How can an astronomer measure the wobble of a star and still not know the true mass of its orbiting planet? Spectroscopy observes only the star’s line-of-sight velocity. The radial-velocity semi-amplitude constrains a projected quantity called Mp sin i; the unseen orbital inclination determines how that quantity relates to true mass.
This guide starts with a toy 14.2 m/s signal around a solar-mass star, calculates Mp sin i≈0.50 Jupiter mass, tests viewing angles and examines the exact mass function, eccentricity and alternative explanations. It also shows when transits and astrometry can remove part of the mass ambiguity.
eduKateSG · The Gold Standard of Astronomy
Choose your reading route
From light to wobble · Chapters 1–3: velocities, semi-amplitude and viewing angle.
Test the mass model · Chapters 4–6: exact formula, additional methods and false signals.
Ask and continue · Chapters 7–8: FAQs, reading map and primary sources.
Chapter index
Observation and geometry · Chapters 1–3
Model and evidence · Chapters 4–6
Questions and sources · Chapters 7–8
A planet and its host star orbit a shared centre of mass. As a consequence, the host star moves toward and away from a distant observer during an orbit. Precision spectroscopy measures shifts in stellar spectral features that can be interpreted as changes in radial velocity along the observer’s line of sight. The measured quantity is not the planet’s mass; it is a series of stellar velocities inferred from the spectrum.
A Keplerian orbit can create a periodic velocity pattern. Its shape and timing depend on orbital period, eccentricity, phase and other parameters. The observed spectrum also contains instrumental calibration, stellar surface activity and measurement uncertainty. A planet interpretation therefore requires a time series and competing explanations, not merely one redshifted spectrum.
The quantity normally called radial-velocity semi-amplitude, K, is the half-range of the modelled velocity variation after accounting for the systemic velocity. If a circular-orbit model has a sinusoidal velocity signal from −14.2 to +14.2 metres per second around its reference velocity, K is 14.2 m/s, not 28.4 m/s. The complete peak-to-peak excursion is 28.4 m/s.
The NASA Exoplanet Archive defines K as half the peak-to-peak variation in stellar radial velocity. It separately records the minimum mass parameter M sin i, orbital eccentricity and period. These are different dimensions of the planetary inference, and their names should not be collapsed into the vague word ‘mass’.
The central question is how a visible stellar motion constrains an unseen companion when the orientation of its orbit is not directly known. That uncertainty is not a measurement error in K alone; it is a geometric degeneracy called the inclination problem.
For a planet much less massive than its host star, a useful circular-orbit scale is K≈28.4 m/s × (Mp sin i/MJ) × (P/1 year)^(−1/3) × (Mstar/MSun)^(−2/3). Here MJ and MSun are Jupiter and solar masses, P is the orbital period, and i is inclination measured relative to the plane of the sky. The coefficient is a rounded astronomical unit conversion for the approximate model.
Choose an entirely fictional system with a host star of one solar mass, a period of one year and zero eccentricity. Suppose spectroscopy supports a semi-amplitude K=14.2 m/s. The other scale factors become one, giving Mp sin i/MJ≈14.2/28.4=0.50.
The observational inference in this toy model is therefore Mp sin i≈0.50 Jupiter mass. That is the projected minimum-mass quantity, not an unconditional measurement of true mass. It assumes the stellar mass, period and eccentricity are represented correctly and that the Keplerian approximation captures the velocity signal.
To check the half-amplitude definition, let the velocity residual relative to the systemic velocity vary from −14.2 m/s to +14.2 m/s. Its peak-to-peak change is 28.4 m/s, yet the K used in the equation is 14.2 m/s. Substituting 28.4 would double the inferred projected mass without justification.
An actual multi-planet star need not exhibit one clean sine wave; multiple orbital terms may overlap. A circular orbit is chosen here solely so the arithmetic can be checked and the inclination degeneracy can be isolated from other modelling questions.
| Input | Toy value | Role |
|---|---|---|
| Host mass | 1.00 solar mass | Assumed stellar mass |
| Period | 1.00 year | Orbital cycle |
| Eccentricity | 0 | Circular model |
| RV semi-amplitude K | 14.2 m/s | Half the modelled peak-to-peak velocity |
| Peak-to-peak velocity | 28.4 m/s | Twice K |
| Projected companion mass | ≈0.50 Jupiter mass | Mp sin i under small-mass approximation |
The observed line-of-sight velocity depends on orbital orientation. Astronomers define i≈90° as an edge-on orbit, giving sin i≈1, while an approximately face-on orbit has i≈0° and produces little radial-velocity component of orbital motion. The same physical orbit can therefore make a different K depending on the viewing direction.
