# Chapter ᚆ(H): The observation car *Teclam Áirme, the Mathematical Compendium — a copy for AI readers of https://www.teclamairme.com/#ch-T8, as of 2026-09-30. Links are relative to https://www.teclamairme.com/fortheaireaders/.* [Guide for AI readers](index.md) · [Its glossary cards](ch08-cards.md) · [Its references](ch08-refs.md) · [Previous chapter](ch07.md) · [Next chapter](ch09.md) *A footnote such as [r-8.3] is reference entry 8.3 and links to it. Section 8 of 28; chapters are named by their Ogham letter, never by number.* ## The train out of the city ᚆ – [Uath](ch08-cards.md#g-og-uath): associated with the whitethorn tree, Section 8 ## Chapter ᚆ(H): The observation car > “Cuius rei demonstrationem mirabilem sane detexi. Hanc marginis exiguitas non caperet. (I have discovered a truly marvelous proof of this, which this margin is too narrow to contain.)” > — Pierre de Fermat · marginal note in his copy of Diophantus’s *Arithmetica*, about 1637; the copy is lost, and the note survives in the edition his son Clément-Samuel printed in 1670 [r-8.19](ch08-refs.md#r-8.19) Formula: ([[∃ n, x, y, z ∈ ℤ⁺, n > 2 : xⁿ + yⁿ = zⁿ]](ch08-cards.md#g-fermat) + ([x ∧ ¬x](ch08-cards.md#g-boolean)))/([[1+1=2] + [√2 ∉ ℚ] + [∞ many primes] + [ℵ₀ < 𝔠]](ch08-cards.md#g-truths) + ([x ∨ ¬x](ch08-cards.md#g-excluded))) Key concepts: Logic: Boolean algebra and the Iverson bracket. It includes symbols you may not have seen before, such as: [[]](ch08-cards.md#g-iverson), [∧](ch08-cards.md#g-boolean), [∨](ch08-cards.md#g-excluded), and [¬](ch08-cards.md#g-logicsyms). Zero here means false: the claim that Fermat was wrong, turned into a single bit, settled at 0 after 358 years. The panoramic view is incredible, but note the steady stream of signals we pass by. At the stadium, “This position is lost” is a claim, and a claim wants checking. The signals along the track ahead are how mathematics records which claims have been settled. Whether the route is open or closed matters a great deal to our safety, and to mathematics. Mathematics has a simple way to record whether a statement is true or false: give it the value 1 or 0. Establishing which value it deserves is another matter. One of the statements ahead kept mathematicians occupied for centuries; happily, they finished before we bought our tickets. The sign we need is the [Iverson bracket](ch08-cards.md#g-iverson). Put a statement in square brackets and read it as 1 if true, 0 if false. The bracket translates a truth value into an ordinary integer. It does not find a proof for us. [r-8.1](ch08-refs.md#r-8.1) Field notes One fraction: two false things above, five true things below. #### Numerator [[∃ n, x, y, z ∈ ℤ⁺, n > 2 : xⁿ + yⁿ = zⁿ]](ch08-cards.md#g-fermat) = 0; Fermat’s Last Theorem as a single bit, which took roughly 358 years to settle. The bracket claims that there exist positive integers n, x, y, and z, with n greater than 2, such that xⁿ + yⁿ = zⁿ. Squares can do this ([3² + 4² = 5²](ch12-cards.md#g-pythag345)), but around 1637 Fermat wrote that cubes and higher powers cannot, and that his marvelous proof would not fit in the margin. Nobody knows whether he really had one, and it seems unlikely. [r-8.4](ch08-refs.md#r-8.4) The bracket asserts exactly the counterexample Fermat said could not exist, so it is 0 exactly when he was right, and he was. Andrew Wiles announced a proof in 1993; a gap emerged, he repaired it with Richard Taylor in 1994, and the completed work appeared in 1995. The route ran through [elliptic curves](ch19-cards.md#g-weierstrass) and [modular forms](ch19-cards.md#g-modularity), which we will meet later. [r-8.6](ch08-refs.md#r-8.6) [x ∧ ¬x](ch08-cards.md#g-boolean) = 0; the [law of non-contradiction](ch08-cards.md#g-boolean), [Boolean algebra](ch08-cards.md#g-boolealg)’s own zero. Here x stands for a statement, ∧ means “and,” and [¬](ch08-cards.md#g-logicsyms) means “not.” Let x say “the track ahead is open”; the expression says the same track is both open and not open, at the same time and in the same sense. Whichever way the signal reads, one half of the claim is false, and an “and” with a false half is false. Unlike the Fermat bracket, this needs no centuries-long investigation: its form settles it. [Boolean algebra](ch08-cards.md#g-boolealg), which George Boole gave logic in 1847, writes true as 1, false as 0, “and” as multiplication, and “not x” as 1 − x; then x ∧ ¬x is x × (1 − x), which is 0 whether x is 1 or 0. [r-8.2](ch08-refs.md#r-8.2) #### Denominator [[1+1=2] + [√2 ∉ ℚ] + [∞ many primes] + [ℵ₀ < 𝔠]](ch08-cards.md#g-truths) = 4; four statements that are established, in a widening progression: arithmetic, the inadequacy of fractions, infinitely many primes, and unequal infinities. 