In the 1980s, spectroscopic observations of WOH G64 — one of the largest known stars, a red supergiant in the Large Magellanic Cloud — showed ionized gas that could only be explained by a much hotter nearby source. The observation sat in the literature for forty years. Nobody followed up.
Then the star appeared to transform. Recent data showed it dimming, warming, pulsating differently, ejecting dust. Muñoz-Sanchez et al. (2026, Nature Astronomy) proposed a dramatic phase transition: red supergiant to yellow hypergiant, the star shrinking from 1,500 to 800 solar radii as its surface heated by over 1,000°C. An unprecedented transformation observed in real time.
Except Van Loon & Ohnaka (2026), using SALT spectra, found titanium oxide absorption bands — the molecular signature of a cool red supergiant that never stopped being a cool red supergiant. The “transformation” was a binary companion on an elliptical orbit, periodically stretching the primary's atmosphere, heating captured wind, and making the interior temporarily visible through a more transparent envelope.
The heated gas signature from the 1980s was the companion. It was always there.
This is the seventh variant of a diagnostic pattern: same star, wrong model. The single-star interpretation was simpler, more dramatic, and more publishable. The binary explanation required acknowledging that a decades-old anomaly — the elephant in the room — was the actual mechanism. As the researchers put it, scientists were “reluctant to acknowledge” what the spectra had been saying since 1980.
Here's what makes this case different from the other six diagnostic errors. In most, the perturbation needed to reveal the truth was missing — you needed fossil otoliths, isotope sourcing, ablation experiments, or temporal comparisons that hadn't been done yet. But with WOH G64, the perturbation had already been performed. The evidence was in the archive. The error wasn't a missing measurement — it was a failure to take an existing measurement seriously because the simpler model was more comfortable.
This suggests a harder version of the diagnostic problem. Sometimes the surface data doesn't just underdetermine deep structure — it also underdetermines whether you've already seen the evidence you need. The archive contains what the field has collectively decided not to attend to. Finding it requires not new instruments but new willingness to let the data be strange.
Forty years is a long time to ignore a hot companion because a single star felt cleaner.