Redshift: what is observation, what is postulate

Why this article exists

The word “redshift” is used across cosmology as if it referred to one thing. It does not. It refers to a chain — an observation, followed by interpretive layers that are not observations but are habitually reported as though they were.

The layers have been conflated for so long that a reader encountering “the galaxy is receding at velocity v” or “the universe is expanding” cannot tell, from the statement alone, where the observation ended and the interpretation began.

This article separates them. Each statement is labelled for what it is. The labels are fixed:

The purpose is not to take a position on which postulates are correct. The purpose is to make it possible — for anyone — to know at every step whether they are looking at the sky or at a layered interpretation of the sky.

The instrument output

1.1 · What the detector registers

A spectrograph pointed at a distant luminous source produces exactly one kind of primary output:

Photon counts at wavelength bins at angular positions on the sky at times of arrival.

This output exists as numbers in a detector file. The numbers do not presuppose any cosmological model, any theory of gravity, any assumption about the universe’s global geometry, or any interpretation of what the photons represent.

1.2 · What “wavelength bins” means

The spectrograph’s wavelength calibration is performed against laboratory reference sources — known atomic emission lines from calibration lamps (typically neon, argon, thorium-argon, or similar) whose transition wavelengths have been measured in terrestrial laboratories under controlled conditions. The calibration does not require any cosmological assumption. It requires only that the laboratory measurements are accurate and that the spectrograph’s optics are stable during the observation.

1.3 · What “angular positions” means

The spectrograph’s angular calibration is performed against reference stars with known positions on the celestial sphere (astrometric catalogs, ultimately tied to very-long-baseline interferometry against distant extragalactic radio sources). The angular position of a photon arrival is a measurement in the detector’s reference frame, subsequently transformed to an astronomical coordinate system.

1.4 · What “arrival time” means

The arrival time is a reading of the observatory’s clock, ultimately traceable to atomic time standards (TAI, UTC, or equivalent). The time coordinate is local to the observatory.

The observation called redshift

2.1 · The computable quantity

When the spectrograph records photons from a distant source, certain wavelengths show intensity patterns (emission lines, absorption lines) that match, in their relative spacing and intensity ratios, the patterns of known atomic transitions measured in the laboratory. For a given identified transition, one computes the ratio:

z = ( λ_observed − λ_laboratory ) / λ_laboratory

where λ_observed is the wavelength at which the pattern appears in the spectrograph’s output, and λ_laboratory is the wavelength of the corresponding transition as measured in terrestrial laboratories.

This ratio is a number. It is arithmetic on two measurements. It does not depend on any cosmological model. It does not depend on any theory of gravity. It does not depend on any assumption about the universe’s geometry or history.

2.2 · The observation, stated without interpretation

For a given distant luminous source, the ratio λ_observed / λ_laboratory is greater than 1 for most observed galaxies — that is, the patterns appear at longer wavelengths than the corresponding laboratory transitions. This fact is what the instruments register when pointed at distant galaxies.

2.3 · What the observation is <em>not</em>

The observation is not:

• That the source is moving.
• That the source is moving away.
• That the source has a velocity.
• That space is expanding.
• That the universe has a history.
• That the source is at any particular distance.
• That time is passing differently at the source.

None of these is contained in the observation. Each is a separate interpretive move, to be audited below.

The constancy of atomic physics

3.1 · What is assumed when a laboratory transition is matched to an observed pattern

The identification of an observed spectral feature with a specific atomic transition rests on one assumption:

The atomic transition measured in the laboratory corresponds to the same atomic transition in the distant source.

This requires that the physics of the atomic transition — the energy levels, the selection rules, the transition probabilities — is the same at the source as it is in the laboratory.

It is supported by the observation that the patterns of spectral lines from distant sources preserve their internal structure — the relative spacings of multiplet components, the intensity ratios within a multiplet, the selection-rule behaviour — in a way consistent with terrestrial atomic physics applied to the same species. It is not independently verified at the source, because no laboratory at the source has been accessed. The term “empirically transferred” names exactly this status: inferred from internal consistencies of the observation, not measured at the target.

3.2 · Why this matters

The assumption is minimal in the sense that it does not import cosmological structure. It is non-minimal in the sense that it is an assumption, not an observation. If the fundamental constants governing atomic physics (the fine-structure constant α, the electron-to-proton mass ratio μ_p, etc.) varied across cosmological space or time, the redshift ratio would receive contributions from that variation, separate from any interpretation of z as motion or expansion.

