Is the Stellar Mass Spectrum Written Into Molecular Filaments Before Stars Form?

Molecular clouds are not uniform fog. In the cold, dusty nurseries where stars are born, gas gathers into filaments - long ridges that can stretch light-years. Dense cores, the seeds of individual stars, often sit inside them.

One of astrophysics' most durable patterns is the stellar initial mass function (IMF), the relative number of stars born at each mass. At the high-mass end, it follows a power law with a slope near 1.35, named after Edwin Salpeter. Why that slope looks similar across very different star-forming regions remains unresolved. A new arXiv preprint asks whether the slope is already present in filaments before stars form.

A Census of Filaments in Dust

Guo-Yin Zhang, Alexander Men'shchikov, and Jin-Zeng Li analyzed Herschel Space Observatory images at wavelengths from 70 to 500 micrometers, where cold dust emits. They studied seven nearby clouds - Taurus, Ophiuchus, Perseus, Orion A, California, IC 5146, and Vela C - at distances of 140 to 920 parsecs. They converted dust emission into gas surface density maps, then used multiscale software called getsf to trace filaments at eight spatial scales from 14 to 216 arcseconds. For each segment, they fit the cross-sectional density profile and measured the linear density, Lambda: mass per unit length. They were measuring the filament linear density function (FLDF), the statistical distribution of that quantity. They compared it with a critical value for an idealized 10-kelvin filament, about 15 solar masses per parsec. Above it, gravity can make a filament fragment into dense cores; astronomers call such filaments supercritical. The authors treat this threshold with a factor-of-two uncertainty, however, because real filaments need not match the idealized model.

A Density That Changes With Scale-and With Cloud

Filaments do not have a single characteristic density. Their median linear density rises approximately linearly with the scale on which they are detected. Combined across clouds, the median climbed from 0.70 solar masses per parsec at the smallest scale to 9.3 at the largest. Individual clouds ranged from about 0.07 to 3.4 at 14 arcseconds, and from 3.2 to 63 at 216 arcseconds. Larger-scale filaments are denser, perhaps because they include extended envelopes or have had more time to gather gas. A census at one scale would miss the hierarchy.

Gravitational stability varies widely. Pooled over all scales, the fraction of filament segments above the nominal critical density was about 7 percent in Ophiuchus and 8 percent in Taurus. Perseus, California, IC 5146, and Orion A were intermediate, at roughly 11, 12, 23, and 26 percent. Vela C reached about 54 percent. At the largest scale, the fraction above the nominal threshold ranged from 14 percent in Ophiuchus to 95 percent in Vela C. This ordering broadly matches independent evidence for how actively those clouds form stars.

The Salpeter Slope Emerges Only From the Whole Hierarchy

The most striking result came from combining every filament segment from every cloud and every spatial scale. That composite distribution of linear densities follows a power law: the number per logarithmic interval falls off as Lambda to the power minus alpha, with alpha between 1.30 and 1.34. The Salpeter IMF slope is 1.35.

The match does not appear at any single scale. Slopes at small scales are shallower, about 0.48 to 0.72. At the largest scale they reach about 1.19 to 1.26. Only the full multiscale ensemble produces the Salpeter-like value. The authors argue this suggests the stellar mass spectrum may be pre-encoded in the filamentary structure of molecular clouds, rather than imposed later.

What Would Link Filaments to Stars

The paper relates the core mass function (CMF) to the filament linear density function (FLDF) through a simple fragmentation model: the CMF slope psi equals (alpha plus eta) divided by (1 plus eta), where eta describes how the typical fragmentation length depends on linear density. If fragmentation length is independent of Lambda, eta is zero, and the CMF directly inherits the filament slope. Nearby-cloud CMFs have slopes around 1.3 to 1.4, consistent with this eta = 0 case. The observed Salpeter-like filament slope could then feed the Salpeter-like core mass function and eventually the IMF.

But this is still indirect: the FLDF and CMF are measured from different physical structures - filament segments and dense cores, respectively - and a direct test requires measuring both distributions in the same clouds.

Where the Filament-Star Link Remains Uncertain

Several uncertainties remain. Statistics weaken toward the smallest scales, where fewer filament segments have reliable profile fits, so the scale-dependent steepening is a trend rather than a precise relation. The critical density assumes an idealized isothermal cylinder, which real filaments need not resemble, and the authors adopt a factor-of-two uncertainty around it. Magnetic fields, turbulence, protostellar feedback, and competitive accretion can all alter how gas turns into stars.

The authors also tested resolution degradation and distance-dependent blending. Blending raises measured linear densities, they find, but does not steepen the slope; distance-binned composite slopes remain consistent with the Salpeter-like value.

There is also a tension. The ALMA-IMF program has reported a combined core mass function slope of about 0.97 across 15 high-mass protoclusters, shallower than Salpeter. Within the paper's framework, that would require a different fragmentation regime, with eta around 0.5, meaning higher-density filaments fragment on longer lengths. The authors note that their measured width-density scaling could be consistent with such a regime, but they call the discrepancy unresolved. And because the study is an arXiv preprint, its conclusions remain subject to peer review.

The central claim is not that the IMF has been proven to originate in filaments. It is that the same power-law tendency seen in stars appears in the gas on scales far larger than any star. If the connection holds, the universality of the stellar mass spectrum would not be something stars impose on their nurseries. It may be written into the filaments from which stars condense.