Look up from a dark place and a faint band runs right around the sky. Turn a telescope on it and the band proves to be a countless crowd of stars.
Why a band? If stars filled the sky evenly, every direction would show the same number. That it forms a band means the arrangement of stars has a direction in which it is thin.
Which is to say we sit inside a flattened crowd of stars and are looking at it from within. Look along the disc and the line of sight runs through stars for a long way, so a great many overlap into a faint glow.
Look at right angles to the disc and the line of sight leaves it almost at once, so few stars lie that way. The band is the plainest evidence there is that we are inside a disc.
Build up the shape seen edge on. First there is a thin disc where the stars and the gas are gathered. The spiral arms ride here as well.
Around the centre it swells thicker than the disc. This is the bulge, packed with old stars.
Outside that is a region spread out in a sphere, the halo. Scattered through it are globular clusters, round gatherings of hundreds of thousands of stars each.
The stars of the halo are older than those of the disc, and they do not turn together in one direction as the disc does. It is the trace of an order: the galaxy gathered as a sphere first, and the flat disc formed afterward.
The look of the band alone does not say where in the disc we sit, because it could equally be the middle. For a long time that is what was believed.
The globular clusters gave the clue. Being scattered through the halo, they can be seen far off without the gas of the disc getting in the way. And their distribution is not spread evenly over the sky: it is heaped toward Sagittarius.
If the globular clusters surround the whole galaxy, then the centre of their distribution is the centre of the galaxy. That they appear heaped to one side means we sit away from that centre.
This is how the position of the Sun was settled: partway out in the disc, some kiloparsecs from the centre. The assumption that we were at the centre came undone by measuring a distribution.
The disc turns. Not, however, the way a record does, with the whole of it sweeping through the same angle.
The nearer a star is to the centre, the less time one lap takes; the farther out, the slower. This is called differential rotation. Turn a row of stars drawn straight out from the centre in the figure, and the inner ones run ahead while the row winds itself up before your eyes.
And now there is trouble. If a spiral arm were a row made of particular stars, a few turns would wind it up and it would be gone. Yet every spiral galaxy holds clear arms.
So an arm is not a thing made of fixed stars. It is a crowded region that turns, with stars passing through it. It is a traffic jam, travelling at a speed of its own that is not the speed of the cars. How the turning itself runs is measured in the article on rotation curves.