The paper had to arrive first

There is a persistent idea that origami is limited by imagination and patience — that a folder good enough, given enough time, could have produced anything at any point in the tradition’s history.

The arithmetic says otherwise, and it is not close.

Where the layers go #

Layers accumulate, and they accumulate worst exactly where a design is most ambitious.

Every fold doubles the paper locally. A flap that has been narrowed four times sits on sixteen layers, and the tip of a long thin point in a complex model may carry sixty-four or more. Thickness accumulates toward the extremities, which is the opposite of convenient, because the extremities are where the fine work is.

So the layer count at the thickest point is a property of the design, and it rises steeply with the number of flaps and with how narrow they are.

There is a second reason the extremities are the binding case. A flap’s tip is not only the thickest paper in the model; it is also the smallest feature, so it is where the ratio of stack to feature size is worst by both measures at once. The design’s most delicate region and its most congested region are the same region.

The working limit #

The other half of the arithmetic is a number folders know without writing down.

A fold stops behaving when the stack becomes comparable to the smallest feature being folded. Below that the paper bends where it should crease, the layers slide against each other, and the model refuses to hold a shape. The working figure is of the order of a few millimetres for a model with features a centimetre or two across.

That is a soft limit rather than a physical constant, and it is soft in a way that does not help: a folder can push past it a little with skill and cannot push past it by a factor of two with any amount.

Why the count doubles rather than adds #

The steepness deserves a paragraph, because it is what makes the constraint bite so hard.

Narrowing a flap is a halving operation: each narrowing folds the flap in on itself and doubles the layers at its tip. So layer count goes as two to the power of the number of narrowings, and the number of narrowings is roughly what makes a point thin.

A base with four narrowings per point sits on sixteen layers. Six narrowings is sixty-four. Eight is two hundred and fifty-six, which is a centimetre and a half of copier paper at a single point and is simply not a fold.

That exponential is the reason the constraint is not negotiable. Doubling the available thickness budget buys exactly one more narrowing, and a designer who wants noticeably finer points needs the paper to get thinner by the same factor again.

The sevenfold difference #

Put the two together and the result is stark.

At a working limit of three millimetres: ordinary copier paper at a hundred microns reaches sixteen layers. Kami, the standard coloured folding paper, at seventy microns reaches thirty-two. Fine washi at forty microns reaches sixty-four. Unryu tissue at eighteen microns reaches a hundred and twenty-eight.

A sevenfold spread in thickness is a threefold spread in the number of doublings available, and doublings are what a design spends. A folder with copier paper and a folder with tissue are not doing the same activity with different materials; they have different sets of reachable models.

The claim this licenses #

The historical consequence follows directly and is unusually clean for this field.

A model requiring sixty-four layers cannot be folded from a substrate that runs out at sixteen, by anybody, ever. So a claim that elaborate work is very old is simultaneously a claim about the paper available at that date — and the paper claim is sometimes checkable when the folding claim is not.

That is a rare thing in a field where most claims rest on a single document: an argument about what was possible, resting on material properties rather than on attestation.

The crease has a width, and it costs too #

There is a second material effect running alongside the stack, and it constrains the opposite end of the design.

A crease in real paper has a radius rather than being a line, and the paper consumed by that radius is lost from the model. The loss per crease is small; the number of creases in a dense design is not, and the fraction of the sheet consumed grows with how finely the design is divided.

So a design has two thickness bills to pay: layers at the extremities, and crease radius across the whole grid. Both fall with thinner paper, which is why the substrate improvement bought so much at once.

None of which is to say skill is irrelevant. A better folder gets closer to the limit and gets there more reliably; what no folder does is move it.

What the tradition actually used #

The papers matter and the tradition’s choices make sense in this light.

Washi, made from long kozo fibres, is thin and unusually strong in tension for its weight. That combination is precisely what the arithmetic asks for — thinness to keep the stack down, tensile strength so a thin sheet survives being creased repeatedly at the same point.

Ordinary Western wood-pulp paper of the same thickness is much weaker, because short fibres give a sheet that tears along a crease. So the two traditions were not working with the same constraint even at equal thickness.

