Stellar Proper Motion: Components, Tangential Velocity and Worked Calculations

How do you convert proper motion to a star's transverse speed?

A large angular drift does not by itself imply a large physical speed. A nearby object can move through a larger angle than a distant object with the same transverse speed. The chart and invented examples below separate those quantities.

Proper motion is an angular rate across the sky. Radial velocity describes motion toward or away from the observer. Their combination requires a consistent reference frame and a distance estimate; one component alone does not determine the full motion.

Separate proper motion from the sky's daily sweep

The ESA explanation of proper motion describes the long-term angular change in a star's position. It is different from the apparent daily rotation of the sky caused by Earth's spin and from the annual positional pattern caused by parallax.

In a short teaching model, a measured coordinate can contain a starting position, a steady proper-motion term and a periodic parallax term. Real astrometry also accounts for the observing geometry and other effects. Measuring two image positions without dealing with those contributions does not automatically isolate proper motion.

Record the catalogue's epoch, coordinate frame and units. “200” with no label could mean 200 mas/year or 200 arcsec/year, which differ by a factor of a thousand. A coordinate date and an observation date are also not interchangeable labels.

Derive the factor 4.74047 from units

A parsec is defined by the angular conversion that makes one arcsecond at one parsec correspond to one astronomical unit in the differential, small-angle relation. Thus 1 arcsec/year at 1 pc is a transverse rate of 1 astronomical unit per year.

NASA JPL's parameter table gives one astronomical unit as 149,597,870,700 metres, a day as 86,400 seconds and a Julian year as 365.25 days. Therefore one Julian year is 31,557,600 seconds, and:

149,597,870.7 km ÷ 31,557,600 s = 4.7404704635… km/s.

This article rounds that conversion to 4.74047. It uses a Julian year, not a particular calendar year's day count. Keep μ in arcsec/year and d in pc for this form of the equation. With μ in mas/year, vt = 0.00474047 μd; with μ in mas/year and d in kiloparsecs, vt = 4.74047 μd.

The numerical coefficient changes when the chosen units change. Its familiar value is not an extra physical property of the star. Writing the units beside every input is the most effective way to avoid a plausible-looking thousandfold error.

Combine right ascension and declination correctly

Let μα* = (dα/dt) cos δ and μδ = dδ/dt. These are local perpendicular components on the celestial sphere. Their magnitude is μ = √(μα*² + μδ²). A component can be negative; the magnitude cannot.

The Astropy Gaia tutorial explicitly identifies the catalogue's pmra field with the cos-declination form. If a field already contains μα*, multiplying it by cos δ again is an error. Other catalogues must be checked against their own column definitions.

For an invented example at δ = 60°, a raw right-ascension rate dα/dt of 120 mas/year gives μα* = 60 mas/year. With μδ = 80 mas/year, the total is 100 mas/year. If the supplied value is already pmra = 60, use 60 directly; do not reduce it to 30.

Right ascension can also be written in time units. One second of right-ascension time corresponds to fifteen arcseconds of coordinate angle before the cos δ projection. That is another conversion to make explicitly, not a reason to apply a factor of fifteen to a catalogue field already measured in angular mas/year.

Six calculations using positive, illustrative parallaxes

Each row uses d = 1000/p with p in mas, then applies the velocity formula. These are invented inputs chosen for clean arithmetic, not measurements of named stars. The direct reciprocal is appropriate for this geometry exercise; a noisy measured parallax needs a justified distance estimate.

In the fourth row, components 120 and −160 mas/year give μ = 200 mas/year. A parallax of 20 mas gives d = 50 pc. Converting μ to 0.2 arcsec/year yields vt = 4.74047 × 0.2 × 50 = 47.4047 km/s.

The first two rows reverse the sign of one component while preserving the magnitude. They have different directions but the same transverse speed. Adding signed components instead of taking their vector magnitude would incorrectly give different answers.

The parallax and distance guide explains why a negative or poorly constrained measured parallax cannot simply be inverted into a trustworthy physical distance. Do not take its absolute value just to make this calculation return a positive number.

