Relativistic Stellar Orbits as Probes

A star on a resolved eccentric orbit around Sagittarius A* is simultaneously a clock, a test mass, and a probe of unseen matter. Its sky position measures orbital geometry; its spectrum measures line-of-sight motion plus gravitational and transverse-Doppler redshift; its secular precession mixes Schwarzschild curvature with distributed mass, companions, and reference-frame drift. The inference is therefore not “fit an ellipse and inspect the residual.” It is a joint, time-tagged forward model for astrometry, spectroscopy, light propagation, and calibration. Multiple stars and observables are valuable because the competing terms have different radial, directional, and temporal signatures.

Required background. Classical Field Theory supplies covariant trajectories, conserved quantities, and perturbative field effects used by the orbit model.

Helpful background. Black-Hole Geometry and Mechanics supplies the Kerr parameters and horizon-scale geometry; Relativistic Matter and Compact Objects identifies possible stellar remnants and distributed mass; Black-Hole Astrophysics provides the source and measurement context; and Relativistic Two-Body Dynamics supplies the approximation regimes needed to separate geodesic, post-Newtonian, and perturbing-body effects.

Relativistic signatures in a stellar orbit

For a test star on a weak-field bound orbit with semimajor axis $a$, eccentricity $e$, and semilatus rectum $p=a(1-e^2)\gg GM_\bullet$, the leading Schwarzschild periapse advance per orbit is

$$\Delta\omega_{\rm S}= \frac{6\pi GM_\bullet}{a(1-e^2)}.$$

It is prograde and lies in the orbital plane. The site units set $c=1$; restoring ordinary units inserts $c^2$ in the denominator. The observable is not an instantaneous rotation of a drawn ellipse. For a highly eccentric star, most sensitivity comes from comparing pre- and post-periapse astrometry while the earlier orbit fixes the incoming direction.

With positive spectroscopic redshift defined as recession, the leading measured shift has the schematic expansion

$$z=\frac{\lambda_{\rm obs}-\lambda_{\rm emit}}{\lambda_{\rm emit}} =v_{\rm los} +\frac{v^2}{2} +\frac{GM_\bullet}{r} +z_0+O(v^3),$$

where velocities are in units of $c$, and $z_0$ absorbs the reference-system velocity and stellar spectral zero point in this schematic distant-observer expression. The velocity and potential are evaluated at emission; a real fit also maps emission time to arrival time. The $v^2/2$ transverse-Doppler and $GM_\bullet/r$ gravitational terms have the same leading size for a Kepler orbit. The 2018 S2 analysis fitted them jointly rather than measuring a pure gravitational-redshift term in isolation GRAVITY Collaboration 2018, §3.1 and Appendix A, article pp. 2–8.

Black-hole spin produces a smaller, orientation-dependent change. Define $a_*=|\boldsymbol J_\bullet|/(GM_\bullet^2)$, and measure the ascending node of a tilted stellar orbit relative to the black hole’s equatorial plane. To leading order, its Lense–Thirring advance about the spin axis is

$$\Delta\Omega_{\rm LT} \simeq4\pi a_* \left[\frac{GM_\bullet}{a(1-e^2)}\right]^{3/2},$$

with spin-directed azimuth defined positive. The apparent node and periapse changes on the sky depend on the spin orientation and stellar orbit; an exactly equatorial orbit has no separately defined node. The quadrupole effect is smaller still in this weak-field expansion. Through the evidence cutoff below, these stellar-orbit spin signatures remain unresolved; a forecast based on a hypothetical shorter-period star is not a detection. Alexander separates the Schwarzschild, spin, and Newtonian perturbations in Alexander 2017, §5.2, author manuscript p. 33.

For an illustrative $M_\bullet=4.0\times10^6M_\odot$, $a=1000\ {\rm AU}$, and $e=0.88$, one has $p=225.6\ {\rm AU}$ and $GM_\bullet/c^2\simeq0.0395\ {\rm AU}$. The Schwarzschild advance is about $0.00330$ radians, or $11.3$ arcminutes, per orbit. Even at $a_*=1$, the spin-relative nodal advance is only about $0.10$ arcminute. These are orbital-element changes for a teaching example, not angular displacements of the star on the observer’s image. Projection and distance must be included before comparing with an astrometric error bar.

