John Rector’s Math Lab

Contents

1. The rules of the lab
2. Experiment 01 — break the rationals
3. How close you can get
4. Almost all of the line is missing
5. The crack is a door
6. The floor plan
7. What the lab is for

01 The rules of the lab

Welcome to the lab. The apparatus is a pencil. The safety equipment is a wastebasket. There is one standing assignment, and it never changes: you are here to break mathematics. Find the number that isn’t there. Find the equation with no answer. Find the picture the rules can’t draw. Push on every wall until something gives.

I should tell you up front how the experiment always ends, because it is the whole reason this lab exists. In twenty-five centuries of recorded attempts, nobody has ever broken mathematics. Not once. What happens instead is stranger: the wall you push on swings open. The number that wasn’t there turns out to be the majority of numbers. The equation with no answer founds a new geometry. The crack you found in the floor turns out to be a staircase, and the staircase goes down further than the building you thought you were standing in.

One point of honesty before we start, because this lab does not run on legend. No one in this story was trying to break mathematics. Every crack in the record was found by someone in the middle of building — a geometer comparing two lengths, a logician shoring up a foundation, an algebraist finishing a formula. The breaking is our assignment, in here, with pencils; history’s cracks were all discovered by accident, mid-construction. That, as we will see by the end, is itself a piece of evidence — and one of the strangest.

Each experiment in this lab is a full lesson, published and numbered, and the floor plan in section 06 maps all of them. This manual runs the first experiment in full — the oldest one, the one that broke the only numbers anyone believed in — and then hands you the keys to the building.

02 Experiment 01 — break the rationals

This assignment is older than the axiomatic method. Aristotle’s report (Metaphysics A5, 985b–986a) is that the Pythagoreans held things to be numbers — and “number,” to a Greek, meant a multitude of units: whole numbers, and the ratios between them. The slogan “all is number” is our compression of his report, but the commitment behind it was real, and it makes the square on your desk a loaded instrument. If all is whole number and ratio, then the diagonal and the side — two of the simplest lengths that exist — must stand in some ratio p/q. All you have to do is find it.

Here is the proof that you can’t, in the form it has been folklore since antiquity. Suppose the ratio exists: diagonal to side as p to q, with the fraction in lowest terms — all common factors cancelled. The Pythagorean theorem does the rest.

Both p and q are even. But we cancelled all the common factors before we started. The assumption ate itself. There is no fraction whose square is 2. Not “none found yet” — none, provably, forever. The oldest trace we have of this argument is Aristotle mentioning it in passing as a familiar example (Prior Analytics I.23, 41a26): the diagonal is incommensurable with the side, he says, “because odd numbers become equal to even ones” if you assume otherwise. The version printed in some manuscripts of Euclid as Elements X.117 is generally judged a later insertion. The proof was common property before anyone thought to sign it.

Notice what the experiment has actually shown, because it is sharper than the slogan. The proof does not directly produce a new number; it produces an absence. No rational number squares to 2. Meanwhile the diagonal is still lying there on your desk, entirely real, drawable with a straightedge in two seconds. Geometry testifies that the length exists; arithmetic proves that no ratio names it. That gap between the drawing and the naming is the crack — and everything else in this lab comes out of it.

The diagonal was never accused of madness. It was accused of being unsayable — a length with no name in the language of ratio.

The Greek vocabulary is worth one careful paragraph, because “irrational” smuggles in exactly the wrong idea. In Plato’s generation such a length could be called arrhētos — unsayable, inexpressible; Plato himself uses the word for a diagonal at Republic 546c. A century later Euclid drained the drama out of it: his term is asymmetros, “without a common measure” — the side and diagonal simply share no unit, however small, that measures both exactly. (Euclid also has a technical term alogos, ancestor of our “irrational,” but his definitions are set up so that it doesn’t even apply to the diagonal of a rational square — a warning about how far the modern word has drifted.) Nothing here means crazy. The Latin translators rendered “without ratio” as irrationalis, the word later grew a psychiatric career on its own time, and the diagonal has been paying for it since. The accusation was never madness. It was namelessness.

And Hippasus? The story goes that the Pythagoreans drowned him at sea for revealing this. Hold the story with tongs. The drowning first appears in sources written some eight centuries after the fact; in the oldest versions the betrayed secret is a different one (the construction of the dodecahedron); and no ancient writer actually credits Hippasus with discovering incommensurability — that attribution is a modern reconstruction. The same tongs apply to the popular “foundational crisis of Greek mathematics”: no ancient source reports a crisis, and the phrase itself was minted in 1928 by mathematicians projecting their own era’s foundations panic backwards. What the Greeks actually did was calmer and more interesting: they kept number and magnitude in separate books — literally; in Euclid, whole numbers live in Books VII–IX and general magnitudes in Book V — and they turned the crack into a research program. Within a generation Theodorus was proving the sides of squares of area 3, 5, and so on incommensurable with the unit, case by case, “up to seventeen square feet — and there, for some reason, he stopped” (Plato, Theaetetus 147d). He started at 3. By then, 2 was old news.

