Earendel: Lens-Magnified High-z Star or Cluster

• Earendel is a highly magnified source at z~6.2, observed near a cluster critical curve that provides unique insights into early cosmic structures.
• Gravitational lensing models yield extreme magnifications (up to >4000) and constrain its source-plane size to as small as 0.005–0.02 pc, favoring compact stellar configurations.
• Spectral energy fits suggest Earendel may be a massive B-type star, binary system, or nascent compact star cluster, probing Population III and globular-cluster progenitor scenarios.

Searching arXiv for papers on Earendel and closely related lensing interpretations. Earendel, formally WHL0137-LS, is a highly magnified, point-like source in the strongly lensed Sunrise Arc at redshift $z\approx 6.2$, behind the galaxy cluster WHL J013719.8−082841. It was identified in Hubble Space Telescope imaging and subsequently re-examined with JWST as an object observed within the first billion years of cosmic history. The central interpretive question has been whether Earendel is an individual star, a very compact multiple-star system, or a compact stellar cluster seen through extreme gravitational lensing. A further question, motivated by its high redshift and uncertain mass, is whether it could be a Population III star. The literature therefore treats Earendel both as an observational outlier and as a test case for the limits of caustic-based source interpretation near cluster critical curves (Welch et al., 2022).

1. Discovery and astrophysical setting

Earendel was identified as an extraordinarily magnified source in the Reionization Lensing Cluster Survey (RELICS), embedded in the $z\approx 6.2$ Sunrise Arc galaxy. The object is remarkable because it is likely not a whole galaxy but either a single star or a small stellar multiple, made visible only because of extreme gravitational lensing by a foreground massive galaxy cluster (Schauer et al., 2022).

The early observational picture relied on a photometric redshift of $z_{\rm phot}=6.2\pm0.1$, and JWST imaging later yielded a combined estimate of $z=6.15^{+0.27}_{-0.34}$, consistent with adopting the fiducial value $z=6.2$ (Welch et al., 2022). A later JWST/NIRSpec PRISM analysis of the Sunrise galaxy obtained a spectroscopic redshift of $z=5.926\pm0.013$ from $[\mathrm{O\,II}]\lambda\lambda3727,3729$, $[\mathrm{Ne\,III}]\lambda3869$, H$\beta$, $[\mathrm{O\,III}]\lambda\lambda4959,5007$, and H$z\approx 6.2$0, slightly lower than the earlier photometric estimate but consistent with the Lyman-break-based interpretation of the stellar continuum (Pascale et al., 7 Jul 2025).

Because Earendel is seen at a cosmic epoch when Pop III stars are usually expected to have been largely replaced by metal-enriched Pop I/II populations, it immediately became relevant to several domains: low-metallicity stellar evolution, early chemical enrichment, the astrophysics of caustic-crossing events, and the feasibility of directly observing compact stellar sources at very high redshift (Schauer et al., 2022).

2. Lensing geometry and compactness constraints

The original force of the Earendel interpretation came from its location on the lensing critical curve and its unresolved morphology. JWST/NIRCam imaging in 8 filters spanning 0.8–5.0 $z\approx 6.2$1m, with 2104 s exposure in each filter, showed that Earendel remains a single unresolved point source in higher-resolution data. The images were reduced to 0.02″ per pixel in the short-wavelength channels, and the residuals after point-source subtraction were consistent with noise in every band (Welch et al., 2022).

This unresolved appearance substantially tightened the compactness argument. Under the smooth macro-lens interpretation, the lower limit on the magnification increased to $z\approx 6.2$2, and the source-plane radius was constrained to $z\approx 6.2$3 pc, or $z\approx 6.2$4 AU; some models allowed even smaller radii down to $z\approx 6.2$5 pc. The lensing analysis used the relation

$z\approx 6.2$6

where $z\approx 6.2$7 is the distance to the critical curve in arcseconds and $z\approx 6.2$8 is model-dependent (Welch et al., 2022).