Competing Newtonian and instrumental effects

A smooth spherical extended mass with nonnegative density produces retrograde apsidal precession in the nearly Keplerian regime. For a power-law density, its order of magnitude may be written

$$\Delta\omega_{\rm ext} \sim-2\pi\frac{M_{\rm ext}(<a)}{M_\bullet} C_\gamma(e),$$

where $C_\gamma(e)$ is a positive, model-dependent factor. The opposite sign helps distinguish it from the Schwarzschild term, but imperfect phase coverage and reference-frame parameters can correlate the two. A compact companion produces non-secular, star-dependent residuals rather than a single smooth radial law. Source confusion can shift a measured centroid abruptly when two infrared sources overlap.

Reference-frame drift is especially adversarial because a low-order translation or velocity of the astrometric origin can imitate part of a long-period orbital residual. Distance $R_0$ and mass $M_\bullet$ are correlated: angular astrometry and physical radial velocities scale differently with distance, so a joint fit is essential. Light-travel time, aberration, barycentric time conversion, and the observer’s velocity must use one convention across instruments.

The decisive comparison is therefore not “relativity versus no relativity.” Credible competing components are Schwarzschild precession, redshift, frame dragging, extended stellar or dark mass, one or more perturbing companions, source confusion, and reference-frame motion. They should enter the same likelihood wherever the data are sensitive, rather than being tested in separate fits with mutually fixed nuisance parameters.

Data release, calibration, and likelihood

This page freezes two named observational products. The redshift example is the GRAVITY Collaboration’s 2018 S2 release: 1992–2018 astrometry and spectroscopy, including the 2017–2018 NACO, SINFONI, and GRAVITY epochs described in the article and Appendix A. The current precession and extended-mass comparison uses the GRAVITY Collaboration’s 2024 analysis: GRAVITY astrometry through 2022 for S2, S29, S38, and S55, legacy spectroscopy, and the stated ERIS radial-velocity additions GRAVITY Collaboration 2024, §2, article pp. 2–3. These are publication-bound datasets, not a timeless amalgam of every Galactic-centre measurement.

Calibration must preserve plate scale, detector orientation, optical distortion, interferometric phase aberrations, the link between infrared and radio frames, source-confusion treatment, spectral line-spread function, wavelength zero point, barycentric timestamps, and cross-instrument covariance. The 2018 paper’s Appendix A explicitly documents point-spread overlap, atmospheric-line wavelength checks, visibility calibration, alternative fitting codes, and systematic-error treatment GRAVITY Collaboration 2018, Appendices A.1–A.7, article pp. 6–8.

Selections must list the accepted epochs, stars, spectral lines, confusion windows, quality flags, radial-velocity cuts, and outlier rule. Nuisance choices should be fixed before testing a relativistic amplitude when practical. Removing confused adaptive-optics epochs is a selection, not a calibration improvement that can be applied silently after inspecting a residual.

For data vectors $\boldsymbol d_i=(\Delta\alpha_i,\Delta\delta_i,v_{{\rm los},i})$ and a joint model $\boldsymbol m_i(\Theta)$, a Gaussian form is

$$\ln\mathcal L(\Theta) =-\frac12\boldsymbol r^T C^{-1}\boldsymbol r -\frac12\ln\det C+\text{constant}, \qquad \boldsymbol r=\boldsymbol d-\boldsymbol m(\Theta).$$

The parameter vector includes orbital elements, $M_\bullet$, $R_0$, astrometric-frame position and drift, spectroscopic offsets, relativistic amplitudes, and competing extended-mass or companion parameters. A posterior or Bayes factor is conditional on priors for all of them. Bootstrap or alternative weighting checks probe sampling sensitivity but do not erase shared calibration systematics.

Reproduction should first recover the published Schwarzschild-amplitude or redshift posterior from the article’s machine-readable epoch tables and declared model, then perform leave-one-instrument-out fits and injections of extended mass, a drifting frame, and a companion. The raw interferometric calibration products and every consortium fitting implementation are not bundled with this page, so it does not claim bitwise reproduction. A reusable reproduction package would need the precise data-product identifiers, calibration pipeline version, covariance construction, priors, and executable orbit/light-propagation software.

Results and limitations through 2026-08-18

The following matrix distinguishes measured terms from model-dependent limits and forecasts. “Consistent with” is not promoted to a universal verdict about every possible compact-object model or gravity theory.

