Astronomy · Measuring distance and motion

Parallax and a known baseline

Build the geometric distance argument from a known separation and two directions, keeping full shifts, annual parallax conventions and noisy fitted coefficients distinct.

Continue in the approved chapter order. Necessary vocabulary, quantities and assumptions are explained in the lesson.

Back to the chapter

After reading, you can

Relate a baseline and measured directions to a geometric distance.
Distinguish the annual parallax parameter from a full opposite-view displacement.
Use parsec and angular units with appropriate uncertainty and model limits.

A direction can change even when the target stays in place

Parallax is an apparent directional change associated with a change of observing position. The target need not move for the two sightlines to differ. A known baseline between observing positions supplies the length scale needed to turn angular information into distance.

Directions alone specify a location on the sky rather than a unique physical distance. For triangulation, the observing positions, their separation and the coordinate convention must be known. The same angular shift corresponds to different distances if a different baseline is assumed.

Astronomical measurements also compare directions with a reference frame. Reference sources need not be literally motionless or at infinite distance. Their own motions and parallaxes, instrument calibration and the observer’s trajectory belong in the actual astrometric model.

The first example below is a deliberately enlarged stationary-target triangle in arbitrary units. It makes the geometry visible before the much smaller angular quantities used in stellar astrometry are introduced.

Separate the half-baseline and the full directional change

Choose observers A = (−1, 0) and B = (+1, 0), and target P = (0, 8), using length unit L₀ and one common coordinate scale. The full baseline AB is 2 L₀. Its midpoint lies 8 L₀ perpendicularly from P.

The symmetry divides the target angle between P→A and P→B into two equal parts. Call either half p_geo and the half-baseline b = 1 L₀. The right triangle gives tan p_geo = b/D, where D = 8 L₀ is the perpendicular midpoint-to-target distance.

Thus p_geo = arctan(1/8) ≈ 7.125016 degrees, and the full opposite-view angle is 2p_geo ≈ 14.250033 degrees. Inverting the exact relation gives D = b/tan p_geo = 8 L₀. Keeping this half-angle equation but substituting the full baseline gives the incorrect value 16 L₀. Substituting the full angle while retaining b = 1 L₀ instead gives b/tan(2p_geo) = 63/16 L₀ = 3.9375 L₀. The latter approaches half the correct distance only in the small-angle limit; the enlarged triangle makes the finite-angle difference visible.

For a sufficiently small angle in radians, tan p_geo is approximately p_geo. Here the approximate distance b/p_geo is about 8.04149 L₀, slightly above the exact value. The intentionally large diagram angle makes that approximation visible; this is not a literal stellar distance or an annual orbit drawn to scale.

A known baseline and the matching half-angle

Scroll sideways to see the full diagram. Its explanation is below.

Original enlarged stationary-target geometry in arbitrary length units L₀, not a literal stellar distance, annual orbit or observation. Observers A = (−1, 0) and B = (+1, 0) have a full baseline of 2 L₀. Its midpoint M = (0, 0) is a reference point, rather than a third observer. Target P = (0, 8) is 8 L₀ perpendicularly above M. Horizontal and vertical lengths use the same scale. Both sightlines meet P, and every angle arc has its vertex at P. Each half-angle p_geo between a sightline and P→M is arctan(1/8), about 7.125016 degrees. The full angle between P→A and P→B is 2p_geo, about 14.250033 degrees. Half-baseline b = 1 L₀ belongs with tan p_geo = b/D, giving D = b/tan p_geo = 8 L₀. Replacing tan p_geo by p_geo in radians gives about 8.041494 L₀ for this deliberately large angle. Marker sizes do not encode physical body sizes. Separate dimension bars label b, full baseline B and perpendicular D. No background star is drawn. The companion conventional parallax table has p = 100, 50, 20 and 10 mas, equivalent to 0.100, 0.050, 0.020 and 0.010 arcseconds, with distances 10, 20, 50 and 100 pc. That inverse-distance convention is distinct from the finite tangent triangle.

The stellar convention uses a one-au parallax scale

An observer following Earth’s orbital motion samples changing positions through the year. The standard annual parallax scale uses a baseline of one astronomical unit, rather than the approximate two-au separation of opposite orbital positions in a simplified circular diagram. The parallax parameter is a semiamplitude scale, not the full opposite-view displacement.

An arcsecond is 1/3600 of a degree; a milliarcsecond, abbreviated mas, is one thousandth of an arcsecond. Always convert the unit before using a reciprocal numerical relation. A measured value of 50 mas is 0.050 arcseconds, not 50 arcseconds.

