Hipparcos–Gaia Proper-motion Anomalies

• Hipparcos–Gaia proper-motion anomalies quantify deviations between short-term catalog motions and long-term motions derived from positional differences over a ~25-year baseline.
• The method employs cross-calibration of Hipparcos and Gaia data with corrections for covariance, frame rotation, and perspective effects to isolate genuine motion anomalies.
• Applications include tracing stellar multiplicity, refining companion mass estimates, and diagnosing systematic errors in astrometric catalogs.

Hipparcos–Gaia proper-motion anomalies are discrepancies between short-term proper motions measured by Hipparcos or Gaia and long-term proper motions inferred from the positional displacement between the two missions over a baseline of about $24$–$25$ years. In the ideal single-star, uniform-motion limit, these quantities should agree after transformation to a common reference frame and after correction for covariance, frame rotation, and perspective effects. In practice, their mismatch is used as a diagnostic of non-linear sky motion, most commonly orbital reflex motion from an unseen companion, but also of reference-frame inconsistencies, sky-correlated catalog systematics, and perspective acceleration. The modern formalism was established through the Hipparcos–Gaia Catalog of Accelerations (HGCA) and related PMa catalogs, then extended to stellar multiplicity surveys, dynamical mass determinations, reference-frame diagnostics, and orbit fitting (Brandt, 2018, Brandt, 2021, Kervella et al., 2018, Kervella et al., 2021).

1. Fundamental definition and astrometric observables

The operational definition of a proper-motion anomaly (PMa) is the vector difference between a short-baseline catalog proper motion and the long-baseline Hipparcos–Gaia proper motion. For Gaia DR2 and Gaia EDR3/DR3 applications, the long-term motion is obtained from the positional difference divided by the mission epoch separation; for a single star it should match the short-term proper motion except for perspective acceleration and measurement or calibration errors (Brandt, 2018, Makarov, 2022).

Quantity	Meaning	Typical epoch usage
Hipparcos proper motion	Short-term mission proper motion	Near $1991.25$
Gaia proper motion	Short-term mission proper motion	Near $2015.5$ for DR2 or $2016.0$ for EDR3/DR3
Hipparcos–Gaia proper motion ($\mu_{\rm HG}$)	Long-term proper motion from positional displacement	Over the $\sim 24.25$- or $24.75$-year baseline

A standard DR3-era geometric construction writes the Hipparcos–Gaia proper motion as

$$\boldsymbol \mu_{\rm HG} = \frac{\boldsymbol r_{\rm G} - \boldsymbol r_{\rm H}}{24.75},$$

with covariance

$$\boldsymbol C_{\mu\rm HG} = \frac{\boldsymbol C_{r\rm H} + \boldsymbol C_{r\rm G}}{24.75^2}.$$

The anomaly is then expressed as a difference vector such as

$25$0

or, in the HGCA notation, as residuals between the Hipparcos proper motion, the Gaia proper motion, and the long-term position-difference proper motion (Makarov, 2022, Brandt, 2018).

Several papers distinguish between “proper-motion anomaly” and “astrometric acceleration.” The latter is common usage, but some authors prefer PMa because the measured quantity is in $25$1, not in dynamical acceleration units (Zhang et al., 2023).

2. Statistical formulation, covariance propagation, and cross-calibration

The modern PMa literature treats the anomaly as a two-dimensional vector with a full covariance matrix. A common significance statistic is

$25$2

where $25$3 is the covariance of the proper-motion difference. In the Pleiades reassessment, the same structure appears as

$25$4

with $25$5 taken as the sum of the Gaia and Hipparcos proper-motion covariance matrices (Makarov, 2022, Makarov, 2022).

The HGCA formalism makes the three proper motions nearly independent by propagating Hipparcos and Gaia positions to their central epochs, thereby reducing position–proper-motion covariance in a given coordinate direction. In the DR2 edition, the final catalog adopts the Gaia DR2 reference frame, locally cross-calibrates the Hipparcos proper motions and the scaled Hipparcos–Gaia positional differences to that frame, and supplies covariance matrices and central epochs suitable for orbit fitting in a $25$6 framework (Brandt, 2018). The EDR3 edition preserves the same architecture but improves the calibration with Gaia EDR3 proper motions, explicit local frame-rotation maps, color- and magnitude-dependent corrections, and an updated Gaia uncertainty inflation factor of $25$7 (Brandt, 2021).

Cross-calibration is central because global rigid alignment is not sufficient. Brandt’s DR2 calibration divides the sky into many regions, fits local offsets or rotations between proper-motion systems, and smooths the results with Gaussian process regression; the EDR3 edition again finds strong evidence for locally variable frame rotations between all pairs of proper-motion measurements (Brandt, 2018, Brandt, 2021). Both HGCA versions also conclude that a linear combination of the two Hipparcos reductions is superior to either one alone, with a weight close to $25$8 van Leeuwen and $25$9 ESA (Brandt, 2018, Brandt, 2021).

