Fifty years in the wrong language

In 1936 Margherita Piazzolla Beloch published a short paper on using paper folding to solve geometric problems. In it is the construction that makes the whole of this subject’s algebra interesting: a single fold that places two given points onto two given lines at the same time, which is a common tangent to two parabolas, which is a root of a general cubic.

That operation is the sixth Huzita axiom. It is the one a straightedge and compass cannot reach. It is the reason folding trisects an angle and doubles the cube and constructs a regular heptagon, and it is the entire content of what this site’s construction field is about.

The subject learned about it in 1991.

What the fold does #

The mechanism deserves restating, briefly, because the rest of the essay is about the cost of not knowing it and the cost is only legible if the thing is understood.

A fold that brings a point onto a line makes the fold line tangent to a parabola: the point is the focus, the line is the directrix, and every point of the fold is equidistant from both. That is the fifth axiom and it is quadratic, which puts it exactly where a compass already is.

Now demand two such conditions at once, with one fold. The fold line must be tangent to two parabolas simultaneously. A common tangent to two conics is a condition of degree three in the slope, so the fold solves a cubic — and there may be one such fold or three, which is precisely the root structure of a cubic.

Beloch’s actual claim #

Beloch’s paper does not present this as a curiosity. It presents it as a method: given a cubic, construct the two parabolas whose common tangent has the required slope, fold, and read the root off the fold line.

She goes further and connects it to Lill’s method — a nineteenth-century graphical technique for finding the roots of a polynomial by bouncing a ray off a path built from the coefficients — which is where the cubic’s coefficients come from geometrically rather than by fiat.

So the 1936 result is not a fragment that somebody later recognised as important. It is the complete argument, with its algebraic context attached, in a form a modern reader recognises immediately.

Why nobody found it #

The obvious explanation is language, and the obvious explanation is not sufficient.

The paper is in Italian. That is a barrier and it is not a wall — mathematicians read Italian papers, and plenty of results from Italian geometry of that period circulated perfectly well. What actually happened is more specific and more instructive.

The paper is in a mathematics venue, addressed to geometers, about a technique for solving classical construction problems. The people who eventually needed it were not geometers working on construction problems; they were people interested in paper folding, coming from origami, arriving at the mathematics from the other side, three decades later. Those two populations had no overlap, no shared journals, and no reason to search each other’s literature.

A result is findable by somebody who knows what to look for. Nobody in the origami community in 1975 knew that a paper about the algebraic reach of ruler-and-compass constructions might contain the thing they wanted, because they did not yet know that was the question.

What it cost #

The counterfactual is worth being concrete about rather than hand-waving.

Had the result been current in, say, 1970, the axiomatic treatment of single-fold operations — what a fold can specify, how many such operations there are, and what algebraic degree each reaches — would have been available at the start of the modern period rather than at its middle. The completeness argument for the seven axioms is essentially an enumeration, and enumerating is much easier once somebody has pointed out that the interesting case exists.

The reach results follow directly. That folding constructs exactly the Pierpont primes’ polygons, that it doubles the cube, that it trisects — each of these is a corollary of the cubic, and each was rediscovered in the 1970s and 1980s by people working without it.

Who Beloch was, and why it matters here #

The biographical note is usually the least interesting part of an essay like this. Here it is load-bearing, because it explains the venue.

Margherita Piazzolla Beloch was an algebraic geometer, working in a strong Italian tradition, on questions about curves and surfaces. The folding paper is not a departure into recreation; it sits inside a line of work about constructions and what they can reach, which was a live question in that school. Folding appears in it as a tool with an algebraic degree, in the same way a compass has one.

That is exactly why the paper is where it is, and exactly why it was invisible. It was written by somebody with no interest in paper folding as a practice, for readers with none either, and its subject is the algebra. Nothing about its title, venue or vocabulary signals to a folder that it concerns them.

The lesson is not that she published in the wrong place. She published in precisely the right place for what she was doing. The mismatch was created later, by the arrival of a community whose interests overlapped hers and whose vocabulary did not.

