Chapter 3 asked whether two bracketings of the operator chain are equal on the nose or merely isomorphic. That question has an older form, and the corpus has been quoting its answer for years without noticing the shape of it. In 1868 Beltrami read Lobachevsky’s plane off a surface of constant negative curvature. The standing question is what that reading transports.
Three relations get called the same in ordinary mathematical speech, and they carry different amounts. Equality on the nose transports everything, including anything stated about the underlying representation. Isomorphism transports every property expressible in the signature, and nothing about the carrier. Interpretation — a structure-preserving map into a model of a different theory — transports only what the dictionary was written to cover, and in one direction.
Beltrami’s is the third, and it is provably not the second. Hilbert showed in 1901 that no complete regular surface of constant negative curvature embeds in $\mathbb{R}^{3}$, so the pseudosphere carries a proper piece of the hyperbolic plane. The two objects are not isomorphic; they cannot be. What the dictionary transports is narrower than either loose phrase suggests: derivations, and therefore contradictions. That is the entire content of the 1868 result, and it is enough to end a two-thousand-year search, because a proof of the parallel postulate would now be a contradiction inside Euclid.
An associator is needed when composition holds only up to isomorphism. A pentagon is needed when the associators themselves must agree. In both cases the mathematical content is not the weakening — it is the obligation to write down what survives it, and to check that the surviving data is consistent.
A model is that obligation at the level of theories. Once “is” has been weakened to “interprets,” every subsequent sentence has to say which side it is about, and the failure mode is mechanical: a theorem proved about the model gets quoted as a theorem about the thing. The pseudosphere has a boundary and the hyperbolic plane does not; a statement about the edge is true of one and meaningless for the other. Nothing in the notation stops you from transporting it anyway.
If the contact formulation of G = U∘F∘K∘C is a model, then every metric-flavoured sentence in the corpus — curvature, geodesic, distance, threshold as a length — needs a named dictionary and a statement of what it transports. None exists.
If it is the thing itself, then the metric vocabulary is not available at all, and its appearances are borrowings. Darboux’s theorem sharpens this rather than softening it: contact structures have no local invariants, so no quantity computed from the structure can distinguish one metric’s geodesics from another’s. What remembers a metric is the choice of form, through its Reeb field. An equivalence that holds for every contact structure is too strong to be informative about any of them.
The positive statement on the other side is a theorem and is worth having exactly because it is narrow: for a Riemannian manifold $(M,g)$, the restriction of the canonical Liouville 1-form to the unit cotangent bundle $S^{*}M$ is a contact form whose Reeb field generates the geodesic flow of $g$. Geodesic motion is Reeb dynamics — on a manifold built from the original one, by a construction that has to be named each time it is used.
The translator framing did not come from this chapter. It arrives with the IMPA notes the series already carries: Nachbin & Tabak, 21º Colóquio Brasileiro de Matemática, Chapter 1 §1.1.1, O Matemático como Tradutor:
O matemático aplicado deve se acostumar a fazer o papel de um tradutor simultâneo, realizando a tradução do problema em palavras — em português — para o problema na linguagem matemática — “em matematiquês” — e vice-versa.
Two things in that sentence this chapter is the theory of. It is simultaneous, so the translator never finishes one side before starting the other — which is how a dictionary written for one direction ends up used in both. And it closes on e vice-versa, the harder half, and the half where a corpus loses things: a theorem proved about the model, carried back and quoted about the thing. The passage is on the IMPA page. The author chapter, and the test of whether that sentence was technical or decorative, is Book 7 · André Nachbin.
Book 3 · Ch.H states it outright: “The Collatz conjecture is visible from within the crystal geometry before it is axiomatic within it. The gap between visibility and proof is a gap in formal language, not in underlying truth.” That is this chapter's subject, asserted about a particular object.
It has both an external test case and a warning, from the same
person. Tao's 2019 Collatz paper closed a gap of exactly that shape by changing the density
the statement was measured in — natural to logarithmic — and those are
demonstrably different instruments rather than a strong claim and a weak one. His own
writing supplies the check: “The point of rigour is not to destroy all intuition;
instead, it should be used to destroy bad intuition while clarifying and elevating good
intuition.” From the inside, a true “I can see it” and a false one are
indistinguishable. See Book 7 · Terence Tao and
book7/ch-tao-verify.py.
The standing bar for Part II is that each volume carries a machine-checked core: a Lean file that elaborates clean against a pinned Mathlib and reports its axioms, or the volume is a reading list with a DOI on it. This chapter does not have one, and saying so is the point of the status line above rather than a caveat under it.
The candidate core is not a rung-30 object, which is the finding this chapter contributes. Tile a closed surface with geodesic triangles and the angle excesses add, because $\alpha+\beta+\gamma-\pi=\iint K\,dA$ is additive over the pieces; carry on until the surface closes and the total is fixed by topology alone at $2\pi\chi$. The discrete form of that is pure counting and needs no differential geometry: for a triangulated closed surface, summing the angle defect $2\pi-\sum\theta$ over vertices gives $2\pi V-\pi F$, and $3F = 2E$ makes that $2\pi(V-E+F)$. A local quantity integrating to a global invariant, with the local side elliptic — that is an index theorem, in its first and oldest instance.
Index theory is rung 28, Volume XI, which WP-82 classifies as FLOOR and which this volume’s own index already says comes first in reading order. XI is not written. Neither is XII. Building a coherence chapter on top of an absent index volume repeats the pattern this corpus keeps correcting, so the entry is recorded here and the work is not claimed here.
Discrete Gauss–Bonnet is the smallest honest core XI could carry: a finite statement, provable from $3F=2E$ and the Euclidean angle sum, with an axiom report the kernel can produce. It is also the formal shadow of the oldest argument in the subject — measure a triangle, repeat, and watch the surface close.
No Lean artifact, so no axiom report, so no Result status. The three-way distinction between equality, isomorphism and interpretation is standard model theory and is not claimed as new; what is claimed is that the corpus has not said which of the three it means when it calls the contact formulation a description of the operator chain. The Reeb–geodesic statement is quoted from the literature and is not formalised here. And the reading of Beltrami as an interpretation rather than an isomorphism rests on Hilbert’s embedding theorem, which constrains surfaces in $\mathbb{R}^{3}$ and says nothing about the abstract hyperbolic plane.