How do astronomers measure distances to stars?

Hold a finger in front of you and alternate closing your left and right eye. The finger seems to jump against a distant wall, although you are not moving it. Hold it farther away and the jump becomes smaller. The wall has not changed; your viewpoint has.

Distances to many stars are estimated using parallax: a change in apparent position caused by the observer’s movement. For annual stellar parallax, different positions of Earth or a space observatory around the Sun act as the two eyes. A known baseline and a measured angle provide a geometric distance estimate, without needing the star’s intrinsic luminosity. ESA: parallax, Bailer-Jones: geometric distances and inference.

A known baseline replaces a journey to the star

To measure the distance to a tree across a river, a surveyor can observe it from two places on the near bank. The separation between observing sites is the baseline. Two directions to the tree form a triangle; knowing its baseline and angles lets us find the long sides without crossing the river.

Stars require a much larger baseline. In approximately half a year, Earth moves to the opposite side of its orbit. In a simplified circular model, those positions are separated by 2 astronomical units, about 300 million kilometers. An astronomical unit, or AU, corresponds approximately to the Earth–Sun distance scale. OpenStax: triangulation and Earth’s orbital baseline.

The star’s apparent direction still changes by a tiny amount. At a fixed baseline, a more distant star produces a narrower triangle and a smaller parallax. This, rather than color or faintness alone, provides the distance.

The parallax angle is half the maximum annual swing

This is where a factor-of-two error easily slips in. Annual parallax $p$ corresponds to a baseline of 1 AU, rather than the full 2 AU orbital diameter. In the idealized geometry, the maximum shift between opposite Earth positions is $2p$ . The distance formula uses half that full swing. OpenStax: the angle definition, Figure 19.6.

In the diagram, imagine a stationary star and a circular orbit. The Sun–star line is perpendicular to the chosen orbital diameter. The angle $p$ sits at the star, between the direction to the Sun and the direction to one Earth position. It lies opposite the 1 AU Earth–Sun segment.

Half the angle corresponds to half the baseline

A: Earth: first observation
B: Earth: about half a year later

p is the angle at the star between the directions to the Sun and one Earth position. The full angle between both Earth directions is 2p.

1 AU runs from the Sun to one Earth position; the whole A–B baseline is 2 AU.

At the same baseline: smaller parallax means greater distance

100 mas → 10 pc
50 mas → 20 pc
25 mas → 40 pc

Schematic right triangle, not a scale drawing or measurements of a particular star. Angles are greatly enlarged and stellar distance compressed. Circular orbit, stationary star, chosen diameter perpendicular to the Sun–star direction; maximum seasonal swing shown. Text describes the same geometry without requiring color discrimination.

This is a deliberately simple projection. On the real sky, the annual parallactic track depends on the star’s direction: for a circular orbit its projection can be an ellipse, circle or line segment. Arbitrary observations six months apart need not give the maximum swing. Actual measurement uses the observatory’s known position at each observation time. Luri and colleagues: observation model, Section 3.1.

The parsec turns an angle into a convenient distance

An arcsecond is 1/3,600 of a degree, written ″. A parsec is a distance unit tied to an annual parallax of one arcsecond; one parsec is approximately 3.26 light-years. A light-year is also a distance, rather than an observing duration. For tiny stellar angles, a simple relation follows: distance in parsecs is approximately one divided by $p$ in arcseconds. OpenStax: parallax and the parsec.

For example, an ideal annual parallax of 0.10″ corresponds to 10 parsecs, roughly 32.6 light-years. A parallax of 0.05″ is half as large, so its distance is twice as large: 20 parsecs. These are teaching examples, rather than new measurements of particular stars.

Branch · With formulasOptional · you can skip this

Where one divided by the angle comes from

In our diagram’s right triangle, $d$ is the Sun–star distance and $b=1$ AU is the short side. Thus $\tan p=b/d$ . The trigonometric formula requires the appropriate angular representation; in the approximation $\tan p\approx p$ , the angle must be in radians. Stellar angles are so small that this approximation is extremely accurate.

After converting to parsecs and arcseconds:

d_{\mathrm{pc}}\approx\frac{1}{p_{\mathrm{arcsec}}}.

Here $d_{\mathrm{pc}}$ and $p_{\mathrm{arcsec}}$ mean numerical values in those particular units. If a catalog gives milliarcseconds, or mas, then 1,000 mas = 1″, and the convenient expression becomes $d_{\mathrm{pc}}\approx1000/p_{\mathrm{mas}}$ . Thus 100 mas = 0.10″ gives 10 pc. Confusing mas with arcseconds creates a thousandfold error. Bailer-Jones: definitions and geometry.

The star is moving too

Two photographs do not separate every cause of displacement. A star has proper motion: a changing sky direction due to its actual motion relative to the solar system. The parallactic component follows the observer’s annual orbit; proper motion over a limited interval can often be described as a smooth, nearly linear trend. On the sky we see their sum. ESA: parallax and proper motion.

That is why Gaia repeatedly measured directions to stars, with processing jointly estimating position, parallax and proper motion. Additional effects, including orbital motion within a binary system and systematic instrumental errors, can also affect the result. A geometric method still requires a measurement model. Luri and colleagues: astrometric data and the source model.

A small parallax calls for a careful answer

For the same angular uncertainty, a weaker parallactic signal is less well determined. A sufficiently precise positive parallax can have a useful inverse-distance approximation. With substantial uncertainty, blind division no longer provides a reliable distance: the transformation is nonlinear, while distance must remain positive. Astronomers infer a distribution of possible distances, accounting for uncertainties and justified additional information. Bailer-Jones, 2015: uncertainty and the inference problem.

Branch · Measurement limitsOptional · you can skip this

Why catalogs can contain negative parallaxes

A catalog parallax is a parameter fitted to noisy direction measurements. With a weak signal, the best such fit can be zero or negative. This is not a negative physical distance, nor a reason to discard the star automatically: the measurement and its uncertainty still contain information. Simply taking the reciprocal of the negative number is inappropriate here. Luri and colleagues: Sections 3.1–3.3.

Nonlinearity appears even with positive numbers. An imaginary parallax of 50 mas gives an inverse value of 20 pc. Values of 45 and 55 mas give about 22.2 and 18.2 pc. Equal angular changes of 5 mas produce unequal distance changes. These two substitutions only show the transformation’s shape; they do not, by themselves, define a statistical confidence or Bayesian interval.

Geometry provides a foundation for farther distances

Try this: in the ideal diagram, the maximum annual swing is 0.20″. Why is the distance not 5 pc? Because 0.20″ is $2p$ . Annual parallax is 0.10″, so the distance is approximately 10 pc. Here we assume no measurement uncertainty and proper motion already separated out.

Parallax reaches the distance through geometry, without traveling to a star or guessing its luminosity. Yet we measure a tiny angle and infer the distance from it. A known baseline, the correct definition of $p$ , units and uncertainty are therefore equally essential.

This foundation also helps check other ways of measuring cosmic distances. Before moving on, revisit parsecs and light-years or angular measurements on the sky.