Webb finds the gas blown out of early dead galaxies mostly falls back

Sodium is an awkward tracer, because the Na I D doublet also appears in the light of the stars themselves. Zhu's group modeled each galaxy's stellar population, subtracted the absorption those stars should produce, then ran thousands of simulated noise tests on what was left to check the signal was real ^[7]. For the confirmed cases they fit the absorption for gas velocity, density and the fraction of the galaxy the outflow covers, and converted those into a mass of gas expelled ^[8].

The reason to chase the cool phase is mass. "In galaxies with both neutral and ionized outflows, the neutral outflow rates are typically 10-100 times higher than the ionized outflow rates," the authors write in the paper, published in Astronomy & Astrophysics on Aug. 14 ^[5][3]. Most outflow studies measure the hot, ionized gas ^[4]. "JWST/NIRSpec has now enabled direct detections of Na I-traced outflows well beyond the local universe," they write ^[6].

"Since most outflows are not likely to escape, their direct role in quenching or maintaining quiescence is uncertain," the authors write ^[15]. They describe the winds as "short-lived fountain cycles rather than ejective quenching events" ^[16]. Returning gas can be used again, and the paper allows that star formation may reignite once the supply is replenished ^[14].

What drives the winds is unsettled ^[18]. Two of the galaxies with excess sodium absorption host an actively feeding black hole, and two galaxies with confirmed feeding black holes show no outflow at all ^[19]. Of the four confirmed feeders, half do nothing measurable to the neutral gas ^[30]. The two most extreme outflows are both linked to active black holes ^[19]. "Their origins are not directly tied to ongoing AGN activity or the most recent starbursts, but they do potentially reflect a mix of past AGN activity and ongoing mergers," the team concludes ^[20].

One design feature limits how far any of this travels. Every galaxy in the sample was selected because it had already stopped forming stars ^[26], so the spectra record these systems after the shutdown, and a faster wind that cleared one of them out earlier would not appear in the data ^[31]. What the measurement constrains is what the winds visible now can do to a dead galaxy. The sample cannot show whether black hole activity killed these galaxies, and the authors themselves point to past AGN activity and mergers ^[20].