1 + 1 = 2; [√2](ch04-cards.md#g-sqrt2) is irrational; [there are infinitely many primes](ch08-cards.md#g-euclidprimes); and [the natural numbers](ch27-cards.md#g-aleph0) form a smaller infinity than [the real numbers](ch27-cards.md#g-cont). Each bracket contributes 1. [r-8.8](ch08-refs.md#r-8.8) The first is the shortest: from [Peano’s axioms](ch08-cards.md#g-axioms) it takes three steps, yet *Principia Mathematica*, Whitehead and Russell’s three-volume work on the foundations of mathematics, does not finish the proof until its second volume. [r-8.14](ch08-refs.md#r-8.14) Each of the other three is proved by showing that its counterexample cannot be built. Write [√2](ch04-cards.md#g-sqrt2) as a fraction in lowest terms, and squaring shows that its top and bottom must both be even, so it was not in lowest terms after all. Take any finished list of primes, multiply them all together and add one: no prime on the list divides the result, so its prime factors are missing from the list. Take any list of the [real numbers](ch01-cards.md#g-reals), and [a diagonal construction](ch27-cards.md#g-diagonal) produces a number that differs from every entry. [r-8.10](ch08-refs.md#r-8.10) [r-8.11](ch08-refs.md#r-8.11) [x ∨ ¬x](ch08-cards.md#g-excluded) = 1; the [law of excluded middle](ch08-cards.md#g-excluded), the partner of the numerator’s zero. The symbol ∨ means inclusive “or.” In the classical two-valued logic used here, a statement is true or its negation is true, so the expression is 1 for every x. [r-8.7](ch08-refs.md#r-8.7) [De Morgan’s laws](ch08-cards.md#g-demorgan) turn “not both” into “at least one”: negating the numerator, x ∧ ¬x, gives x ∨ ¬x. [r-8.16](ch08-refs.md#r-8.16) In 1937 Claude Shannon showed that [Boole’s algebra of truth values](ch08-cards.md#g-boolealg) is also the arithmetic of switching circuits: wire switches to act as “and,” “or,” and “not,” and logic acquires physical form. [r-8.3](ch08-refs.md#r-8.3) The only fraction is (0 + 0)/5: 4 + 1 = 5. [r-8.18](ch08-refs.md#r-8.18) **Fine print.** Reading a claim and establishing it are different jobs: a claim nobody has settled is not 0 just because its value is unknown; its value is simply not known yet. Only settled statements stand below the line, and each of the five here is established as 1. Truth also differs from proof. [Gödel’s incompleteness theorems](ch08-cards.md#g-godel) of 1931 show that any consistent list of [axioms](ch08-cards.md#g-axioms) strong enough for arithmetic, and plain enough that a machine could check it, leaves some arithmetic statements it can neither prove nor refute. That is a limit of the axioms, not a third truth value between 0 and 1. [r-8.12](ch08-refs.md#r-8.12) What this chapter adds to the formula: (0 + 0)/5 = 0 The two zeros and five ones are tiny answers with very different amounts of mathematics behind them: one zero took 358 years, the other takes a glance at its own shape. As important as the signals are for our journey, the locomotive is doing all the work. I know the engineer; let’s go up front and see how much we can ask a machine to do. Watch and read - [Fermat’s Last Theorem - Numberphile](https://www.youtube.com/watch?v=qiNcEguuFSA) Numberphile, Simon Singh: the margin, the 358 years, and Wiles, told by the man who wrote the book - [The Heart of Fermat’s Last Theorem - Numberphile](https://www.youtube.com/watch?v=ua1K3Eo2PQc) Numberphile, Simon Pampena: what modularity means and why it settled the theorem - [Fermat’s Last Theorem](https://en.wikipedia.org/wiki/Fermat%27s_Last_Theorem) Wikipedia - [Iverson bracket](https://en.wikipedia.org/wiki/Iverson_bracket) Wikipedia