Science · 6 min read
Why Measurements of the Hubble Constant Still Disagree
A comparison of early-universe inference, the local distance ladder, independent checks, and unresolved systematics.

Why Measurements of the Hubble Constant Still Disagree
One expansion rate, several routes to it
The Hubble constant, written H₀, describes the present relation between the recession velocity of a sufficiently distant galaxy and its distance. Its conventional unit—kilometers per second per megaparsec—states how much faster cosmic expansion appears for each additional megaparsec of separation. Measuring H₀ is not the same as timing a car on a road. Astronomers infer distances and velocities from light that has crossed an evolving universe, using physical models and calibrated sequences of observations.
Two mature routes give persistently different answers. Analyses of the early universe, especially the cosmic microwave background measured by the Planck satellite and interpreted within the standard flat ΛCDM cosmological model, infer a value near 67.4 kilometers per second per megaparsec. A local distance ladder led by the SH0ES collaboration, using Cepheid variable stars to calibrate type Ia supernovae, reported 73.04. Their quoted uncertainties are small enough that the estimates do not comfortably overlap.
This is the “Hubble tension.” The measurements do not show two expansion rates operating in different places. They show that methods connecting different observations to the same parameter disagree under their stated calibrations and models. The unresolved question is whether an unrecognized systematic error, a statistical fluctuation, incomplete cosmological physics, or some combination explains the gap.
The early-universe inference
About 380,000 years after the big bang, cooling matter allowed electrons and nuclei to form neutral atoms, and photons began traveling relatively freely. The cosmic microwave background, or CMB, preserves a patterned map of temperature and polarization from that period. The angular sizes and amplitudes of its acoustic features depend on the contents, geometry, and expansion history of the universe.
Planck did not observe today’s galaxies receding and read H₀ directly. Its collaboration fitted a six-parameter ΛCDM model to the CMB power spectra. Within that model, the physical scale of sound waves in the early plasma and their observed angular scale constrain the cosmic parameters. The final Planck analysis reported H₀ = 67.4 ± 0.5 kilometers per second per megaparsec for base ΛCDM.
That precision is conditional. ΛCDM assumes, among other things, general relativity on cosmological scales, a spatially flat universe in the base model, cold dark matter, a cosmological constant, and a specified early-universe particle content. The model fits a wide set of data remarkably well, which is why the inference carries weight. But adding new components or changing early expansion can shift the inferred sound horizon and H₀.
Baryon acoustic oscillations measured in the later distribution of galaxies provide a related standard ruler. Surveys such as the Dark Energy Spectroscopic Instrument, or DESI, measure the ruler’s apparent scale across redshift. BAO strongly tests expansion history, but an H₀ inference usually requires calibrating the ruler through early-universe physics or combining BAO with another absolute scale. Describing BAO as a wholly model-free measurement of H₀ would therefore be misleading.
The local distance ladder
The distance ladder constructs an absolute scale through overlapping measurements. Geometric anchors—such as parallax, water-maser distances, and distances inferred from eclipsing binaries—calibrate the luminosities of Cepheid variable stars. Cepheids then calibrate nearby type Ia supernovae. Similar supernovae observed farther into the smooth Hubble flow connect distance to recession redshift.
Each rung transfers a calibration, so the analysis tries to observe anchors, Cepheids, and supernovae with consistent instruments and methods. The SH0ES team’s 2022 comprehensive Hubble Space Telescope analysis reported H₀ = 73.04 ± 1.04 kilometers per second per megaparsec. Its authors tested alternative selections, metallicity treatments, crowding corrections, and anchors. The result remained above the Planck ΛCDM inference.
Potential systematics receive sustained scrutiny. Cepheids live in crowded stellar fields, and unresolved neighbors can bias brightness measurements. Their period–luminosity relation depends modestly on chemical composition and wavelength. Supernova samples require corrections for light-curve shape, color, host environment, and survey selection. Peculiar velocities—motion caused by local gravity rather than cosmic expansion—matter for nearby galaxies.
