# Riess, Casertano, Yuan, Bowers, Macri, Zinn, Scolnic — A Comprehensive Measurement of the Local Value of the Hubble Constant with 1 km s⁻¹ Mpc⁻¹ Uncertainty (2021)
**Adam G. Riess** and SH0ES team (Space Telescope Science Institute / Johns Hopkins University). arXiv:2012.08534, *ApJ* 908 L6 (2021) and follow-up *ApJ* 934 L7 (2022). (Paper uses Gaia EDR3 parallaxes combined with HST Cepheid photometry to calibrate $H_0$.)
---
## Overview
The SH0ES team uses Gaia EDR3 parallaxes for 75 Milky Way Cepheids combined with HST spatial-scanning photometry to anchor the extragalactic distance ladder, yielding $H_0 = 73.2 \pm 1.3$ km s⁻¹ Mpc⁻¹. The paper explicitly marginalises over a global Gaia parallax zero-point offset, and finds **+14 µas residual** after applying the Lindegren EDR3 correction — i.e., the L21 correction **overcorrects** the Cepheid sample by 14 µas. This residual is the footprint of the sample-dependence of Gaia's zero point ([[2024_Ding_GalacticPlane_Parallax_Bias]], [[2025_Ding_Bright_Star_Parallax_Bias]]). The result is 4.2σ in tension with the Planck CMB value of $H_0$.
---
## PDF p.1 — Abstract (key numbers)
![[Riess2021_CepheidH0-001.png]]
> "We find $H_0 = 73.2 \pm 1.3$ km s⁻¹ Mpc⁻¹, a 4.2σ difference with the prediction from Planck CMB observations under ΛCDM. We expect to reach ~1.3% precision in the near term from an expanded sample of ~40 SNe Ia in Cepheid hosts."
---
## PDF pp.3–11 — The Lindegren overcorrection
![[Riess2021_CepheidH0-003.png]]
![[Riess2021_CepheidH0-011.png]]
From the SH0ES methodology:
> "We find an indication of overcorrection of the parallax offset for Cepheids, with a median difference of 15 µas in the residuals" after applying Lindegren 2021.
> [!critical] L21 overcorrects Cepheids by +14 to +15 µas
> Riess et al. apply the official Lindegren EDR3 parallax correction to their 75 Cepheid sample and find a median residual of +14 µas (earlier 2021 paper reports +15 µas). Overcorrection means the published Gaia parallaxes, plus the Lindegren correction, are too small by 14–15 µas. If uncorrected, the Cepheid distances are systematically too large, which biases $H_0$ downward by a small but measurable amount. SH0ES compensates by marginalising over a global offset term.
---
## Why Cepheids need different PZPO than quasars
- Cepheids are concentrated in the Galactic plane (where quasars are sparse).
- Cepheids are bright (G < 11 mostly; quasars are G > 16).
- Cepheids are variable stars (chromaticity changes with phase).
Each of these factors takes Cepheids out of the regime where Lindegren's quasar-based correction is properly calibrated. The 14 µas residual is consistent with:
- Ding 2024 galactic-plane offset of −31 µas vs L21's −17 µas.
- Ding 2025 orbital-parallax bright-star offset of −39 µas.
- Bobylev 2026 VLBI offset of −38 µas.
> [!note] The +14 µas Cepheid residual fits the bright-star PZPO pattern
> If the true bright-star Gaia PZPO is around −30 to −39 µas (as the bright-star cross-checks consistently find), but the L21 correction only applies −17 µas, then the residual is +13 to +22 µas. This is exactly what SH0ES measures at +14 µas. The Cepheid result is not an anomaly; it is a data point in the bright-star PZPO cluster.
---
## The Hubble-constant tension
| Method | $H_0$ (km/s/Mpc) |
|---|---|
| SH0ES Cepheids + SNe Ia (this paper) | 73.2 ± 1.3 |
| SH0ES 2022 follow-up | 73.04 ± 1.04 |
| Planck CMB + ΛCDM | 67.4 ± 0.5 |
| Tension | 4.2–5σ |
The tension is "one of the most important open problems in cosmology" (widely quoted).
---
## Implications for the Gaia parallax calibration
- If the Cepheid PZPO is actually −39 µas (Ding 2025) instead of −17 µas (Lindegren), the SH0ES Cepheid distances shrink by a small amount, which increases $H_0$ slightly.
- Conversely if Riess is right about the Cepheids, then L21's global recipe is simply wrong for the Cepheid sample.
- Either way, the Hubble-constant value depends on which PZPO the analyst chooses.
- The ~1 km/s/Mpc uncertainty in $H_0$ maps to ~1% in distance, which is ~20 µas in parallax at 1 kpc — the same scale as the PZPO disagreements.
---
## Summary
| Claim | Riess et al. (2021) |
|---|---|
| Lindegren EDR3 correction gives accurate Cepheid parallaxes | No. +14 µas residual — overcorrection. |
| The Hubble constant tension is resolved | No. 4.2σ. |
| $H_0$ is independent of Gaia PZPO choice | No. Explicitly marginalised over a global PZPO. |
| Gaia EDR3 is a clean distance anchor for Cepheids | Not yet. Residual PZPO of ~15 µas exists in the Cepheid sample. |
---
## Citation
Riess, A. G., Casertano, S., Yuan, W., Bowers, J. B., Macri, L., Zinn, J. C., Scolnic, D., 2021. "Cosmic Distances Calibrated to 1% Precision with Gaia EDR3 Parallaxes and Hubble Space Telescope Photometry of 75 Milky Way Cepheids Confirm Tension with ΛCDM." *Astrophysical Journal Letters* 908, L6. arXiv:2012.08534.
Follow-up: Riess, A. G. et al., 2022. *ApJ* 934, L7.
Local copy: .
---
## See also
- [[2021_Lindegren_EDR3_Parallax_Bias]]
- [[2024_Ding_GalacticPlane_Parallax_Bias]]
- [[2025_Ding_Bright_Star_Parallax_Bias]]
- [[2026_Bobylev_VLBI_Gaia_Zero_Point]]
- [[2021_Groenewegen_Parallax_ZeroPoint_Offset]]