The anomaly can be converted into a tangential-velocity signal through

$1991.25$0

which is widely used to express the reflex motion in physical units and to connect PMa amplitudes with companion mass–separation constraints (Kervella et al., 2021, Zhang et al., 2023).

3. Reference-frame distortions and Hipparcos systematics

A major development in the field is the recognition that Hipparcos–Gaia PMa is not purely astrophysical. The proper-motion systems of carefully filtered common stars are not statistically consistent within the given formal errors, and the difference field contains a large sky-correlated component. Fitting the vector field with $1991.25$1 vector spherical harmonics complete through degree $1991.25$2 reveals a median vector length of about $1991.25$3 in the Gaia-minus-Hipparcos field and a rigid spin of about $1991.25$4 (Makarov, 2022).

In that decomposition, the first three degree-$1991.25$5 magnetic harmonics represent the rigid spin. For the Gaia-minus-Hipparcos field, the fitted components are

$1991.25$6

with combined amplitude about $1991.25$7. By contrast, Gaia proper motions and HG proper motions built from Hipparcos and Gaia positions are explicitly consistent by construction at a level of $1991.25$8, although higher-degree distortions remain (Makarov, 2022).

This systematic floor has direct implications for PMa interpretation. It implies that an observed $1991.25$9 can contain a sky-correlated catalog component in addition to genuine barycentric acceleration. The Pleiades case provides a localized demonstration. Reassessing the classic Hipparcos Pleiades parallax discrepancy with Gaia EDR3, Makarov found that the same reduction weakness that biased the cluster parallax also biased its proper motions. For the $2015.5$0 common Pleiades stars, the raw Gaia EDR3 minus Hipparcos weighted mean differences were

$2015.5$1

and, after subtraction of the fitted systematic vector field at the Pleiades location, the residual proper-motion difference became

$2015.5$2

Those values are statistically close to the corrections inferred earlier from Hipparcos Intermediate Astrometry Data, strongly suggesting that the Pleiades parallax error was also a proper-motion error produced by small-scale, sky-correlated distortions in the Hipparcos reduction (Makarov, 2022).

A plausible implication is that PMa studies operating near the $2015.5$3 regime must distinguish local catalog distortions from true orbital curvature rather than treating Hipparcos as a perfectly clean first epoch.

4. Perspective acceleration, observing-window smearing, and anomaly diagnostics

Not every proper-motion anomaly is caused by a companion or by catalog systematics. A distinct kinematic contaminant is perspective acceleration: if a star has nonzero radial velocity, its distance changes, and therefore its angular motion on the sky changes even under rectilinear three-dimensional motion. In the HTPM treatment this is expressed through a radial proper motion

$2015.5$4

and the induced proper-motion bias from an RV error $2015.5$5 follows

$2015.5$6

The bias scales with baseline, proper motion, inverse distance, and radial-velocity error, so nearby fast movers are the most sensitive cases (Bruijne et al., 2012).

Perspective terms were already recognized as essential in the joint Hipparcos–Gaia solution proposed for HTPM. That framework incorporated Hipparcos data as prior information inside AGIS, introduced the scaled model of kinematics, and defined a goodness-of-fit statistic

$2015.5$7

with $2015.5$8 sensitive to deviations from uniform space motion caused, for example, by binaries with periods of $2015.5$9–$2016.0$0 years (Michalik et al., 2014).

A second limitation is observing-window smearing. Hipparcos and Gaia proper motions are not instantaneous velocities but averages over finite observing windows. In the Kervella DR2 and later PMa analyses, the relevant windows are $2016.0$1 days for Hipparcos, $2016.0$2 days for Gaia DR2, and $2016.0$3 days for Gaia EDR3. This attenuates anomalies for orbits with periods comparable to or shorter than the observing window, reducing sensitivity below roughly $2016.0$4 days in the Cepheid/RR Lyrae PMa analysis and producing explicit sensitivity spikes in later RV+PMA mass determinations (Kervella et al., 2019, Kervella et al., 2021, Piccinini et al., 14 Jan 2026).

A third limitation is long-period suppression: for periods much longer than the Hipparcos–Gaia baseline, subtracting the long-term proper motion removes part of the orbital signal. This suggests that PMa is naturally most informative for companions with periods longer than the observing windows but shorter than, or not vastly longer than, the $2016.0$5- to $2016.0$6-year baseline (Kervella et al., 2018, Zhang et al., 2023).

5. Astrophysical uses: multiplicity censuses, binaries, and companion masses

The most extensive use of Hipparcos–Gaia PMa has been as a multiplicity tracer. In the Gaia DR2 nearby-star survey, Kervella et al. analyzed $2016.0$7 stars within $2016.0$8 pc and found that $2016.0$9 have $\mu_{\rm HG}$0, while the PMa catalog for the broader Hipparcos sample provided PMa for $\mu_{\rm HG}$1 of the Hipparcos catalog (Kervella et al., 2018). The Gaia EDR3 extension improved the PMa precision by a factor of $\mu_{\rm HG}$2, reaching a median tangential-velocity anomaly precision of $\mu_{\rm HG}$3, identifying $\mu_{\rm HG}$4 Hipparcos stars with significant PMa $\mu_{\rm HG}$5, $\mu_{\rm HG}$6 with common-proper-motion bound candidate companions, and $\mu_{\rm HG}$7 Hipparcos stars exhibiting at least one signal of binarity when PMa, CPM, and RUWE are combined (Kervella et al., 2021).