These properties were initially taken to strongly disfavor an extended star cluster. A young massive cluster should have begun to show structure at JWST resolution, whereas Earendel remained unresolved; by contrast, a tight multiple-star system remained fully compatible with the size limits. HST monitoring across four epochs spanning just over two years showed no statistically compelling variability, although the range of measurements was suggestive at the $z\approx 6.2$9 to $z_{\rm phot}=6.2\pm0.1$0 level. That behavior was described as consistent with a microlensing picture in which the magnification should usually remain within a factor of about two (Welch et al., 2022).

3. Stellar and multiple-system interpretations

Assuming that the light is dominated by a single star, stellar-atmosphere fits to the JWST photometry yielded an effective temperature

$z_{\rm phot}=6.2\pm0.1$1

with best fits often near $z_{\rm phot}=6.2\pm0.1$2 K depending on the atmosphere grid. Under the magnification constraints used in that analysis, the delensed bolometric luminosity was inferred to be

$z_{\rm phot}=6.2\pm0.1$3

a range identified with very luminous B-type giants or even luminous blue variable-like objects (Welch et al., 2022).

The spectral-energy distribution was, however, not straightforwardly consistent with a single-star interpretation. The broadband shape appeared to show both a pronounced Balmer break and a steep ultraviolet slope, features described as hard to reconcile with a single-star SED. A two-star example fit, involving one cooler and one hotter component, could reproduce the data better, but the parameter space was too large and the number of photometric points too small to render that scenario conclusive (Welch et al., 2022).

Later discussion sharpened this tension. The cluster-based reinterpretation notes that earlier JWST/NIRCam photometry did not fit a single-star model well, but could be matched by a binary with two $z_{\rm phot}=6.2\pm0.1$4 stars, one hot $z_{\rm phot}=6.2\pm0.1$5 and one cooler $z_{\rm phot}=6.2\pm0.1$6. That account uses the mismatch between point-source stellar models and the observed continuum shape as part of the motivation for reconsidering Earendel as a compact stellar population rather than a lone star (Pascale et al., 7 Jul 2025).

A persistent misconception is that the unresolved morphology alone proves a single-star origin. The observational record supports a narrower claim: Earendel is best explained as an individual star or very compact multiple-star system, not as a resolved extended source under the original smooth-lens assumptions (Welch et al., 2022).

4. Population III hypothesis

A distinct line of analysis asks whether Earendel could be a Population III star, formed from pristine hydrogen-helium gas before heavy-element enrichment. The relevant calculation combined three ingredients: the probability that the host halo contains metal-free star-forming pockets, assumptions about the IMF, and stellar lifetimes as a function of mass (Schauer et al., 2022).

The environmental estimate began with the Sunrise Arc stellar mass of $z_{\rm phot}=6.2\pm0.1$7. Assuming a star-formation efficiency of $z_{\rm phot}=6.2\pm0.1$8 and a baryon fraction of $z_{\rm phot}=6.2\pm0.1$9, the host halo mass was inferred to have a lower limit of $z=6.15^{+0.27}_{-0.34}$0, placing it in the post-reionization regime considered by Liu & Bromm (2020). The adopted Pop III occupation probability was

$z=6.15^{+0.27}_{-0.34}$1

so only about $z=6.15^{+0.27}_{-0.34}$2 of such host halos were expected to contain metal-free star-forming regions (Schauer et al., 2022).

For Pop II/I stars, the analysis used a Larson-type IMF,

$z=6.15^{+0.27}_{-0.34}$3

with $z=6.15^{+0.27}_{-0.34}$4 and a lower mass cutoff of $z=6.15^{+0.27}_{-0.34}$5. For Pop III stars, two limiting IMFs were considered: a Larson-type IMF with $z=6.15^{+0.27}_{-0.34}$6, and a log-normal limiting case written as

$z=6.15^{+0.27}_{-0.34}$7

Pop III stellar lifetimes were taken from Schaerer (2002) and Marigo et al. (2001), while Pop II lifetimes were approximated for a $z=6.15^{+0.27}_{-0.34}$8 population (Schauer et al., 2022).

The result was explicitly mass dependent. Existing data already required Earendel to be massive, above roughly $z=6.15^{+0.27}_{-0.34}$9, but the exact mass remained uncertain. For the Larson-type Pop III IMF with $z=6.2$0, the Pop III probability stayed nearly constant at about $z=6.2$1–$z=6.2$2 across the allowed mass range. For the more top-heavy log-normal Pop III IMF, the probability exceeded $z=6.2$3 beyond $z=6.2$4 and reached about $z=6.2$5 at $z=6.2$6 (Schauer et al., 2022).