The constancy of atomic physics is the boundary between direct observation and interpretation.

The Doppler postulate

4.1 · The interpretive move

The ratio z is habitually interpreted as a line-of-sight velocity via the Doppler relation:

• For non-relativistic sources: z ≈ v / c
• For relativistic sources: 1 + z = √( (1 + β) / (1 − β) ), where β = v/c.

This interpretation is not contained in the observation. It is a postulate layered onto the observation.

The Doppler interpretation selects one specific physical mechanism (source motion) to explain the wavelength ratio. Other mechanisms are logically possible: gravitational potential differences between source and observer, historical “tired-light” proposals, variations in atomic physics, or alternative structures that produce wavelength ratios. The Doppler interpretation is chosen by convention, not forced by the data.

4.2 · Hubble’s own position

The discoverer of the redshift–distance correlation, Edwin Hubble, was explicit that the velocity interpretation was a postulate and not the observation.

Across his 1929 paper, his 1935 Tolman collaboration, the 1936 monograph The Realm of the Nebulae, the 1937 Rhodes Lectures, the 1942 American Scientist article, and the 1953 George Darwin Lecture, Hubble maintains a consistent distinction: the redshift is an observation; the velocity interpretation is a postulate, to be adopted or rejected on grounds other than the observation itself.

The person who discovered the redshift–distance correlation did not endorse the Doppler interpretation as physical fact, and said so, in his own published work, across a quarter-century.

4.3 · What the Doppler interpretation requires

For the interpretation “z means velocity” to hold, one must additionally commit to:

• that the source’s rest frame and the observer’s rest frame are inertially connected in a well-defined way over cosmological baselines;
• that the relativistic Doppler relation applies at the scales in question (which requires a specific spacetime structure between source and observer);
• that no other mechanism contributes to z.

None of these commitments is in the observation. Each is an additional postulate.

The expansion postulate

5.1 · The further interpretive move

When the Doppler interpretation is combined with the observation that more distant sources have larger z, one obtains the correlation:

z increases with inferred distance.

This correlation itself is an observation (given the distance-ladder assumptions required to infer distance — see §7). It does not, by itself, specify what causes the correlation.

The standard interpretation is that the correlation reflects cosmic expansion — that space itself is expanding, that the wavelength of photons in flight is stretched by this expansion, and that the apparent recession of distant sources is not motion through space but a global expansion of spatial geometry.

1. General relativity as the correct theory of gravity at cosmological scales. This is an extrapolation from the regimes where GR has been tested (solar system, binary pulsars, strong-field phenomena at LIGO scales, gravitational lensing at certain scales) to scales 10–15 orders of magnitude larger. Extrapolation, not independently verified at cosmological scales before application.
2. The FLRW metric as the correct solution class of GR. FLRW is selected from the infinite solution space of Einstein’s field equations by imposing a specific set of symmetry assumptions (see §6). Selected solution class, selected by assumption, not derived from observation.
3. The cosmological principle — spatial homogeneity and isotropy at sufficiently large scales. Assumption about the global structure of the universe made from a single vantage point within it. Not verifiable from any single vantage point.

5.2 · What the expansion postulate is <em>not</em>

The expansion is not observed. The wavelength ratio z is observed. The interpretation of z as a manifestation of expanding space is a three-layer postulate (GR + FLRW + cosmological principle) applied to the observation.

A reader who sees the statement “the universe is expanding” and takes it to be an observation has been misinformed about the epistemic status of the claim.

FLRW — the full list of embedded assumptions

FLRW is not a neutral background. It is a symmetry-selected solution class loaded with explicit assumptions. Every item below is a separate commitment. The stack is a package deal.

Symmetry assumptions

1. Spatial homogeneity. At any given cosmic time, all spatial points are equivalent.
2. Spatial isotropy. At any given cosmic time, all spatial directions are equivalent from any point.
3. Global cosmic time. A preferred foliation of spacetime into spacelike hypersurfaces labelled by a time parameter.
4. Trivial topology of spatial sections. Conventionally assumed; non-trivial topologies (toroidal sections, etc.) are mathematically permitted but excluded.
5. Constant spatial curvature on each time slice.

