Fe II · measured products, uncertainty and reproducible evidence
Fe.json v1.223 · 30 Fe II products · 160 total
Solar Fe I anchor: 7.477 ± 0.052
HARPS · Reference Grade · Amarsi 3D-NLTE · n = 21. The selected Fe I product anchors this Solar appendix; products and engines are not averaged together.
Kitt Peak · Synth · 1D-LTE · Gerber · Reference Grade · n = 12
Uncertainty budget closed for six of eight rows (RYA-1226); two Fe I Kitt Peak/molecfit rows keep a legacy bar pending their microturbulence bracket. On the migrated rows most of the quoted systematic is a conservative profile-width term carried from a measurement in another band.
HARPS · Synth · 1D-NLTE · Bergemann · Deep Grade · n = 2
How to read this forest. Each measured row is a separate abundance product for an instrument, holding, line pool and treatment; engines are never averaged together. Solid bars show statistical uncertainty from the line measurements; wireframes show systematic uncertainty from the product’s adopted error budget, which does not average away with more lines. Green and gold regions mark the Asplund and Lodders literature values and uncertainties, not pass/fail gates. Grades describe the line pool’s provenance and selection, not the quality of an engine. Reddish-orange mean-3D / Frankenstein rows are experimental and not adopted. Unmeasured model rows remain N/A.
1D-LTE uses a one-dimensional atmosphere with atomic populations in local thermodynamic equilibrium. 1D-NLTE allows departures from that equilibrium; the named implementation supplies the correction or synthesis treatment.
Synthesis fits the spectral line profile. See Methodology for the full model, line-selection and uncertainty procedures.
Fe II traces singly ionized iron and is retained as an ionization and consistency diagnostic alongside Fe I, not an independent headline to average with it. The current resolved pool contains strong, saturation-sensitive lines and documented blend exclusions; the diagnostics below distinguish those limitations from model and holding differences.
Molecular opacity INCLUDED — OH A-X, NH, CH and CN violet, 6,223 VALD lines as Turbospectrum lists with the dominant isotopologue (RYA-1207). The band previously carried none: every molecular list iSpec ships begins at 400 nm. Measured PAIRED against a molecules-off run on the same code and pool, the lever is -0.047..-0.055 dex on Fe I and -0.276..-0.332 on Fe II, with chi2r median 81.8 -> 67.4 and control windows containing no molecular line bit-identical. RYA-1209 additionally curated out Fe I 3721.272 (systematic:K07 — gf attributed to a VALD3 reference that publishes no per-line uncertainty, no NIST accuracy class, so it never met the DEEPGRADED tier's lab-gf definition); the Fe I pool is 57 candidates and the budget clears at gf rung 3 (cited lab), 55/55 GF-LAB, RMS 0.0509 dex. RESIDUAL DEFICIT: the correction still UNDER-corrects — RYA-1204 measured synth/obs above 1 on 24 of 40 windows after it — so the band remains short of opacity by an amount not yet measured on this pool, and these values are a floor rather than a converged answer. sigma_syst is unchanged at 0.1122 because it is dominated by the 0.10 dex pseudo-continuum term, not by gf: the true continuum is never observed here. Not adopted for the headline. The headline is the per-band best-available product, selected from this feed by the publication generator rather than fixed as a constant: RYA-850 promoted the graded laboratory-gf pool to the PRIMARY reported value, and RYA-819 retired the earlier constructed anchor after showing its Magic-2013 correction wrong in both magnitude and shape. RYA-161: nothing calibrated to a target.
4 experimental products remain available in the Fe I product values and downloads.
RYA-1197 Part A: the Frankenstein mean-<3D> route is HELD FROM HEADLINE publication until the Bride validates it. It inflates A(Fe) by roughly +0.10 dex against the 1D siblings and exists on VIS/KP only, so it is not a comparable measurement of the same quantity. It stays in the feed deliberately: the LTE/NLTE mean-3D PAIR is the NLTE differential on ONE atmosphere (RYA-542), which is the diagnostic it is for. Not a headline value.
Committed proof census: 120 line/holding records; 100% have observed depth > 0.70. Median observed log10(EW[mÅ]/λ[Å]) = -1.252. In consistent Å units this is -4.252. These diagnostics characterize the audited saturated pool, not every newly added product.
Download PDF — Solar Fe II appendix
6686 graded per-line measurements exported. Evidence for 22 graded products is missing or does not reproduce the feed; 0 exported lines lack an unambiguous gf match. The download is explicitly incomplete until those upstream records are supplied.
Generator scripts/generate_fe_publication.py v1.0.0
Science source commit 74217457b5a004eac983f9e34eb23d5fc4b89fd7
Fe.json v1.223 · feed timestamp 2026-09-24T16:11:18Z
Generated 2026-09-24T16:34:16+00:00
Feed SHA-256 140e81ffc805708f97f1a1caede872c40cd5eef3e05580211ecf0ee6a0e72728