# Independent interferometric observable confirms black hole shadow in M87.

- **DOI:** [10.1038/s41586-026-11241-9](https://doi.org/10.1038/s41586-026-11241-9)
- **Rating:** Sound (3.5 out of 5)
- **Verdict:** Closure-invariant imaging strengthens the case for M87*’s compact ring across observing years, but shared data and conditional modelling limit the independence of its confirmation.
- **Framework version:** v5.13
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-11-16
- **Evaluation page:** https://www.nabu.science/paper/10.1038%2Fs41586-026-11241-9

## Key claims

1. We reconstruct the black hole shadow image and unequivocally reproduce a ring-like morphology, with an angular diameter of 42.2 ± 1.5 μas
   - Support: Tentative
   - Type: Primary finding
   - Basis: multi-epoch ring recovery does not establish categorical certainty or complete reconstruction uncertainty
   - Location: Abstract
2. a black hole mass of 6.5 ± 0.5 billion solar masses assuming nominal scaling parameters
   - Support: Supported
   - Type: Secondary
   - Basis: diameter-based mass inference remains conditional on borrowed scaling calibration and distance
   - Location: Abstract; Inferring the black hole mass
3. [M87* is] observed at an inclination angle of approximately 20° from the approaching jet
   - Support: Tentative
   - Type: Secondary
   - Basis: preferred inclination depends on the selected simulation family and spin assumptions without quantified inference uncertainty
   - Location: Abstract
4. In our work, when higher-quality datasets are considered, the sensitivity [of M87* image morphology] to field of view disappears, contrary to predictions based on the traditional hybrid mapping approach
   - Support: Tentative
   - Type: Secondary
   - Basis: field-of-view robustness is demonstrated within this imager without a matched hybrid-mapping comparison
   - Location: Interferometric imaging, paragraph 4
5. In all epochs, we measure a mean [M87* ring] diameter and standard deviation of d = 42.3 ± 1.1 μas from forward geometric modelling of closure invariants
   - Support: Supported
   - Type: Secondary
   - Basis: the reported diameter summarizes fits conditional on a ring-based geometric model
   - Location: Morphological analysis, paragraph 4

## Strengths

- **Ring recovery spans years and imaging windows** Reconstructions cover eight observing epochs in 2017, 2018 and 2021 and three fields of view. The best-coverage days consistently recover a compact ring, extending reconstruction-based support across observing years.
  - Dimension: contribution
  - Location: Interferometric imaging; Figure 1; Table 1
- **Targeted checks address specific reconstruction choices** The analysis varies reference stations and training-library composition and reports a frozen-model unblinding protocol. These checks address identifiable sources of reconstruction sensitivity without requiring hand-selected image initialization at inference.
  - Dimension: methodological rigour
  - Location: Methods, Closure invariants and GenDIReCT imaging
- **Discussion distinguishes stable from unstable features** The morphological analysis separates relatively stable diameter and orientation measurements from method-sensitive width, asymmetry and central brightness. It also engages published non-ring reconstructions and competing dynamical mass estimates.
  - Dimension: positioning
  - Location: Morphological analysis; Figure 2; Interferometric imaging

## Limitations

- **Headline certainty exceeds the conditional evidence** The abstract's unequivocal morphology language is stronger than warranted by ambiguous lower-quality epochs and generative diversity that is explicitly not Bayesian uncertainty. Its approximately 20° inclination is a preferred match within selected simulations, not a model-independent angle measurement.
  - Dimension: reporting
  - Location: Abstract; Methods, GenDIReCT imaging; Morphological analysis; Figure 4
- **Mass inference lacks method-specific scaling calibration** The mass estimate provisionally borrows α ≈ 11.0 rather than calibrating GenDIReCT on synthetic GRMHD data. The stated calibration range alone gives masses of 6.0–6.7 billion solar masses, limiting the independence of the physical inference.
  - Dimension: methodological rigour
  - Location: Methods, Inferring the black hole mass
- **Geometric agreement does not exclude non-ring alternatives** The complementary xsringauss fit assumes a ring and uses the same closure-invariant observations. Its agreement on diameter supports consistency within that model but is not an independent comparison against non-ring morphologies.
  - Dimension: methodological rigour
  - Location: Morphological analysis; Figure 3; Methods, Geometric modelling

## Reliability checks

Caveats (4 of 4 checks).

- **Results check:** Noted
  - ring-like morphology, with an angular diameter of 42.2 ± 1.5 μas: "from our non-parametric image reconstructions" (Morphological analysis, geometric-model diameter comparison) vs "The mean and standard deviation across all epochs for the ring diameter is listed as a summary statistic." (Table 1 caption). Using the eight rounded epoch diameters gives a mean of 42.25 μas and sample SD approximately 1.38 μas, compared with the reported 42.2 ± 1.5 μas. → Conclusion holds
  - Minor reporting notes (2):
    - Descriptive: "Perfect azimuthal symmetry corresponds to a value of" (Methods, Image-domain feature extraction, text following Equation 8). Equation 8 divides the modulus of the first angular mode by that same complex mode, producing a unit-modulus integrand where defined and an undefined ratio at perfect symmetry rather than the stated normalized asymmetry. → Conclusion holds
    - Descriptive: "blurring in the image domain with a circular Gaussian kernel is a pointwise multiplication in the Fourier domain by" (Methods, Geometric modelling, paragraph beginning Each image component). The inline Gaussian blurring factor has an imaginary exponent, whereas Equation 11 specifies the real Gaussian damping required for blurring. → Conclusion holds
- **Reporting and transparency:** Clean. Public EHT datasets and reference software are identified. Availability and competing-interest declarations deferred to online content are outside the supplied material and do not establish a reporting concern.
  - Data availability: declared · Code availability: declared
- **Reference integrity:** Noted
  - 54 references in manuscript
  - 51 of 53 checkable references found in an index
  - 1 are books, websites or datasets — counted, but not index-checkable
  - Could not be verified automatically — queued for manual review:
    - Hills, R. The James Clerk Maxwell Telescope (United Kingdom/Netherlands 15-metre antenna). In European Southern Observatory Conference and Workshop Proceedings Vol. 22 (eds Shaver, P. A. & Kjar, K.) 63–70 (1985). (Ambiguous: title ratio 0.55 via crossref_bibliographic sits between the divergence (0.30) and match (0.70) thresholds; too close to call either way, so it is recorded as unresolved rather than judged.)
    - Paszke, A. et al. PyTorch: an imperative style, high-performance deep learning library. In Proc. 33rd Conference on Neural Information Processing Systems (NeurIPS 2019) (Curran Associates, 2019). (Crossref bibliographic top candidate ratio 0.38 — no plausible match)
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Last checked:** 2026-11-16
