0
What "Observable" Means
A sphere of finite information inside something far larger — possibly infinitely larger
The observable universe is not the Universe. It is the region from which light has had time to reach us in 13.8 billion years — a sphere ~93 billion light-years across, centered on us only in the trivial sense that every observer sits at the center of their own. Beyond it, space almost certainly continues: identical physics, more galaxies, possibly without end. The boundary is not a wall but a horizon — and cosmology has several distinct ones that popular accounts routinely conflate:
Particle horizon — the boundary of everything we can have ever seen (46.5 Gly comoving). Hubble radius — where recession reaches c (14.4 Gly); not a limit on seeing. Event horizon — the boundary of everything we can ever see happen (16.5 Gly today). Visibility limit — the asymptotic particle horizon (~62 Gly): the final extent of all possible astronomy.
Two themes run through this sheet. First, the horizons are calculable to the percent level — they are integrals of the measured expansion history, not philosophy. Second, the unobservable is not beyond science: curvature, topology, and inflation each let observations inside the horizon place rigorous bounds on what lies outside it. What we cannot do — even in principle — is ever check those regions directly. The discipline is knowing exactly where that line sits.
As on the companion cosmology, Big Bang, gravity, and multiverse sheets, every equation is paired with a plain-language reading of what it physically asserts, a Use in Research column with key references, and each section ends with the open unknowns. Toggle the Dark theme at top-right.
I
The Horizon System
6 equationsFour distances, four different questions. All are integrals over the expansion history a(t) — measure the cosmological parameters and every horizon follows. Davis & Lineweaver's classic paper untangled the confusions; these are its equations.
| Name | Equation | Variables | Use in Research |
|---|---|---|---|
| Comoving Distance | \[ D_C = c\int_0^z \frac{dz'}{H(z')} \]
The ruler distance to an object measured in coordinates that expand with the Universe — the distance "now," with expansion factored out. The master quantity from which every other cosmological distance is built. |
H(z) = expansion rate history; z = redshift |
What you actually compute from any survey redshift before making a 3D map; BAO measurements are calibrations of precisely this integral.
Key referencesHogg (1999); Davis & Lineweaver (2004); Planck Collaboration (2020).
|
| Particle Horizon | \[ D_P(t) = a(t)\,c\!\int_0^t \frac{dt'}{a(t')} = 46.5\;\text{Gly today} \]
The proper distance to the farthest matter whose light could have reached us since t = 0 — the true edge of the observable universe. It is 46.5, not 13.8, Gly because the space the light crossed kept stretching behind it. |
a(t) = scale factor; integral runs from the Big Bang |
Defines the causal patch: the total mass, galaxy count, and information content of "the universe" in any quantitative statement all mean "within D_P."
Key referencesRindler (1956); Davis & Lineweaver (2004); Conselice et al. (2016).
|
| Hubble Radius & Hubble's Law | \[ v_{\rm rec} = H_0\,d;\qquad R_H = \frac{c}{H_0} = 14.4\;\text{Gly} \]
Recession speed grows linearly with distance, passing c at the Hubble radius. Nothing breaks: this is space stretching, not motion through space — special relativity's speed limit applies to the latter only. |
R_H = Hubble radius; v_rec = recession velocity |
The scale separating sub- and super-horizon physics in perturbation theory, and the source of the most persistent misconception in cosmology — that the Hubble sphere bounds what we can see. It does not.
Key referencesHubble (1929); Harrison (1991); Davis & Lineweaver (2004).
|
| Cosmic Event Horizon | \[ D_E(t) = a(t)\,c\!\int_t^\infty \frac{dt'}{a(t')} \approx 16.5\;\text{Gly today} \]
The mirror of the particle horizon: the farthest distance from which a signal sent today will ever reach us, given accelerating expansion. Events beyond it are happening — and are permanently outside our future. |
integral converges only because dark energy accelerates a(t) |
The quantity dark energy created: in a decelerating universe the integral diverges and everything is eventually seen. Λ makes patience insufficient — the event horizon is the price of acceleration.