Figure 01

The whole apparatus: one square, one diagonal

• the assumption: all lengths are ratios
• what survives: the diagonal

03 How close you can get

A good lab never stops at “it broke.” It measures the break. If no fraction lands on the diagonal, the next question is: how close can fractions get? And here the experiment turns beautiful, because the answer has a precise shape in both directions.

Chasing √2 with fractions produces a ladder of best approximations — the convergents: 1/1, 3/2, 7/5, 17/12, 41/29, 99/70, 239/169, 577/408, each one the best possible rational at its size. Babylonian scribes knew stretches of this ladder; 17/12 and the superb 577/408 fall out of ancient recipes. Climb it and watch both what improves and what refuses to:

Read those two lines together, because together they are the finding. Fractions with denominator q can land within half of 1/q² of the diagonal — the convergents do, every one of them, infinitely often, forever. And no fraction, ever, of any size, lands closer than a third of 1/q². The door opens a crack and never further. You can approach √2 for the rest of time and the gap in front of you, measured in units of 1/q², never drops below a fixed width. The diagonal is not hiding. It is fenced — and the fence has a surveyed position: q² times the error converges to 1/(2√2) ≈ 0.3536, a constant as characteristic of √2 as its digits are.

Figure 02

The ladder of convergents: each step buys the same slice of forever

04 Almost all of the line is missing

So the rationals miss one point. One point seems survivable — patch it and move on. The second half of the experiment is discovering that the situation is precisely the reverse. It is not that the rationals miss a point here and there. It is that the rationals, all of them together, amount to almost nothing, and the “exceptions” are almost everything.

Here is the measurement, and it is one of the cleanest in mathematics. The rationals can be listed — first by denominator, or any systematic sweep: a first, a second, a third, one list catching every fraction eventually. Now pick any small budget of length — say ε. Cover the first rational on the list with an interval of length ε/2, the second with ε/4, the third with ε/8, halving forever:

Every rational number is under a patch, and the patches total at most ε — which you chose, and could have chosen smaller. The entire rational tribe, dense as it is, fits inside a total length of one millimeter, or one nanometer, or less. In the language of measure: the rationals have length zero. Throw a dart at the interval and the probability of hitting a rational is exactly 0. Two cautions belong in the same breath, because philosophers reliably fall down both stairs. Probability zero is not impossibility — the dart lands somewhere, and whatever point it hits also had probability zero. And “length zero” is not “sparse”: the rationals remain dense — infinitely many of them between any two points you name. A set can be everywhere and still be almost nothing. Density is about where; measure is about how much; and the two come apart.

Figure 03

Covering every fraction with half a unit of tape

• tape over the rationals, total ≤ ε
• what survives: the uncovered line

How much is missing? Cantor answered in 1874, in his first uncountability proof — and not with the famous diagonal, which came only in 1891. The 1874 argument is a trap: hand it any list of real numbers whatsoever and it walks along the list closing nested intervals, cornering a real number your list provably missed. No list catches the line. The rationals can be listed; the line cannot; so the unnameable points are not only most of the line by length, they are a strictly bigger infinity by count. (The paper’s advertised punchline was even better: since the algebraic numbers can be listed, transcendental numbers must exist — existence proved wholesale, without exhibiting a single one.)

And what exactly are these missing points? Here the lab report must be honest about dates, because it is the most commonly fudged fact in the story. The Greeks discovered incommensurable magnitudes — lengths. The irrational numbers are an 1872 invention. The parity proof shows no fraction squares to 2; to conclude “therefore there is a number √2 and it is irrational,” you need a number line with no gaps — and that property, completeness, is an axiom, not a theorem. Nothing about fractions forces it; the rationals satisfy every rule of arithmetic and order while remaining riddled with holes. Someone had to decide the line was full. In 1872, Richard Dedekind published the cut construction he had first worked out on November 24, 1858, while wondering what calculus lectures actually rest on; the same year, Cantor published a different construction from Cauchy sequences. (Charles Méray had, in fact, gotten there in 1869, to little notice; the year belongs to him as much as anyone.) Two men, two entirely different repairs, one identical structure at the end. File that coincidence; it becomes evidence in section 05. Kronecker’s famous scoff at the whole enterprise — “the whole numbers the dear Lord made; everything else is the work of man” — reaches us secondhand, from an unpublished 1886 lecture via a colleague’s memory seven years later: a quip at tongs’ length, not a doctrine. But it does state the question this lab keeps open on every wall: made, or found?