The astronomical unit, symbol au, is defined as exactly 149,597,870,700 metres. The parsec, symbol pc, is exactly (648000/π) au. Under the conventional inverse-distance parallax parameterization, distance in parsecs equals one divided by parallax in arcseconds.

This unit convention should be distinguished from the exact finite right-triangle tangent formula. At stellar parallax angles the small-angle geometry agrees at the precision of the examples here. An exact unit definition does not make an actual observer trajectory perfectly circular or eliminate uncertainty in positions and measured angles.

Convert the angle before taking the reciprocal

For original positive precise parallax inputs, p = 100, 50, 20 and 10 mas correspond to 0.100, 0.050, 0.020 and 0.010 arcseconds. The conventional reciprocal distances are 10, 20, 50 and 100 pc.

Smaller parallax at the same defined baseline corresponds to a larger inferred distance. This does not mean the more distant source must look fainter in every band: intrinsic luminosity and intervening material affect brightness separately.

A real annual fit combines many epochs and the observer’s actual position, with source direction, proper motion and relevant corrections. The parallax pattern depends on direction on the sky. One image pair at arbitrary dates is not automatically a measurement of twice the annual parallax parameter.

The reciprocal table illustrates the distance conversion after suitable measurement and modelling. It is not a catalogue selection rule, a mission-performance statement or a guarantee that every fitted value can be inverted accurately.

Original conventional annual-parallax conversion examples; input angles are stipulated positive values, not observations.
p / mas	p / arcsec	d / pc
100	0.100	10
50	0.050	20
20	0.020	50
10	0.010	100

A fitted coefficient and a physical distance have different limits

For a sufficiently precise positive corrected parallax with a suitable error model, first-order propagation gives relative distance standard uncertainty approximately equal to relative parallax standard uncertainty. A stipulated p = 50 mas with standard uncertainty 1 mas gives d = 20 pc and an approximate distance standard uncertainty 0.4 pc.

The reciprocal is nonlinear. Inverting the illustrative input values 51 and 49 mas gives about 19.6078 and 20.4082 pc, unequal offsets from 20 pc. These transformed endpoints demonstrate the nonlinearity; without a probability or coverage model they are not a confidence or credible interval.

With large relative uncertainty, the asymmetry can be substantial and a simple symmetric distance error can mislead. A noisy fitted parallax coefficient can also be negative even though physical distance is positive. Taking its absolute value, clipping it to zero or automatically discarding it is not a general inference method.

Distance inference should use the measurement model, calibration and selection information appropriate to the actual data. No universal signal-to-noise cutoff or prior is prescribed here. The next lesson separates the repeating annual contribution from a source’s ongoing angular drift.

Worked examples

Recover the declared triangle distance

Which inputs recover D in the symmetric A–B–P example?

Use half-baseline b = 1 L₀ and half-angle p_geo = arctan(1/8).
The exact right-triangle relation is D = b/tan p_geo.
This gives D = 8 L₀; the full sightline angle is twice p_geo.

Use the half-baseline and its corresponding half-angle in this equation. Full quantities require the matching geometry.

Keep the reciprocal tied to its unit

What conventional distance follows from stipulated positive p = 20 mas?

20 mas = 0.020 arcseconds.
d/pc = 1/0.020.
The result is 50 pc under the stated annual-parallax convention.

50 pc. Directly taking 1/20 as parsecs would use the wrong angular unit.

Try the reasoning

Is annual parallax generally the full opposite-view shift in the simplified circular diagram?
Show the answer and reasoning
No.
The standard parameter uses the one-au scale; the full opposite-view change in the symmetric diagram is twice its semiamplitude.
Does the exact au definition make the measured observatory baseline uncertainty zero?
Show the answer and reasoning
No.
Unit definition and knowledge of actual observing positions are different matters.
Does a negative noisy fitted parallax establish a negative physical distance?
Show the answer and reasoning
No.
The coefficient is estimated from noisy data; distance needs a suitable positive-distance inference model.
Is replacing tan p by p exact for every angle?
Show the answer and reasoning
No.
It is a small-angle approximation with p expressed in radians; the enlarged triangle shows its finite difference.

What to carry forward

A known baseline and directions supply a geometric distance scale.
Full shifts, semiamplitudes and angular units must match the equation used.
Noisy parallax inversion needs a measurement and uncertainty model rather than an unrestricted reciprocal rule.

References

Teaching resources and research supporting this lesson’s explanations and examples.