Key referencesRindler (1956); Loeb (2002); Davis & Lineweaver (2004).
|
| Conformal Time | \[ \eta = \int_0^t \frac{c\,dt'}{a(t')} \]
A time coordinate in which light always travels at 45° on a spacetime diagram, no matter how space expands. In conformal coordinates the entire causal structure of the Universe — every horizon on this sheet — becomes straight lines. |
η = conformal time; total elapsed: ~46.5 Gly equivalent |
The working time variable of CMB and inflation theory: sound horizons, horizon crossing of perturbations, and the famous conformal diagrams of our past light cone are all drawn in η.
Key referencesPenrose (1963, diagrams); Mukhanov (2005); Planck Collaboration (2020).
|
| Distance Duality | \[ D_L = (1+z)\,D_C = (1+z)^2 D_A \]
One object, three distances: luminosity distance (from its brightness), comoving distance (the map), angular-diameter distance (from its size). The (1+z)² relation between them is a theorem in any metric theory — and is itself testable. |
D_L = luminosity dist.; D_A = angular-diameter dist. |
The bookkeeping that makes surveys consistent — and the origin of the strange fact that beyond z ≈ 1.6, more distant galaxies look bigger: D_A turns over because the Universe was smaller when their light left.
Key referencesEtherington (1933); Bassett & Kunz (2004); Hogg (1999).
|
Open unknowns · The Horizon System
H₀ Sets Every Horizon
Which Hubble constant should the horizons be computed with?
The 5σ disagreement between local (73) and CMB-inferred (67.4) H₀ propagates into every distance on this sheet at the ~8% level. Until the tension resolves, the size of the observable universe itself carries a systematic uncertainty rarely quoted.
Superhorizon Imprints
Do the CMB's large-angle anomalies point to structure beyond the horizon?
The low quadrupole, hemispherical power asymmetry, and axis alignments persist from WMAP through Planck at the ~3σ level. Pre-inflationary relics or superhorizon perturbations could cause them — or they are flukes of cosmic variance. No consensus exists.
Does the Event Horizon Exist?
If dark energy is dynamical, is anything permanently unreachable?
The event horizon exists only if acceleration continues forever. DESI's hints of evolving dark energy reopen the question: a decaying w(z) could dissolve the horizon entirely, making the "permanently unobservable" merely "not yet observed."
II
Seeing to the Edge
4 equationsThe particle horizon is the fundamental limit; the practical limits sit closer in, and differ by messenger. Photons hit a wall at recombination — but neutrinos and gravitational waves pass straight through it, carrying the universe's earliest information in principle.
| Name | Equation | Variables | Use in Research |
|---|---|---|---|
| The Wall of Last Scattering | \[ z_{\rm rec} \approx 1090,\quad t_{\rm rec} \approx 380{,}000\;\text{yr} \]
Before recombination the Universe was an opaque plasma — photons scattered off free electrons every few thousand years of travel. The CMB is the glowing inner surface of that fog: the absolute limit of electromagnetic astronomy, 0.3 Gly inside the particle horizon. |
z_rec = recombination redshift; visibility function width Δz ≈ 80 |
Everything in observational cosmology beyond z ≈ 1100 must be inferred, not imaged. The CMB sphere is simultaneously our deepest photograph and a curtain over the first 380,000 years.
Key referencesPeebles (1968); Zel'dovich et al. (1969); Planck Collaboration (2020).
|
| Neutrino & GW Horizons | \[ z_\nu \sim 6\times10^9 \;(t\sim1\,\text{s});\qquad z_{\rm GW} \to \text{inflation} \]
Different messengers decouple at different epochs, so each has its own last-scattering surface. Neutrinos stream free from t ≈ 1 second; gravitational waves were never in equilibrium at all — a primordial GW background would image the universe at 10⁻³⁵ s. |
z_ν = neutrino decoupling; CνB temperature today = 1.95 K |
The roadmap for seeing past the CMB wall: the cosmic neutrino background (CνB) and inflationary B-modes are the only known signals from the first second — both are active experimental targets.
Key referencesFollin et al. (2015); PTOLEMY Collaboration (2019); BICEP/Keck (2021).
|
| Lookback Time vs. Distance | \[ t_L = \int_0^z \frac{dz'}{(1+z')H(z')} \;\neq\; \frac{D_C}{c} \]
"Looking back 13.4 billion years" and "33 billion light-years away" describe the same galaxy without contradiction: the first is the photon's travel time, the second the stretched distance to where the source is now. Expansion divorces the two permanently. |
t_L = lookback time; compare D_C from row I.1 |
The conversion every high-z discovery announcement juggles — and the reason "the most distant galaxy" has three defensible answers (lookback, comoving, light-travel) that differ by a factor of ~2.5.
