection's other answer to how much smaller it gets gives those five tilings five different numbers.

8 figures
the period cell of the Miuraone period, with its neighbours round it2 interior vertices in the cell7 crease pieces drawnperiod 1.000 × 2.000one column wide and two rows high, because the zigzag returns after twothe cell is a rectangle of ordinary paper until somebody says its edges are one edge

The corrugation that closes on itself

A Miura cell crosses one crease per period in one direction and four in the other. So one of its two directions has a parity condition that half the sizes fail, and the other direction cannot fail at any size — the same sheet, the same drawing, and two gluings that behave completely differently.

7 figures
nodes of search per panel, as the rim goesthe grid ×1 cut1.004 nodes · 4 panels · 4 lettersthe grid ×2 cut1.009 nodes · 9 panels · 12 lettersthe grid ×2 cyl x1.177 nodes · 6 panels · 10 lettersthe grid ×2 cyl y1.177 nodes · 6 panels · 10 lettersthe grid ×2 torus1.506 nodes · 4 panels · 8 lettersthe grid ×3 cut1.0016 nodes · 16 panels · 24 lettersthe Miura ×1 cut1.006 nodes · 6 panels · 7 lettersthe Miura ×1 cyl y1.255 nodes · 4 panels · 6 lettersthe Miura ×2 cut1.0015 nodes · 15 panels · 22 lettersthe Miura ×2 cyl x1.1011 nodes · 10 panels · 18 lettersthe Miura ×2 cyl y1.0813 nodes · 12 panels · 20 lettersthe Miura ×2 torus1.2510 nodes · 8 panels · 16 lettersthe Miura ×3 cut1.0028 nodes · 28 panels · 45 lettersthe Miura ×3 cyl y1.0425 nodes · 24 panels · 42 lettersthe Yoshimura ×1 cut0.9110 nodes · 11 panels · 12 lettersthe Yoshimura ×1 cyl y1.139 nodes · 8 panels · 10 lettersthe Yoshimura ×2 cut0.9728 nodes · 29 panels · 36 lettersthe Yoshimura ×2 cyl y1.0024 nodes · 24 panels · 32 lettersthe Yoshimura ×3 cut0.9653 nodes · 55 panels · 72 lettersthe Yoshimura ×3 cyl x1.0042 nodes · 42 panels · 60 lettersthe Yoshimura ×3 cyl y0.9445 nodes · 48 panels · 66 lettersthe Yoshimura ×3 torus1.0036 nodes · 36 panels · 54 lettersfewer panels to divide by, and the same argument to settle

One node per panel, with the rim gone

A rectangle of repeating pattern cut out of the plane costs exactly one node of search per panel, on every family and at every size. Take the rim away and the total falls and the cost per panel rises, because the letters that were removed were the ones that could not be wrong.

6 figures
the folded period of the Yoshimuradrawn periods across the top123456the turnsame way upslides240°yesno120°yesnoyesyes240°yesno120°yesnoyesyesa turn of 240° comes back to nothing after three of them, and that is the folded periodthe drawing repeats every one, which is what makes it a tessellation

The turn a column costs

The Yoshimura's drawing repeats every column. Folded flat, it does not: the fold carries one column onto the next by a turn of two hundred and forty degrees, so the folded state repeats every third column and not before. A pattern has two periods and only one of them has ever been written down.

6 figures
the folded period of the Yoshimuradrawn periods across the top1234567the turnsame way upslides240°yesno120°yesnoyesyes240°yesno120°yesnoyesyes240°yesnoa turn of 240° comes back to nothing after three of them, and that is the folded periodthe drawing repeats every one, which is what makes it a tessellation

The period nobody measured

Every repeating pattern in this collection has its drawn period recorded, because a drawing cannot be generated without one. Its folded period is recorded nowhere, and on one of the families measured the two differ by a factor of three — which means the number that has always been quoted is the wrong one for anything about the folded object.

7 figures
a square twist, 2×2 cellsone drawing, four sheetscutoutgluedacrossgluedalonggluedboth waysvertices16161616free letters40363632panels25202016V − E + F1000the vertex row is the control: identifying edges can neither make nor destroy a vertexand Euler's number is the cheapest check that the gluing did what it says

A tessellation on a cylinder

A twist tessellation has been drawn here as a patch and as a torus, and never as anything in between. Gluing one pair of a cell's edges gives the family its first sheet with exactly two edges — the shape every folded tube actually has, and the only object in the collection that can say whether the rim's cost is linear in how much rim there is.

7 figures
the case the corner search cannot seeedges clear of every vertex, and a crease through a corner anywaythe corner is where four edges meeta crease piece ending there has no partneron any one of themand Euler's count comes out −1the cure is a nudge along a gap the vertex search had already cleared

The seam that is not a symmetry

Gluing a cell's edges looks like a symmetry of the drawing and is not. It is an instruction about which points of the paper are the same point, the drawing has to agree with it along the whole of a glued edge, and a rectangle that is not a period of the pattern does not glue at all — which turns out to be the only real restriction on which cylinders exist.

6 figures
the period cell of the Miuraone period, with its neighbours round it2 interior vertices in the cell7 crease pieces drawnperiod 1.000 × 2.000one column wide and two rows high, because the zigzag returns after twothe cell is a rectangle of ordinary paper until somebody says its edges are one edge

A metamaterial with no edge

A folded metamaterial's properties are quoted per unit cell, because a material is supposed to be the same everywhere and a cell is supposed to stand for the whole of it. Every cell this collection has measured has been cut out of a patch, with a rim round it — and a rim is the one place a repeating material is not like itself.

7 figures
do the pleat equations close round every loop, as drawn and under three linear mapstrivially means every edge's equation is one at both ends; otherwise the largest disagreement round a loopas drawnshearedstretchedgeneralthe square gridcloses, triviallycloses, triviallycloses, triviallycloses, triviallythe triangular gridcloses, triviallycloses, triviallycloses, triviallycloses, triviallythe honeycombcloses, triviallycloses, triviallycloses, triviallycloses, triviallythe rhombille tilingclosesoff by 1.76off by 2.18off by 2.96the elongated triangular tilingcloses, triviallyoff by 3.14closes, triviallyoff by 4.77the maps: a shear of 0.3, a stretch of 1.5 along one axis, and the matrix [1.3, 0.4; −0.2, 0.9]

Closing the loops is not folding

The twist construction propagates one equation along every edge of a tiling, and it can only work where the equations agree round every loop. Asked which irregular tilings pass, a linear map gives a clean answer: the square grid, the triangular grid and the honeycomb pass under every shear and stretch tried, because each edge has a half-turn symmetry that makes its equation exactly one at both ends, and a half-turn survives any linear map. The rhombille passes only as drawn. But passing is not folding. On every one of those images — including the ones whose loops close exactly — the construction produces a pattern that fails the angle condition at every turn tried. The loops were a necessary condition all along, and the construction needs something the tilings' images do not give it.

5 figures
does the construction fold, with one side distance a vertex and with one a sideat a turn of 0.42 radians; the second construction's pleats run at -0.5 radians from their own edgesas drawnshearedstretchedgeneralone a vertex / one a sidethe square gridfolds/foldsno/foldsno/foldsno/foldsthe triangular gridfolds/foldsno/foldsno/foldsno/foldsthe honeycombfolds/foldsno/foldsno/foldsno/foldsthe rhombille tilingfolds/foldsno/foldsno/foldsno/foldsthe elongated triangular tilingfolds/foldsno/foldsno/foldsno/foldsthe maps: a shear of 0.3, a stretch of 1.5 along one axis, and the matrix [1.3, 0.4; −0.2, 0.9]

One number where the corners wanted four

The twist construction gives a vertex a single side distance, and every account of these patterns does the same — it is what rotate-and-shrink means. The conditions never asked for it. Written out, the corner condition is one linear equation per pleat crease in the distances taken one per edge, so a degree-four vertex carries four unknowns against two independent equations. Given them back, the twelve sheared and stretched tilings that refused to fold all fold.

6 figures
each side of the twist, divided by the edge it facesone vertex of the the square grid under [1.3, 0.4, -0.2, 0.9]1.0520.3871.0520.387the weightsthe same edges, scaled by them, end to endthe 4 weighted edges close to 3e-16 of their own total length, and the two ends of every edge agree to 3e-15

Every twist writes an equilibrium

Divide each side of a twist polygon by the length of the edge it faces. The polygon closing says those numbers, weighted onto the edges, balance at the vertex; the pleat matching says the two ends of an edge agree on the number. Together they are a positive equilibrium stress — the thing a tiling has when it is the plan of a spider web — and the construction has been writing one at every vertex without being asked for it.

5 figures
how much more the folded sheet draws in one way than the otherthe ratio of the two principal factors of the collapse, fitted to the twists' positions before and after foldingas drawnshearedstretchedgeneralthe square grid1.00004.41391.00003.5417the triangular grid1.00001.00001.00001.0000the honeycomb1.00002.81795.38555.0597the rhombille tiling1.00002.81795.38555.0597the elongated triangular tiling1.00004.41391.00003.5417one is a similarity — the folded sheet is the flat one scaled and turned, with no direction preferred

The sheet draws in crooked

Every twist tessellation measured here has collapsed by a similarity: the folded sheet is the flat one scaled and turned, the same way in every direction. The patterns that exist on sheared and stretched tilings do not. Ten of the fifteen images fold by a map with two different principal factors, up to five and a third to one — and the prediction that said which ten, made from the weights the pattern writes on its edges, is wrong in both directions.

6 figures
each family's dial, run, in the plane of the two directional factorsmeasured on the folded state of every pattern, not predicted from its rule1234561234along the patternacross itthe accordionthe square twistthe Miura foldthe Yoshimuraa point on the lower edge leaves one direction alone; a point on the dashed diagonal draws in equally both ways

The plane the five points were in

Five corrugations measured at one setting each gave five points, and the space between them was left as an open question: forbidden, or merely unvisited. Every one of those patterns has a dial nobody turned. Turned, they trace curves — the accordion's is a line with integers on it, the square twist's is the diagonal and nothing else, and the Miura's turns round on itself, so a steeper slant stops buying a smaller sheet.

8 figures
the cross factor of a leaf corrugation against its zigzag anglemeasured off the folded state's own extent, at every angle drawnno change0.600.680.760.840.921.00zigzag angle, radians0.981.001.02below one between 0.7 and 0.88 radians, at worst 0.98555 — and the areal factor never falls below 1.360

The direction that gets longer

A shrink factor below one is a direction in which the folded sheet is bigger than the flat one, and the leaf corrugation has one. The cause is not the taper and not the angle: a corrugation's folded extent across its own creases is a constant of the cell, the same number at two rows and at ten, so the cross factor is the sheet's height divided by a fixed length — a straight line through the origin that crosses one at a definite row count.

