Emily Noyes Vanderpoel's documented "good" dyad/triad hue groupings (Color Problems, 1902, Plate XXXVI "Good Dyads or Pairs" and Plate XXXVII "Good Triads"), served straight from the build artefacts vanderpoel/vanderpoel_dyads.json and vanderpoel_triads.json.

Given a hex: snaps it to the nearest of Vanderpoel's eleven real hue-family labels, resolves that hex against the archive, and checks whether the label is a member of any documented dyad or triad. Each OTHER member is returned with its own nearest real archive match (CIEDE2000, same machinery colour_harmonies uses), so a caller gets a named archive colour with provenance rather than a bare 1902 label string.

Unlike hay_contrast, this carries no proportional ratio -- only Vanderpoel's own categorical "good" editorial judgement, printed verbatim on the plates. Member names are sourced from a transcriber's digitisation_alt_text (an accessibility description of the plate image), not verified printed captions -- every response says so in member_source_caveat, and each grouping additionally carries its own record's wording of that caveat.

Returns groupings: null (not an empty list, not a substituted weaker answer) when there is nothing documented -- for a near-neutral input, where hue is not meaningfully defined and Vanderpoel's hue-family judgements do not apply, or where the nearest label is a member of no dyad or triad. With today's 11-label / 10-grouping dataset every label IS in at least one grouping, so in practice the near-neutral case is the one that fires. Silence is the honest answer, not a forced match -- same principle as hay_contrast's covered:false.

Example: {"hex": "#CC1A14"}