The other limit, for comparison #

There is a second thickness argument in this subject and the two are worth holding apart.

How many times a sheet can be halved is bounded by the length of the sheet, because paper is consumed at the closed end as the square of the layer count. That bound is about a single sheet folded repeatedly in one direction, and its answer is a length rather than a count.

This essay’s bound is different: it is about the stack at the thickest point of a model with many independent folds, and its answer is a layer count. The two share the observation that thickness is the binding constraint and they bind on different quantities.

The reach is a logarithm #

The doublings can be turned into a formula, and doing so explains why the constraint feels as immovable as it does.

A stack of $2^{k}$ layers is $2^{k}t$ thick and has to stay under the working limit $\ell$. So the number of narrowings a paper permits is

$$k \le \log_{2}\!\left(\frac{\ell}{t}\right)$$

At a three-millimetre limit that is 4.9 narrowings for copier paper, 5.4 for kami, 6.2 for fine washi and 7.4 for unryu tissue. Rounded down, because half a narrowing does not exist: four, five, six and seven.

So the whole spread of substrates — a factor of five and a half in thickness, from the paper in an office printer to the thinnest sheet anybody folds — is worth three narrowings. That is what centuries of papermaking bought, and it is a small number because it sits inside a logarithm.

Which says how much each improvement is worth #

The logarithm also prices the marginal decision the last section describes, and prices it slightly higher than the intuition does.

A sheet twenty per cent thinner has $\ell/t$ larger by a factor of 1.25, and $\log_{2}1.25$ is 0.32 — about a third of a narrowing rather than a fifth. Three such improvements compound into one whole extra narrowing, which is why folders chase margins that look absurd from outside.

It also says why size is the weak lever. Scaling a model up multiplies $\ell$ rather than dividing $t$, and it enters the same logarithm: doubling the sheet buys exactly one narrowing, and a tenfold enlargement — which no one can handle — would buy 3.3. Thinner paper and bigger paper are the same purchase at different prices, and only one of them is available in quantity.

Why one material change bought two things #

There is a reason the substrate improvement mattered more than this arithmetic alone suggests, and it is that thickness appears in both of the essay’s bills.

Halving the sheet’s thickness doubles the layer budget, which is one narrowing. It also halves the crease radius, since the radius of a fold scales with the material’s thickness — so the paper lost across the whole grid halves as well, and a design can be divided more finely before the losses matter.

One change to the substrate relaxes both constraints at once, and they are constraints on opposite ends of the design: the stack binds at the extremities and the crease radius binds across the field. That is unusual. Most improvements in a designed object trade one limit against another, and this one moves both in the same direction, which is why the history reads as a step change rather than as a gradual accumulation of technique.

What thin paper costs #

Thinness is not free, and the trade explains why folders do not simply use the thinnest thing available.

A thinner sheet is weaker, tears more easily at a crease, and holds a shape less well — paper’s memory is one of the four idealisations, and a very thin sheet has less of it. Foil-backed tissue exists precisely to resolve that: a metal layer supplies the memory the tissue lacks, at the cost of behaving nothing like paper.

So the substrate constraint is a two-sided one. Too thick and the stack binds; too thin and the model will not stand up. The papers folders converge on are the ones that sit in the window, and the window has moved as manufacturing has.

Size is not a way out #

The obvious escape is to fold bigger, and it is worth closing off because it nearly works.

Scaling a model up by a factor of ten with the same paper multiplies every feature by ten while the stack stays the same, so the ratio that binds improves by ten. That is a genuine gain and it is how large display pieces get made.

What it costs is everything about handling. A two-metre square of tissue cannot be turned over on a table, cannot be creased accurately along its diagonal by one person, and tears under its own weight when lifted. The practical ceiling on sheet size is set by the folder’s arms and by the paper’s tear strength, and it arrives well before the thickness constraint has been escaped.

So size buys perhaps one extra narrowing in practice, against the three or four that thinner paper buys. It is a real lever and a short one.

What a manufacturing date buys #

The date-bearing claim is about when paper of a given thickness and strength could be made in quantity, and it is a claim about industry.