5 figures

Rigid folding

Panels and hinges instead of paper — the version that scales to solar arrays and stents.

the pattern as it isevery edge keeps its length6.7e-16the same pattern, moved by 0.01and one of them cannot1.7e-21e-181e-161e-141e-121e-101e-81e-61e-41e-21largest change in any edge length, in panel widthswhat an isometry has to do, and what it manages5 × 4 panels, at 50% folded, every edge of both comparedthe moved pattern is fitted the best single panel its own edge lengths allow before being folded at allso the gap is not a bad choice of panel — it is what is left when the best choice has been made

Panels instead of paper

Flat-foldability asks whether a pattern can reach a flat state. Rigid-foldability asks whether it can get there without any face bending on the way. The second is much stronger, and everything that gets manufactured lives inside it.

5 figures
050100150-100100how far the vertex is drivenfold anglesectors60° 90° 120° 90°crease 1: Mcrease 2: Vcrease 3: Mcrease 4: Mcrease 2 is the odd onethree agree, one does notKawasaki holdsand it reaches flatfound by the linkage,not by the theorem

What the vertex does on the way

A four-crease vertex is a linkage on a sphere. Solving its closure gives the fold angles at every moment, and two theorems that are usually proved about the flat state turn up in the answer without being put there.

6 figures
zero thicknesspanels meet exactly4 layers of real materialeach fold has to clear the ones belowhinge offset to the surfacethe panel rotates about the right linea crease pattern describes a surface with no thicknessand everything anybody builds has some

The sheet has a thickness

Every crease pattern describes a surface with no thickness. Everything anybody builds has some, and getting it around a corner is the central problem of turning origami into hardware.

6 figures
61°108°travel before contact110.0°measured by contact testpanel thickness22% of the panel lengthwhat it gives upmaterial at the crease, sothe panel is thinnest whereit is worked hardesta zero-thickness pattern says the panels meet along a line; nothing that is built does

Getting thickness round a corner

There are half a dozen ways to build a fold in a panel that has depth, and the useful way to arrange them is not by what the cross-section looks like. It is by what each one gives away.

6 figures
-10010000.20.40.60.81how far the vertex is drivenstored energybranch oneMVMMbranch twoMVVVboth run downhillfrom the flat state,and end at zeroso the energy does notprefer either branch —the noise decides

Paper that folds itself

A self-folding sheet has to supply the fold and then choose what to fold into. The second half is where these things fail, and no amount of torque helps, because the two outcomes are equally downhill.

6 figures
Miura solar array17×Space Flyer Unit, 1995airbag folding25×stored for years, opens in 30 msheart stentthreaded through an arterystarshade11×26 m disc, 2.5 m launch tubemap foldthe original problempackeddeployedthe ratio is what is bought; one degree of freedom is what makes it reliable

Folding that gets built

Solar arrays, airbags, stents and starshades. The requirement is always the same — large in use, small in transit, along a path nobody has to trust to chance — and folding is what answers it.

6 figures
60° / 90°all 4 reached30° / 120°all 4 reached45° / 45°2 of 8 reached50° / 70°all 4 reached80° / 55°all 4 reachedsectorseach square is one assignment the theorems allowfilled — a rigid motion arrives there · open — a flat state with no path to itthe gap opens where two sectors are equal, and nowhere else on this listbig-little-big has nothing to forbid there — the linkage still does

A state no motion reaches

Flat-foldability asks whether a folded state exists. Rigid-foldability asks whether there is a path to it. The two sets are different, and the difference can be counted on a single vertex.

6 figures
grown symmetricallyoffset to one sidepanels overlap over 4% of their areawhich is the jam every thick-panel design meetspanels do not overlap at alland the hinge axes have not movedwhich is not free: fold it tighter and this offset runs out toosheet 0.16 panel-lengths thick, folded to 100° — the overlap is measured from the geometry

Panels with somewhere to go

Every way of giving a folded panel real thickness costs something. Tachi's offset-panel technique costs the least interesting thing there is — it stops the panels being a surface, and leaves the hinges exactly where the zero-thickness pattern put them.

7 figures
05101520253000.050.10.150.20.250.3creasesdrift, in panel widthsthe same way each timeat randomeach crease 0.5° out · 40 strips averaged for the random casethe drift is a composition of reflections, and would be the same on paper

Error is folded too

A folded position is a composition of reflections, and a reflection in a line that is slightly off turns everything beyond it by twice as much. So an error does not stay where it was made — and whether it grows with the crease count or with its square root depends on whether it is the same error every time.

8 figures
20° a crease0.50 turns of papernothing touching anything34° a crease0.85 turns of papernothing touching anything36° a crease0.90 turns of paper1 pair through one another50° a crease1.25 turns of paper5 pairs through one anotherone strip of 10 panels, seen end-onit laps itself at 36.0° a crease, which is where its cross-section closesevery panel is the same length in every frame; the only thing changed is how far each crease is turned

Paper through paper

Every test the subject has for rigid folding is a statement about a neighbourhood, and a neighbourhood cannot see the far side of the sheet. So a pattern can satisfy all of them while driving one panel straight through another, and the sharpest witness has no interior vertex in it at all.

9 figures
40°71°travel before contact72.1°measured by contact testpanel thickness22% of the panel lengthwhat it gives upa gap, and with itstiffness and a pattern nolonger quite the one onpapera zero-thickness pattern says the panels meet along a line; nothing that is built does

Nowhere to put the error

Paper takes a misplaced crease and spreads it along its whole length as a curvature nobody notices. A panel is flat by definition and cannot, so the error arrives at the hinge — and the room to receive it is a length that has to be drawn, is paid for in fold angle, and has to grow with the crease count.

9 figures
1.7e-6 at 1e-61e-61e-51e-41e-31e-21e-61e-51e-41e-31e-2how far each vertex of the flat pattern was moved, in panel widthslargest edge-length errorfitted slope0.9999over four decadesat a displacement ofexactly zero the erroris 6.7e-16, which iswhere the arithmeticstops and not wherethe geometry does5 × 4 panels at 50% folded, with one set of displacement directions scaled across the decadesa slope of one is the claim: the failure is first order, so no displacement is small enough to be free

The only pattern that moves

A rigid motion is not a generic property of a folded pattern. Move one vertex of a Miura by a thousandth of a panel and the sheet has no isometric folded position of that kind at all — and the amount by which it fails is first order in the displacement, so no move is small enough to be free.

8 figures
the opposite crease, ×1the odd crease, ×0.26794910how far the vertex has foldedsectors 60° · 90° · 120° · 90°constant to 8.2e-13 over 199 points of the motionand equal to cos((α+β)/2) ÷ cos((α−β)/2), which the solver never forms

The vertex is geared

A rigid four-crease vertex has one degree of freedom, which says that one number decides everything and not how. The how is a fixed ratio: the tangents of the half fold angles at two creases stay in constant proportion for the whole of the motion, and the proportion is a function of the sector angles and nothing else.

8 figures
parallel columnsKawasaki to 3e-14°columns fanning by 5.2°Kawasaki to 5e-14°columns fanning by 9.2°Kawasaki to 5e-14°the mountain-and-valley letters are read off the motion rather than drawn, and then put past Maekawa

The family the Miura belongs to

Move one vertex of a Miura and the sheet has no rigid folded position at all — which leaves the obvious question unanswered. What else moves? A row of paper reflected in each of a fan of lines is flat-foldable for nothing at all, and whether it also folds rigidly turns out to be a condition on a table of cosines: it has to be a column of numbers times a row of numbers.

8 figures
the ratio of each column's cosine to the first column'scol 1col 2col 3col 4col 5the sheet that folds1.0000-1.08221.0000-1.08221.00001.0000-1.08221.0000-1.08221.00001.0000-1.08221.0000-1.08221.0000row 1row 2row 3one row moved by 14 per cent1.0000-1.08221.0000-1.08221.00001.0000-1.13971.0000-1.13971.00001.0000-1.08221.0000-1.08221.0000row 1row 2row 3worst disagreement between rows: 4e-16 before, 0.0575 after

Where an error goes

A misplaced crease in a folded sheet has to be paid for somewhere, and this subject has two answers already — the error is folded too, and the hinge is where it ends up. There is a third. In a quadrilateral mesh a mistake in one row has no consequence in that row at all: it is felt by the columns, which is to say by every other row on the sheet.

8 figures
flat-foldable at every vertex, all threeworst Kawasaki residual 0e+0 radiansone vertex moved -30 per centloop residual 2.4e-1the Miuraloop residual 5.8e-14one vertex moved 40 per centloop residual 3.6e-1every one of these is developable and flat-foldable at every interior vertex, exactly

The condition that is not flat-foldability

Take away the assumption that one crease family runs straight through every vertex and ask what makes a quadrilateral mesh move. It is not flat-foldability. There is a one-parameter family of meshes, every one of them developable and flat-foldable at every vertex to machine precision, and exactly one member of it folds — the Miura. Slide a single vertex along the ray that keeps every condition exact and the sheet stops moving, first order in the displacement.

8 figures
one crease decided, and how much of the sheet followsthe flat-folding conditions, propagated2 of 12the rigid-folding conditions, propagated12 of 12and the rigid propagation leaves 1 consistent set of fold angles

One crease decides the sheet

Fix one crease of a flat-folding problem, propagate every condition the subject has, and three creases out of a hundred and fifty-eight follow. Fix one fold angle of a rigid one and every crease on the sheet follows, with a single consistent answer. The same experiment, two questions, opposite answers — and it is why a self-folding sheet needs one biased vertex rather than one per vertex.

8 figures
how far the fold closes, against where the hinge sitslower faceupper facemid-surface180°closing upwardclosing downwardtapering the panels buys most of it back: 166° of the 180°, less twice the taper

Thickness has a sign

Swap every mountain for a valley and back again. Kawasaki does not notice, Maekawa gets the same condition the other way round, the lemma still asks the two creases to differ, and the layers come out mirrored. Every theorem on this site is blind to which side of the paper it is looking at — and a hinge in a panel with depth is not. The fold closes one way and jams at nothing at all the other.

9 figures
mismatch 0.066 radiansmismatch 8.5e-14 radiansevery vertex of both is developable and Kawasaki-exact to the last bit a double holds

Solving every face at once

A quadrilateral mesh that folds rigidly has to close round every one of its faces, and the rung that built the general mesh could close one. Four of them at once resisted a descent that drove each free length to its own root, because closing a loop is a condition on several lengths together — and solving them jointly finds a sheet with no two vertices alike that folds, and a surface of them sixteen dimensions wide.

9 figures
the solved mesh, cut wrong byworst mismatch left, radians0.017 mm0.00210.051 mm0.00580.169 mm0.01990.508 mmno closure at alla Miura, cut wrong by0 per cent2e-141 per cent6e-155 per cent4e-15

Solved is not built

A mesh that folds because an equation holds and a mesh that folds because one crease family runs straight through every vertex are not two examples of the same thing. Cut a Miura's every dimension five per cent wrong and it still folds exactly. Cut a solved general mesh a fifth of a millimetre wrong on a 150 mm sheet and the closure is gone.