Matter-content assumptions

1. Perfect-fluid stress–energy tensor. No anisotropic stresses, no heat conduction, no viscosity.
2. Comoving matter. Matter is at rest in the preferred frame on average.
3. Specified equations of state for each matter component (dust: p = 0; radiation: p = ρ/3; cosmological constant: p = −ρ; dark energy: p = wρ).
4. Non-interacting components. Total stress–energy is a sum of independently-conserved species.

Dynamical assumptions

1. General relativity is the correct gravitational theory at cosmological scales.
2. The Einstein–Hilbert action (or equivalent) with no higher-curvature terms.
3. Minimal coupling of matter to geometry.
4. Fixed fundamental constants — G and c do not vary across cosmic time.
5. Equivalence principle holds at all scales and epochs.

Averaging, initial conditions, observer location

1. Commutativity of averaging and dynamics. Backreaction assumed negligible. Non-commutation is generic for non-linear equations; it is being set aside by assumption.
2. Homogeneity scale. Above ~100 Mpc (contested) the universe is effectively homogeneous.
3. Smooth initial conditions (or, alternatively, inflation must be invoked separately to “explain” the smoothness).
4. No preferred direction in initial conditions.
5. We occupy a typical location. Our vantage is representative.
6. The comoving frame is realised by the CMB rest frame. The CMB dipole is read as our peculiar motion, not as an intrinsic directional structure.
7. (No hidden twentieth) — every modeller must, if they wish to claim FLRW, subscribe to the nineteen above. Any slip in any item renders the subsequent inference framework-conditional on that specific slip.

When “FLRW” appears in a sentence, the sentence imports the full stack.

The distance ladder

7.1 · What “distance” is, in cosmology

No distance in extragalactic astronomy is measured directly. Every quoted distance is the output of a chain of calibrations, each rung of which inherits the assumptions of the rungs below it:

1. Parallax — geometric, assumption-light, good only to ~kpc.
2. Cepheid variables, calibrated against parallax; assumed to have a stable period–luminosity relation across host galaxies and metallicities.
3. Type Ia supernovae, calibrated against Cepheids; assumed to behave as standard candles across cosmic time and environments.
4. Tully–Fisher, fundamental plane, surface-brightness fluctuations — each calibrated against lower rungs.
5. At the highest redshifts: distances inferred from FLRW applied to the measured z, which is circular with respect to the expansion postulate.

7.2 · What a “distance to a galaxy” statement contains

When a galaxy is said to be “at distance D”, the statement contains the observation (angular position, flux, spectrum), the assumption that the calibration chain applies to this object, and the interpretive layer (Doppler + expansion + FLRW, if z is involved in the inference). The number D is the output of the chain. It is not a direct measurement.

7.3 · The Hubble tension, epistemically

The so-called “Hubble tension” — the ~5σ disagreement between H₀ values inferred from the CMB (via ΛCDM) and from the local distance ladder (via Cepheids + Type Ia SNe) — is a disagreement between two outputs of the same interpretive machinery applied to two different datasets, within the same framework. It is not a tension in the universe. The universe has whatever behaviour it has. The disagreement is a failure of the framework to return consistent values of one of its own parameters from two of its own inference chains.

Framework-internal inconsistency. Reported, in the popular register, as “tension in the universe” — a grammar that inverts the model–reality relation.

The cumulative stack

To convert “we detected photons at wavelengths λ_obs at angular position θ at time t” into “the galaxy at distance D is receding at velocity v because the universe is expanding at rate H₀”, the following stack must be accepted:

Each layer above 2 is a postulate. Each postulate is habitually reported in the register of observation.

A statement like “the universe is 13.8 billion years old and 68% dark energy” is a statement whose every term is conditional on every layer of the stack. It is not an observation.

What is empirically backed, what is not

Empirically backed

• Photon counts at wavelength bins at angular positions at arrival times. (Layer 0.)
• Laboratory calibration of wavelengths against terrestrial atomic transitions. (Layer 1b.)
• The computed ratio z for identified transitions. (Layer 2, conditional on 1a.)
• The correlation: z increases with observational distance proxies (Hubble’s original finding; refined and extended).
• CMB photons arrive with a near-perfect blackbody spectrum at ~2.7 K. (Layer 0, combined with Planck’s law.)
• Distant sources appear at systematically different wavelengths than nearby sources.