Variable-star populations provide a distinct application. For Galactic classical Cepheids and RR Lyrae stars, PMa was used as a purely astrometric companion diagnostic by comparing $\mu_{\rm HG}$8 and $\mu_{\rm HG}$9 with the long-term $\sim 24.25$0. In a sample of $\sim 24.25$1 Cepheids, $\sim 24.25$2 binaries were identified from PMa and $\sim 24.25$3 additional candidates were reported; among $\sim 24.25$4 tested RR Lyrae stars, $\sim 24.25$5 showed significant PMa and $\sim 24.25$6 were additional candidates. The authors concluded that the binary fraction of Galactic Cepheids is probably above $\sim 24.25$7, while the RR Lyrae binary fraction is at least $\sim 24.25$8 (Kervella et al., 2019).

For orbit determination, PMa has become a bridge between radial-velocity $\sim 24.25$9 detections and true masses. Using the HGCA together with RVs, Kiefer-like ambiguities have been revisited in several systems, while more direct joint RV+astrometry modeling has yielded precise masses for long-period companions. In HD 92987, a Gaia–Hipparcos astrometric signal of $24.75$0 showed that the nominal $24.75$1 RV companion is instead a $24.75$2 star in a near-polar orbit. In HD 221420, the same methodology raised the minimum mass to $24.75$3, leaving the object in the substellar regime (Venner et al., 2021). More elaborate direct modeling of Hipparcos and Gaia positions and proper motions, with RVs, also recovered low-mass companions such as HD 190360 b with $24.75$4 and $24.75$5 yr (Feng et al., 2021).

PMa has also become useful in exoplanet-host selection. A volume-limited catalog of $24.75$6 transiting-planet hosts or candidates within $24.75$7 pc with significant HGCA anomalies was constructed to examine stellar multiplicity around S-type transiting planets. In that sample, TOIs with significant PMa showed nearly four times more false positives due to eclipsing binaries than TOIs with marginal anomalies, and the differential-velocity distribution suggested that planets are more likely to form in systems with low-mass substellar companions or stellar companions at wider separation (Zhang et al., 2023).

6. Interpretation, caveats, and extensions beyond PMa-only analysis

Hipparcos–Gaia PMa is best understood as a calibrated multi-epoch kinematic residual, not as a self-sufficient companion classifier. Several papers explicitly warn that statistical distributions of accelerations from HGCA should be interpreted with caution because of residual spurious matches, heavy tails in Gaia uncertainties at low precision, contamination from blended binaries, and imperfect individual error modeling (Brandt, 2018). Likewise, the EDR3 HGCA emphasizes that the catalog is intended primarily as a follow-up and orbit-fitting tool, not as a final physical classification of all residuals (Brandt, 2021).

This caution has motivated methods that go beyond PMa-only fitting. The calibrated G23H catalog combines HGCA, cross-calibrated Gaia DR2 and DR3 proper motions, DR3–DR2 scaled position differences, Hipparcos intermediate astrometric data, Gaia astrometric excess noise, and Gaia radial-velocity variability constraints into a joint likelihood implemented in Octofitter. The resulting framework recovered all $24.75$8 tested ORB6 stellar binaries at high significance, found independent evidence for $24.75$9 of $$\boldsymbol \mu_{\rm HG} = \frac{\boldsymbol r_{\rm G} - \boldsymbol r_{\rm H}}{24.75},$$0 tested Jovian systems, and showed that Gaia-only curvature within the mission baseline can break degeneracies inherent to proper-motion-anomaly or excess-noise modeling alone (Thompson et al., 30 Jan 2026).

A related line of work combines RV data with PMa and explicitly examines the sensitivity curve and observing-window spikes. In that framework, some supposed brown dwarfs or low-mass stars were reclassified as likely planets, while other systems remained in the brown-dwarf or stellar regimes; the authors stressed the impact of smearing, sensitivity spikes, and multi-companion contamination on inferred true masses (Piccinini et al., 14 Jan 2026).

The broader historical trajectory begins with HTPM and the joint-solution program, where Hipparcos information was to be folded directly into AGIS rather than simply differenced against Gaia catalogs. That program already anticipated that Hipparcos–Gaia baselines would improve proper motions well before Gaia-only solutions matured and would enable detection of long-period binary and exoplanetary candidates (Michalik et al., 2014). The later PMa catalogs, HGCA calibrations, and composite Bayesian orbit-fitting frameworks collectively suggest that Hipparcos–Gaia proper-motion anomalies are most powerful when treated as one component of a multi-timescale astrometric inference problem, with explicit modeling of catalog systematics, perspective effects, and finite-window averaging rather than as a raw difference between two catalog proper motions.