Mistral on 20 November 2026

Hexarchon Re-Derivation: Fan Geometry and Cell Inventory

Status key: [P] proven · [D] definitional · [O] open

This page re-derives, with every coordinate step shown, the plate fan and cell inventory of the regular octahedron (the “hexarchon”) under its four internal plates. It replaces section 1 of the 19 Nov page and corrects two lemmas (1.1’s meaning, 1.2’s reflection classification). Section 7 states what this page settles and what it deliberately does not.


0. Objects and notation [P]

Octahedron O = {|x|+|y|+|z| ≤ 1}. Circumradius R = 1 (vertices ±e₁, ±e₂, ±e₃); edge a = √2; volume V(O) = (√2/3)a³ = 4/3. Center C = (0,0,0).

The twelve edge-midpoints of O are m(±eᵢ, ±eⱼ), i ≠ j, i.e. the points (±1/2, ±1/2, 0) and permutations, with all sign choices.

The four plates (planes through C):

  • P₁ : x+y+z = 0, normal n₁ = (1,1,1)
  • P₂ : x+y−z = 0, normal n₂ = (1,1,−1)
  • P₃ : x−y+z = 0, normal n₃ = (1,−1,1)
  • P₄ : −x+y+z = 0, normal n₄ = (−1,1,1)

The four normals are pairwise at 60° (nᵢ·nⱼ = 1 for i ≠ j), so they form a regular tetrahedron of directions. Because a plane n·x = 0 equals its negation (−n)·x = 0, the “other” tetrahedral class of normals (two or three minuses) describes the same four planes: the plate family is the complete self-dual tetrahedral set. No choice of orientation was made.

Write h₁ = x+y+z, h₂ = x+y−z, h₃ = x−y+z, h₄ = −x+y+z. Note the linear dependence h₄ = h₁ − h₂ − h₃.


1. The second moment [P]

Clean one-line derivation. The cross-section of O at coordinate x is {(y,z) : |y|+|z| ≤ 1−|x|}, a square (diamond) of area 2(1−|x|)² and total volume 4/3. So the marginal density of X is f(x) = (3/2)(1−|x|)² on [−1,1], and

E[X²] = 3∫₀¹ x²(1−x)² dx = 3·(1/30) = 1/10.

By symmetry ⟨r²⟩ = 3E[X²] = 3/10, and

∫_O r² dV = V·⟨r²⟩ = (4/3)(3/10) = 2/5. ∎

The “5” is the length-5 dimension of the second moment (∫₀¹ u⁴ du = 1/5, R⁵ scaling). It is not a pentagon and not the Aristotle count.


2. The plate hexagons [P]

Claim 2.1. Each Pᵢ ∩ O is a regular hexagon whose six vertices are edge-midpoints of O, and each of the 12 edge-midpoints lies on exactly 2 plates.

Proof for P₁. P₁ ∩ O consists of points with |x|+|y|+|z| ≤ 1 and x+y+z = 0. The plane cuts exactly the six edges of O joining a positive-axis vertex to a negative-axis vertex (edges ±eᵢ–±eⱼ with i ≠ j and opposite signs): on each such edge the plate value runs linearly from ±1 to ∓1, hitting zero at the midpoint. The six other edges of O (same-sign pairs) have midpoints with h₁ = ±1 ≠ 0 and are cut at no other point since the plate value is affine along each edge. So P₁ ∩ O has as its boundary the six midpoints m(e₁,−e₂), m(e₁,−e₃), m(e₂,−e₃), m(−e₂,e₃), m(−e₁,e₃), m(−e₁,e₂), forming a hexagon with circumradius |m| = √2/2 and side √2/2 (e.g. |(1/2,0,−1/2) − (0,1/2,−1/2)| = √2/2): a regular hexagon.

Counting: 4 plates × 6 sides = 24; 12 edge-midpoints, each lying on exactly 2 plates (checked directly: e.g. (1/2,1/2,0) lies on P₃ and P₄, since h₃ = h₄ = 0 there). 24 = 12 × 2, and by the tetrahedral symmetry every midpoint has the same incidence. ∎

So the plate family and the edge-midpoint set are mutually incident, 2-to-2. This is the fixed frame on which everything below is built.


3. Cell inventory, constructed [P]

The four plates divide O into conv (open) regions; the inventory comes from the sign patterns of (h₁,h₂,h₃,h₄).

Claim 3.1 (count). There are exactly 14 bounded regions: 6 corner cells and 8 tetrahedra, every one having C = (0,0,0) as a vertex.

Proof. Using h₄ = h₁ − h₂ − h₃, the sign of h₄ is determined (given the other three) only by the two patterns (s₁,s₂,s₃) = (+,−,−), which force h₄ > 0, and (−,+,+), which force h₄ < 0. The remaining six patterns leave h₄ free. Total regions: 2 + 6·2 = 14.