Not empirically backed, but assumed

• That z is caused by motion (Doppler). Postulate.
• That z is caused by cosmic expansion. Postulate — requires GR + FLRW + cosmological principle.
• That FLRW describes the universe’s global geometry. 20-item assumption stack.
• That distance-ladder calibrations apply unchanged across cosmic time. Chain of assumptions.
• That atomic physics is unchanged across cosmological space and time. Empirically transferred; not independently verified at the source.
• That the CMB dipole is purely kinematic (our motion) rather than an intrinsic feature of the universe. Postulate, chosen to preserve isotropy.
• That dark matter, dark energy, and inflation are entities with the properties required to close the framework’s residuals. Postulates with no independent verification in the sense required to ground them as entities.

The test for empirical backing

A quantity is empirically backed if and only if its value can be determined from the observation without passing through an interpretive layer. By this criterion:

• z is empirically backed.
• “velocity of a galaxy” is not.
• “expansion of the universe” is not.
• “distance to a galaxy at high z” is not.
• “age of the universe” is not.
• “composition of the universe” is not.

This does not mean these quantities are wrong. It means they are framework-conditional. Their values change if the framework changes. Their empirical status is contingent on the framework’s correctness.

What became of anomalies

Once the framework is in place, observations that do not fit the framework’s predictions are habitually relabelled:

• The CMB dipole, aligned with the axis of the quadrupole–octupole alignment, aligned with the axis of hemispherical power asymmetry, aligned with the cold spot’s direction: “anomaly.”
• H₀ disagreement between CMB and local ladder: “tension.”
• σ₈ disagreement between CMB and weak lensing: “tension.”
• Lithium abundance disagreement between BBN prediction and observation: “problem.”
• JWST observations of mature, well-formed galaxies at z > 10: “puzzle.”
• Radio / quasar number-count dipole amplitude exceeding the kinematic prediction by 2–5× (Secrest et al. 2021 and follow-ups): “anomaly.”

Each of the items above is a direct observation. Each is in the sky. Each is being relabelled in a vocabulary that subordinates the observation to the framework’s authority to label it.

Observations do not have anomalies. Models have anomalies.

To call an observation “anomalous” is to grant the model ontological priority over the direct measurement. This inverts the relation between representation and what is represented. A parametric model does not have the authority to label reality as error. If the model and the observation disagree, the disagreement is information about the model.

The minimal epistemic commitment

The minimal commitment required to speak about redshift — to use the word without smuggling in interpretive content — is:

1. Detector outputs exist — photons are being counted at wavelengths at positions at times.
2. Laboratory physics is reliable — terrestrial atomic transitions are measured correctly.
3. Atomic physics is approximately constant across the space and time being surveyed (a transferred assumption, acknowledged as such).

From these three commitments, one can compute z for identified transitions and report the correlation of z with angular position and with observational distance proxies. That is the empirical content of “redshift.”

Every further claim — velocity, expansion, distance at high z, age, composition, evolution, fate — requires additional postulates. Each postulate can be named and audited individually. None is contained in the observation.

The observation is z. The rest is what has been built on top.

The word “redshift” in professional and popular usage conflates: a direct observation (wavelength ratios); a laboratory-backed measurement (wavelength calibration); a minimal transferred assumption (atomic physics constancy); the Doppler postulate; the expansion postulate; the FLRW metric’s twenty-item assumption stack; the distance ladder’s chain of calibrations; and the full ΛCDM concordance framework.

A reader cannot tell, from a sentence like “galaxy A at z = 0.5 is 5 billion light-years away and receding at 150 000 km/s due to cosmic expansion,” which parts are observations and which are outputs of the stack. All of it is reported in the grammatical register of fact.

The purpose of this article is to restore the distinction. Not to argue that any particular layer is wrong. To make it possible, at any moment, for any reader, to answer the question: at this step, am I looking at the sky, or am I looking at an interpretation of the sky?

Without that distinction, the enterprise that calls itself cosmology cannot be audited. With it, every claim becomes traceable to its epistemic source.

This article is part of the Mathematical Research Institute of Physical Reality’s ongoing audit of the assumption stack of modern cosmology. Every factual claim is traceable to the primary sources cited above. No claim depends on any framework-internal inference.