Axion string networks — the topological defects left over from symmetry breaking in the early universe — are supposed to evolve at a specific rate. This rate determines how many axions they produce, which constrains the axion's mass to a narrow window around a millielectronvolt.
Note (2026-11-07): the abstract says friction between the thermal bath and the axion string drastically changes the network for lower f_a values in DFSZ-like models, delays the onset of scaling and increases the axion energy density, and that the authors 'argue' that in addition to the standard m_a of about 1 meV, an axion mass of about 0.1 eV also reproduces the dark matter density. The essay says the window shifts and that 0.1 eV was previously excluded; the abstract adds a second mass, and the general lesson about cosmological exclusions is mine.
In DFSZ-like models, the strings interact with the thermal bath through tree-level fermion-axion couplings. This produces friction. The strings don't evolve freely — they're dragged by the plasma they move through.
Friction delays the scaling dynamics. The strings take longer to reach their equilibrium configuration. While they're delayed, they radiate more axions. The total axion energy density increases.
One dissipative mechanism — thermal friction on cosmic strings — shifts the viable axion mass range by two orders of magnitude. Masses around 0.1 eV, previously excluded, now reproduce the observed dark matter density.
The lesson: the mass window was never measured directly. It was calculated from a model of how strings evolve. The model assumed free evolution. Add friction, and the window moves. The constraint was on the model, not on the axion.
This is the general vulnerability of any “excluded by cosmology” argument. The exclusion depends on a chain of assumptions about the early universe's dynamics. Break one link — here, the assumption of frictionless string evolution — and the entire exclusion shifts.