CATALOGUE VALIDATION AND VISUAL-BINARY SENSITIVITY
Checked 3 October 2026

DATA PROVENANCE
Independently retrieved the original public CDS catalogue J/A+A/511/L10,
including ReadMe, table3.dat (53 UVES rows), table4.dat (47 FIES rows), and
stars.dat. URLs and SHA-256 hashes of the received bytes are recorded in
catalogue_verification.json. This is an independent source-data comparison;
the binary sensitivity reuses the original estimator implementation.

SOURCE COMPARISON
All 94 supplied object IDs, instrument choices, and all 658 primary numerical
inputs (FeH, MgFe, SiFe, NiFe, U, V, W) match CDS exactly after normalising
spaces/underscores in object names. Six duplicate measurements occur across
Tables 3 and 4: G20-15, HD148816, HD179626, HD189558, HD193901, HD194598.
Retaining Table 3/UVES produces 94 unique stars and follows Table 4 footnote b
of Nissen & Schuster (2010). The velocity cut yields 78 stars and leaves 16
below or equal to the 180 km/s threshold.

Four fields unused by the primary estimator differ for HD193901:
Field      Supplied/original arXiv Table 3    CDS Table 3
Teff       5656 K                            5650 K
xi         1.3 km/s                          1.2 km/s
CaFe       0.23 dex                          0.22 dex
TiFe       0.11 dex                          0.10 dex
This version discrepancy has no effect on any primary analysis input. The
supplied input CSV is preserved. The CDS-derived companion CSVs show the
independently retrieved values; agreement is exact for the seven primary
analysis variables. Eleven parenthesised source population labels are equivalently
encoded with an 'uncertain-' prefix in the supplied CSV.

MEASUREMENT-ERROR PROVENANCE
Section 3 of the source paper gives differential one-sigma precisions of
0.03-0.04 dex for Fe/H, Mg/Fe and Si/Fe, and 0.01 dex for Ni/Fe. Section 4
gives average halo velocity errors of (12,16,9) km/s. The analysis's
0.035-dex midpoint and independent Gaussian perturbations are assumptions,
not per-star uncertainty/covariance data supplied by the original paper.

VISUAL-BINARY SENSITIVITY
The source's Figure 1 caption identifies G112-43 and G112-44 as components
of a visual binary. Excluding either component separately or both jointly
tests whether this known dependent pair drives the estimator contrast.
The original total-speed >180 km/s sample is selected first; metallicity tertiles and
instrument-by-tertile strata are recomputed after exclusion. The same
original standardisation, GC and KSG estimators are used. Each case has
2000 fresh full-velocity-vector permutations within these strata and a
fixed seed recorded in binary_exclusion_sensitivity.csv.

Exclusion       N     GC MI     GC p      KSG5 MI   KSG5 p
G112-43         77    0.142725  0.667166  0.186725  0.015492
G112-44         77    0.148815  0.617691  0.191805  0.015992
Both            76    0.145097  0.646177  0.207223  0.002499
MI values are in nats. These are post hoc sensitivity results, not a
preregistered analysis. KSG k={3,4,5,6,8,10} remains above its own 95th
null percentile in all three cases; Gaussian-copula does not. The minimum
excess across the estimator family remains negative in every case.
This verifies persistence of the estimator contrast for this one known
visual pair; it does not establish independence of all stars or remove
other sample-selection, error-covariance, or spectroscopic-binary effects.

REPRODUCTION
From the archive root:
python catalogue/verify_catalogue.py
python catalogue/verify_catalogue.py --binary-sensitivity

CATALOGUE FOLDER CONTENTS
verify_catalogue.py, CDS_ReadMe, CDS_stars.txt, CDS_table3.txt, CDS_table4.txt,
catalogue_verification.json, cds_all_100_measurements.csv,
cds_verified_94_stars.csv, binary_exclusion_sensitivity.csv, this file.

PRIMARY REFERENCES
Nissen, P.E., Schuster, W.J. (2010). Astronomy & Astrophysics 511, L10.
https://doi.org/10.1051/0004-6361/200913877
https://arxiv.org/pdf/1002.4514 (source text checked: Sections 3-4, Figure 1
caption, Tables 3-4; publisher PDF access returned HTTP 403).
https://cdsarc.cds.unistra.fr/viz-bin/ReadMe/J/A%2BA/511/L10?format=html&tex=true
https://doi.org/10.26093/cds/vizier.35119010