8 figures
creaseworst amplification of an error in itc:0:11.76c:0:21.77c:0:31.74c:1:11.76c:1:21.77c:1:31.74c:2:11.76c:2:21.77c:2:31.74c:3:11.76c:3:21.77c:3:31.74r:1:01.00r:1:11.00r:1:21.00r:1:31.00r:2:01.58r:2:11.58r:2:21.58r:2:31.58r:3:01.65r:3:11.65r:3:21.65r:3:31.65the best crease is 1.77 times better than the worst, and it is on the sheet's edge

Which crease to push

Deciding one fold angle settles every other one on a quadrilateral mesh, which is what makes a self-folding sheet buildable with a single actuator. It leaves a question that sounds like an afterthought: which crease. Driving each of a mesh's twenty-four in turn gives twenty-four different answers to how far an error in it travels — and on the sheet that repeats one vertex, it gives several answers to what shape the sheet takes.

9 figures
5 of 6 solved meshes are solid at every angle sampledthe bar is the deepest interpenetration found anywhere in the motion, in panel widthsmesh 11closes to 9e-14solid at every anglemesh 17closes to 1e-121.18 — panels 3:0 and 3:2, 2 steps apartmesh 19closes to 6e-14solid at every anglemesh 23closes to 5e-14solid at every anglemesh 27closes to 2e-12solid at every anglemesh 71closes to 4e-12solid at every angle

Closing is not building

A quadrilateral mesh solved so that every loop closes to within a millionth of a radian is a mesh whose fold angles are consistent. It is not necessarily an object. One of the six solved here drives a panel through another at every angle of its motion — there is no part of the fold at which it could be made of solid panels — and the pair that crosses is two steps apart in the sheet, where nothing evaluated at a vertex could see it.

8 figures
00.20.40.60.81-7-6-5-4-3-2-1error in every length, millimetres on a 150 mm sheetclosure mismatch (powers of ten)across the surface of solutionsthe direction the closure's own derivative points ina direction chosen without regard to the surfacewhich has a component along bothalong the surface of solutionsone of the twelve directions the equations do not seethe same step costs 5,130 times as much one way as the other

A tolerance is a direction

Cut a solved mesh a fifth of a millimetre wrong and its closure is gone. That is true of the errors it was tried with and false of errors in general: the solutions form a surface sixteen directions wide, an error along it costs five thousand times less than the same error across it, and the fifth of a millimetre is the allowance in one direction out of twenty.

9 figures
deepest pile on the shelf: 60 layersin office copier paper, that is 18.5 mm of paper to find at one creaseThe Yoshimura pattern60 layers · 18.5 mm · 59× the two-layer allowanceThe waterbomb tessellation32 layers · 9.7 mm · 31× the two-layer allowanceThe Miura fold16 layers · 4.7 mm · 15× the two-layer allowanceThe tapered corrugation16 layers · 4.7 mm · 15× the two-layer allowanceThe square twist9 layers · 2.5 mm · 8× the two-layer allowanceThe preliminary base8 layers · 2.2 mm · 7× the two-layer allowanceThe hexagon twist7 layers · 1.9 mm · 6× the two-layer allowanceFold and cut — the triangle7 layers · 1.9 mm · 6× the two-layer allowancetwo layers

The pile, not the panel

Every technique for building a fold out of panels with depth is drawn, described and priced at one crease between two panels. A folded model has two layers nowhere except at its last fold: the printed patterns here reach eight, sixteen, thirty-two and sixty, and the length a thick panel has to find at those creases is not the published allowance but fifty-nine times it.

8 figures
the height is the allowance in millimetres on a 150 mm sheetthe horizontal axis is the fold angle of the driven crease, in radians0.425 mm0.033 mm0.32.4fold angle of the driven creasea budget of 0.02 radians on a 150 mm sheet12.9 times less allowance at the closed endthe mesh is most forgiving where it is doing least

The allowance is spent at the end

A tolerance on a solved mesh was priced at one fold angle, because that is where a tolerance is priced. The surface of solutions turns out not to move as the sheet folds — the free directions at a third of a radian are the free directions at two and a half, to twelve figures — and the price of leaving it rises by a factor of thirteen along the way.

8 figures
the height is the worst amplification anywhere on the sheetone line per crease; the horizontal axis is the driven crease's own fold angle3.010.32.5fold angle of the driven creasea mesh with no two vertices alike6 of 6 creases are worst near the flat sheet0 steps refused as branch changes

The hardest instant

Driving one crease of a quadrilateral mesh settles every other one, and an error in the driven crease arrives elsewhere multiplied. That multiplier was measured once, at one fold angle. Followed along the whole motion it is worst at the flat sheet on twenty of twenty-four creases — and on the Miura the measurement has to refuse to answer.

8 figures
the bar is the share of the population with a folded stateevery pattern in all four passes every condition at every interior vertexthe printed patterns4 of 80 cannot be placed · 0 cannot be ordered · 4 undecidedtwist tessellations2 of 125 cannot be placed · 2 cannot be ordered · 3 undecidedquadrilateral meshes2 of 60 cannot be placed · 4 cannot be ordered · 0 undecidedfold-and-cut patterns5 of 70 cannot be placed · 0 cannot be ordered · 2 undecidedundecided is a real answer here and is not rounded toward either side

A collision is an order

Paper passing through paper is treated here as a thing that happens during a motion and is caught by watching for it. At the flat state it is not an event at all: it is the absence of an ordering, and it can be proved rather than observed. Four of the six quadrilateral meshes this site solves for rigid folding place perfectly and admit no ordering of their nine panels — so every one of them must pass through itself, and none of them was ever driven to find out.

8 figures
the bar is sheet-widths of crease per layer of compactionshorter is a better exchange rate, and the order is nothing like the order aboveThe Yoshimura pattern0.2314.0 of crease · 60.0 layersThe waterbomb tessellation0.4514.3 of crease · 31.5 layersThe preliminary base0.604.8 of crease · 8.0 layersThe Miura fold0.676.2 of crease · 9.2 layersThe tapered corrugation0.796.6 of crease · 8.3 layersFold and cut — the triangle1.451.7 of crease · 1.2 layersThe square twist1.564.7 of crease · 3.0 layersThe hexagon twist1.886.1 of crease · 3.3 layersa corrugation pays less per layer than a base does, and the difference is not small

Fourth of eight, and still not chosen for it

A deployable is sold on compaction: large in use, small in transit. Measured, the pattern that actually gets built converts folding into compaction at 0.67 sheet-widths of crease per layer, which is fourth of the eight printed patterns — nearly three times worse than the Yoshimura, which nobody deploys, and nearly three times better than the hexagon twist, which nobody deploys either. The ranking does not pick out the pattern that flew from anywhere on the shelf, and that is the finding.

6 figures
the sectors are 40°, 140°, 40°, 140° — each row is one way the vertex can start to foldcrease 1crease 2crease 3crease 4mode 10.7100.7102 of the four creases movemode 200.7100.712 of the four creases movethe numbers are the four fold angles' ratios to one another as the vertex leaves the flat state

Two mechanisms at one point

Two creases drawn across each other cannot fold flat — Maekawa's count refuses them at every angle. They move perfectly well as rigid panels, and they move in two ways: bend along one line while the other stays flat, or the reverse. Every other developable vertex of degree four has two ways too, and in both of them all four creases move together at a fixed ratio. The crossing is the case where the two motions have nothing to do with each other.

7 figures
the bar is the vertices the drawing has and the list does notThe preliminary base09 listed · panels closeThe Miura fold035 listed · panels closeThe square twist016 listed · panels closeThe hexagon twist022 listed · panels closeThe Yoshimura pattern045 listed · panels closeFold and cut — the triangle011 listed · panels closeThe tapered corrugation040 listed · panels closeThe waterbomb tessellation041 listed · panels closethe square grid, assembled064 listed · panels closethe triangular grid, assembled1282 listed · panels 1.73 apartthe honeycomb, assembled1884 listed · panels 2.00 apartthe rhombille tiling, assembled12138 listed · panels 1.86 apartthe elongated triangular tiling, assembled576 listed · panels 1.73 apartevery pattern with a bar has panels that cannot be placed, and every pattern without one places exactly

How deep is a crossing

A crossing is a verdict with no middle: two creases either pass through one another or they do not, and the first makes a pattern unfoldable while the second leaves it untouched. Measured on the patches where they occur, the shallowest crossing runs 0.16 mm past the end of the crease it meets, on a sheet 150 mm across. Five of the forty-seven are under half a millimetre, which is thinner than the line a pencil draws.

7 figures
the bar is the nodes the ordering search visitedthe letters are consistent on every one of these, so the one-pass test says nothing about any of themmesh 37,4739 panels · 7,473 nodes · no order existsmesh 58,0079 panels · 8,007 nodes · no order existsmesh 89,3469 panels · 9,346 nodes · no order existsmesh 111,0159 panels · 1,015 nodes · an order existsmesh 141449 panels · 144 nodes · an order existsmesh 199,0629 panels · 9,062 nodes · no order existsa red bar is a pattern with no folded state, found only by visiting every ordering it might have had

Two refusals that refuse differently

Four of the six developable quadrilateral meshes this collection solves have no ordering of their nine panels — they must pass through themselves, and a search over every ordering proves it. On all four, the letters agree with themselves perfectly. The linear proof and the exponential search are not a fast test and a slow one: they answer different questions, and neither contains the other.

8 figures
the bar is how many letterings of the mesh can have their panels stackedout of every labelling of its twelve creases, enumeratedmesh 3016 pass every vertex · 16 agree with themselves · arrived refusedmesh 5032 pass every vertex · 32 agree with themselves · arrived refusedmesh 8832 pass every vertex · 32 agree with themselves · arrived refusedmesh 11832 pass every vertex · 32 agree with themselves · arrived foldablemesh 141416 pass every vertex · 14 agree with themselves · arrived foldablemesh 19416 pass every vertex · 16 agree with themselves · arrived refusedtwo of the meshes have none at all, and two more were refused only at the lettering they came with

Refused at one lettering

Four of six quadrilateral meshes here have no arrangement of their nine panels — established by searching every ordering, at the labelling each mesh arrived with. Enumerate every labelling instead and two of the four fold perfectly well at a different one. What was reported as a fact about four meshes is a fact about two meshes and two labellings.

6 figures
the bar is how many letterings of the mesh can have their panels stackedout of every labelling of its twelve creases, enumeratedmesh 3016 pass every vertex · 16 agree with themselves · arrived refusedmesh 5032 pass every vertex · 32 agree with themselves · arrived refusedmesh 8832 pass every vertex · 32 agree with themselves · arrived refusedmesh 11832 pass every vertex · 32 agree with themselves · arrived foldablemesh 141416 pass every vertex · 14 agree with themselves · arrived foldablemesh 19416 pass every vertex · 16 agree with themselves · arrived refusedtwo of the meshes have none at all, and two more were refused only at the lettering they came with

A search with nothing to reorder

One search on a crease pattern costs eighty steps or fifteen thousand depending on the order it takes its decisions in. The other search on the same crease pattern costs 1,188,571 steps whatever order it is given — twelve permutations of the panels, twelve identical counts. The difference between them is one line of code that neither has and one has.