The analysis also modified the Pop II IMF by applying an exponential cutoff above $z=6.2$7,

$z=6.2$8

while leaving the Pop III IMF unchanged. Under that assumption, Earendel becomes more likely Pop III than Pop II at $z=6.2$9 for the log-normal Pop III IMF and at $z=5.926\pm0.013$0 for the Larson-type Pop III IMF; the transition mass was stated to be of order $z=5.926\pm0.013$1. In conservative Larson-type Pop III families with slope varying from $z=5.926\pm0.013$2 to $z=5.926\pm0.013$3 and characteristic mass down to $z=5.926\pm0.013$4, Earendel becomes more likely Pop III if its mass exceeds $z=5.926\pm0.013$5. The paper therefore concluded that Earendel is most likely a metal-enriched Pop II object, but that a Pop III probability ranging from roughly $z=5.926\pm0.013$6 to $z=5.926\pm0.013$7 remains possible depending on mass and IMF assumptions (Schauer et al., 2022).

5. Lensing-systematics revisions and the challenge to the single-star claim

Subsequent work questioned whether the original size and magnification inferences were overly dependent on a smooth macro-lens model. One study introduced tidally truncated NFW subhalos with masses in the range $z=5.926\pm0.013$8–$z=5.926\pm0.013$9, plus a uniform negative-mass disk to conserve mass, into the local lens near Earendel. The total deflection field was written as

$[\mathrm{O\,II}]\lambda\lambda3727,3729$0

with ray equation $[\mathrm{O\,II}]\lambda\lambda3727,3729$1. In that framework, subhalos corrugate the critical curve and replace a single smooth fold caustic with a network of smaller fold caustics, weakening the direct inference from “single unresolved image” to “single star” (Ji et al., 2024).

The quantitative revision was that the earlier smooth-model size bound of $[\mathrm{O\,II}]\lambda\lambda3727,3729$2 pc should be relaxed by a factor of a few to ten, allowing source sizes of $[\mathrm{O\,II}]\lambda\lambda3727,3729$3 pc and therefore permitting a compact star cluster. The same analysis found that subhalos could induce an astrometric perturbation of $[\mathrm{O\,II}]\lambda\lambda3727,3729$4, which was stated not to contradict observation. A single-star interpretation remained possible, but it was no longer forced by the lensing-size argument alone (Ji et al., 2024).

A more direct challenge came from a JWST joint strong- and weak-lensing reconstruction of the cluster WHL J013719.8-08284 at $[\mathrm{O\,II}]\lambda\lambda3727,3729$5. Using the hybrid code MrMARTIAN, a free-form grid, and parametric TNFW halo components, that study revised the multiple-image identifications in the Sunrise Arc and combined them with a weak-lensing catalog of 1183 background galaxies total, corresponding to source densities of 132.7 arcmin$[\mathrm{O\,II}]\lambda\lambda3727,3729$6 in module A and 99.2 arcmin$[\mathrm{O\,II}]\lambda\lambda3727,3729$7 in module B. The accepted strong-lensing models had lens-plane scatter below $[\mathrm{O\,II}]\lambda\lambda3727,3729$8, with a best-performing model at $[\mathrm{O\,II}]\lambda\lambda3727,3729$9 and $[\mathrm{Ne\,III}]\lambda3869$0 (Scofield et al., 11 Apr 2025).

In that reconstruction, Earendel’s magnification fell to $[\mathrm{Ne\,III}]\lambda3869$1–$[\mathrm{Ne\,III}]\lambda3869$2, dramatically lower than earlier literature values of $[\mathrm{Ne\,III}]\lambda3869$3–$[\mathrm{Ne\,III}]\lambda3869$4. The critical curve in the viable models lay about $[\mathrm{Ne\,III}]\lambda3869$5–$[\mathrm{Ne\,III}]\lambda3869$6 away and crossed the arc near image $[\mathrm{Ne\,III}]\lambda3869$7 rather than at Earendel itself. On that basis, the authors argued that Earendel is not well supported as a single star and is more plausibly part of a compact stellar system or a small star cluster / globular-cluster progenitor (Scofield et al., 11 Apr 2025).

