DR1 — First public data release

SpaceCatalog DR1 — First public data release

Released 2026-08-21. DOI: not yet deposited.

Every distance carries the method that produced it, every modelled stellar parameter carries the quality cut that admitted it, and every galaxy distance is computed under one stated cosmology.

What is in this bundle

• objects.csv.gz — 223,448 objects, one row each.
• designations.csv.gz — 520,680 catalogue identifiers, keyed on the object slug.
• relations.csv.gz — 3,396 object-to-object relations (double-star pairs, and what orbits what).
• sources.csv — the 16 upstream catalogues, with the licence each publishes under and the date each was retrieved.
• source_disagreements.csv — 82 places two upstreams describe one object differently, and what this release kept.
• manifest.json — every count, checksum and cut in machine-readable form.
• SHA256SUMS — a checksum per file.

Licence

The pipeline and the site are Apache-2.0 (LICENSE). The data is not one licence: every row names its source in the source column, and sources.csv gives that source's licence. The bundle mixes CC-BY-4.0, CC-BY-SA-4.0, ODbL, CC0 and public-domain rows. Attribution is required for the CC-BY and CC-BY-SA ones; the share-alike terms reach anything derived from them.

Release notes

DR1 is the first state of the catalogue meant to be cited. There is no
release before it, so what follows describes what a row is rather than
what changed.

Every object names the catalogue it came from and the licence that
catalogue publishes under, and every distance names the method behind it.
For an exoplanet system that method is the archive column and the paper the
archive credits for it — which for 280 microlensing systems is a Bayesian
estimate from a galactic model with no parallax measured anywhere in it,
and not the parallax the label used to claim.

Modelled stellar parameters are gated rather than republished. Gaia's
photometric temperatures, radii, masses and ages are withheld from 4,592
stars whose astrometric solution failed the standard binarity test
(RUWE > 1.4, Lindegren et al. 2021), and the 17,874 kept values that fall
in a regime the model is known to handle poorly — too hot, or classified as
a pair — say so on the row.

Doubt is carried in the numbers rather than in prose. Published
uncertainties are stored as [minus, plus] pairs; a distance derived from
a redshift carries the redshift's own measurement error together with the
±4% the expansion rate is uncertain by, added in quadrature. For the 31,510
galaxies whose redshift is photometric rather than spectroscopic, that bar
is twelve times wider than what earlier versions of these pages showed.

Where a redshift is small enough that the choice of reference frame decides
whether there is a Hubble flow to measure at all, no distance is given in
either frame: 93 galaxies, several of them well known, carry no distance
here rather than one that was right by accident.