Key referencesHogg (1999); Carniani et al. (2024, JADES).
|
| The Inventory Within | \[ N_{\rm gal} \sim 2\times10^{12};\quad N_{\rm baryon} \sim 10^{80};\quad N_\gamma \approx 411\,\text{cm}^{-3} \]
The observable universe is finite and countable: roughly two trillion galaxies, 10⁸⁰ protons, and a billion CMB photons per baryon. "The universe" in any quantitative statement means this census — the contents of one causal patch. |
counts within the particle horizon; photon density from T = 2.725 K |
The denominator of cosmology: baryon-to-photon ratio η ≈ 6×10⁻¹⁰ anchors nucleosynthesis; the total galaxy count calibrates extragalactic background light and merger-history models.
Key referencesConselice et al. (2016); Fukugita & Peebles (2004); Planck Collaboration (2020).
|
Open unknowns · Seeing to the Edge
The Dark Ages Gap
Can the 21-cm signal open the unseen epoch between the CMB and the first stars?
Between z ≈ 1100 and z ≈ 30 nothing shines — yet neutral hydrogen's 21-cm line redshifted to tens of MHz could map it in 3D, with more modes than every galaxy survey combined. EDGES' contested 2018 detection and upcoming lunar farside arrays make this the decade's frontier.
Touching the First Second
Will the cosmic neutrino background ever be detected directly?
1.95 K neutrinos carry ~10⁻⁴ eV of kinetic energy — below every existing detection threshold. PTOLEMY's tritium-capture concept is the only proposal on the table, and it requires control of systematics nobody has demonstrated.
Primordial Gravitational Waves
Is there an inflationary GW background waiting at r just below 0.03?
Detection would image the universe at 10⁻³⁵ s and fix the inflation energy scale. Non-detection at r ~ 0.001 (CMB-S4, LiteBIRD reach) would kill the classic large-field models. Either outcome reshapes early-universe physics.
III
Horizons in Motion — The Edge Is Not Static
4 equationsDark energy makes the horizon system dynamic in a cruel direction: the visible universe grows while the reachable universe shrinks. Cosmology has an expiration date — and these equations compute it.
| Name | Equation | Variables | Use in Research |
|---|---|---|---|
| The Future Visibility Limit | \[ D_{P}(t\to\infty) = c\int_0^\infty \frac{dt}{a(t)} \approx 62\;\text{Gly comoving} \]
Wait forever, and the particle horizon converges to a finite comoving radius — the total extent of everything any observer here will ever see. All possible astronomy, for all time, is bounded by this sphere. |
converges because Λ-dominated a(t) grows exponentially |
The ultimate completeness statement: today's 46.5 Gly will grow to 62 Gly and stop. The fraction of all-time-visible galaxies already visible is (46.5/62)³ ≈ 42% — most of what can ever be discovered is still on its way.
Key referencesDavis & Lineweaver (2004); Loeb (2002); Krauss & Starkman (2000).
|
| Galaxies Crossing Over | \[ z_{\rm EH} \approx 1.8:\;\; \text{events at}\; z {\gt} z_{\rm EH}\;\text{now are forever unseen} \]
The event horizon currently sits at redshift ~1.8. Galaxies beyond it remain visible — their old light keeps arriving — but everything happening in them from now on is causally lost to us. They are exiting our future while remaining in our past. |
z_EH = redshift of today's event horizon (≈16.5 Gly) |
A genuinely strange consequence worth internalizing: roughly 97% of the galaxies in the observable universe are already beyond two-way contact. A probe launched today at any speed could reach only the nearest ~3% by volume.
Key referencesLoeb (2002); Busha et al. (2003); Davis & Lineweaver (2004).
|
| The End of Cosmology | \[ t \sim 10^{11}\,\text{yr}:\;\; \lambda_{\rm CMB} {\gt} R_H,\;\; z_{\rm gal}\to\infty \]
In ~100 billion years, every galaxy outside the (by then merged) Local Group redshifts beyond detectability and the CMB stretches longer than the horizon itself. Observers then will see one island galaxy in seemingly static, empty space — and will have no observational route to discovering the Big Bang. |
R_H = future Hubble radius (~17 Gly, constant under Λ) |
A humbling calibration of cosmic epistemology: the evidence for expansion is itself transient. We live in the window when the Universe's history is legible — a window that closes.
