Circulatory & Musculoskeletal Spokes
Two more spokes promote inherited adjacencies to declared seams. Leukocyte effector populations, rooted at bone-marrow haematopoiesis (RUNX1), point one-way into the circulatory vasculature; the same haematopoietic root is housed in the musculoskeletal marrow niche. Both consume no sibling value; engine hash byte-identical; the niche identity is a named candidate, not claimed.
The seam layer now carries five declared seams under one 2×sha256 (9357f23b…), still importing ZERO sibling code. Circulatory: the leukocyte effector populations (rooted at bone_marrow_hematopoiesis, RUNX1; spinodal 0.585385, barrier 0.437252) are a one-way pointer OUT to vascular transport — the immune volume owns the populations, the circulatory volume owns the absolute transport scale [O]. Musculoskeletal: the haematopoietic root (same organ) is housed in the marrow niche — a one-way pointer now, with a shared-substrate-identity upgrade (niche threshold = 0.585385) NAMED but not claimed awaiting a live musculoskeletal run. Both pointers land on a real in-package organ; neither feeds the hashed engine core.
From inherited adjacency to a declared spoke
The hub picture of chapter 16 listed two relationships that were already present through inherited primitives but had not yet been promoted to explicitly declared seams: the circulatory volume carries the vasculature on which immune cells traffic, and the musculoskeletal volume holds the bone compartment that houses haematopoiesis. This chapter promotes both to declared entries in the single source of truth (inherited/cross_references.json), each as a one-way POINTER — the weakest, safest seam kind, which states where something travels and consumes no sibling value. The hub now has five spokes, not three, and the discipline is unchanged: the gated package imports zero sibling code, and the engine emit() determinism hash is byte-identical because nothing here feeds the hashed core.
Circulatory: leukocyte trafficking is a one-way pointer
The leukocyte effector populations of this volume — the ON-committed clones of the R19 switch, rooted at the haematopoietic organ bone_marrow_hematopoiesis (master gene RUNX1, spinodal 0.585385, barrier 0.437252) and selected through the adaptive lymphoid compartment — travel through the circulatory vasculature to reach peripheral tissue. That is the seam: a one-way pointer OUT from the immune effector populations to the circulatory transport layer. The pointer carries no circulatory snapshot and consumes no circulatory value, so it is a directional statement, not a numeric dependency. The ownership split is explicit: this volume owns the effector populations (its own measured R19-committed output), while the circulatory volume owns the absolute vascular transport scale — vessel geometry, blood flow, margination and extravasation kinetics — which stays [O] here. Because the two volumes share one byte-identical R19 substrate, neither re-derives the other's part; but unlike the gut seam this is deliberately not a closed-form identity, only a pointer, because no equality between an immune quantity and a circulatory quantity is asserted.
Musculoskeletal: the marrow niche, with a named identity upgrade
The haematopoietic origin of every immune cell — the same organ bone_marrow_hematopoiesis (RUNX1, spinodal 0.585385, barrier 0.437252), the stem-cell-to-lineage branching that is the root of the whole system — is physically housed in the bone-marrow niche, the bone compartment owned by the musculoskeletal volume. This chapter declares that relationship as a one-way pointer: the immune volume owns the haematopoietic primitive, the musculoskeletal volume owns the absolute bone-niche scale (trabecular architecture, niche capacity, marrow volume), which stays [O] here.
There is a stronger claim available, and the honest move is to name it without asserting it. If the musculoskeletal volume's marrow-niche reading rides the same RUNX1 substrate, then the niche's haematopoietic-commit threshold would equal this volume's bone_marrow_hematopoiesis spinodal 0.585385 in closed form — a shared-substrate identity exactly analogous to the gut seam, where the IBD induction threshold equals luminal antigen plus the spinodal. But verifying a shared-substrate identity requires the live sibling engine, the way the gut identity is checked against digestive's live relapsing course. The musculoskeletal volume is not on disk in this session, so the identity cannot be verified live. The chapter therefore declares the weaker one-way pointer now and records the identity as a NAMED CANDIDATE with a stated promotion path (add the volume to the live harness, confirm substrate drift zero and the niche threshold equality) and a stated falsifier (if the niche threshold did not reduce to the haematopoietic spinodal, the identity candidate is dead — the pointer survives, because a pointer asserts no equality).
Why a pointer consumes nothing, and what is owned elsewhere
A one-way pointer is the safest cross-package construct in the program precisely because it consumes no sibling value. The immune engine computes the effector populations and the haematopoietic root with zero circulatory or musculoskeletal input; the sibling volumes read only where those populations and that root sit in their own tissue frame. The firewall is enforced, not asserted: a scan of every Python file in the gated package returns zero sibling imports, and the emitted emergence state carries no sibling key. Each new pointer lands on a real in-package organ — bone_marrow_hematopoiesis — verified against this volume's own substrate, so neither pointer is a dangling reference. The owner split is recorded for every cross-package quantity: the absolute vascular transport scale belongs to the circulatory volume, the absolute bone-niche scale to the musculoskeletal volume, and both are openly [O] here with their owner named.
What this is — and is not
This chapter is architecture and bookkeeping, not new physics. It adds no measurement to the thirty-five stress targets and changes no number in the engine; it declares two structural relationships and states honestly which is a pointer, which is a candidate identity, and what absolute scale is owned elsewhere and is therefore [O] here. The cross-package statements are direction-and-structure claims — not medical advice, not a validation of VP theory. The seam layer carries its own 2×sha256 (9357f23b…), separate from the engine emit() hash, which stays byte-identical.