The James Webb Space Telescope offers sharper infrared imaging of Cepheid fields and access to other stellar distance indicators. Comparisons have not produced a single agreed resolution. Some analyses find that JWST observations support the Hubble calibration; work led by Wendy Freedman combining the tip of the red-giant branch, carbon stars, and Cepheids has favored a value closer to the high 60s or about 70. Differences in samples, calibrations, and combination methods remain under examination.
Independent checks and proposed explanations
Other techniques are valuable because their errors differ. The tip of the red-giant branch uses a feature in stellar evolution rather than pulsating stars. Megamaser disks can yield geometric galaxy distances from their orbital structure. Time delays between multiple images of a gravitationally lensed variable source constrain combinations of distances. Standard sirens use gravitational-wave signals, with redshift information, to infer the scale.
These methods have not yet converged at a precision and consensus that forces one main route to yield. Some results sit nearer the local ladder, some nearer the CMB inference, and many have wider uncertainty intervals. Gravitational-lens estimates depend on models of lens mass and structures along the line of sight. Standard-siren precision depends on event counts, detector calibration, viewing geometry, and host redshifts.
If measurement errors are not responsible, new physics must alter the connection between early and late observations while preserving other successful predictions. Proposed ideas include extra energy density before recombination, additional relativistic particles, changing dark-energy behavior, or modified gravity. Early dark energy can reduce the inferred sound horizon and raise the CMB-compatible H₀, but fits must also satisfy CMB details, BAO, nucleosynthesis, structure growth, and supernova data. Improvements in one statistic can worsen another.
There is no accepted new-physics model that resolves the tension across the full evidence base. The space of proposals is partly a measure of the constraint: changing expansion at one epoch affects several observables. A plausible account needs predictive gains beyond introducing enough flexibility to move one parameter.
What remains unknown
Several findings are established. The Planck maps and the local distance observations are high-quality measurements. Base ΛCDM fits the CMB and many other cosmological data well. Within that model, Planck infers a lower H₀ than the Cepheid–supernova ladder measures. Reanalyses have not made the discrepancy disappear through one acknowledged arithmetic error.
The interpretation remains contested. Researchers disagree about the residual size of stellar-photometry systematics, the best way to combine distance indicators, and whether model extensions improve the total fit enough to count as evidence for new physics. Quoted statistical significance can change with dataset choices and assumptions about correlated uncertainty; it is not a direct probability that ΛCDM is false.
Future progress can come from more geometric anchors, independent supernova calibrators, larger standard-siren samples, improved lens modeling, and cross-survey observations of the same stars. New CMB and galaxy surveys will test proposed model changes with observables other than H₀.
The boundary is therefore sharper than the solution. Astronomers can identify which inferential chains disagree and quantify much of their uncertainty. They cannot yet say whether the missing term belongs to an instrument model, an astrophysical calibration, or cosmology. Calling the tension evidence of a problem is justified. Calling it a discovery of new physics is not.
Sources
4 sources, 3 of them with a link you can follow.
- Planck Collaboration, “Planck 2018 Results. VI. Cosmological Parameters,” Astronomy & Astrophysics 641 (2020): A6, doi:10.1051/0004-6361/201833910.
- Adam G. Riess and colleagues, “A Comprehensive Measurement of the Local Value of the Hubble Constant with 1 km/s/Mpc Uncertainty from the Hubble Space Telescope and the SH0ES Team,” Astrophysical Journal Letters 934 (2022): L7, doi:10.3847/2041-8213/ac5c5b.
- Wendy L. Freedman, “Measurements of the Hubble Constant: Tensions in Perspective,” Astrophysical Journal 919 (2021): 16, doi:10.3847/1538-4357/ac0e95.
- Particle Data Group, “Cosmological Parameters,” Review of Particle Physics (2022 and subsequent online updates), section on the Hubble constant.
