Astronomers propose a 1,000-hour JWST survey to read galaxies’ stellar histories

A proposed observing campaign would spend roughly 1,000 hours of James Webb Space Telescope time collecting unusually deep spectra of distant galaxies. Its aim is not simply to detect more galaxies, but to measure the faint stellar light needed to reconstruct how they formed, enriched their stars with chemical elements and, in some cases, stopped forming stars.

The proposal comes from Adam C. Carnall and a large collaboration in a November 4 preprint. It combines a scientific case, an empirical exposure-time estimate and an illustrative survey design. It does not report a completed survey, and its preprint status should not be mistaken for established peer-reviewed validation.

Reading beyond the brightest lines

Many existing observations with JWST’s Near Infrared Spectrograph, NIRSpec, target emission lines: bright spectral features that reveal properties of gas. Measuring a galaxy’s continuum—the underlying light across a range of wavelengths—and the absorption features superimposed on it generally requires longer observations.

That additional information matters because stellar age, chemical composition and dust can produce similar changes in a galaxy’s overall colours. Individual spectral features help disentangle those effects. According to the authors, deep continuum measurements could constrain when stars formed and how their chemical composition changed, rather than providing only a snapshot of the gas at the time of observation.

Absorption features also carry information about stellar motions, which can constrain dynamical masses and merger histories. Features produced by neutral gas can reveal inflows and outflows, helping researchers investigate whether gas expelled by active galactic nuclei contributes to shutting down star formation.

The authors identify further opportunities involving the compact sources called little red dots, early star formation and cosmic reionization. These are proposed applications, not problems solved by the new analysis.

Why spectral resolution matters

The team advocates NIRSpec’s medium-resolution gratings, with a resolving power of about 1,000. Resolving power describes how finely an instrument separates neighbouring wavelengths.

NIRSpec’s lower-resolution prism covers a broad wavelength range and remains valuable for confirming redshifts and studying extremely faint galaxies. But the paper shows how it can blend features that medium-resolution observations separate, including stellar hydrogen absorption lines and important emission-line groups.

The distinction is practical: detecting a broad feature does not necessarily provide enough detail to infer the underlying physical properties reliably. The authors therefore argue that prism observations cannot generally replace medium-resolution spectra for the investigations they describe.

Turning existing data into a survey estimate

To estimate the required observing time, the researchers analysed spectra from the EXCELS survey. They measured average signal-to-noise per pixel over a specified rest-frame optical interval, retaining galaxies with more than half of that interval covered. They then compared those measurements with apparent brightness in the NIRCam filter closest to a rest-frame wavelength of 0.55 micrometres.

Their empirical relationship links continuum signal-to-noise to brightness and exposure time. Across the EXCELS observations, the authors find that the same relationship describes different redshifts and gratings. Its exposure-time dependence follows a square-root scaling: longer integrations improve signal-to-noise, but with diminishing returns.

The planning target is a signal-to-noise ratio of about 20 per pixel at medium resolution. Drawing on work cited in the paper, the authors associate that quality with roughly 0.1-dex precision in stellar ages and metallicities. This is a target for suitable analyses, not a demonstrated precision for every future survey galaxy.

Their estimate indicates that galaxies at a rest-optical apparent magnitude of 24.5 would need approximately 20 hours per grating. They estimate that this brightness limit would provide 90 percent mass completeness above roughly 20 billion solar masses across redshifts 1.5 to 5.

A blueprint with unresolved choices

The sketch combines ten deep pointings, each taking about 80 hours including overheads, with sixteen shorter pointings of about 12 hours. The shorter observations would target brighter galaxies while maintaining a similar signal-to-noise goal. NIRSpec’s microshutter array would allow many targets to be observed simultaneously.

A central motivation is statistical power. The authors argue that small, unevenly selected existing samples cannot adequately establish how stellar ages, metallicities and other properties vary with mass and redshift.

Background subtraction remains a significant technical choice. A shared “master” background could reduce random noise and roughly double the number of simultaneous targets compared with conventional arrangements. However, it introduces systematic uncertainties that may not diminish with longer exposures. The authors recommend tests on similarly deep datasets before committing substantial observing time.

The paper’s conclusion is therefore a case for investment, not a guaranteed scientific return: build a systematically selected reference sample while JWST’s distinctive infrared spectroscopy remains available, and validate the observing strategy needed to make that sample reliable.