Eclipsing spectroscopic binaries from Gaia DR3 orbits and TESS: survey data products (92 systems at periods of 13–119 days)

Notice (4 November 2026). The paper that these data products accompany was withdrawn from New Astronomy before review, after an audit found that its description of the method does not match the analysis as it was run on several points (among them the propagation of the Gaia period uncertainties and some of the criteria that define the sample). The detections are not in question, but the sample and its description will change. A corrected paper and a corrected version of this record are in preparation; until then, use the counts given here with caution. Data products of the survey of eclipses at the conjunctions predicted by Gaia DR3 single-lined spectroscopic orbits in TESS light curves, accompanying the paper "Companions at 13–119 days from Gaia DR3 spectroscopic orbits and TESS: masses and radii for 54 eclipsing systems found at the predicted conjunctions" (Chopard-Vilhem 2026, submitted to New Astronomy). 9 389 orbits searched in up to three TESS sectors each; 221 stars with two consistent primary dips and 562 with one; the candidates absent from the variability catalogues vetted on every available sector; 92 eclipsing systems established at periods of 13–119 days (20 beyond 30 d), 88 of them not identified as eclipsing when their list was deposited, 54 with companion masses and radii from detached-binary fits. The record includes an integrity and reproduction check (check_z4.py) that recomputes the paper's funnel from the deposited files without a network connection, and a check of every number of the paper's text (check_numbers.py). Contents: the selection of Gaia DR3 SB1 orbits observed by TESS with a conjunction predictable to one day (selection_v2.parquet); every predicted conjunction tested, with the depth and significance of the deepest dip and the control level (conjunctions.jsonl); the candidate stars with their star-level evidence and automatic catalogue checks (candidate_stars.jsonl); the batch vetting of the candidates absent from the variability catalogues — difference imaging, dips at predicted conjunctions in every sector, eclipse ephemerides, Gaia primary parameters, companion masses, verdicts (vetting_summary.csv, vet_*.json); the eclipse fits of the confirmed systems (fits_table.csv, fit_*.json); the completeness, archive and literature checks; the tables and figures of the paper with the scripts that draw them; and the online material of the paper (unconfirmed candidates, predicted primary eclipses of December 2026 to December 2027, the literature search behind its novelty statement, the cross-match with the independent list Eclipses by Appointment). Light curves are not included: they are the SPOC, TESS-SPOC and QLP products on MAST. CC BY 4.0. Version 1.1.0 supersedes 1.0.0 (doi:10.5281/zenodo.22998329): the funnel counts only the 9 389 selected stars (788 flagged, 561 single-dip); the candidate file has one line per source with grades recomputed from the conjunction rows; Table 3 uses the DR3 eccentricity for four partial eclipses whose free fit departs from it by more than 3σ; the software is lctriage 1.2.0 (doi:10.5281/zenodo.21916566, all versions). The data rows are unchanged. Version 1.1.1 changes only the software reference inside README.md and CITATION.cff: version 1.1.0 (doi:10.5281/zenodo.23003140) cited lctriage 1.2.0 by a version DOI that was never registered. Version 1.2.0 supersedes 1.1.1 (doi:10.5281/zenodo.23003581). The mass–radius sample of the paper now requires a reduced chi-square of the detached-binary fit of at most 10: six fits above it, previously counted as partial eclipses, are counted as poorly fitted, and the sample has 54 systems (34 full and 20 partial eclipses) instead of 60; check_z4.py applies the rule. Table 3 is regenerated accordingly: three partial eclipses are computed with the DR3 eccentricity, with a/R1 printed rescaled together with R2; K1 is given in km s−1. Figures 4 and A.1 are redrawn. The data rows are unchanged. Version 1.3.0 is the version cited by the paper submitted to New Astronomy; it supersedes 1.2.0 (doi:10.5281/zenodo.23023549). The survey, vetting and fit files are unchanged. The conjunction registry holds 74 rows written twice by 33 stars processed twice in a row; it is deposited as written, and every count now uses one row per conjunction (42 910 conjunctions; 783 stars with a flagged conjunction). One system of the earlier list, TIC 95244521, had entered the candidates only through a dip written twice: it is no longer counted, and the record has 92 systems (Table 2 and Figures 1, 2, 5 and A.1 updated). The archive and literature checks of the 93 are completed (ADS full text included). Table 3 gives, for each fit, the in-eclipse rms of the residuals over the out-of-eclipse scatter, and marks the eight fits above 5, which stay in the sample. Figure 3 is drawn in vector form from the light curves. New: the completeness on the TESS Ten Thousand catalogue (Kostov et al. 2025), the online tables, the literature search behind the novelty statement, the cross-match with Eclipses by Appointment (against the TIC numbers of that list at a fixed commit, with the SHA-256 of the original file), check_numbers.py (110 numbers of the text), and the bibliography of the paper. See README.md for details.