⚜ PRINCIPIA ORTHOGONA · Book 7 · The Scientists · Atiyah ← Conley · the third candidate
Book 7 · Michael Atiyah · 2026-09-19 · closes WP-82 §3b

The Space Has No Room For It

WP-82 asked a sharp question: is there a K-theory class whose pairing is constant along this helix? It named three candidates. Conley closed the third. The other two close together, and for one reason — the manifold retracts to a circle, and a circle has exactly one ℤ in its K-theory, in odd degree, measuring winding.
Methodhomotopy type by hand; K-groups cited
book7/ch-atiyah-verify.py, five blocks, exit 0
Resultboth remaining candidates close
and the invariant that does survive is the winding number, 1
TierK-groups of S¹ are standard
the retractions are elementary and shown
An index does not move. That was WP-82's whole objection to e−4π, and it left the better question behind: if the multiplier moves, is there something on this manifold that does not? The answer is yes, there is exactly one thing, and it is not a contraction rate.

1 · What the space is

The dm³ manifold is M = {(r, θ, z) : r > 0} with α = dz − r²dθ. The constraint r > 0 is not a convenience — the contact form is non-degenerate exactly there, and θ is only a coordinate where r ≠ 0. So

M = (ℝ² − {0}) × ℝ

which deformation retracts, in two elementary steps, onto a circle: collapse the ℝ factor in z, then retract the punctured plane radially onto r = 1. M ≃ S¹.

2 · What a circle has

Standard, cited not derived $\widetilde{K}^0(S^1) = 0$ — every complex vector bundle over a circle is trivial.
$K^1(S^1) \cong \mathbb{Z}$ — generated by the winding, or determinant, class.

That is the entire K-theory of this space. One ℤ, in odd degree, and what it measures is how many times a loop goes round. There is no even-degree room at all.

3 · Candidate two closes: the relative class

WP-82's second candidate was a relative class on the pair (M, {z ≤ c}). Write A = {z ≤ c} = (ℝ² − {0}) × (−∞, c]. The inclusion A ↪ M is a deformation retract — slide the half-line onto the line — so it is a homotopy equivalence, and therefore

K*(M, A) = 0.

There is no non-zero relative class to pair with anything. The candidate does not fail to give e−4π; there is nothing there at all.

4 · Candidate one closes: the asymptotic class

The first candidate was the z → ∞ limit as a boundary or asymptotic index. The limit is clean — block [2] of the verify script confirms λ(z) = −2(1 − e−z) → −2 and the multiplier → e−4π = 3.487342356 × 10−6, so the drift really does vanish in the limit. That much of the intuition was right.

But the end itself is again (ℝ² − {0}) ≃ S¹. Even degree is empty. The one available class lives in K¹ and evaluates to a winding number — and on Γ, where θ̇ = 1 over T* = 2π, that number is 1. It is an integer, it is constant along the helix, and it is deformation invariant: block [4] scales ṙ → k·ṙ across nine orders of k, moving the multiplier by more than three hundred orders of magnitude, and the winding number does not move at all.

The answer to WP-82's question Yes, there is a class whose pairing is constant along the helix. It is the winding number, and it equals 1. It is not e−4π, it is not related to e−4π, and it carries no information about transverse contraction — because a transverse contraction rate is not the kind of thing K-theory of a circle can hold. The multiplier was never going to be an index of this space. It was looking for a home in a building with one room, already occupied.

5 · The gap underneath, which is larger

ch-conley recorded that “index theorem” appears in five chapters of this corpus and “Fredholm” in none. That is the deeper problem and this chapter sharpens it rather than solving it.

An index in Atiyah–Singer's sense is not a property of a space. It pairs a K-theory class with an elliptic operator, and its content is that the analytic index — dim ker − dim coker of a Fredholm operator — equals a topological expression. This corpus has never named an operator. Sections 3 and 4 show the topological side is empty; the analytic side was never set up. Both halves of the theorem are missing, and only one of them was ever noticed.

So Volume XI's core — WP-82's own bar, “an index, computed, verified, and reported by the kernel” — does not fail here. It has not been attempted, because attempting it begins with writing down a Fredholm operator, and none exists in the corpus to write down. Weibel's K-book, 576 pp, is now held and addressed; the index-theorem half of rung 28 needs a different book and an operator.

6 · Status

7 · References

  1. WP-82 §3b — the drift computation and the three candidates · ch-conley — the third, closed.
  2. book7/ch-atiyah-verify.py — five blocks, standard library only, exit 0, 2026-09-19.
  3. Weibel, The K-book: an introduction to Algebraic K-theory, 576 pp, held — sha256 in docs/floor-texts.tsv. The topological K-theory used here is standard and is not taken from it.