Under the low-companion-mass approximation, if the velocity fit determines Mp sin i≈0.50 Jupiter mass, the true companion mass is Mp≈0.50 MJ/sin i. At i=90°, sin i=1, so Mp≈0.50 MJ. At i=30°, sin i=0.5, so Mp≈1.00 MJ. At i=10°, sin i≈0.17365, so Mp≈2.88 MJ.
This is the minimum-mass trap: it is permissible to write ‘a companion with Mp sin i of about 0.5 Jupiter masses’. It is not permissible to silently rewrite that as ‘a companion with a true mass of exactly 0.5 Jupiter masses’ when inclination is unknown.
For a fixed observed projected mass the true mass increases as the orbit moves toward face-on. In the limiting face-on case the radial velocity from a Keplerian circular orbit tends toward zero for a fixed true companion mass. An RV detection alone cannot usually establish a unique inclination, so an arbitrarily tiny sin i cannot be inferred just by stretching the measured RV curve.
This table is intentionally algebraic. When the companion is no longer much smaller than its star, the mass in the denominator of the exact Keplerian relation also matters, and simply dividing a quoted approximate minimum mass by sin i becomes less accurate. The small-mass assumption is good for our Jupiter-scale teaching scenario around a solar-mass star.
| Assumed inclination | sin i | Approximate true companion mass |
|---|---|---|
| 90° | 1.0000 | 0.50 MJ |
| 60° | 0.8660 | ≈0.58 MJ |
| 30° | 0.5000 | 1.00 MJ |
| 10° | ≈0.17365 | ≈2.88 MJ |
A Keplerian radial-velocity fit supports the exact binary mass-function relationship f=(P K³/2πG)(1−e²)^(3/2)=(Mp sin i)³/(Mstar+Mp)². The left-hand side depends on the measured period, semi-amplitude and eccentricity plus the gravitational constant. The right-hand side links that observable combination to the physical masses and inclination.
In the small-companion-mass limit, Mstar+Mp≈Mstar, which leads to the simplified power-law scale used earlier. That approximation is not equally good for a giant companion around a much lighter host or when the companion has stellar-scale mass.
The formula also shows why eccentricity matters. For fixed masses, period and inclination, K contains the factor 1/sqrt(1−e²). If eccentricity e=0.60 instead of zero, the factor is 1/0.80=1.25. For an observed K held fixed, the inferred low-mass Mp sin i scale would instead include sqrt(1−e²)=0.80 relative to an otherwise identical circular solution.
Thus the toy projected mass of 0.50 MJ under e=0 would become approximately 0.40 MJ if K and the other assumptions were kept fixed but a valid fit required e=0.60. This is a sensitivity test, not a claim that an arbitrary eccentricity can be selected while preserving a physically acceptable fit to a time series.
Host stellar mass also matters. If the host mass were reassessed from 1.00 to 1.20 solar masses with K, P and e unchanged, the low-mass inferred Mp sin i increases by a factor 1.20^(2/3)≈1.129, from 0.50 to about 0.56 MJ. A precise-looking planet mass cannot be more secure than the underlying stellar and orbital measurements.
The NASA Exoplanet Archive’s calculation notes include the full K dependence and explicitly discuss when a small-planet-mass approximation is being used. Preserve those assumptions whenever a mass number is reported.
| Changed model assumption | Approximate Mp sin i | Why it changes |
|---|---|---|
| One-solar-mass circular base case | 0.50 MJ | Reference case |
| e=0.60, same K and host mass | ≈0.40 MJ | sqrt(1−e²) factor |
| Mstar=1.20 MSun, e=0 | ≈0.56 MJ | Mstar^(2/3) factor |
When a planet transits its star, the orbital plane is usually observed close to edge-on, so a light-curve model can constrain orbital inclination relative to the line of sight. Combining a validated transit inclination with radial velocity can turn a projected Mp sin i measurement into a much better estimate of true Mp.
This does not mean every transiting planet is at exactly 90° inclination. The transit geometry depends on the star and planet radii, orbital separation and eccentricity. A carefully modelled light curve, rather than the single word ‘transit’, supplies the relevant inclination evidence.
Astrometry provides another possible path. Instead of measuring the motion component along the line of sight, it measures apparent sky-plane motion of the host star. A sufficiently informative astrometric orbit can constrain inclination and the companion’s mass scale when combined with spectroscopy and an appropriate stellar model.
These methods have different sensitivities and systematics. A transit constrains one viewing geometry but may be altered by blending and stellar variability. An astrometric solution can be affected by unresolved companions, observational scanning and model selection. The Gaia RUWE and unresolved-binary guide explains why a poor astrometric fit is not by itself an orbital solution.