8 figures
12% folded42% folded72% folded95% foldedno face bends anywhere in the motion — which is what makes it a mechanism rather than a fold

The motion has no letters to choose

A flat-folding search picks a letter for every crease and can pick badly. A rigid folding does not pick anything: the fold angles are real numbers, determined by the panels through equations that have a solution or do not. Replacing a discrete choice with a continuous solve removes every ordering question at once, and introduces a failure of its own.

8 figures
panels with nothing below them, and where they aresquare ×1125 panels, 16 of them touching the edge · all 1 at the edgesquare ×2281 panels, 32 of them touching the edge · all 2 at the edgesquare ×33169 panels, 48 of them touching the edge · all 3 at the edgetriangular ×1369 panels, 39 of them touching the edge · all 3 at the edgetriangular ×25233 panels, 79 of them touching the edge · all 5 at the edgehexagonal ×1469 panels, 39 of them touching the edge · all 4 at the edgehexagonal ×27233 panels, 79 of them touching the edge · all 7 at the edgehexagonal ×310493 panels, 119 of them touching the edge · all 10 at the edgeelongated ×12105 panels, 48 of them touching the edge · all 2 at the edgeelongated ×23369 panels, 96 of them touching the edge · all 3 at the edgethe sheet these letters belong to has no such panel at all

An order with no least element

Enumerating every way a folded pattern can be stacked works by building upward from a panel with nothing below it. The smallest square twist patch has exactly one such stacking and takes eleven thousand steps to find it. The pattern that patch was cut from has no panel with nothing below it at all, so the enumeration has nothing to start from — and the sheet is perfectly well stacked anyway.

8 figures
one period of the square grid's twist tessellationa ring is where a crease leaves and returns on the far side40 crease pieces → 32 creases25 drawn panels → 16 panels16 vertices, every one interiorV − E + F = 0mountainvalleyraw edge

Two panels that are one panel

Paper cannot pass through paper, and every test for it compares pairs of panels. On a glued sheet two pieces of the drawing can be the same piece of paper — so a test that does not know the identification either reports a collision between a panel and itself, or misses one where the sheet meets itself round the loop.

6 figures
the period cell of the Miuraone period, with its neighbours round it2 interior vertices in the cell7 crease pieces drawnperiod 1.000 × 2.000one column wide and two rows high, because the zigzag returns after twothe cell is a rectangle of ordinary paper until somebody says its edges are one edge

A mechanism that closes on itself

A rigid-foldable pattern is a mechanism: panels as rigid plates, creases as hinges, and a motion counted by degrees of freedom at each vertex. Close the sheet into a tube and the mechanism has to come back to itself after a circuit — a constraint that is not at any vertex and that the degree-of-freedom count does not see.

6 figures
Miura solar array17×Space Flyer Unit, 1995airbag folding25×stored for years, opens in 30 msheart stentthreaded through an arterystarshade11×26 m disc, 2.5 m launch tubepackeddeployedthe ratio is what is bought; one degree of freedom is what makes it reliable

The tube that gets built

Every folded structure that leaves a laboratory is a sheet joined to itself — a boom, a stent, a bellows, an airbag, a packed antenna. The mathematics has been done on flat rectangles for the whole history of the subject, and the object is a cylinder, which is a different sheet with different counts and a condition the rectangle does not have.

6 figures
61°108°travel before contact110.0°measured by contact testpanel thickness22% of the panel lengthwhat it gives upmaterial at the crease, sothe panel is thinnest whereit is worked hardesta zero-thickness pattern says the panels meet along a line; nothing that is built does

Thickness round a closed loop

Real panels have thickness, and every technique for accommodating it works by shifting a hinge off the ideal crease by a small amount. On a flat sheet the shifts accumulate outward and end at the edge. On a closed sheet they accumulate round a loop and have to come back to where they started, which is a condition none of the techniques was designed to satisfy.

6 figures
the bar is the share of the footprint at least half as deep as its deepest pilethe note is how deep that pile is and how much of the footprint it coversThe preliminary base100.0%uniform · deepest 8 layers over 99.7%The Yoshimura pattern100.0%uniform · deepest 60 layers over 100.0%The waterbomb tessellation96.6%uniform · deepest 32 layers over 96.6%The tapered corrugation87.9%graded · deepest 16 layers over 0.9%The Miura fold75.5%graded · deepest 16 layers over 11.9%The hexagon twist24.5%island · deepest 7 layers over 24.5%The square twist17.5%island · deepest 9 layers over 17.4%Fold and cut — the triangle3.0%island · deepest 7 layers over 2.9%uniform: the pile is the pattern · island: the deep region is a patch · graded: deepest on a sliver, half as deep nearly everywhere

Three kinds of pile

A thick-panel technique is priced at the deepest pile a pattern has, and the depth of that pile says nothing about where it is. Mapped over the folded footprint, the printed patterns fall into three kinds. On a uniform pile the deepest count is the whole footprint — sixty layers everywhere on the Yoshimura. On an island it is a patch and the rest is shallow. On a graded pile it is a sliver — under one per cent of the tapered corrugation — while nearly nine tenths is at least half as deep, and the Miura, the pattern that gets built, is graded.

8 figures
the bar is the share of the paper in panels that lie over more than one deptheach panel's folded image is laid on the depth map and the depths under it are countedThe preliminary base0.0%uniform · 8 panels · 1 depth under eachThe Yoshimura pattern0.0%uniform · 65 panels · 1 depth under eachThe waterbomb tessellation0.0%uniform · 52 panels · 1 depth under eachFold and cut — the triangle90.0%island · 7 panels · 1 to 3 depths under eachThe hexagon twist92.5%island · 13 panels · 1 to 4 depths under eachThe square twist94.2%island · 9 panels · 1 to 4 depths under eachThe Miura fold100.0%graded · 24 panels · 3 to 4 depths under eachThe tapered corrugation100.0%graded · 28 panels · 4 depths under eacha panel over one depth can be given one thickness; a panel over several cannot

A panel is not the unit of depth

A thick-panel design gives each panel a thickness, an offset or a taper, so a graded pile could be met panel by panel only if every panel's folded image lay over one depth. On the Miura none does. Every one of its twenty-four panels lies over three or four of the four depths its pile takes, and on the tapered corrugation every panel lies over all four. The uniform piles are the opposite — every panel of the Yoshimura, the waterbomb and the preliminary base lies over exactly one depth — which is why panel-by-panel techniques look adequate on the patterns they are drawn for. On the Miura the steps between depths cross the middle of panels, and they run parallel to the panels' own sides.

8 figures
how many folded states each crease of a four-by-four Miura leavesdriven to 0.6 radians, with every consistent assignment enumerated rather than the first eight1 state4an actuator belongs on one of these2 states82 states, so the sheet has a choice4 states44 states, so the sheet has a choice8 states88 states, so the sheet has a choicethe four that leave one are c:3:2, c:3:3, r:3:2, r:3:3 — all of them at the same corner of the sheet

Only four creases decide a Miura

Driving one crease of a rigid quadrilateral mesh settles every other one — except that on the pattern everybody builds it often does not. Enumerated properly, four of a four-by-four Miura's twenty-four creases leave exactly one folded state and the other twenty leave two, four or eight. A mesh whose vertices all differ leaves one from every crease. The ambiguity is not a property of quadrilateral meshes; it belongs to the symmetry.

7 figures
0123400.010.020.030.040.05gearing between the two creasesradians of errormoved at the firstleft at the secondworst at a gearing of onetwo actuators disagreeing by 0.05 radians, equal stiffness · the sheet settles where the stored energy is least

Two drivers and one freedom

Two actuators on a sheet with one degree of freedom are two commands for one number, and if they disagree by a hundredth of a radian the sheet cannot satisfy both. Where it settles is decided by the gearing between the two creases: a strongly geared pair absorbs the disagreement and leaves a quarter of it standing, while a weakly geared pair keeps ninety per cent. The loosest coupling is the expensive one, which is the opposite of what coupling usually means.

5 figures
folding, priced by how often it has to happena crease of radius ρ strains its outer fibre by t ⁄ 2ρ, and a material takes less strain the more often it is askedcycles it must survivesmallest hinge radiusbest fold countpacking it reachesagainst once10.05087.643.8100.15827.713.93× worse1000.5008.7604.38010× worse1,0001.5812.7701.38532× worsesheet 10, thickness 0.1, fatigue exponent 0.5 · the radius goes as N^0.5 and the packing as N^−0.5

What a second deployment costs

Every folded structure this field builds deploys once. The reason is a power law: a hinge asked to survive more cycles cannot be as sharp, a blunter hinge takes more surface out of the sheet, and the fold count that packs best falls as the cycle count to a fatigue exponent. A structure required to work a thousand times packs thirty times worse than one required to work once, and the exponent decides how fast rather than whether.

6 figures
what a finer pattern costs in confidenceeach hinge working 0.999 of the time, against a target of 99 per cent for the whole deploymenthingesthe system openseach hinge needssuccesses to show itand the article itself899.2%0.9987442,385cannot be tested2497.6%0.9995817,154cannot be tested6094.2%0.99983317,885cannot be tested12088.7%0.99991635,769cannot be tested30074.1%0.99996689,422cannot be testedr consecutive successes put a 95 per cent lower bound of 0.05^(1⁄r) on a hinge, and a flight article deploys once

The crease count is a reliability budget

A deployment that needs every hinge to work is the hinge reliability raised to the crease count, so the fineness that buys compaction spends the probability of getting it. At a thousandth of a chance of a hinge failing, sixty hinges give a 94 per cent deployment and three hundred give 74. The fold count that maximises expected compaction is well below the one that maximises compaction — and demonstrating the result takes tens of thousands of successful tests on an article that itself deploys once.

6 figures
the energy two actuators store in a disagreement, pair by pairenergy k₂δ² ⁄ (1 + k₂g² ⁄ k₁); the last column is the second crease's holding stiffness, k₂ + k₁⁄g², kept when k₂ is cut to a tenthgearingbalancing k₂ ⁄ k₁equal stiffnesssecond ten times stifferten times softerstiffness kept0.31610.010.915.000.09992%0.5393.440.772.560.09780%0.7841.630.621.400.09466%0.9341.150.531.030.09258%1.0001.000.500.910.09155%1.7610.320.240.310.07632%a mesh with no two vertices alike, driven at c:0:1 · energy stored in the fight, in units of δ² times the first actuator's stiffness

A gearing reflects stiffness squared

Two actuators on a sheet with one freedom disagree, and the sheet settles where their stored energy is least. With unequal stiffnesses the answer depends on them only through k₂g² ⁄ k₁ — the second actuator, seen from the first crease, is a spring of stiffness k₂g², the gearing entering squared as a gear train reflects any stiffness. That settles which actuator to make compliant. On a rigid mesh's loosest pair a second actuator ten times stiffer than the first stores fifty times the fighting energy of one ten times softer, and softening it gives up only 8 per cent of how firmly that crease is held, because the first actuator already holds it ten times over through the gearing. On the tightest pair softening saves four times the energy and gives up 68 per cent of the hold. Compliance is cheap exactly where the fight is expensive.