Evidence or status	Result at the cutoff	Scope and limitation
S2 spectroscopy through 2018	A combined transverse-Doppler plus gravitational-redshift contribution was detected in a joint orbit fit GRAVITY Collaboration 2018, §§2–3, article pp. 2–5	The two terms are not independently isolated; wavelength calibration, reference velocity, and orbital geometry are shared nuisances
S2 astrometry through 2019	Prograde in-plane Schwarzschild precession was reported from the combined GRAVITY, NACO, and spectroscopic fit GRAVITY Collaboration 2020, §§2–4	The inference depends on the astrometric frame, confusion handling, epoch weighting, and the extended-mass model
Four-star analysis through the 2024 release	Additional GRAVITY and ERIS data strengthened the Schwarzschild-precession inference and constrained smooth extended mass under power-law and Plummer profiles GRAVITY Collaboration 2024, §§2–4, article pp. 2–8	The mass bound is profile- and prior-dependent; granular remnants and companions need different forward models
Frame dragging and quadrupole	No robust stellar-orbit detection is established at this cutoff GRAVITY Collaboration 2024, §1	Forecasts depend on discovering and characterizing shorter-period stars, spin orientation, confusion, and calibration domain

The established evidence supports a relativistic Schwarzschild term in the named S-star analyses. It does not establish a stellar measurement of spin, a unique distributed-mass profile, or the absence of every perturbing companion. Those questions remain open because different formulations and source populations can fit below current signal-to-noise and resolution.

Adversarial inference checks

First, perform leave-one-instrument-out and leave-one-periapse-window-out analyses. A relativistic amplitude that depends on a single calibration stream or a post-selected cluster of epochs has underestimated systematic uncertainty. Second, inject a retrograde smooth-mass term, a localized companion perturbation, and linear frame drift into synthetic data with the real cadence and covariance. Recovering the relativistic amplitude without bias is more discriminating than obtaining a small residual on the original sample.

Third, fit several stars simultaneously. Schwarzschild precession follows $[a(1-e^2)]^{-1}$, frame dragging follows a steeper scale and an orientation, smooth extended mass follows its enclosed-mass profile, and a reference-frame drift is common on the sky. Multiple radial scalings and orbital planes can separate terms that one star cannot. The analysis must retain shared $M_\bullet$, $R_0$, and frame parameters rather than multiply nominally independent posteriors.

The page stops at astrophysical orbit inference. Generic alternative field equations and global tests of gravity belong to their dedicated treatments; importing them here without a calibrated stellar forward model would not resolve the observational degeneracies.

What could change the interpretation

The interpretation could change after a new public Galactic-centre astrometric or spectroscopic release changes the epoch set or covariance, after a shorter-period star obtains sufficient phase coverage, or after a robust extended-mass, companion, reference-frame, frame-dragging, or quadrupole result changes the competing model space. A material calibration revision or withdrawal also affects the interpretation even if the central numerical estimate remains similar.

Exercises

A single-star fit finds a small prograde apsidal residual. Adding a smooth extended-mass component increases the inferred prograde relativistic amplitude, while allowing linear reference-frame drift broadens it. A second star on a different orbital plane is available. Which formulation should be fitted, and what pattern would count against interpreting the first residual as frame dragging?

Solution

Fit both stars in one model with shared $M_\bullet$, $R_0$, frame position and drift, while giving each star its own orbital elements. Include Schwarzschild, smooth extended-mass, and spin terms simultaneously, with the spin vector shared. Extended mass contributes retrograde apsidal shifts determined mainly by each radial range; frame drift is a common sky motion; frame dragging has a steeper semilatus-rectum dependence and orientation-dependent nodal and apsidal components. A signal confined to one star, disappearing under plausible frame choices, or lacking the correlated orientation pattern in the second orbit counts against a spin interpretation. The correct outcome may be an upper limit rather than a detection.

References

• Alexander, Tal. 2017. “Stellar Dynamics and Stellar Phenomena Near a Massive Black Hole.” Annual Review of Astronomy and Astrophysics 55: 17–57. DOI.
• GRAVITY Collaboration. 2018. “Detection of the Gravitational Redshift in the Orbit of the Star S2 Near the Galactic Centre Massive Black Hole.” Astronomy & Astrophysics 615: L15. DOI.
• GRAVITY Collaboration. 2020. “Detection of the Schwarzschild Precession in the Orbit of the Star S2 Near the Galactic Centre Massive Black Hole.” Astronomy & Astrophysics 636: L5. DOI.
• GRAVITY Collaboration. 2024. “Improving Constraints on the Extended Mass Distribution in the Galactic Center with Stellar Orbits.” Astronomy & Astrophysics 692: A242. DOI.