Key referencesKrauss & Scherrer (2007); Adams & Laughlin (1997); Loeb (2011).
|
| Redshift Drift | \[ \dot z = (1+z)H_0 - H(z) \;\sim\; \text{few cm/s per decade} \]
Watch one quasar long enough and its redshift changes as the expansion accelerates in real time — the only direct, model-free measurement of cosmic acceleration, requiring no standard candles or rulers at all. |
ż = redshift change per unit observer time (Sandage–Loeb signal) |
The horizon dynamics of this section, made into an observable: positive drift at low z is dark energy operating live. A detection would be the first cosmological measurement performed in the time domain rather than down the light cone.
Key referencesSandage (1962); Loeb (1998); Liske et al. (2008, CODEX/ANDES).
|
Open unknowns · Horizons in Motion
Is the Future Horizon Secure?
Does evolving dark energy rewrite the far-future forecast?
Every number in this section assumes Λ is constant. DESI's w₀wₐ hints, if confirmed, change the visibility limit, the end-of-cosmology timeline, and possibly eliminate the event horizon — the entire eschatology of this sheet is one measurement from revision.
Drift Detection Feasibility
Can cm/s spectroscopic stability actually be held for 20 years?
Redshift drift demands wavelength calibration (laser frequency combs), instrument continuity, and source stability across decades — systematics budgets at levels never sustained. Whether the first detection comes from ELT optical or SKA radio is open.
Civilizational Window
How typical is our observational epoch?
We can see the CMB, the acceleration, and the galaxy population simultaneously — observers 10× earlier or later could not. Whether this "coincidence" demands anthropic explanation or is unremarkable selection is a live argument in cosmological foundations.
IV
Bounding the Unobservable
5 equationsWhat lies beyond the horizon cannot be seen — but it can be constrained. Curvature, topology, and inflation each convert measurements made inside our patch into rigorous statements about the whole. The bounds run one direction only: the Universe keeps proving itself bigger than required, never smaller.
| Name | Equation | Variables | Use in Research |
|---|---|---|---|
| Radius of Curvature | \[ R_{\rm curv} = \frac{c/H_0}{\sqrt{|\Omega_k|}};\qquad \Omega_k = 0.0007\pm0.0019 \]
If space is curved, geometry inside the horizon reveals the curvature radius of the whole. Planck+BAO find flatness to a tenth of a percent — so if the Universe does curve back on itself, it does so on a scale at minimum hundreds of times the part we see. |
Ω_k = curvature density (Planck 2018 + BAO); R_curv = curvature radius |
The strongest direct lower bound on the size of everything: |Ω_k| < 0.004 (2σ) gives R_curv ≳ 230 Gly — a closed universe must contain ≳ 10²–10³ observable-universe volumes. If exactly flat or open: infinite.
Key referencesPlanck Collaboration (2020); Vardanyan et al. (2009); Knox (2006).
|
| Topology & Matched Circles | \[ \text{no matched circle pairs} \;\Rightarrow\; L_{\rm cell} \gtrsim 0.97\,d_{\rm LSS} \approx 27\;\text{Gpc} \]
A universe with compact topology — finite but unbounded, like a 3-torus — would show the same CMB circles in different sky directions: light wrapping around. Searches find none, so any repeating cell is larger than essentially the whole observable universe. |
L_cell = fundamental-domain size; d_LSS = last-scattering diameter |
The only direct probe of global shape. A small finite universe was a serious live option (the "soccer-ball universe" once claimed to fit the low quadrupole); circle searches and Planck likelihood analyses closed it.
Key referencesCornish, Spergel & Starkman (1998); Luminet et al. (2003); Planck Collaboration (2015, XVIII).
|
| Inflation's Minimum Stretch | \[ N = \ln\frac{a_{\rm end}}{a_{\rm start}} \gtrsim 60 \;\Rightarrow\; \text{whole} \geq e^{3(N-60)}\times\text{observable} \]
Solving the horizon and flatness problems requires at least ~60 e-folds of inflation — and 60 is only the minimum that makes our patch work. Every e-fold beyond it multiplies the true Universe's volume by e³ ≈ 20. Generic models overshoot by enormous factors. |
N = number of e-folds; excess e-folds inflate the unobservable |
The theoretical engine of the "vast unobservable": in most inflationary potentials N is far above 60 (often absurdly so), and in eternal-inflation regimes the volume is unbounded. The observable patch is then a Planck-scale speck inflated to visibility.