6 figures
a 4-by-4 Miura, every crease driven in turn, at 10 angles along the motionevery consistent assignment enumerated at each crease, with both configurations found at every vertexanglecreases × states they leavethe creases that decide it0.24×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:30.44×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:30.64×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:30.84×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:31.04×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:31.24×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:31.64×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:32.04×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:32.44×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:32.84×1 8×2 4×4 8×8c:3:2 c:3:3 r:3:2 r:3:3every row is the row above it: the census is a property of the pattern, not of how far it has folded

The deciding set does not move

A driven Miura leaves several folded states from most of its creases and exactly one from a few, and those few are where an actuator belongs. It was reported that the few change along the motion — four of twenty-four at 0.6 radians, fourteen at 0.8 — and that a five-by-five sheet had a crease leaving fifteen states where every other count was a power of two. Mapped at twenty angles from 0.1 to 3.0 radians on three sizes of sheet, neither survives. Every crease leaves the same number of states at every angle, every number is a power of two, and the same creases decide the sheet throughout. The changes were the vertex solver losing one of a vertex's two configurations on 138 of 8,640 solves, and the configurations it lost can be carried exactly from an angle where it finds both.

5 figures
one freedom, or severala stuck hinge or a failed actuator loses its own module and nothing elsemodulesall of it opensshare expectedat least 90%at least 75%173.3%73.3%73.3%73.3%272.6%85.2%72.6%72.6%570.4%93.2%70.4%96.0%1067.0%96.1%94.4%99.4%2060.6%97.5%98.7%100.0%5044.8%98.4%100.0%100.0%10027.1%98.7%100.0%100.0%300 hinges at 0.999 each, split evenly · each module's actuator works 0.99 of the time

Splitting a sheet buys area, not certainty

A folded deployable with one freedom needs every hinge and its one actuator, and three hundred hinges at 0.999 each open all the way 73 per cent of the time. Split the same hinges among ten separately driven modules and a stuck hinge costs only its own module: the share of the area expected to open rises to 96 per cent, and the chance of at least nine tenths of it rises to 94. The chance of all of it falls, to 67 per cent, because every freedom added is an actuator added. So freedoms, actuators and reliability trade in a definite way: one freedom is the best design only for a mission that is worthless without its whole area, and for any mission that can live with less, several freedoms win by a margin that no improvement in the hinges matches.

5 figures
tests needed for each layer of compaction, pattern by patterna test campaign grows with the hinge count, so the pattern with fewest hinges per layer is cheapest to trustpatternhingeslayersper layertests per layerby testsby lengthThe preliminary base88.01.002981st3rdThe Yoshimura pattern8660.01.434272nd1stThe waterbomb tessellation7631.52.417203rd2ndThe square twist123.03.981,1854th7thThe Miura fold389.24.121,2285th4thFold and cut — the triangle61.25.051,5056th6thThe tapered corrugation458.35.401,6117th5thThe hexagon twist183.35.541,6508th8thtests are consecutive successes demonstrating 99 per cent for the whole pattern at 95 per cent confidence

The pattern cheapest to trust

Demonstrating that a one-shot deployment will open takes a number of successful tests proportional to its hinge count, so the pattern that needs fewest tests for what it delivers is the one with fewest hinges per layer of compaction. That criterion is a count nobody computes, and computed on the printed shelf it ranks the patterns differently from crease length per layer: the preliminary base is first, at exactly one hinge per layer, and the square twist rises from seventh to fourth. As patterns are refined the difference sharpens. The waterbomb settles at 2.47 hinges a layer and the Yoshimura at 1.47, but the Miura climbs without levelling — 1.32 at two cells a side, 5.69 at eight — so every finer Miura costs more tests for each layer it adds, and the pattern that gets built is the only one of the three that gets dearer to trust as it gets finer.

5 figures
the other fault a drawing can have, counted and measureda stub is a crease with a free end; the distance is how far from the rim it stoppedpatcheswith a crossingwith a stubstubsdistinct depthsshallowestdeepest120761566270.27 mm35.1 mm0.27 mm35.09 mm66 stubs27 depthsdistances at the 150 mm these patterns print at; the scale is logarithmic because the range is a factor of 128

A stub is never alone

A crossing is a crease running past another and it has a depth. A stub is a crease that simply stops, and it has one too — how far from the rim it stopped, which is also how much shorter than a crease it is. Measured across a hundred and twenty drawings: sixty-six stubs, from 0.27 mm to 35 mm at printed size, every one of them paired with another at exactly the same distance, and not one on a drawing that did not already have a crossing.

6 figures
drawings carrying each fault, of those drawna junction is split in two: the fault, and the rim ending the reading also calls a junctionextended, of 120clipped, of 120a crossing760two creases pass through each othera stub150a crease stops in the middle of the papera crease ending on a crease00the junction as a faulta crease ending on the rim120120the junction as the reading counts ita fragment25a crease too short to seeevery stub is on a drawing with a crossing; every clipped fragment is on a drawing with nothing else wrong

Two faults, not four

A drawing departs from its crease list in four named ways — a crossing, a stub, a junction and a fragment — and a checker tests for all four. Counted side by side on two hundred and forty drawings, two of the tests find nothing the others do not. No crease anywhere ends on another crease: every junction the reading finds is a crease meeting the rim, which is where creases are meant to end. Every stub is on a drawing that already has a crossing. What is left is two independent faults, and the clipped construction, which makes no crossings, still makes the second — creases a thousandth of a millimetre long, in pairs.

6 figures

Curves and material

Curved creases, developable surfaces, and everything the zero-thickness sheet was lying about.

the patternconcentric arcs, alternatingwhat the sheet doesa shape with no flat state at allthe curve is in the crease; the saddle is the paper refusing to stretch

A crease that curves

Bend a crease and the paper either side is forced into a shape nobody creased. The flat-folding theorems say nothing about it, because they are statements about straight creases meeting at a point.

6 figures
the patternwhat the sheet doesconcentric arcs with their rulings drawn as radii; the metric matches the flat sheet to 5e-7so nothing here is stretching — every point of the surface is where folding alone can put itthe flat-folding theorems say nothing about any of this: they are about straight creases meeting at a point

The sculptors got there first

Curved-crease folding produced its best objects decades before anybody could compute one. The surfaces were made by hand, the ruling lines that determine them were not calculated until much later, and the mathematics has been catching up ever since.

6 figures
cylinderreachablecurved one way onlyconereachablecurved one way, from a pointsphereunreachablecurved two ways — impossiblesaddleunreachablecurved two ways — impossible

What a flat sheet can become

A sheet that cannot stretch cannot become a sphere. That much belongs to differential geometry; what belongs to folding is the three ways round it — seams, curved creases, and a few percent of stretch — and what each one costs.

7 figures
no thicknesslayers add up; a 64-grid model is millimetres thick at the coreno stretchpaper stretches a little, which is why wet-folding works at allcreases are linesa crease has a radius; sharp folds tear and soft ones springperfect memorypaper relaxes, so a model opens slightly the moment it is put downthe theorems are exact statements about a sheet nobody has ever folded

Four things that are not true

Zero thickness, no stretch, creases that are lines, and perfect memory. Every theorem on this site rests on all four, every one of them is false, and the interesting engineering is exactly where each fails.

6 figures
5%10%15%15°30°45°60°75°90°strain the rim must takehow much of a sphere the cap coversdry paper1% of stretchreaches 14°damp paper3% of stretchreaches 24°wet-folded6% of stretchreaches 35°a hemisphere needs 36% and nothing made of cellulose is going to supply it

Paper that stretches on purpose

Wet-folding breaks the assumption every theorem of flat folding rests on, deliberately. It does not repeal the geometry — it buys a few percent of strain, and a few percent of strain is worth about twenty degrees of sphere.

6 figures
ρ = 0.12 mmarc 0.377 mm, stack advances 0.240 mmlost per crease (π − 2)ρ = 0.1370 mmone crease, at its real radiuson a 150 mm sheet8 × 81.0 mm — 0.6%16 × 162.1 mm — 1.4%24 × 243.2 mm — 2.1%32 × 324.2 mm — 2.8%48 × 486.4 mm — 4.3%lost along every line of the grid, in both directionswhich is why an ambitious grid is folded from thin paper

The crease has a radius

A fold does not go through a line. It goes round a small arc, and the arc uses more paper than the stack advances by — a fraction of a millimetre per crease, and several millimetres across a grid, which is why an ambitious tessellation comes out short.

7 figures
2030405060708000.511.522.5cone half-angle β, degrees — 90° is the unfolded sheetcurvature, in units of 1/rκ, in space1/(r sin β) — unboundedκ_g, in the surface1/r — flat, at every anglethe isometry, as a lineκ_n, out of the surfacecot β / r — all of the gainκ² = κ_g² + κ_n² to 3e-15true at every sample,not only at the ends

One curve and one number

Folding cannot change how curved a crease is within the surface — that is what an isometry means. Everything a curved fold produces is the curvature it adds out of the surface, and one number controls all of it.

6 figures
24681012-202halvingsmetres of paper (powers of ten)297 mm — 6 folds, 64 layers1 m — 7 folds, 128 layers10 m — 8 folds, 256 layers100 m — 10 folds, 1024 layers1200 m — 12 folds, 4096 layerspaper 0.1 mm thick · L = (πt/6)(2ⁿ + 4)(2ⁿ − 1)the loss is the paper that goes round the closed end, and it doubles twice per fold

How many times can it be halved

The folklore says seven, and the folklore is a statement about one sheet of paper. What actually binds is arithmetic: every halving doubles the layers and the paper spent at the closed end grows as the square of the layer count, so the length needed for twelve folds is nearly a kilometre.

6 figures
straight creasescurvature 1.4curvature 2.6the tangents are the same in every panelsectors 60.0°, 120.0°, 120.0°, 60.0°they sum to 360.0°, and alternately to 180.0° and 180.0°which is Kawasaki, on tangents rather than on lines

Where curved creases meet

A curved-crease design looks like a smooth object and its constraints are not smooth. They live at the finitely many points where creases cross, and at each of those the conditions are about the creases' tangent directions — the curvature does not appear in them at all.