05 The crack is a door

Now run the lab’s signature move, the one every experiment in this building performs at least once. You have been reading the result as: the diagonal fails to be a number. Reverse the direction of explanation. The diagonal never failed at anything. The rationals failed to contain the diagonal. The structure — the plane, the square, the length — was there first, entire; the fractions were a notation that covered less of it than anyone knew. The proof did not break mathematics. It broke a notation’s claim to be all of mathematics. What we call the “discovery of irrational numbers” was the discovery that reality outran the naming system — and the repair, the completed line of 1872, is bigger than the crack by any measure you like: the patch turned out to be almost the whole line.

Once you see the shape of that event, you see it is not an event. It is the event, on repeat, and the rest of this lab walks the repetitions. The equation x² + 1 = 0 has no answer on the completed line — push, and the crack opens into the plane: the imaginary unit, which turns out to be not a new kind of number so much as the quarter-turn wearing a letter, and the engine of circles, growth, and eventually quantum mechanics. And that crack, remarkably, is the one that closes the mine it opened: over the complex numbers, every polynomial equation — complex coefficients included — has its full quota of roots. The Fundamental Theorem of Algebra says this particular method of breaking mathematics, the unsolvable equation, retires at ℂ. (New numbers were manufactured after that — Hamilton’s quaternions among them — but under different pressures; the polynomial mine was exhausted.) A century ago the pattern reached the foundations themselves: Russell, mid-construction on the logic of sets, found the crack in naive set theory; Gödel, working at Hilbert’s own program, found that no consistent system of the required kind proves everything true about arithmetic — and each crack, precisely described, became a theorem and then a discipline: axiomatic set theory in the one case, proof theory and the theory of computation in the other. (A door that is not open, for any theologian reaching for it: incompleteness is a theorem about arithmetically definable sentences in consistent formal systems. It does not show minds exceed machines, and it does not show truth exceeds God’s proof — it constrains formal systems, nothing else.)

So the honest form of the lab’s slogan is this: every crack ever found in mathematics has ended as a theorem. Not “survived,” not “was patched” — became subject matter. The break, precisely described, is the new mathematics. That is a strange property for a human artifact to have, and it is the standing exhibit for this lab’s deepest claim — the one argued across every room in the building: that mathematics is not a description we drape over the world but the structure the world is. Math is not modeling. Math is why.

06 The floor plan

Eleven experiments and an epilogue, each a full published lesson. This is the reading order — not the order they were written, but the order that breaks things properly. Every room states its assignment the same way: what you are trying to break, and what survives the attempt.

1. The Diagonal — this manual

   Try to break

   the claim that whole numbers and their ratios are all there is.

   What survives

   the diagonal — and a line so full that the fractions on it amount to length zero.

2. The Line Cannot Do Its Own Arithmetic

   Try to break

   the number line’s self-sufficiency: build √2 without ever leaving the line.

   What survives

   the plane — even the line borrows a dimension it does not have.

3. How Straight Lines Make a Circle

   Try to break

   the wall between straight and curved: make a circle out of nothing but straight steps.

   What survives

   multiplication by i as a quarter turn — the complex plane becomes the lab’s home ground.

4. e Is Where You Land If You Don’t Turn

   Try to break

   compounding: grow continuously and try to escape the number 2.71828…

   What survives

   e — the fixed destination of all growth that never turns.

5. You Only Feel the Real Part

   Try to break

   your own senses: feel your velocity in a moving car.

   What survives

   the derivative ladder — the body reads force alone, and half the rungs are perpendicular to feeling.

6. Mass Cannot Turn You

   Try to break

   the axis rule: feel momentum, just once.

   What survives

   momentum on the unfeelable axis — and the quantum momentum operator, with its i, waiting there already.

7. A Frame Is an Angle

   Try to break

   simultaneity: make two observers agree about “now.”

   What survives

   the invariants — relativity as a theory of what does not depend on you.

8. Rotation Is What Perpendicular Does

   Try to break

   the mystery of i: find something spooky in the square root of minus one.

   What survives

   a one-line proof: perpendicular change is rotation, and i is the quarter turn wearing a letter.

9. The i Is in the Equation

   Try to break

   the quantum: take the i out of Schrödinger’s equation and keep the physics.

   What survives

   the i, load-bearing — the exact boundary between classical and quantum, drawn in one symbol.

10. The Parts Do Not Determine the Whole

    Try to break

    reduction: describe two particles completely by describing each one completely.

    What survives

    entanglement as the ordinary condition of composite things — separability is the rare exception.

11. Collapse Is Not a Rotation

    Try to break

    measurement: make the update rule a smooth rotation like everything else.