Handmade papers can be very thin, and were: fine washi is old. What is recent is consistency — sheets of uniform thickness, in useful sizes, reliably, at a price that makes ruining one acceptable. A folder who must succeed on the first attempt is not going to attempt a hundred-step model.

So the constraint that changed is not only thinness but cheapness and uniformity, and those arrive with industrial papermaking. That is the sense in which the complex tradition is downstream of manufacturing.

What the generator asserts #

The figure is arithmetic and its checks are correspondingly simple, which is worth being clear about.

Every paper must have a positive measured thickness. The stack for each layer count is computed rather than tabulated. And the substantive assertion: the thinnest paper in the comparison must reach strictly more layers than the thickest, and the papers must span more than a factor of three in thickness — because a figure whose whole claim is that the substrate decides would be arguing nothing if the substrates were all alike.

That last check is the one that could fail on a badly chosen set of papers, and it is the reason the set is not chosen for tidiness.

The same constraint, at industrial scale #

The argument is not confined to art, and the engineering version is where it is most consequential.

Every folded structure that gets built meets the same arithmetic with the sheet replaced by a panel, and the constraint is far harsher: a solar array’s panels are millimetres thick rather than microns, so the layer budget is single digits and thickness accommodation becomes the central design problem rather than a nuisance at the extremities.

So the history and the engineering are the same argument at different scales. Paper folding got its complexity from a substrate that became thin; deployable hardware cannot, and therefore had to invent geometry instead.

The idealisation, named #

The model here treats a stack as n sheets thick and real folded paper is not so obliging.

Layers compress: sixty-four sheets of tissue under a folder’s fingers occupy less than sixty-four times one sheet, by a factor that depends on the paper and the pressure. Layers also do not distribute evenly — a real model has regions of two layers and a small region of sixty-four, and the binding constraint is local rather than global.

Both effects make the true limit softer and neither changes the direction or the order of magnitude. The number to trust here is the ratio between papers, which is what the essay’s argument rests on, rather than the absolute layer count for any one of them.

What a folder is optimising #

There is a last observation about the trade that explains a piece of practice which otherwise looks like fussiness.

Competition and exhibition folders are famously particular about paper, often making their own, and the standard reading is aesthetic — a nice surface, the right colour. The arithmetic suggests something else is going on. A folder choosing between two sheets is choosing between two layer budgets, and a twenty per cent reduction in thickness is a fifth of a doubling, which at the margin is the difference between a design working and not.

That is a quantitative decision dressed as a preference, and it is made by people who have never written the arithmetic down. It is the same relationship the diagram notation has to the simple-fold machine model: practice arriving at the right answer to a question nobody had posed formally.

What this cannot settle #

The argument bounds what was possible and says nothing about what was done.

Thin, strong paper existing at a date does not mean anybody folded sixty-four layers of it. The constraint runs one way only: it rules things out and licenses nothing. So the honest form is elaborate work could not have preceded the paper, and not it followed it.

That is weaker than it first looks and is still useful, because the claims in circulation are mostly of the form this rules out.

A bound that runs the right way #

It is worth ending the argument by noting what kind of claim it is, because this field has few of them.

Most historical reasoning in this subject is positive: a source says a thing, therefore the thing was so. That reasoning is only as good as the source, and the sources here are thin.

This one is negative: the arithmetic rules a class of models out, whatever any source says. A negative bound does not need the record to be complete, does not care what failed to survive, and cannot be overturned by a manuscript turning up — only by a measurement of a sheet of paper being wrong.

That makes it the most robust claim in the field, and it is worth noticing that its robustness comes from being about material rather than about people.

Where this goes next #

Sideways, the sheet has a thickness is the same constraint read as an engineering problem rather than as a historical one, and getting thickness round a corner is what happens when the sheet becomes a panel.

The surprising connection: this is the only argument in the whole history field that does not depend on a document. Every other claim here is as good as its sources and no better; this one is as good as the arithmetic, which is exact, and as good as the measured thickness of a sheet of paper, which anybody can check with a micrometer. In a field built on survivals, the strongest evidence turns out to be a material property.