9 figures
the sheeta wedge of 60° marked for removal60°56.4°what it closes intoa cone of half-angle 56.44°83.3% of the turn is left, and the sine of the half-angle is that same fractionthe circles of latitude are the disc's own, arriving shorter than a flat sheet would needno fold can do this: folding moves paper about and never alters how much of it surrounds a point

What one cut buys

A fold moves paper about and cannot change how much of it surrounds a point. A cut can, and that one difference is the whole of what this site's founding rule is worth. Take a wedge out and the sheet closes into a cone; let one in and it has more paper than the plane will accept.

9 figures
010203040-3-2-10how far open — degrees for the cut sheet, the same fraction of the motion for the foldPoisson's ratiocut into squares−1 everywhere, exactlya fold, slant 0.35-0.12 at the startand without limit at the endthe two cross onceand agree nowhere elsethe flat line is a finite difference of two measured widths, taken the same way as the curve beside ita material made of matter cannot change its Poisson's ratio as it deforms; a material made of geometry can

Bought with holes

A Miura-folded sheet gets wider as it is pulled, and by how much depends on its panels and on how far it happens to be folded. A sheet cut into squares joined at their corners does the same thing and holds the value at exactly minus one, everywhere in its motion — the same property, bought with different geometry, and paid for in holes.

8 figures
the creaseno panels at all128.3° of turning4 segments8 panels128.3° of turning8 segments16 panels128.3° of turning16 segments32 panels128.3° of turningone crease, cut into flat pieceseach panel keeps its width across the crease and loses length along it as the count risesa finer approximation is a better picture and the same total kink, spread over more joints

A curve has no panels

A rigid folding is a finite list of flat pieces joined along lines. A curved crease has no such list, and refining one does not help: the kink at each joint falls as one over the segment count, and the total of the kinks does not fall at all, because it is a constant of the curve.

8 figures
folded 8 times, then unfolded33 interior vertices, all of degree 433 of 33 satisfy Kawasakithe folding is the reason, not the drawing45 creases drawn at random485 interior vertices, all of degree 40 of 485 satisfy Kawasakisame count, same sheet, nothing folded

The creases a sheet gives itself

A crease pattern drawn at random satisfies the flat-folding condition at essentially none of its vertices. A sheet crumpled at random satisfies it at every single one, on every seed, at every size — and the reason is a tautology that is very easy to miss.

8 figures
24681012foldsinterior verticesfacetscrease lengththe median facet falls from 2.6e-1 to 4.0e-3 of the sheet

Every facet is a layer

Fold a sheet at random as many times as patience allows, then count three things: the creases it carries, the facets they cut it into, and the layers in the stack. The last two are the same number, always, and it is one more than the first — so how deep a crumpled sheet folds can be read off the flattened pattern without folding anything.

8 figures
cuts at 85% of the pitch0.15 of material between one cut and the nextcut length, as a fraction of the pitchligament6 piecesone piece at every length up to 97%and the count is worked out from the cuts, not measured off a picture

One cut short of falling apart

Everything a cut sheet can do is bought out of the material between the end of one cut and the start of the next. That material shrinks to nothing in a straight line as the cuts grow, and the sheet stays in one piece the whole way down — until the instant it does not, and then it is in six.

9 figures
a disc, with a vertexa ring, with noneone interior vertex, 3 creases at itodd degree, so they do notno interior vertices at alland the panels still do notboth refuse: two routes round the sheet leave a panel 1.87 sheet-widths apart

A cut that removes no paper

Cuts in this subject are graded. Take a wedge out and the angle at a point falls by exactly the wedge; take twice as much and it falls twice as far. A hole is not like that. Its effect on what the sheet can do is the same whether it is a tenth of the paper or a ten-thousandth, and it is the same because it is not a quantity at all.

8 figures
50% of the stack, 9 foldsthe whole stack, 9 folds13 interior vertices of odd degree13 creases with a loose end before planarisingrefused by the first condition it is put pastno odd vertex anywhereno loose end anywhereand every vertex satisfies Kawasaki

The crease that stops in the middle

A sheet folded flat at random writes a crease pattern that satisfies every condition in the subject, everywhere. Leave one layer behind on each fold — one layer out of a dozen — and it stops writing crease patterns at all: the creases stop in the middle of the paper, and a crease with a loose end is a thing no flat folded sheet can have.

8 figures
the crease, its rulings, and the line where they crossan elliptical crease, rulings at 52° from the tangentthe shortest ruling0.0998the tightest radius0.1274their ratio0.7833nothing on the crease pattern marks this line, and no amount of paper moves it

Where the rulings run out

A curved fold's surface is made of straight lines leaving the crease, and the lines are not parallel, so they cross. Past the first crossing there is no surface: two points of the paper have been sent to one point of space. The boundary is a curve nobody drew, no crease pattern shows it, and it sits at the sine of the ruling angle times the crease's own tightest radius — on every curve tried.

9 figures
two models, and the sheet that has carried both12 crossings, of which 10 cannot fold flatThe preliminary baseThe hexagon twistthe sheet after bothno crease has moved and none has been added; what is new is where they cross

The sheet remembers

Perfect memory is the fourth idealisation, and the least examined of the four. It is usually read as a complaint that paper will not lie flat again; the large half is the opposite. A sheet folded once is no longer blank, so folding a second model into it is folding the union of two patterns — and a union folds flat only where every new crease meets every old one at a right angle.

8 figures
patterncrease length near the grainbest / worstceilingThe preliminary base4 crease directions21% / 21%45°The Miura fold3 crease directions54% / 46%35°The square twist2 crease directions0% / 0%45°The hexagon twist3 crease directions30% / 0%30°The Yoshimura pattern3 crease directions29% / 0%30°Fold and cut — the triangle6 crease directions35% / 11%29°The tapered corrugation3 crease directions42% / 0%33°The waterbomb tessellation3 crease directions21% / 0%45°

The fifth thing that is not true

Four idealisations underlie every theorem here and each has had an essay. There is a fifth and it has never been named, because it is invisible in exactly the way the others are not: paper has a grain, no theorem in the subject mentions a direction, and so nothing in the whole apparatus can tell a folder which way up to lay the pattern down.

9 figures
17%7 circles fit; the middle 17 per cent cannot carry anyspacing 0.06, rulings at 0.9 radians to the crease

The gap between two curves

The rulings leaving a curved crease are not parallel, so they cross, and the surface exists only as far as the first crossing. That bound is usually read as a limit on how far a design extends outward. It is not: the paper between two curved creases has to be reachable from both, so the bound bites hardest where the circles are smallest, and a concentric pleat has a hole in the middle that no sheet size removes.

9 figures
uncut: 4 vertices inside the paper, 6.3% of letterings admittedcut a crease with an interior vertex at each end25.0% admitted2 vertexes released · 4× the share · 4 such creases, all alikecut a crease that already reaches the edge12.5% admitted1 vertex released · 2× the share · 8 such creases, all alikea released vertex is one the four conditions no longer reach, and each is worth a factor of two

A cut is a licence

What a cut buys is usually described in words — freedom, release, a shape a fold cannot reach. It can be counted, and the unit is vertices. Cutting one crease of a square twist turns two interior vertices into vertices no theorem applies to, and the share of letterings the pattern admits goes up by a factor of two for each vertex released: exactly, on every cut tried.

8 figures
the bar is the buried creases a crumple of that depth writestwo sheets at each depth, from two streams2 folds3 creases · 1 pieces3 folds8 creases · 2 pieces4 folds17 creases · 128 pieces5 folds22 creases · 512 pieces6 folds45 creases · 2.68 × 10^8 pieces7 folds61 creases · 1.09 × 10^12 pieces

The decision a crumple has taken

A sheet crumpled at random satisfies every condition in the subject, because it just folded. It also wrote itself a lettering — one of very many the pattern admits — and it is now in a piece of that space it cannot leave: at six folds a crumpled sheet carries thirty-nine buried creases, which is half a million million million pieces, and every change it admits stays inside one of them.

8 figures
The square twist, one crease at a timethe bar is the number of pieces, and a cut anywhere reduces itbefore any cut16 pieces · 0 vertices releasedcutting a crease that reaches the edge4 pieces · 1 vertices releasedcutting a buried crease2 pieces · 2 vertices released

A cut is not local

Cutting one crease of a square twist takes its letterings from sixteen mutually unreachable pieces to two. The cut crease is one of the four that were settled when the pattern was drawn — and it takes two others with it, because the vertices it releases were the far ends of those. Even a cut along a crease that was never settled quarters the count.

8 figures
the bar is the average number of distinct folded statesevery printed pattern on this site has exactly one, and none of its swaps is legal2 folds1.1724 of 24 measured · 2 of 71 swaps legal3 folds1.3824 of 24 measured · 2 of 164 swaps legal4 folds2.1619 of 24 measured · 6 of 335 swaps legal5 folds2.336 of 24 measured · 0 of 121 swaps legala refused row is a sheet with too many panels to search, and refusals are counted rather than dropped

The crumple keeps its options

Every crease pattern this site prints has exactly one folded state and not one of its thirty-nine available rearrangements is legal. A sheet creased by folding it at random four times has an average of 2.16 folded states, one of them has nine, and six of three hundred and thirty-five rearrangements are legal. The sheet nobody designed is the one with room left in it.

8 figures
the bar is millimetres of crease at the printed sizea pattern with more creases is not always a pattern with more folding in itThe preliminary base724 mm8 creases · printed at 150 mmThe Miura fold1,049 mm38 creases · printed at 170 mmThe square twist704 mm12 creases · printed at 150 mmThe hexagon twist916 mm18 creases · printed at 150 mmThe Yoshimura pattern2,380 mm86 creases · printed at 170 mmFold and cut — the triangle258 mm6 creases · printed at 150 mmThe tapered corrugation1,057 mm45 creases · printed at 160 mmThe waterbomb tessellation2,290 mm76 creases · printed at 160 mmsix point seven metres of crease on a sheet seventeen centimetres across

How much line is on the paper

A crease pattern is described by its creases: how many, at what angles, in what arrangement. What a folder spends is length. The Yoshimura this site prints has eighty-six creases and 2,380 millimetres of folding on a sheet seventeen centimetres across; the Miura has thirty-eight creases and 1,049, and the fold-and-cut triangle has six and 258 — and the two counts do not rank the eight printed patterns the same way.

8 figures
each number is the folding in that band against the paper in it — one is an even spreadthe rimband 2band 3band 4the middleThe preliminary base0.550.720.991.694.96The Miura fold0.431.410.751.941.67The square twist0.660.881.182.030.94The hexagon twist0.600.871.242.440The Yoshimura pattern0.791.011.091.211.77Fold and cut — the triangle00.201.313.446.79The tapered corrugation0.611.241.390.731.63The waterbomb tessellation0.890.931.260.951.34bands are equal in depth and not in area: 36% · 28% · 20% · 12% · 4% of the sheet, from the rim inward

Where the length sits

A pattern's folding length is a total, and a total says nothing about where the work is. Divide each printed sheet into bands by distance from its own edge and the answer separates the patterns by kind: a traditional base carries five times its share of folding in the middle 4% of the paper, a tessellation carries between 0.9 and 1.4 everywhere, and a twist unit carries none at all at its centre. On all eight, the outermost band carries less than its share.

