NASA Exoplanet Archive, confirmed planets with both radius and period: 21,524 usable records.
The measurement
Radius histogram, 0.1 R⊕ bins. Counts through the region of interest:
| Radius (R⊕) | Count |
|---|---|
| 1.4 | 1,072 |
| 1.6 | 1,014 |
| 1.7 | 1,001 |
| 1.8 | 845 |
| 1.9 | 756 |
| 2.0 | 742 |
| 2.2 | 850–1,016 |
| 2.5 | 772–859 |
There is a real deficit centred at 1.9–2.0 R⊕, bounded by a super-Earth peak near 1.4–1.7 and a sub-Neptune rise at 2.2. That is the radius valley, and its position is consistent with the published value of roughly 1.8–2.0 R⊕.
The correction
My first automated valley-finder returned 4.55 R⊕ for the whole sample and 0.65 R⊕ for long-period planets. Both are bin-edge artefacts of a rule that searched the full range and locked onto noise. The histogram above is the real measurement; the automated numbers were discarded rather than reported.
This is the third time this session that a check caught an error the first pass would have published.
Duplicate bins
The distribution CSV contains two rows for several bins (e.g. two entries near 1.4 and 2.0). That is a float-rounding artefact in bin assignment, not two populations. The counts above merge them. The CSV is left as-generated so the bug is visible rather than silently cleaned.
What this is and is not
It is a successful measurement of a known feature of nature — a check that the archive and the pipeline recover real astrophysics. It is not a discovery: the valley was found in 2017 and its interpretation (photoevaporation and core-powered mass loss) is an active literature.
Limits
- Detection bias: small, long-period planets are under-detected, which distorts the histogram shape.
- The archive is a heterogeneous compilation of surveys with different selection functions.
- No stellar-radius or insolation correction; period is a crude proxy.
- The valley position depends on bin width and on the radius precision of each detection method.