juliensimon/wds-double-stars

---
license: cc-by-4.0
pretty_name: "Washington Double Star Catalog"
language:
- en
description: "The Washington Double Star Catalog (WDS) is the world reference catalog for visual double and multiple star systems, maintained by the US Naval Observatory. Double stars are essential for determining"
task_categories:
- tabular-classification
- tabular-regression
tags:
- space
- double-star
- binary
- wds
- usno
- astrometry
- astronomy
- open-data
- tabular-data
- parquet
size_categories:
- 100K<n<1M
configs:
- config_name: default
data_files:
- split: train
path: data/wds_double_stars.parquet
default: true
---

# Washington Double Star Catalog

<div align="center">
<img src="banner.jpg" alt="A youthful globular star cluster observed by the Hubble Space Telescope" width="400">
<p><em>Credit: NASA/ESA/Hubble</em></p>
</div>

*Part of a [dataset collection](https://huggingface.co/collections/juliensimon/astronomy-datasets-69c24caf2f17e36128946743) on Hugging Face.*

## Dataset description

The Washington Double Star Catalog (WDS) is the world reference catalog for visual double and multiple star systems, maintained by the US Naval Observatory.

Double stars are essential for determining stellar masses -- the most fundamental property of a star -- and for testing stellar evolution models. The WDS traces its lineage back to Sherburne Wesley Burnham's catalog of 1906 and has been continuously maintained at the US Naval Observatory for over a century, incorporating astrometric measurements from visual micrometry, speckle interferometry, adaptive optics, long-baseline optical interferometry, and space-based observations (Hipparcos, Gaia).

The catalog includes both gravitationally bound physical pairs (true binaries) and optical doubles -- chance alignments of unrelated stars along the same line of sight. Distinguishing between the two requires common proper motion analysis or, ideally, measurement of orbital curvature over a sufficient arc of the orbit.

For physical binaries with well-determined orbits, the combination of angular separation, orbital period, and parallax yields dynamical masses through Kepler's third law. These direct mass measurements are the gold standard for calibrating the mass-luminosity relation and testing stellar structure models across spectral types from O-type supergiants to late M-dwarfs. The position angle and separation measurements recorded at the first and last epochs encode information about orbital motion: systems showing significant changes in these quantities over the observing baseline are strong candidates for orbit determination.

The WDS encompasses an enormous diversity of systems, from pairs separated by fractions of an arcsecond -- resolvable only by interferometric techniques -- to wide common proper motion companions separated by arcminutes or more. Wide binaries (separations > 1000 AU) are particularly interesting as probes of the Galactic gravitational potential.

This dataset is suitable for **tabular classification, tabular regression** tasks.

## Schema

| Column | Type | Description | Sample | Null % |
|--------|------|-------------|--------|--------|
| `wds_id` | str | WDS designation in format 'HHMMM+DDMMM' encoding the J2000 position (e.g. '00055+5258'); the standard identifier for double stars in this catalog | 00000+0044 | 0.0% |
| `discoverer_code` | str | Standard WDS discoverer code + sequence number (e.g. 'STF2272' = Struve discovery #2272); identifies the original survey or observer | SKF1115 | 0.0% |
| `components` | str | Component pair designation (e.g. 'AB' = primary+secondary, 'AC' = primary+tertiary); systems may have multiple entries for different pairs | AB | 73.0% |
| `first_observation_year` | int64 | Year of the earliest published observation of this pair (fractional year, e.g. 1889.0) | 2001 | 0.0% |
| `last_observation_year` | int64 | Year of the most recent published observation (fractional year); span indicates how long the pair has been monitored | 2015 | 0.0% |
| `n_observations` | int64 | Total number of published position-angle/separation measurements for this pair | 3 | 0.0% |
| `position_angle_first` | int64 | Position angle of secondary relative to primary at epoch of first observation (degrees, 0-359, measured East from North); change over time may reveal orbital motion | 33 | 0.0% |
| `position_angle_last` | int64 | Position angle at epoch of last observation (degrees, 0-359); compare with position_angle_first to detect orbital motion | 30 | 0.0% |
| `separation_first_arcsec` | float64 | Angular separation between primary and secondary at first observation (arcseconds); decreasing separation may indicate an inclined orbit | 0.8 | 0.0% |
| `separation_last_arcsec` | float64 | Angular separation at last observation (arcseconds); null if only one observation exists | 1.0 | 0.0% |
| `magnitude_primary` | float64 | Visual (V-band) magnitude of the primary star; null if not measured | 17.8 | 0.1% |
| `magnitude_secondary` | float64 | Visual (V-band) magnitude of the secondary star; null if not measured; difference = magnitude contrast | 17.8 | 1.1% |
| `spectral_type` | str | MK spectral type of the primary component (e.g. 'G2V', 'K0III'); null if not cataloged | M1 | 53.1% |
| `ra_deg` | float64 | Right ascension of the primary star, ICRS J2000.0 (degrees, 0-360) | 359.999375 | 0.0% |
| `dec_deg` | float64 | Declination of the primary star, ICRS J2000.0 (degrees, -90 to +90) | 0.7257499999999999 | 0.0% |

## Quick stats

- **157,371** double/multiple star systems
- Observations spanning **-1** to **2025**
- **157,371** with measured separation
- **73,828** with spectral type

## Usage

```python
from datasets import load_dataset

ds = load_dataset("juliensimon/wds-double-stars", split="train")
df = ds.to_pandas()
```

```python
from datasets import load_dataset

ds = load_dataset("juliensimon/wds-double-stars", split="train")
df = ds.to_pandas()

# Systems with large separation change (orbital motion)
has_both = df.dropna(subset=["separation_first_arcsec", "separation_last_arcsec"])
has_both["sep_change"] = abs(has_both["separation_last_arcsec"] - has_both["separation_first_arcsec"])
movers = has_both.nlargest(20, "sep_change")
print(movers[["wds_id", "separation_first_arcsec", "separation_last_arcsec", "sep_change"]])

# Separation distribution
import matplotlib.pyplot as plt
df["separation_last_arcsec"].dropna().clip(upper=100).hist(bins=100)
plt.xlabel("Separation (arcsec)")
plt.ylabel("Count")
plt.title("WDS Angular Separation Distribution")
plt.show()
```

## Data source

https://vizier.cds.unistra.fr/viz-bin/VizieR-3?-source=B/wds/wds

## Related datasets

- [juliensimon/hipparcos-catalog](https://huggingface.co/datasets/juliensimon/hipparcos-catalog)

- [juliensimon/gcvs-variable-stars](https://huggingface.co/datasets/juliensimon/gcvs-variable-stars)

- [juliensimon/open-star-clusters](https://huggingface.co/datasets/juliensimon/open-star-clusters)

> If you find this dataset useful, please consider [giving it a like](https://huggingface.co/datasets/juliensimon/wds-double-stars) on Hugging Face. It helps others discover it.

## About the author

Created by [Julien Simon](https://julien.org) — AI Operating Partner at Fortino Capital. Part of the [Space Datasets](https://julien.org/datasets) collection.

## Citation

```bibtex
@dataset{wds_double_stars,
title = {Washington Double Star Catalog},
author = {juliensimon},
year = {2026},
url = {https://huggingface.co/datasets/juliensimon/wds-double-stars},
publisher = {Hugging Face}
}
```

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)