6 figures
the bar is how far apart two routes to one panel end up, before the cut3 creases1.751.75 apart · cut open, 0e+05 creases1.591.59 apart · cut open, 0e+07 creases1.431.43 apart · cut open, 0e+09 creases1.321.32 apart · cut open, 0e+011 creases1.241.24 apart · cut open, 0e+0after one cut from the hole to the rim, every one of them places to rounding — with no crease changed

A cut that reaches the edge

A ring of paper with three creases running from its hole to its rim satisfies every condition the subject has — vacuously, because it has no interior vertex at all — and cannot be folded: its panels take no two colours and the two routes to one of them end up 1.75 sheet widths apart. One cut from the hole to the edge, crossing no crease and changing no letter, and it folds exactly. The cut removes an adjacency, which is the one thing neither a fold nor an edge can do.

6 figures
the letters a folding gives a sheet always agree — these are the ones it might have had insteadfolded from seed 7folded from seed 11folded from seed 2300.2500.5000.750110203040panels in the folded sheetshare of redrawn letterings that agreeeach point is one sheet folded a given number of times, and the horizontal axis is what that produced

The letters a crumple was given

A sheet creased by folding it and folding it again arrives with a mountain-valley labelling that cannot be wrong, because a folding produced it. Nothing about the pattern protects it: reletter the same creases and the share of labellings whose letters agree falls from every one of forty at eight panels to eleven of forty at forty-one. The foldability of a crumple is a fact about its history, not about its drawing.

8 figures
the bar is the cuts after which the panels still place at allnone of them clears the contradiction, and no cut of a buried crease leaves a sheet that placessquare0 of 8484 cut, one at a time · 0 still place · 60 are buried and none of those doeselongated0 of 106106 cut, one at a time · 0 still place · 74 are buried and none of those doeshexagonal14 of 142142 cut, one at a time · 14 still place · 100 are buried and none of those doestriangular2 of 142142 cut, one at a time · 2 still place · 100 are buried and none of those doesa cut along a crease removes no paper — the two panels are still there and are no longer joined

One cut removes one arc

A crease pattern whose letters contradict themselves has, in principle, an obvious smallest repair: cut one crease and the statement it was making goes away. Cut every crease of four tessellation patches in turn — four hundred and seventy-four cuts — and sixteen of them leave a sheet whose panels still land anywhere at all. A cut gives the paper a freedom, and a sheet with a freedom in it has no folded state to order.

8 figures
the bar is how many nodes the search visitedone sheet crumpled deeper and deeper, its letters rechosen each time4 folds1716 panels · 34 of 40 random letterings agree · 1 backtracks5 folds1918 panels · 34 of 40 random letterings agree · 1 backtracks6 folds3435 panels · 15 of 40 random letterings agree · 0 backtracks7 folds3839 panels · 19 of 40 random letterings agree · 0 backtracks8 folds7271 panels · 11 of 40 random letterings agree · 2 backtracksthe share that agrees falls by more than half along this ladder; the search's cost tracks the panels and nothing else

Rare is not hard

Crumple a sheet deeper and the share of its labellings that agree with themselves falls from thirty-four in forty to eleven. The number of steps a search needs to find one of them does not move at all: it stays at about one per panel, with no backtracking, the whole way down. How often an answer turns up at random and how much work it takes to find one are different quantities, and a crumpled sheet is where they come apart.

6 figures
each mark is one crease, ranked shortest to longest10⁻⁵10⁻⁴0.0010.010.1a factor of 498, and no crease in itlength, as a fraction of the sheet's side142 creases, rankedthe 12 in magenta are drawn, counted, lettered and put through every theorem, and none of them is visible

The shortest crease is not a crease

A crease pattern's density is usually quoted as total crease length over sheet area, which treats a metre of folding as a metre whether it arrives as one long line or ten thousand short ones. Reading the lengths individually instead finds twelve creases on a printed patch that are shorter than a wavelength of light.

8 figures
each point is one pattern: panels across, nodes up00202040406060one node per panelnodes visitedpanels3 folds to 8 folds, and not one backtrack anywhere in the family

A crumple has no tail

The least structured crease pattern this collection can produce is a sheet folded at random and flattened. Its consistent letterings get rarer as it deepens — thirty-four of forty down to eleven — and finding one costs one step per panel from beginning to end, with no wrong guess anywhere. Disorder and difficulty turn out to be unrelated quantities.

7 figures
how much a cut adds to a crease count, and to a crease lengthsquare ×150.0% too many12 creases counted for 8 · length 12.675 a unit either waysquare ×225.0% too many40 creases counted for 32 · length 12.675 a unit either waysquare ×316.7% too many84 creases counted for 72 · length 12.675 a unit either waytriangular ×141.7% too many34 creases counted for 24 · length 16.938 a unit either waytriangular ×220.8% too many116 creases counted for 96 · length 16.938 a unit either waytriangular ×313.9% too many246 creases counted for 216 · length 16.938 a unit either wayhexagonal ×141.7% too many34 creases counted for 24 · length 17.691 a unit either wayhexagonal ×220.8% too many116 creases counted for 96 · length 17.691 a unit either wayhexagonal ×313.9% too many246 creases counted for 216 · length 17.691 a unit either wayelongated ×130.0% too many52 creases counted for 40 · length 14.654 a unit either wayelongated ×215.0% too many184 creases counted for 160 · length 14.654 a unit either wayelongated ×310.0% too many396 creases counted for 360 · length 14.654 a unit either wayrhombille ×125.0% too many60 creases counted for 48 · length 23.514 a unit either wayrhombille ×212.5% too many216 creases counted for 192 · length 23.514 a unit either wayrhombille ×38.3% too many468 creases counted for 432 · length 23.514 a unit either waythe length is exact because the two halves of a divided crease add back up

A count is not a length

Cut a rectangle out of a tessellation and it reports fifty per cent more creases than the pattern has, then twenty-five, then seventeen — converging on the truth from above and never reaching it. The crease length per unit area it reports is exact at every size, because the two halves of a divided crease add back up. One measurement survives the cut and the other does not.

8 figures
what each sheet's shape costs in conditionsa square0 loopsχ = 1 · no loop that cannot be shrunka slit from the rim0 loopsχ = 1 · the same paper, topologicallyone hole1 loopχ = 0 · one parity conditiona cylinder1 loopχ = 0 · the same sheet as one holetwo holes2 loopsχ = -1 · two independent conditionsa torus2 loopsχ = 0 · two conditions, no rim at alla slit inward from the rim changes nothing, and a closed cut changes everything

A cut is surgery

Two cuts that look identical on the paper do completely different things to the sheet. A slit run inward from the rim changes nothing at all; a closed cut in the middle removes a disc and leaves a sheet carrying a condition it did not have before. What separates them is not the length of the cut or how much paper it removes.

7 figures
two holes, two conditionsone loop is odd — the sheet refuses2 out to the left, 1 to the right, 1 betweenround the left hole: 3 creases, oddround the right hole: 2, evenround both: 3, oddno two-colouring exists0 interior verticesa loop round one hole says nothing about a loop round the other

Two holes are two conditions

One hole in a sheet of paper gives one loop that cannot be shrunk and one parity to satisfy. Two holes give two, and they are independent: an arrangement of creases can satisfy the condition round one hole and fail the condition round the other, and the sheet refuses on the strength of the one it failed.

6 figures
four conditions with nowhere to holddevelopabilityholds, at 0 verticesKawasakiholds, at 0 verticesMaekawaholds, at 0 verticesbig-little-bigholds, at 0 vertices6 of 12 of these bands have no flat folded stateand only the panel colouring can see it

A crease with no vertex to belong to

Crease density is measured as length of line per area of paper, and everything else about a crease is measured at the vertex it runs into. A band of paper has creases that run from one edge to the other and meet nothing, so it has density and no vertices at all — and it still refuses to fold.

6 figures
one rectangle, glued four waysa disc, two cylinders and a torus — from one drawing4 edges lefta disc2 edges lefta cylinder, across2 edges lefta cylinder, alongno edges lefta torusthe same rectangle and the same creases in all four, and nothing in the drawing says which is whichmatching arrowheads mean the two edges are one edge of the paper

The sixth thing that is not true

Five idealisations underlie every theorem here and each has had an essay: no thickness, no stretch, creases that are lines, perfect memory, no grain. There is a sixth, it is more basic than any of them, and it is the one nobody has ever thought to name — the paper is a disc.

6 figures
what each sheet's shape costs in conditionsa square0 loopsχ = 1 · no loop that cannot be shrunka slit from the rim0 loopsχ = 1 · the same paper, topologicallyone hole1 loopχ = 0 · one parity conditiona cylinder1 loopχ = 0 · the same sheet as one holetwo holes2 loopsχ = -1 · two independent conditionsa torus2 loopsχ = 0 · two conditions, no rim at alla slit inward from the rim changes nothing, and a closed cut changes everything

The cut that changes nothing

A slit goes right through the material and leaves the sheet exactly the object it was. A closed cut removes almost no paper and produces a different sheet with a condition it did not have. Kirigami is made almost entirely of the first kind, which is why every result about it survives the distinction untouched.

6 figures
curved tucksthe disc is the flat sheet; the dark lines fold each tuck under, the light line folds it in half8 curved tucks, a cap of 90°hidden at the rim: 36.3%rim thickness on average: 1.571 sheetsthe tuck widens as the cube of the radius

A tuck keeps what a gore cuts

A flat disc gathered into a spherical cap has more circumference than the cap, and a gore removes the excess while wet-folding stretches it away. A tuck folds it under, which keeps the sheet whole and turns the excess into thickness. At the rim of a gathered cap the paper is α ⁄ sin α sheets thick on average — π⁄2 for a hemisphere — and a simple tuck is three, so single tucks reach a cap of 130.6° before they run into one another. And because a sphere's circles fall short of a plane's as the cube of the radius, a tuck that follows the sphere widens as the cube too: its edges are curves.

7 figures
the bar is the worst gap between the hiding straight tucks do and the hiding a sphere needsa cap of 90°, as a share of what the rim hides — every tuck straight, started at evenly spaced radiifrom 1 radius36.9%straight from the centrefrom 2 radii12.3%2.99 times smaller than 1from 4 radii3.3%3.73 times smaller than 2from 8 radii0.8%3.93 times smaller than 4from 16 radii0.2%3.98 times smaller than 8a broken line through a smooth curve is out by the curvature times the square of the spacing

A straight tuck is a cone point

A tuck with straight edges hides length in proportion to how far past its start it has gone, which is a cone's law and not a sphere's. Started at the centre, straight tucks make a cone. Started at several radii, they hide length in a broken line that follows a sphere's cubic, and the worst shortfall falls as the square of the number of starting radii: 36.9 per cent of the rim's hiding from one start, 12.3 from two, 3.3 from four, 0.8 from eight. Every start is three creases at a point, which is a vertex that cannot fold flat — and it is exactly where the gathered sheet's curvature goes.