A short illustrative check: if the independently measured inclination is 87°, sin i≈0.9986. An Mp sin i of 0.50 MJ then implies a mass of about 0.5007 MJ under the same low-mass approximation. The difference from the minimum is small but not exactly zero.
It is therefore more informative to say how inclination was obtained and what uncertainty it carries than to announce that radial velocity ‘measures true mass’. A combined result may justify that statement, while an RV-only solution generally does not.
A periodic apparent stellar velocity can be caused or distorted by stellar spots, pulsations and magnetic activity. The observed spectral line profile can change as different rotating surface regions move into view. The resulting signal may correlate with brightness or chromospheric-activity indicators and can imitate some aspects of orbital radial velocity.
Instrument zero points, wavelength-calibration drift and changing data-reduction pipelines also matter. When combining data from instruments with distinct reference velocities, the systemic offsets should be represented in the fit rather than assumed identical. A small apparent long-period signal can otherwise arise from an unmodelled instrumental difference.
Period searches must address observation cadence. A recurring pattern in the data can be an alias from the sampling schedule. The companion daily aliases and observing cadence guide examines this broader statistical problem.
An orbit model should be tested against time-series residuals. A highly precise K estimate is not reliable evidence for a planet if the model fails independent observations, if activity diagnostics tell a different story, or if systematic errors dominate the assumed uncertainty.
Likewise, a transit-like dip and a periodic velocity signature must be checked for consistency with the same physical orbit and host star. A blended eclipsing binary can sometimes imitate one or both observables if the source identification is inadequate.
Before publication, preserve observation times and standards, wavelength calibration, stellar-mass evidence, period, K, eccentricity, fitted noise terms, quality metrics and the mass provenance field. ‘Minimum mass’ and ‘true mass’ are different fields in the NASA Exoplanet Archive precisely because the distinction matters.
Does K equal the full peak-to-peak velocity range? No. K is the semi-amplitude of a fitted radial-velocity variation. A circular sinusoid with K=14.2 m/s moves through a peak-to-peak range of 28.4 m/s.
Does radial velocity alone measure Mp exactly? Usually not. With an unknown inclination it constrains a projected minimum-mass quantity, subject to the stellar and orbit model.
Why is inclination 90° called edge-on? Under the astronomical convention the orbital plane is viewed close to edge-on at i≈90°, giving sin i near one and a large line-of-sight velocity component.
Does a minimum mass of 0.50 Jupiter mass mean the body must be a planet? Not by itself. Additional evidence and the actual mass range may matter for classification, especially if inclination is unknown.
Will a transiting companion always give its exact mass? No. A well-measured inclination can help substantially, but uncertainties in stellar properties, orbit, measurements and possible contamination remain.
Can an apparently periodic RV curve be caused by stellar activity? Yes. Surface features and spectral-line variability must be investigated with appropriate indicators and independent data.
Does eccentricity change the semi-amplitude relationship? Yes. The Keplerian expression includes 1/sqrt(1−e²), so eccentricity is part of the inference.
Can two RV surveys disagree on a planet mass? Yes. Different instruments, epochs, orbital models and stellar parameters can lead to different estimates; compare their mass provenance and uncertainties.
Does the mass function require Mp much smaller than Mstar? No. The exact expression includes Mstar+Mp. The commonly used simple scaling assumes Mp is much smaller than its host.
CHAPTER 8 OF 8
8. Learn how to read an exoplanet catalogue without inventing precision
The main lesson is to distinguish an observation from the physical parameter inferred through a model. Spectroscopy supplies time-dependent line-of-sight stellar velocities. A Keplerian model extracts K, period and eccentricity. Stellar mass and inclination then connect those observables to companion mass.
Try changing only the inclination in the synthetic example: K=14.2 m/s, P=one year, e=0 and a solar-mass host. The projected companion mass remains about 0.50 MJ, while the inferred true mass changes from 0.50 MJ at 90° to 1.00 MJ at 30°. The difference is geometry, not an unexplained change in the measured stellar velocity.
The subject library includes The Gold Standard of Astronomy, Vetting Exoplanet Transit Candidates, Barycentric Transit Timing, and The Gold Standard of Astrophysics. See also Learning Transfers into Life and How X Works.
For authoritative terminology and calculations, consult NASA Exoplanet Archive: Predicted Observables and Radial Velocity, NASA Exoplanet Archive: Planetary Systems Column Definitions, and NASA Exoplanet Archive: EXOFAST Output Parameters.
Source and scope note, 23 October 2026: the host star, K values, orbital angles and companion masses used above are synthetic educational values. The simple 28.4 m/s scale is rounded and uses the low-mass approximation; no actual planet is identified by this exercise.