    What survives

    a theorem that it cannot be — the geometry fixes everything about collapse except whether it happens.

12. “I Want” Is the Derivative of “I Am”

    Try to break

    the lab’s own boundary: carry the vocabulary to selves and desire — flagged as analogy, run for what it predicts.

    What survives

    the decomposition: every want has a component that grows you and a component that turns you.

07 What the lab is for

This lab was built for philosophers and theologians learning mathematics — people who can handle rigor but were never shown the why, and who came for the metaphysics that most math courses truck out the back door. So the standing method, in every room: the geometry is taught as established, and the philosophy is taught as an argument I find persuasive — with the seam between them flagged every time. The mathematics in this manual is simply true; the reading of it — that the world does not obey a mathematical structure but is one, with Tegmark as its sharpest modern statement in a lineage running back through Wigner’s “unreasonable effectiveness” to Plato — is an argument. Physics is compatible with that reading and suggestive of it; it does not prove it, and an object-oriented realist can accept every theorem in this building while holding that structure rides on things rather than being them. The lab does not hide that. The lab is where you go to weigh it.

And the standing assignment stands. Every room ends the same way this manual is about to: with the instruction to argue the other side, at full strength, and to say what it costs. A lab that cannot state the case against itself is a showroom. Bring a pencil.

§ The ledger

What is load-bearing

The parity proof; both approximation bounds (every convergent within 1/(2q²), no fraction ever within 1/(3q²), the fence at 1/(2√2) — all verified numerically in the sandbox); the ε-cover and measure zero; Cantor 1874 by nested intervals; completeness as an axiom. Every claim in the floor plan links to the room where it is derived in full.

What is convention

The word “irrational” — an accident of Latin translation for “without ratio.” Starting the story from the rationals at all. And the repair itself has two interchangeable forms: Dedekind’s cuts and Cantor’s sequences build the same structure by different scaffolding — which one you use is taste, which is your first hint that the structure, not the construction, is the real object.

Where the shorthand breaks

“Breaking math” is theater — nothing in mathematics ever broke; assumptions about mathematics broke, and no one in the historical record was even trying: every crack was found mid-construction by someone building. Also mind the dates: the Greeks discovered incommensurable magnitudes — lengths — not irrational numbers. The numbers, and the completed line they live on, are an 1872 decision.

Where I would push back on myself

“Every crack became a door” is curated history — survivor’s framing. And the strongest single counterexample cuts at my own thesis: set theory’s repair was not forced. Zermelo’s axioms were chosen — partly to defend his own well-ordering theorem — and rivals existed then and exist now. A repair you choose looks like invention, not discovery. Exercise 5 hands you that blade; use it.

§ Exercises

1. Run the experiment on √3. Adapt the parity proof to show no fraction squares to 3 (you will need: if p² is divisible by 3, so is p). Then find exactly where the same proof fails for √4 — the step that breaks is the lesson.
2. Climb one more rung. The convergents obey pₖ₊₁ = 2pₖ + pₖ₋₁, and likewise for q. Compute the rung after 577/408, verify the 1/(2q²) bound numerically, and evaluate q²·|√2 − p/q| to six decimal places. You should recognize the number you get.
3. Spend a smaller budget. Write out the first five patches of the cover for ε = 0.1, with their lengths and running total. Then explain, to a skeptic who keeps saying “but the rationals are everywhere,” how a set can be dense in the line and still fit under one centimeter of tape.
4. Find the 1872 moment. The parity proof shows no fraction squares to 2. State exactly what additional principle you need before you may say “there is a number whose square is 2” — and identify the precise step in your own argument where the axiom of completeness enters. The proof is short; locating the axiom in it is the exercise.
5. Argue the other side. Make the strongest case that this manual’s cracks are not discoveries but stipulations. Hilbert, to Frege (December 29, 1899): if axioms do not contradict each other, “they are true, and the things defined by them exist” — existence is consistency, so the completed line was legislated, not found. Carnap (1950): adopting the reals is adopting a linguistic framework; “do irrationals exist?” is a practical question about frameworks, not a factual one. Wittgenstein: “The mathematician is an inventor, not a discoverer” (Remarks on the Foundations of Mathematics I §168 — though he would reject the terms of the whole contest, which is its own option). Then price the position: why do stipulations adopted for internal convenience keep fitting physics (Wigner’s problem)? Why did Dedekind and Cantor, inventing freely, invent the same structure in the same year? And why is the space of possible “inventions” so brutally narrow — Frobenius: the only finite-dimensional associative division algebras over the reals are ℝ, ℂ, and the quaternions? Invention with no degrees of freedom looks remarkably like discovery. Argue it anyway, and say which cost you would pay.