AI-generated illustration for Galileo Whispers of generic microwave feedhorns, representing the CMB-measurement context used to constrain cosmic initial conditions. This is not a named instrument, a map of adiabatic or isocurvature modes, or evidence of an isocurvature detection.
The adiabatic vs. isocurvature distinction concerns two different kinds of initial perturbation in the early universe. An adiabatic perturbation preserves the relative number-density ratios of the particle species. A density isocurvature perturbation changes relative composition while initially compensating the total density in the appropriate description. These are possible initial conditions, not two components that observations have established must both be present.
CMB measurements strongly support predominantly adiabatic initial conditions. Searches for isocurvature constrain particular alternatives and mixtures; they do not provide one universal percentage limit that applies to every mode, scale and correlation.
“Everything changes together” needs a careful definition
The early universe contained radiation, ordinary matter, dark matter and neutrinos. A local perturbation could put more of several ingredients in one region, but equal fractional changes in their energy densities are not generally adiabatic.
For nonrelativistic matter and photons, the linear adiabatic relation is δ_m = 3δ_γ/4, where the subscripts m and γ identify matter and photons, and each δ is the density change divided by that component’s background density. Photons’ energy density and number density depend differently on temperature; this is why the factor is three-quarters rather than one.
The 2025 Particle Data Group CMB review, section 29.4, gives this relation. Equivalently, the matter-to-photon number ratio has no spatial perturbation. “Adiabatic” here classifies relative entropy perturbations between species. It is not a claim that no physical process in cosmic history exchanged heat.
Follow two explicitly simplified examples
First suppose the photon energy density in a region is 1% above its background value. An accompanying matter perturbation of 0.75% satisfies the linear adiabatic relation. Assigning matter the same 1% energy-density contrast would instead leave a relative-composition perturbation.
Second, imagine an initial snapshot with 20 arbitrary energy-density units in matter and 80 in radiation. Add 0.2 unit of matter and remove 0.2 unit of radiation. The total remains 100 units. Matter rises by 1%, while radiation falls by 0.25%. This is an illustration of compensated density changes, not a measured cosmic composition.
The relative-entropy combination S_mγ = δ_m − 3δ_γ/4 is then 1% − 3(−0.25%)/4 = 1.1875%, so the perturbation is not adiabatic. The arithmetic distinguishes absolute density changes from fractional ones: equal and opposite absolute changes do not imply equal fractional changes when the backgrounds differ.
These examples use small, linear perturbations evaluated consistently at one initial time and omit other species. They teach the density-mode distinction, not a complete relativistic solution. They should not be applied unchanged to neutrino velocity isocurvature or every compensated baryon–dark-matter mode.
The CMB records how the initial conditions evolved
Before the universe became transparent, photons and ordinary matter participated in acoustic oscillations. Different initial conditions start and drive those oscillations differently, leaving different patterns of temperature and polarization across angular scales.
The important clue is therefore more than the presence of a warm or cold patch. Researchers compare the pattern of peaks and correlations with model predictions. Hu and Sugiyama’s perturbation analysis identifies acoustic structure as a way to discriminate initial conditions, alongside the roles of matter content, thermal history and geometry. It does not make a single peak a unique label for one cosmic origin.
Our introduction to the CMB gives the broader setting. The distinction here is the relationship among primordial ingredients, rather than the later temperature change caused when ancient photons cross a galaxy cluster or an evolving gravitational potential.
Why a constraint needs its model attached
The Planck Collaboration’s 2019 manuscript, published in the Planck 2018 results series, defines cold-dark-matter, baryon and neutrino density modes, as well as a neutrino velocity mode. Section 9 separately treats correlations and compensated baryon–dark-matter perturbations. These alternatives do not all create the same observable CMB signature.
A quoted limit must therefore state which mode was tested, whether it can correlate with the adiabatic perturbation, which scale defines the amplitude, which data were used and what confidence interval is reported. An upper bound on primordial power is also different from a bound on a contribution to observed temperature variance.
The absence of a favored isocurvature component in those analyses supports the adiabatic description within the models tested. It does not prove that every conceivable relative-composition fluctuation is impossible. “Cosmic seeds” names the initial pattern that later evolves; adiabatic and isocurvature describe the relationships within that pattern, with observations deciding which mixtures remain viable.