Does the Universe have a centre or an edge?

Maps of the observable Universe often place Earth at the centre, with the most distant accessible regions surrounding it. Have we found a special location from which everything flew apart?

In the standard homogeneous cosmological model, there is no preferred centre of expansion or material edge of space. The observable region, however, has an observer-dependent centre and a horizon limiting access to signals. That horizon is not a wall around the whole Universe. We must distinguish properties of the model of space from limits on our view. PDG: cosmological principle and horizons, §§22.1, 22.2.4.

Why does recession not make our location special?

Imagine a homogeneous grid in which all large-scale distances between comoving points have increased by a factor of 1.5. Seen from point A, neighbouring points have moved farther away. Choose point B instead, and the corresponding distances from it have also increased by 1.5. The rule does not depend on the chosen address.

Homogeneous expansion works in the same way: another observer following the overall flow obtains the same large-scale description. An ordinary explosion has a centre and fragments flying into an external space. Here no special launch site is needed. Ned Wright: comparing observers A and B.

This does not assert that every cosmic region is literally identical. There are stars, voids and clusters; neighbouring galaxies may approach each other. Homogeneity is a large-scale approximation in a model. Applying it to all space beyond our observations extrapolates that model, rather than directly surveying every possible place.

Why are we at the centre of the visible region, then?

Light takes time to reach a receiver. In standard cosmic history, signals from some comoving positions have not had time to reach us. The idealized limit of this access is called the particle horizon. It depends on the observer, the cosmic time and the expansion history.

Compare two observers in the same model at the same cosmic time. The diagram shows two centres of observable regions, not two origins of the Universe.

Another observer, another centre of the observable region

A: solid circleB: dashed circle

S is inside circle A but outside circle B.
T is inside circle B but outside circle A.
U belongs to the overlap of the two regions.

Schematic two-dimensional slice of a homogeneous flat model at one cosmic time. Observers A and B follow the overall expansion; equal illustrative circles mark their particle horizons. Points are comoving positions, not events photographed simultaneously. No radius in kilometres or light trajectories is shown. Circle lines mark limits of signal access, not walls; the picture frame is not an edge of the Universe either.

Must point S outside B’s circle be “nowhere”? No: in this diagram it lies within A’s region. Passing beyond one observer’s view does not erase a position in space. The circles are not material shells that could collide or burst.

By “observable Universe”, we mean our accessible region. Practical visibility also depends on the medium’s opacity and on instruments. Ancient thermal light does not provide an ordinary photograph through the early opaque plasma. An ideal signal-propagation horizon and the most distant actual photograph are different concepts. Davis and Lineweaver: particle horizons and observations, §§2–3.4.

Branch · The geometry of light travelOptional · you can skip this

Why is the radius not simply light speed multiplied by age?

The current distance to a distant comoving position and the travel time of its ancient light are different quantities. Space expanded while the light travelled. The scale factor $a(t)$ accounts for this.

For a flat model normalized to $a(t_0)=1$ , the current proper radius of the particle horizon is

$R_{\rm p}(t_0)=c\int_0^{t_0}\frac{dt}{a(t)}.$

Here $c$ is the local speed of light in vacuum, $t_0$ the final cosmic time and $a$ dimensionless. The integral has units of time, so the result is a length. The lower limit 0 is a formal boundary of the chosen classical history, not proof of a physical singularity.

For an instructional example, take a flat model containing only nonrelativistic matter, with $a(t)=(t/t_0)^{2/3}$ . The integral gives $R_{\rm p}=3ct_0$ , rather than $ct_0$ . That is a result for this simplified model, not an estimate of the real Universe’s radius. The formula demonstrates why expansion history matters. The photon’s local speed remains $c$ . PDG: radial light travel and horizons, equation 22.32, Davis and Lineweaver: definitions, Appendix A.

The limit on past signals is not every other cosmic boundary

A different question is whether a signal sent now from a distant comoving position will ever reach us. Models with limited future access have a cosmological event horizon. Its existence and size depend on future expansion. Seeing a galaxy’s ancient light today does not guarantee that we will observe every future event there.

There is also the Hubble sphere, with radius $c/H$ : on its surface the rate of increase in proper distance equals $c$ . Here $H$ is the expansion parameter at the chosen time. This sphere is generally not a visibility horizon. That relation alone cannot make everything beyond it unobservable. None of these boundaries is an edge of matter stretched around us. Davis and Lineweaver: distinguishing the three boundaries, §2.

Must a space without an edge be infinite?

A sphere’s surface has finite area but no edge on the surface: you can keep walking without crossing a boundary. The geometric centre of the ball is not a point on that two-dimensional surface.

This is only an analogy showing a logical possibility: “finite” need not mean “enclosed by a wall”. Our space has three dimensions. We have not established that it is the surface of an ordinary ball or that it must be embedded in an external physical space. Infinite space is also possible in cosmological models. Physics FAQ: limits of the sphere analogy.

Placing us at the centre of a map of accessible signals is natural. Inferring a centre of all space from it is a mistake. Global size and geometry remain separate questions; spatial curvature helps explain which properties can be determined from within.