A halo of 3.7 billion solar masses that holds 3 million solar masses of neutral hydrogen is running a gas fraction of about 0.08 percent [10][1]. That ratio, rather than the missing starlight, is the measurement carrying the interpretation, because the mechanism under test is a halo that kept a trace of gas through reionization and then had it heated by the ultraviolet background from quasars and stars until it could never collapse into stars [6]. Chen and colleagues at the National Astronomical Observatories, Chinese Academy of Sciences, report that this large gas-to-halo ratio closely matches theoretical predictions for such systems [10][4].
To qualify as a reionization-limited H I cloud, an object has to be compact, roughly spherical, show narrow spectral lines with velocities around 20 km/s, and show no stellar component at optical wavelengths [7]. The velocity criterion is the one a single dish handles well, since line widths come out of spectra. Two of the four are statements about shape, and shape depends on angular resolution rather than on collected flux.
FEASTS points at galaxies, so the volume searched is galaxy neighbourhoods. Two passing objects across 55 targets works out to roughly 0.04 candidates per galaxy field [9][2]. That is a detection rate, not a space density, and the prediction at issue is the abundance of low-mass halos across the universe [8]. The pair cannot yet be weighed against that number.
The same pointing strategy explains the awkwardness. Faint gas is easiest to detect close to a bright galaxy a survey has already targeted, and gas pulled out of a galaxy also ends up close to it. The ambiguity is built into where the search looks. What breaks it is resolution: the paper closes by saying higher-resolution radio observations will be needed to tell a genuinely starless dark-matter halo from the remnant of a galactic collision [14].
The published account names FEASTS as the data source without stating a sensitivity threshold [15], so how much of this find belongs to FAST's collecting area and how much to where the survey chose to look is not something that account settles. What is on the table is narrower and still worth having: two objects whose gas-to-halo ratio behaves the way a reionization-limited cloud should, described by the team that found them as compatible with the framework rather than as a confirmation of it [11].
