[dan-garden]

Closes #3. Item 14 is in FusionVault/geo5d#1.

This gives an absolute gravimeter its tide correction and tide systems, puts Stray et al.'s gradiometer in the tunnel example, and gives gravitics an equivalence-principle module with real limits. Every test value was read from its source: LongmanTide's code and tests, the Mäkinen, Touboul, Fayet, Schlamminger and Asenbaum PDFs, and IERS TN36 chapters 6–7. One commit per item; the findings beyond the report are on #3.

Item	What landed	Checked against
8 metrology::{GM_MOON, EARTH_ROTATION}	Done in Phase 0 (D9)	—
9 gravimetry::tides::longman	Longman's closed form, as LongmanTide codes it, taking a Julian epoch	LongmanTide's vector to 3×10⁻¹¹ mGal (the issue asked for < 10⁻⁹). Averaged over 18.61 years it is δ₂ times the permanent tide, to 0.2 µGal
10 gravimetry::tides::{delta, gamma}	δₙ = 1 + (2/n)hₙ − ((n+1)/n)kₙ; γ = 1 + k − h	Longman's 1.1575; IERS nominal (TN36 Table 6.3, ch. 7) 1.15495 ≈ 1.16; Mäkinen's 0.69
11 gravimetry::tides::permanent	Mäkinen's Eq. 18 in Cartesian form, (W_T, g_T) along the normal	His Eqs. 19–21 to their printed digits; −30.485 µGal spherical; the 9.9 cm geoid offset; A from IERS H₀
12 Gradiometer in examples/buried_tunnel.rs	Diffraction order done in Phase 0 (D2); the example adds Stray's single-diffraction gradiometer	152 E and 17.7 mrad over the tunnel against Stray's 150 E and 17.5 mrad; the baseline difference equals the point gradient at the geometric-mean depth to 10⁻¹²
13 gravitics::equivalence	eta; TESTS (MICROSCOPE Ti–Pt and Pt–Pt, Eöt-Wash, Stanford) with Test::combine; eta_from_coupling and coupling_bound with Mediator; sphere_form_factor	2.746×10⁻¹⁵; α/(Gu²) = 1.2536×10³⁶; Fayet's Eqs. 24–25 for B, L and B − L at both spins (1.07 and 1.28×10⁻²⁵ unrounded); his Tables IV and V, through yukawa_factor and boson_mass_ev
14 geo5d test vectors	In FusionVault/geo5d#1	JILA, Wisconsin, solar redshift, Cassini, de Sitter, Mercury 42.98″, S2

A defect found and documented: moving C̄20 to TideSystem::MeanTide makes the tide potential part of the Earth's own field (∝ r⁻³ instead of r²). That is right only on r = R. Its gravity is +45.7 µGal at the equator against the true −30.5, so the doc now sends mean-tide gravity to tides::permanent, and a test pins the factor −3/2. The doc also now attributes the −4.1736×10⁻⁹ offset to EGM2008, as TN36 quotes it. IERS's own A₀H₀k₂₀ gives −4.2007×10⁻⁹.

Where the plan changed:

• combine is a method on the published-results registry rather than a free hypot.
• coupling_bound takes the mediator's spin and picks the right side of MICROSCOPE's dissymmetric limits itself.

Release (after merge, with your go-ahead): compatible z-bumps: geopotential 0.3.5 (doc only), gravimetry 0.7.1, gravitics 0.5.5, physica 0.3.4.

Checks: 424 workspace tests, clippy -D warnings on all targets, cargo doc with -D warnings, and the 30 research check scripts.

🤖 Generated with Claude Code