The difference between lost and unread #

This is worth separating from the parallel-discovery cases, because it is a different failure and has a different fix.

Husimi, Justin and Kawasaki arriving at the vertex conditions independently is what happens when a field has no venue: the results existed in several places and none of them could see the others. Nothing was lost; the same thing was found repeatedly.

Beloch is not that. There is one paper, it is correct, it is complete, it is in a normal journal with a normal date, and it sat there. The failure is not of publication or of proof but of search — and search failures are the ones that scale badly, because the amount of correct published mathematics grows and the ability of any individual to survey it does not.

How the rediscovery actually happened #

The 1991 route to the same operation went the other way round, which is the detail that makes the two accounts fit together.

Huzita was not searching the algebraic literature and finding Beloch. He was enumerating: asking what alignments of existing points and lines a single fold can be specified by, and working through the combinations. The sixth operation falls out of that enumeration as a case — two points, two lines — and it falls out whether or not anybody knows it is cubic. Its algebraic significance is then noticed afterwards.

So the two discoveries approach the same object from opposite ends. Beloch starts from the algebra and asks what tool reaches it; Huzita starts from the tool and asks what operations it has. Both are complete arguments and neither requires the other.

That is worth knowing because it defuses a natural but wrong reading of this essay — that the field was careless. Two independent, valid routes to a result, half a century apart, is what a subject looks like when the question is natural from more than one direction.

The obstruction was knowing the question #

The sharpest way to state what went wrong is that the missing thing was not the answer.

Somebody in 1975 with access to every library in the world and unlimited time would still not have found Beloch’s paper, because finding it requires searching for the algebraic degree reachable by a single fold, and that phrase is the answer’s own vocabulary. The question in that form did not exist in the origami community until the community had started to think axiomatically — which is to say, until it was already about to derive the result itself.

That is a genuinely uncomfortable observation, because it means the fifty-five years were not straightforwardly wasted. A result whose search nobody can formulate is not, in a practical sense, available at all.

Three dates, and two different silences #

The gap is usually quoted as one number, and it is two, with different causes — which is the whole practical content of the story.

1837. Pierre Wantzel proved that trisecting a general angle is impossible with straightedge and compass, and the proof works by showing the problem is a cubic while the tools reach only repeated square roots. That proof poses the question this essay is about, in its own vocabulary: if the obstruction is a degree, which tool clears it?

1936. Beloch answers it. A fold reaches degree three, and here is the construction.

1991. The answer reaches the people who most wanted it.

So the first silence is ninety-nine years long and the second fifty-five, and they are not the same kind of thing. The first is a question that was well posed and simply not asked of paper — nobody thought a sheet was a construction instrument, so nobody put it in the sentence Wantzel’s proof leaves open. The second is a question asked loudly, by a community that had the vocabulary for it, against an answer already in print.

Only the second is a search failure. The first is a failure of imagination and it is the more excusable of the two, because there was nothing to find.

Which query would have worked #

That split says something about how to look, and it is more useful than the general observation that fields do not read each other.

Searching the method fails: “fold” in a geometry index of 1975 returns bookbinding. Searching the result fails differently — “common tangent to two parabolas” finds the paper and is a phrase only somebody holding the answer would type.

What sits between them is the problem. Beloch’s paper is framed as solving the classical construction problems, so a query on trisection by paper folding, or simply on trisection with an eye for unusual instruments, lands on it. That query needs no knowledge of parabolas, no knowledge of cubics and no Italian beyond the title.

A shared goal is the thing two communities have in common when they share no vocabulary, and it is therefore the right handle. Neither the technique nor the theorem crosses the gap; the problem does, because it is what made both fields interested in the first place.

What the modern name records #

Humiaki Huzita presented six axioms at the first international meeting on origami science, held in Ferrara in 1989 and published in 1991. Koshiro Hatori added the seventh in 2001, and Robert Lang later proved the list complete.