The same angular rate at different distances

At a fixed proper motion, doubling distance doubles the inferred transverse speed. At a fixed distance, doubling proper motion has the same effect. The graph makes both relationships visible without implying that a particular star changes its distance while keeping its motion constant.

For a useful comparison, object A with 200 mas/year at 25 pc and object B with 50 mas/year at 100 pc both have vt = 23.70235 km/s. The fourfold difference in angular rate is offset by a fourfold inverse difference in distance.

The horizontal axis is a length and the vertical axis is a speed. A slope on this plot represents a selected angular rate after unit conversion; it is not the star's acceleration. These distinctions matter when a straight line is mistaken for a physical trajectory.

Add radial velocity only in the same frame

When vr is the radial component and vt is the transverse magnitude in the same frame, v = √(vt² + vr²). Using the fourth example with vr = −30 km/s gives a total speed of approximately 56.10 km/s. The negative sign identifies approach under the usual positive-receding convention.

Changing the radial sign changes the direction of motion, not the total magnitude in this calculation. A missing radial velocity is not evidence for zero. Without that component, the measured transverse speed is only part of the full speed.

A heliocentric, barycentric, local-standard-of-rest or Galactocentric velocity must be labelled. Transforming to another frame can change components and total speed. The simple square-root expression combines components; it does not itself perform a frame transformation or reconstruct a Galactic orbit.

Move a position to a nearby epoch carefully

For a small displacement away from the poles, a first-order local approximation is Δδ = μδΔt and Δα = μα*Δt/cos δ. Convert mas to degrees by dividing by 3,600,000. Here δ in the denominator is the starting declination, held fixed for the approximation.

This example starts at α = 120°, δ = 60°, with μα* = 60 and μδ = 80 mas/year. After ten years, the east and north tangent-plane offsets are 600 and 800 mas, whose magnitude is 1000 mas, or one arcsecond. The raw right-ascension coordinate change is 1200 mas because cos 60° = 0.5.

The table is not a precision long-term ephemeris. Near a pole, division by cos δ becomes unstable and coordinate angle is a poor substitute for a vector calculation. Large time intervals, radial motion and changing distance also require more careful propagation.

Astropy's space-motion documentation describes propagation using straight-line, constant velocity in an inertial frame. It also states that an omitted radial velocity is assumed to be zero. That software assumption must not be mistaken for a measurement. For scientific use, carry the available full astrometric solution and its epoch.

Track uncertainty through both motion and distance

For small independent errors in μ and d, first-order propagation gives σv/v ≈ √[(σμ/μ)² + (σd/d)²]. For μ = 100 ± 2 mas/year and d = 50 ± 5 pc, the central speed is 23.70235 km/s and the estimated fractional uncertainty is about 10.20%, or roughly 2.42 km/s.

The rows vary the inputs to show sensitivity. Their outer values are not a confidence interval. Combining simultaneous upper or lower inputs is different from defining a joint statistical probability region.

Even the proper-motion magnitude has an uncertainty calculation. For nonzero μ with components a and b, its first-order variance is (a²σa² + b²σb² + 2ab Cov(a,b))/μ². With a = 60, b = 80 and independent component uncertainties of 2 mas/year each, σμ = 2 mas/year.

If distance and motion are correlated, their covariance also contributes. Near zero proper motion, linear propagation of a vector magnitude can be unreliable. If distance was inferred from uncertain parallax, use the distance inference and covariance information appropriate to the data, rather than assigning a symmetrical error by convenience.

A reproducible catalogue worksheet

1. Identify the catalogue release, source and reference epoch.
2. Record the coordinate frame and each column's units.
3. Check whether the RA motion already contains cos δ.
4. Combine the two perpendicular components into μ.
5. Use a justified positive distance estimate with its uncertainty.
6. Convert units and compute vt.
7. Add radial velocity only when available in the same frame.
8. State approximations if propagating the position to another epoch.

These steps connect a moving point on the sky with a physical speed while keeping the assumptions visible. They also explain why apparent brightness, annual parallax, proper motion and daily rising or setting answer different questions about the same star.

Frequently asked questions