Key referencesGuth (1981); Liddle & Leach (2003); East et al. (2016).
|
| Cosmic Variance | \[ \frac{\Delta C_\ell}{C_\ell} = \sqrt{\frac{2}{2\ell+1}} \]
We observe one sky — one random draw from the ensemble of possible universes with the same physics. On the largest scales there are only a few independent samples (5 for the quadrupole), so a ~63% uncertainty is irreducible by any instrument, forever. |
C_ℓ = angular power spectrum; ℓ = multipole; 2ℓ+1 samples per ℓ |
The fundamental epistemic limit of living in one causal patch: the largest-scale properties of even the observable universe can never be measured better than this. Planck already hit the cosmic-variance floor for ℓ ≲ 1500.
Key referencesAbbott & Wise (1984); Knox (1995); Planck Collaboration (2020).
|
| The Information Bound | \[ S_{\rm dS} = \frac{k_B\,A_{\rm EH}}{4\,\ell_P^2} \approx 2.9\times10^{122}\,k_B \]
The de Sitter horizon created by dark energy carries an entropy — the holographic maximum information content of our causal patch: ~10¹²² bits. Whatever the unobservable Universe holds, what we can ever access, compute, or record is finite and this is the ceiling. |
A_EH = event-horizon area; ℓ_P = Planck length |
The deepest sense in which "observable" is bounded: not just in distance but in total information. This number anchors holographic cosmology, the cosmological-constant coincidence (Λ ~ 1/S), and bounds on future computation.
Key referencesGibbons & Hawking (1977); Bousso (2002, review); Egan & Lineweaver (2010).
|
Open unknowns · Bounding the Unobservable
Finite or Infinite?
Is the Universe actually infinite — and can that ever be established?
Flatness to 0.1% is consistent with exactly flat (infinite), barely closed (finite), or barely open (infinite). With the Ω_k ~ 10⁻⁵ fluctuation floor, observation may be permanently unable to decide — possibly the largest fact about reality that is unknowable in principle.
Copies and Recurrences
If space is infinite with random initial conditions, does everything recur?
An infinite ergodic universe contains every finite configuration infinitely often — exact copies of this room ~10^10^115 m away, by Tegmark's estimate. Whether that arithmetic is physically meaningful or an abuse of probability on infinite spaces is genuinely contested.
Bubble Collision Scars
Could another universe have bruised ours?
If eternal inflation spawns bubble universes, an early collision would leave a disk-shaped CMB cold/hot spot with distinctive polarization. Dedicated searches in WMAP/Planck found candidate disks but nothing significant — the one direct observational handle on "outside," still null.
Is It Still Science?
What epistemic status do claims about the unobservable hold?
Inflation + quantum mechanics predicts regions beyond the horizon; the prediction is untestable region-by-region but follows from testable theory. Where physics ends and metaphysics begins here — Ellis vs. Carroll, falsifiability vs. abduction — is an active methodological fight, not a settled one.
Observable-universe reference values: particle horizon 46.5 Gly (diameter 93 Gly ≈ 28.5 Gpc); Hubble radius c/H₀ ≈ 14.4 Gly (z ≈ 1.5); event horizon ≈ 16.5 Gly (z ≈ 1.8); future visibility limit ≈ 62 Gly comoving (2.4× today's volume); CMB at z = 1090, D_C ≈ 45.5 Gly, D_A ≈ 42 Mly; lookback 13.8 Gyr; contents: ~2×10¹² galaxies, ~10⁸⁰ baryons, 411 photons/cm³, ~10⁵⁴ kg; Ω_k = 0.0007±0.0019 → R_curv ≳ 230 Gly; topology cell ≳ 27 Gpc; inflation N ≳ 60 e-folds; cosmic variance √(2/(2ℓ+1)) (63% at quadrupole); information ceiling S_dS ≈ 2.9×10¹²² k_B vs. ~10¹⁰⁴ k_B used; end of cosmology t ~ 10¹¹ yr; redshift drift ~6 cm/s per decade at z = 4.