The names are attached to that sequence, and by the argument of the previous rung they are recording something real: that is when the operations entered general circulation in a form the field could use. What they are not recording is 1936.

The correct formulation, which this site uses, is that the sixth operation is Beloch’s fold and the axiom list is Huzita’s and Hatori’s. That splits the credit along the seam where the work actually divides — one person found the operation, others found that there were exactly seven of them — and it happens to be historically accurate as well.

A result can be recovered; a decade cannot #

There is a small mercy in this story, which is that mathematics is unusually robust to this kind of loss.

Beloch’s result was not damaged by being unread. When it was found again the paper was still correct, still complete, and still readable, and the field simply absorbed it and adjusted the attribution. Compare a lost experimental technique or a lost craft skill, neither of which survives being forgotten.

What does not recover is the sequence. The fifty-five years in which the subject did not have the result are years in which it asked different questions, went in different directions, and reproved things. That is not recoverable and it is invisible, because a counterfactual field leaves no trace.

What a search failure costs a field now #

The general form of this story is worth extracting, because the conditions that produced it have not gone away.

Beloch’s paper was invisible because the field that needed it did not share vocabulary with the field that had it. That is not a problem indexing solves. A full-text search for “fold” in 1975 would have returned nothing useful; a search for “common tangent to two parabolas” would have found the paper and would only have been run by somebody who already had the answer. The barrier is conceptual, and the modern equivalents are everywhere: results in one applied literature that would settle a question in another, phrased so that neither community can query the other.

This subject has at least one further instance. The diamond pattern a crushed cylinder falls into was published in aeronautics in 1951 and entered the folding literature much later, by the same mechanism and for the same reason — a correct, findable result, addressed to readers who were not the ones who needed it most.

What this figure asserts #

The construction is re-run rather than illustrated, and the assertions are worth naming because they are what makes the essay’s claim checkable at all.

The cubic’s real roots are found by bracketing and bisecting to two hundred halvings, which is well past double precision and finds all three when there are three rather than assuming a count. For each root the generator then computes the two tangency conditions independently and requires them to agree to a part in a thousand million. A fold line that merely looked tangent would fail that, and a figure drawn from coordinates that looked right would fail it immediately.

So the picture is Beloch’s argument executed, not a diagram of it. That is the only way this site knows how to make a historical claim about a construction: run the construction.

What the picture cannot show #

The figure draws the construction and it cannot draw the thing the essay is about.

A tangency is visible. A fifty-five-year silence is not, and there is no honest way to put it on a diagram — an empty region of a chart is a claim about what does not exist, which is exactly the claim no record can support. The bars in the attribution figure come as close as this site gets, and even they draw two dated points and a line between them rather than the absence itself.

So the essay’s central quantity is the one quantity none of its figures contains. That is worth admitting plainly rather than dressing a gap in a chart and letting it look measured.

The idealisation, named #

The essay treats “the earliest source anybody can point at” as the date of the result, and the phrase is doing work.

Beloch’s paper is the earliest anybody currently points at. Nineteenth-century work on the algebra of geometric constructions is dense and largely unread, Lill’s method is of 1867, and the possibility that some earlier author made the connection between a common tangent and a folded line is entirely open. This field has already had one fifty-five-year surprise; a second would be unremarkable.

So the fifty-five years is a lower bound on the gap and not a measurement of it — which is the standing condition of everything in this field.

Where this goes next #

A record is not a proof closes the attribution ladder by stating what none of these four essays can do. Sideways, where the cubic comes from is the mathematics without the history, and why the list stops at seven is the completeness argument that Beloch’s operation is the interesting member of.

The surprising connection, saved for last: the thing that made Beloch’s paper hard to find is the same thing that makes it important. It is a paper about what algebraic degree a construction tool reaches, filed under classical geometry — and folding turned out to be a construction tool. The subject did not fail to find an origami paper. It failed to notice that it was doing algebra.