6. Star-cluster and globular-cluster-progenitor interpretation

The most explicit cluster-based reinterpretation used deep archival JWST/NIRSpec PRISM spectroscopy of the Sunrise arc and fitted the rest-UV through optical continuum with simple stellar population (SSP) models from BPASS, BC03, and FSPS. The modeling adopted an instantaneous-burst history, added nebular emission via CLOUDY, used the flexible Salim et al. attenuation law, and imposed a Gaussian stellar-redshift prior centered on $[\mathrm{Ne\,III}]\lambda3869$8 with width $[\mathrm{Ne\,III}]\lambda3869$9 (Pascale et al., 7 Jul 2025).

A central empirical result was that Earendel’s continuum is well described by an SSP, with goodness of fit nearly equivalent to that of another distinct Sunrise knot, $\beta$0, which is described as confidently a star cluster. Across the libraries, the inferred parameters for Earendel were intermediate ages

$\beta$1

very low extinction

$\beta$2

and a metal-poor stellar population

$\beta$3

For the BPASS fit specifically, the paper reported $\beta$4, $\beta$5, and $\beta$6 before a white-noise inflation factor was applied (Pascale et al., 7 Jul 2025).

The interpretation depends primarily on the continuum rather than on absorption features, because the PRISM resolution is low and the lines are weak or absent. The authors argue that the UV/optical slope and Balmer break already favor an evolved, metal-poor stellar population more naturally associated with a compact star cluster than with a single hot star. The near-identity of the Earendel and $\beta$7 continua is central to that case: since $\beta$8 is already accepted as a cluster, its spectroscopic similarity provides an empirical template for Earendel (Pascale et al., 7 Jul 2025).

The same study places Earendel and $\beta$9 in the context of metal-poor globular-cluster progenitors at $[\mathrm{O\,III}]\lambda\lambda4959,5007$0. Their ages and metallicities were described as consistent with the age–metallicity relation seen in local globular clusters and in the E-MOSAICS simulations. This suggests a possible continuity between compact high-redshift clusters and present-day metal-poor globular clusters, although the authors explicitly retained caveats tied to lensing uncertainty, broad age posteriors, and the fact that spectroscopy alone does not formally rule out a single star or binary (Pascale et al., 7 Jul 2025).

7. Earendel in the broader theory of extremely magnified stars

Earendel occupies a defining place in the emerging class of extremely magnified stars and star-like sources near cluster caustics. Later JWST work on two candidates at $[\mathrm{O\,III}]\lambda\lambda4959,5007$1 behind MACS J0647.7+7015 explicitly framed them as the second highest-redshift examples to date after Earendel. In that comparison, Earendel remained the benchmark object at $[\mathrm{O\,III}]\lambda\lambda4959,5007$2 against which new caustic-crossing candidates were assessed (Meena et al., 2022).

This broader literature emphasizes that such detections are governed not only by the smooth macro-lens but also by microlensing by intracluster stars and, potentially, by subhalos. An analytic model for ultra-high magnification events generalized the standard caustic result to a microlensed cluster environment and treated Earendel as an archetypal Icarus-like system. In that framework, the high-magnification tail obeys

$[\mathrm{O\,III}]\lambda\lambda4959,5007$3

modulated by the number of independent microlens critical curves and by finite-source suppression. The model explicitly notes Earendel as a case observed essentially on the macro-critical curve, with magnification $[\mathrm{O\,III}]\lambda\lambda4959,5007$4, and interprets its detectability as the product of large background magnification, a microlensing caustic network, and strong threshold dependence of the event probability (Kawai et al., 2024).

The broader significance of Earendel therefore extends beyond its own unresolved morphology. If it is a single star, it probes the upper end of stellar luminosity and possibly even metal-free star formation at $[\mathrm{O\,III}]\lambda\lambda4959,5007$5. If it is a compact star cluster, it provides parsec-scale information on stellar populations and possible globular-cluster progenitors in the early universe. In either case, Earendel has become a focal object for testing how confidently one may infer astrophysical source class from a lone, highly magnified image near a cluster critical curve.