7 figures
the bar is the worst shortfall in hidden length, as a share of what the rim hidesa cap of 90°, straight tucks started at evenly spaced radii and at the best radii for the same count2 starts, evenly spaced12.3%2 starts, best spaced8.5%31% smaller3 starts, evenly spaced5.8%3 starts, best spaced3.7%36% smaller4 starts, evenly spaced3.3%4 starts, best spaced2.0%38% smaller8 starts, evenly spaced0.8%8 starts, best spaced0.5%40% smaller16 starts, evenly spaced0.2%16 starts, best spaced0.1%41% smallerthe best radii give every stretch between starts the same worst error, which crowds them toward the rim

Crowd the tucks toward the rim

Straight tucks started at several radii follow a sphere's hidden length in a broken line, and evenly spaced starts leave a worst shortfall that falls as the square of their number. Evenly spaced is not the best spacing. A sphere's hiding bends hardest near the rim, so the best starts crowd outward — on a hemisphere, four of them at 0.36, 0.60 and 0.80 of the radius — and leave 38 per cent less error than four evenly spaced. As the count grows the saving closes on 42 per cent, a limit set by the square root of how the sphere's hiding bends; on a shallow dish it approaches five ninths. To follow a hemisphere within one per cent takes six rings of tucks instead of eight, and within a tenth of a per cent eighteen instead of twenty-four.

6 figures
accuracy against thicknessthe pile at a start is the gathering's own mean layers there plus the two a tuck addsplacementstarts atdeepest pilemean pileevenly spaced0.20, 0.40, 0.60, 0.803.323.14placed for equal error0.31, 0.51, 0.68, 0.843.373.19a start is three sheets where it sits, over a gathering already 1.57 sheets thick at the rim

Crowding outward costs almost nothing

Placing the tuck starts for equal error crowds them toward the rim, where the gathered paper is already at its thickest, and the obvious worry is that the accuracy is bought with depth. Measured, it is not: on a hemisphere the crowded placement's deepest start sits in 3.37 sheets against the even placement's 3.32, because a start is three sheets of its own and the gathering beneath it is only one and a half.

6 figures
the same arithmetic three waysm divisions leave a residual of f ⁄ m inside each piece, whatever the divisions are made ofthe material givesdivisions neededas goresas tucksas curved creases2.0%1919 cuts, 59.7 of seam19 tucks, 3 sheets deep19 creases, no cut and no pile5.0%88 cuts, 25.1 of seam8 tucks, 3 sheets deep8 creases, no cut and no pile10.0%44 cuts, 12.6 of seam4 tucks, 3 sheets deep4 creases, no cut and no pile20.0%22 cuts, 6.28 of seam2 tucks, 3 sheets deep2 creases, no cut and no pilecap of 90°, rim excess 36.3% · the count is ⌈f ⁄ ε⌉ in every column; only the cost of a division changes

Three answers, one count

Seams, curved creases and a few per cent of stretch are the three ways round the sphere, and a tuck is a fourth. All four dispose of one quantity — the excess circumference a flat disc has over the sphere's circle — and all four dispose of it by dividing the circle. So the number of divisions needed is the same whichever answer is chosen: nineteen for a hemisphere in a material that gives two per cent, eight at five, four at ten. What differs is what a division costs, and one of the four runs out.

8 figures
00.20.40.60.811.21.41.600.10.20.30.4arc from the pole (radians)excess, as a share of the circle7 rings5.0% stretchfirst at 0.35last at 0.98of the way to the rima ring goes in wherever the residual excess would otherwise pass what the material takes

Where a ring of divisions belongs

A pattern that divides the circle everywhere as finely as its rim requires is over-divided for most of its radius, because the excess grows from nothing. Putting a ring of new divisions in wherever the residual would otherwise pass what the material takes gives seven rings on a hemisphere at five per cent of stretch, at 0.35, 0.50, 0.62, 0.72, 0.81, 0.90 and 0.98 of the way out — and the first of those sits where a completely different criterion put its first tuck start.

6 figures
the length a pattern reports, and the length a sheet of paper hassix of the eight are built on a unit square, so on those the distinction does not arisepatternown widthraw lengthreportedon paperThe Miura fold6.3739.36,6791,049out by 6.37xThe tapered corrugation1.187.81,2431,057out by 1.18xThe preliminary base1.004.8724724the sameThe square twist1.004.7704704the sameThe hexagon twist1.006.1916916the sameThe Yoshimura pattern1.0014.02,3802,380the sameFold and cut — the triangle1.001.7258258the sameThe waterbomb tessellation1.0014.32,2902,290the samea builder working in cells returns a pattern several units across, and not dividing by that is the whole of the error

A length needs a scale

These essays measure crease length, and a crease length is a length in the pattern's own coordinates. Six of the eight printed patterns are built on a unit square, so their coordinates are sheet widths and the distinction never arises. Two are not — a Miura laid out as six cells of unit width spans 6.37 — and on those two the shelf multiplied by the printed size without dividing by the width. The Miura's folding length was reported as 6,679 millimetres and is 1,049, and the same pattern's printable sheet has carried the right number all along.

5 figures
metres of crease a square metre: what each pattern asks for, and what each paper allowsthe pale bars are patterns and the dark ones are paperswhat foil-backed tissue allows641026 µm, a crease 0.16 mm acrosswhat washi allows416740 µm, a crease 0.24 mm acrosswhat kami allows238170 µm, a crease 0.42 mm acrosswhat copier paper allows1667100 µm, a crease 0.60 mm acrossThe waterbomb tessellation89printed at 160 mmThe Yoshimura pattern82printed at 170 mmThe tapered corrugation41printed at 160 mmThe hexagon twist41printed at 150 mmThe Miura fold36printed at 170 mmThe preliminary base32printed at 150 mmThe square twist31printed at 150 mmFold and cut — the triangle11printed at 150 mma crease occupies about 6 sheet thicknesses, so the closest two creases can be laid is that, and the ceiling is its reciprocal

The density a paper allows

Every density these essays measure is a quotient a pattern hands over, and nothing has asked what the paper's own answer is. It has one: a crease occupies a band a few thicknesses across, so two creases closer than that are not two creases, and a sheet of a given thickness carries a largest density. Copier paper allows 1,667 metres of crease a square metre and the densest pattern on the printed shelf asks for 89 — a factor of nineteen below the worst paper's ceiling. The material is not what limits a crease pattern's density at any fineness anybody folds.

6 figures
crease density, the closest two creases come without meeting, and the two multipliedthe closest approach is between creases with no vertex in common, measured as segmentspatternm a m²closest mmproductThe Yoshimura pattern8224.52.02The waterbomb tessellation8920.01.79The square twist3136.11.13The Miura fold3625.10.91The hexagon twist4122.10.90The tapered corrugation4118.60.77parallel creases give a product of exactly one; a pattern above one carries more length than its own spacing would suggest

A paper limits spacing, not density

The density ceiling put a sheet's limit at parallel creases a crease-width apart, 1⁄w of crease a square metre, and claimed no arrangement carries more. Crossing families do: they meet at vertices, which is shared ground, and never come closer than w anywhere else. Measured over every pair of creases that do not share a vertex, density times closest spacing settles at about 1.9 for both the Miura and the waterbomb, so each can be folded nearly twice as fine as the density bound said — 262 cells a side on copier paper for the Miura rather than 139, and 141 for the waterbomb rather than 74.

8 figures
the crease each pattern has least room on, once the ground near its ends is creased twicea narrow sector pushes the overlap out along both creases, as one over the sine of the anglepatternnarrowest sectorreach, both endsthat crease, mmroom to shrinkThe tapered corrugation65.9°2.1020.416×The waterbomb tessellation45.0°2.4128.320×The hexagon twist60.0°2.1525.520×The Yoshimura pattern60.0°2.3128.320×The Miura fold69.9°2.0626.722×The square twist90.0°2.0036.130×Fold and cut — the triangle58.2°1.1828.640×The preliminary base45.0°1.4175.088×reach is in band widths; room to shrink is the crease's length over that reach, on copier paper with a band 0.6 mm wide

A vertex creases the paper twice

Two creases that meet share ground near the point, and their bands overlap out along each of them to w⁄sin θ for a sector angle θ below a right angle. Add the overlap at both ends of a crease, and a crease no longer than that is overlap from end to end — a crease only in the drawing. That third bound binds before the spacing does: the finest Miura on copier paper is 135 cells a side by its vertices against 262 by its spacing, and the finest waterbomb 82 against 141. Both land within a tenth of the density bound, which had the wrong argument and nearly the right number.

6 figures
how far each of a curved fold's two surfaces reaches before its rulings crossas a share of the crease's own tightest radius of curvatureone surfacethe othershare boundeda circular crease at 0.40.389100% / 0%a circular crease at 0.70.644100% / 0%a circular crease at 10.841100% / 0%a circular crease at 1.30.964100% / 0%an elliptical crease at 0.40.389100% / 0%an elliptical crease at 0.70.644100% / 0%an elliptical crease at 10.841100% / 0%an elliptical crease at 1.30.964100% / 0%a parabolic crease at 0.40.389100% / 0%a parabolic crease at 0.70.644100% / 0%a parabolic crease at 10.841100% / 0%a parabolic crease at 1.30.964100% / 0%a wave at 0.40.3890.38950% / 50%a wave at 0.70.6440.64450% / 50%a wave at 10.8410.84150% / 50%a wave at 1.30.9640.96450% / 50%a dash is a surface whose rulings never converge, which is a surface with no boundary of this kind at all

Only one side can run out

A curved fold has two surfaces and every reach ever computed here has been one of them. The closed form's denominator is the crease's curvature plus the rate the ruling angle turns at, and crossing to the other surface negates both — so at any point of any crease at most one of the two surfaces can have its rulings converge. A crease that never changes the way it bends therefore has a surface with no such boundary at all, anywhere along it.

5 figures
what happens when the ruling angle is not constantthe share of the crease each surface is bounded over, and how far it reaches thereturning rateone surfaceits reachthe otherits reach0.00100%0.29560%0.25100%0.23500%0.50100%0.19090%0.75100%0.14840%0.90100%0.11990%1.00100%0.09930%1.1086%0.077314%2.95101.2580%0.041320%1.17461.4075%0.000025%0.7250the crease is a circular crease and the angle runs 1.4 plus the rate times a sine, so the rate is how fast it turns against how fast the crease bends

An angle that turns faster than the crease

Which of a curved fold's two surfaces runs out is decided by a sum of two rates — how fast the crease bends and how fast the ruling angle turns — and every measurement so far has set the second to zero. Let it turn and it carries the sign on its own: past a rate of exactly one, a crease of unchanging curvature bounds both of its surfaces, which no constant angle on that crease can do. Below that rate the turning costs reach without changing anything else.

6 figures