The Sky as a Sphere
Look up on a clear night and the sky does not look like a sphere. It looks like a dome — a ceiling, with points of light stuck to it, all seemingly the same distance away.
That impression is wrong about the universe and exactly right about what you can measure. Astronomy takes it seriously enough to build a model on it.
The model
The celestial sphere is an imaginary sphere of unlimited radius, centred on the observer, onto whose surface every object in the sky is projected.
Every star, planet, the Sun and the Moon are treated as sitting on that surface. Not near it, not at various depths behind it — on it.
The sphere has no radius worth quoting. Calling it "unlimited" is a way of saying the radius never enters any calculation, because the model never uses it.
What the model keeps, and what it throws away
The celestial sphere keeps direction and discards distance. That is the entire trade, and it is worth being blunt about the cost.
Three objects sit at three genuinely different distances. They lie in the same direction, so the sphere records them at a single point, and the model has nothing whatever to say about which is nearer.
This sounds like a serious loss. In practice it is close to free, for two reasons:
- Distance is not what you were asking. "What was overhead at that moment?" is a question about direction. Feeding distance into it would be answering a question nobody asked.
- For the stars, the loss is not even measurable by eye. The stars are so far away that moving right across the Earth's orbit shifts their apparent positions by no more than a second or two of arc, even for the nearest of them. To an observer on the ground, they behave exactly as though they were painted on a sphere.
It is a surface, not an object
The most common misreading is treating the celestial sphere as a thing that exists — a shell somewhere out past the planets.
It is a coordinate surface. It exists in the same sense that lines of longitude exist: as an agreed framework for naming positions, not as material you could ever collide with. The sphere is scaffolding for describing directions, and the scaffolding is centred on whoever is doing the describing.
Which means — and this is why the model belongs in the Locations category — the celestial sphere is yours. Its centre is you. Change the location in the dropdown and you have not adjusted a setting; you have moved the centre of the sphere.
Why the sphere appears to turn
Watch for an hour and the whole sphere appears to rotate, carrying every star with it, as one rigid piece.
It does not. The Earth turns, and the sphere is what that looks like from a platform you cannot feel moving. The model describes the appearance faithfully and says nothing about the cause.
That is deliberate rather than primitive. A chart is a record of a sky as seen from one place at one moment, so a model built around the observer is the one that fits. Knowing the Earth goes round the Sun does not make the observer-centred description wrong, any more than knowing about tectonic plates makes a street map wrong.
Where the approximation does break
Being honest about the limit: the model's "distance doesn't matter" assumption holds beautifully for stars and acceptably for planets. It strains for the Moon.
The Moon is close enough that observers in different places genuinely see it in slightly different directions — by up to about a degree, roughly two of the Moon's own widths. That effect is called parallax, and it is the reason the applications offer a choice of coordinate origin at all. Until that category, the sphere can be taken at face value.
What comes next
A sphere with nothing marked on it is not yet useful. Before anything can be drawn on it, one distinction has to be settled — because the circles that follow are not all the same kind of circle.