Directed by the sign pattern of the region containing it, the cells start at the six vertices of O and the eight facet-directions (±1,±1,±1) of the cube: 6 + 8 = 14, all regions accounted for. O and the plates are centrally symmetric, so the cells fall into 7 antipodal pairs: 3 corner pairs and 4 tetrahedral pairs. ∎

Claim 3.2 (corner cells are regular octahedra). The cell at e₁ = (1,0,0) is

K₁ = {h₁, h₂, h₃ > 0, h₄ < 0} ∩ O.

In the variables s = y+z, d = y−z, the four conditions read s > −x, d > −x, d < x, s < x, i.e. |s| < x and |d| < x with x > 0; since max(|s|,|d|) = |y|+|z|, this is simply the cone x > |y|+|z|. Its boundary planes are x = ±y±z, which are exactly the four plate planes. The l₁ cone exits O along |y|+|z| = 1−x, x = |y|+|z| ⇒ x = 1/2: the exit cross-section is the square Q₁ = {(1/2, ±1/2, 0), (1/2, 0, ±1/2)}. So

K₁ = bipyramid with apexes (1,0,0) and (0,0,0) over the square Q₁.

All twelve edges of this bipyramid have length √2/2 (apex-to-corner: √(1/4+1/4); square side: √2/2), so K₁ is a regular octahedron of scale 1/2, volume √2·(√2/2)³/3 = 1/6. Its eight faces alternate: four on the boundary of O (triangles (e₁, m, m′) on the four facets at e₁), four on the plates. ∎

Claim 3.3 (tetrahedra). The cell about direction (1,1,1) is T₊ = conv(0, (1/2,1/2,0), (1/2,0,1/2), (0,1/2,1/2)): apex at the center, base the medial triangle of the facet x+y+z = 1. Its volume is (1/6)|det| = (1/6)(1/4) = 1/24. By the antipodal symmetry every one of the 8 cells about a cube-corner direction is a tetrahedron of volume 1/24, apex at C, base the medial triangle of one of the 8 facets. (The same holds at, e.g., direction (1,1,−1): its cone is {x > 0, |y+z| < x, y−z > x}, whose exit on O is the triangle conv((1/2,1/2,0), (0,1/2,−1/2), (1/2,0,−1/2)) — again the medial triangle of the facet x+y−z = 1, again with apex C.) ∎

Claim 3.4 (volume closure, exact). 6·(1/6) + 8·(1/24) = 1 + 1/3 = 4/3 = V(O). The inventory is complete and the cells are disjoint. The corner octahedra occupy 3/4 of O, the tetrahedra 1/4. ∎

Remark (the centerpoint). Every one of the 14 cells has C as a vertex: the four plates meet only at C, so C belongs to the closure of all cells. The inventory entry “1 centerpoint” is exactly this shared vertex, not a cell.

Second-moment cross-check. K₁ is the scale-1/2 octahedron centered at (1/2,0,0): ∫{K₁}|x|² = (2/5)(1/2)⁵ + (1/4)(1/6) = 1/80 + 1/24 = 13/240; six cells give 13/40. For a tetrahedron with vertices 0, p₁, p₂, p₃, E|x|² = (1/10)(Σ|pᵢ|² + Σ{i<j} pᵢ·pⱼ) = (1/10)(3/2 + 3/4) = 9/40, so ∫ = (1/24)(9/40) = 3/320; eight cells give 3/40. Sum: 13/40 + 3/40 = 2/5 — the second moment obtained a second, independent way through the cell decomposition. Consistent with Section 1. ∎


4. The fan and the alternation (corrected Lemma 1.1) [P]

Fix P₁ with hexagon vertices in cyclic order v₁…v₆ = m(e₁,−e₃), m(e₂,−e₃), m(−e₁,e₂), m(−e₁,e₃), m(−e₂,e₃), m(e₁,−e₂), and fan triangles Tᵢ = conv(C, vᵢ, v_{i+1}) (indices mod 6).

Claim 4.1 (what each sector is). Across each plate sector Tᵢ, exactly two cells meet: a corner octahedron and a tetrahedron. Specifically, vᵢ is the midpoint of an O-edge and v_{i+1} of the next; the two edges share the O-vertex uᵢ, and the corner octahedron at uᵢ meets, on the other side of the plate, the tetrahedron whose base is the medial triangle of the facet carrying vᵢ and v_{i+1}. (For T₁: corner at (0,0,−1); tetrahedron on facet x+y−z = 1. Verified by evaluating the four h-signs at the sector centroid, which give (+,+,−,−): a tetrahedral pattern; the flipped pattern (−,+,−,−) is the corner pattern of (0,0,−1).) ∎

Because of Claim 4.1, the fan’s 2-coloring cannot be “which cell type adjoins the sector” — every sector adjoins one of each. The genuine alternation is in the corner cells’ positions:

Claim 4.2 (alternation, corrected 1.1). The corner apexes u₁…u₆ run through all six vertices of O, and consecutive ones alternate between the two inscribed complementary tetrahedra

Q⁻ = {(−1,0,0), (0,−1,0), (0,0,−1)} and Q⁺ = {(1,0,0), (0,1,0), (0,0,1)}.

For P₁ the sequence is uᵢ = (0,0,−1), (0,1,0), (−1,0,0), (0,0,1), (0,−1,0), (1,0,0): signs −, +, −, +, −, +. Each sector’s corner cell therefore carries a sign class, and the sign class alternates around the fan. ∎

Proof. The uᵢ are the endpoints shared by consecutive edges eᵢ, e_{i+1} of the 6-cycle of O-edges cut by the plate. In O’s edge graph K₂,₂,₂ (complete tripartite on the antipodal pairs {±e₁}, {±e₂}, {±e₃}), a 6-cycle visiting all six vertices passes each part twice, in antipodal pairs (uᵢ, u_{i+3}). Composing with the plate’s own reflection symmetry (which fixes the hexagon setwise and maps uᵢ → u_{7−i} for the sector indexing above, alternating the class), the visits must alternate between Q⁻ and Q⁺. Directly: reading the cycle above confirms −,+,−,+,−,+. ∎

Remark. This is the stella octangula structure: the two orbits of 3 corner octahedra are matched to the two tetrahedra Q⁺, Q⁻ inscribed in O. This is the geometric content the 19 Nov page was reaching for; “three triangles of type A and three of type B” holds, but the coloring is the dual-tetrahedral class of the corner cell, not the cell type on one side of the sector.


5. Stabilizer (corrected Lemma 1.2) [P]

Claim 5.1. The symmetry group of the colored fan (hexagon plus the Claim 4.2 coloring) is D₃, order 6.

Proof. The 60° rotation sends the coloring A→B at every sector (Claim 4.2), so it is not a symmetry; the 120° rotation sends Q⁻→Q⁻ and Q⁺→Q⁺ and is a symmetry. Reflections: the axis through the midpoints of opposite sides of the hexagon fixes two opposite sectors setwise (so fixes their colors) and swaps the remaining sectors in same-color pairs (T₂↔T₆, T₃↔T₅ in the indexing above): a symmetry. The axis through opposite vertices of the hexagon maps each sector to its neighbor: A↔B everywhere: not a symmetry. Surviving group: ⟨120° rotation, one side-axis reflection⟩ = D₃, order 6. ∎

Correction note. The 19 Nov page states the reverse classification (“reflections through axes through opposite vertices preserve the coloring, the reflections through edge-midpoints do not”). It is the side-midpoint axes that preserve and the vertex axes that swap. The conclusion (D₃, not D₆) is unchanged, and Corollary 1.3 (the breaking D₆ → D₃ is forced by the cell combinatorics, not chosen) stands.


6. Orbits [P]

Under D₃ (Claim 5.1): the 6 corner octahedra fall into two orbits of 3 (their apexes lie on Q⁻, Q⁺ — Claim 4.2); the 8 tetrahedra fall into two orbits of 4, according as their facet lies in the plate-normal class {n₁,n₂,n₃,n₄} = Q⁺-facing tetrahedron of directions or the antipodal class. This is the static orbit structure of the 19 Nov page’s Corollary 1.4, now derived rather than asserted.


7. What this page settles, and what it does not

Settled here [P]: the plate-hexagon incidence (Claim 2.1); the 14-cell inventory with exact volumes 1/6 and 1/24 and exact closure 4/3 (Claims 3.1–3.4); the second moment 2/5 by two independent routes; the fan alternation as dual-tetrahedral class with a graph-theoretic proof (Claim 4.2); the D₃ stabilizer with the corrected reflection classification (Claim 5.1); the orbit structure 3+3, 4+4.

Not settled here: the dynamic identification of the corner/tetrahedron orbit ratio 4/(4+6) = 2/5 with the second moment — that is an arithmetic coincidence between small orbit counts and the moment integral, not a derived identity, and stays [O]. The ratio map r ↦ 1+1/r and its attractor φ remain as on the 19 Nov page [P], but the claim that interface alternation produces Fibonacci-type addition (Definition 2.1 there) is a definitional choice, and the product 2φ/5 is a hypothesis: it gets its own page, where the observed quantity it names must be specified before any falsification condition can attach to it.

###