The William Herschel Double Star Catalogs Restored

Bruce MacEvoy


Charles Messier is still honored by name through the continued use of his catalog numbering for deep sky objects that in many cases he did not discover. Yet the groundbreaking observational work of William Herschel in both deep sky and double star astronomy has been almost entirely obliterated through the incorporation of his discoveries into the catalogs and publications of later observers, and under their names.

This distortion of the historical record has been remedied in part by the recent popularization of the "Herschel 400" list of deep sky objects, which is an observing checklist drawn from the New General Catalog that developed from Herschel's 2446 deep sky discoveries. But nothing comparable has been available to celebrate Herschel's equally remarkable achievement in double star astronomy.

One of the great unanswered questions in 18th century astronomy was also one of the most basic: how far away are the stars? Herschel's double star survey had as its goal the measurement of interstellar distances using a method first suggested by Galileo Galilei. If two stars could be found lying very close to the same line of sight from earth, but very distant from each other in space, the distance between them might appear to shift back and forth as the Earth orbited around the sun. This parallax could be used to triangulate the distance to the stars.

In the event, stellar distances were established by careful positional measurements of single nearby stars. But Herschel's observations allowed him to prove the unanticipated fact that many double stars actually orbit each other, demonstrating for the first time the generality of Newton's Laws of Motion and Universal Gravitation outside the solar system.

My interest to restore credit for Herschel's double star discoveries motivated me to obtain his three double star catalogs — the Catalogue of Double Stars (1782) with 269 entries, Catalogue of Double Stars (1784) with 484 entries, and On the Places of 145 New Double Stars (1821) — and compile from them the Herschel 500 list of double stars which he was the first to observe and measure, or discovered independently of other observers.

Based on that work, I have now restored all of Herschel's discoveries, so far as these can be identified from his catalog entries with modern reference materials.

My thanks to P.J. Anway for providing me with a pdf copy of Dreyer's edition of the catalogs, and to William Hartkopf of the USNO for helpfully answering many queries.

Reconstructing the Lists

I relied on the edition of Herschel's catalogs found in The Scientific Papers of William Herschel (1912), edited by J.L. Dreyer. These were the source for Herschel's original measurements and observing notes.

My source for current double star information was the November, 2010 edition of the Washington Double Star Catalog (WDS) compiled and maintained by the United States Naval Observatory (USNO) in Washington, DC. The WDS provides the now standard star catalog and component labels, current celestial coordinates and positional and magnitude measurements.

I began in every case with Herschel's given location instructions, which in the first two catalogs (1782/84) are either (1) labeled star identifications or (2) star hopping directives from an easily identified reference star. In the 1812 addendum, these are supplemented by measured right ascension and polar declination for most (but not all) entries.

Labeled stars are almost always identified by the system of Bayer Greek letters and/or Flamsteed numbers — beta Cygni or 61 Cygni — which are still widely used by astronomers today. These labeled stars comprise roughly 45% of Herschel's discoveries, and provide highly reliable star identifications.

However, these had to be identified in WDS by means of modern double star catalog designations. I first consulted the Cambridge Double Star Atlas (CDSA, 2009) by James Mullaney and Wil Tirion. Where that reference did not suffice, I referred to the Millennium Star Atlas (MSA, 1997) by Roger Sinnott and Michael A.C. Perryman for precise celestial coordinates, and used those to identify the system in WDS. Either way, I compared Herschel's catalog description to the star magnitude and "first" positional measurements in the WDS (which typically date from Herschel's era) to confirm that they matched within reasonable margins of error.

A second group of stars were unlabeled stars located "near" a Bayer/Flamsteed reference star. Again, I first looked in the CDSA for the labeled object, then confirmed it with reference to the MSA and WDS; or I used the MSA or Stellarium as a surrogate low power telescope to identify the star and determine its coordinates, identify the system in WDS, and confirm it against Herschel's description. This group, comprising about 45% of all entries, are in general very high probability identifications.

A third group of stars, roughly 10% of the total, had to be identified by methods that were in some way problematic or conjectural:

1. The reference star was a Flamsteed star whose number was not reproduced in CDSA or MSA. In these cases I referred to the online copy of Francis Baily's edition of Flamsteed's British Catalogue of Stars (1835), then precessed the 1690 equatorial coordinates given there to epoch J2000. Baily was also my source to identify stars with lowercase letter designations (e.g., f Orionis).

2. The identifying information was iconographic, referring to the mythological constellation figures found in Flamsteed's Atlas Cœlestis. (For example: Multiple. It is a spot over the right fore-foot [of Monoceros]; 4 or 5 small stars within one minute.) While most of these references are ceremonial and superfluous, in some cases they are all we have to identify the star in question.

I referred to the charts in Jean Fortin's reprint of Flamsteed's original atlas, published in London in 1776 and available online at the Linda Hall Library of Science, Engineering & Technology. (Fortin's Atlas Cœlestis very closely follows the imagery of Flamsteed's original published in 1726 and was very likely the edition used by Herschel.) I then used CDSA and MSA, as before, to identify the star.

3. The identifying information used Hevelius or Harris labels, or was baldly nondescriptive and undirective (for example: H V 28: Double. It is a star near beta Cephei.). In these cases I resorted to a brute force search: first sorting the entire WDS catalog to identify all double stars within a specific area of sky in relation to the reference star (or, in three cases, within the boundaries of an entire constellation). I then sorted the selected entries by position angle, separation and magnitude, as appropriate, to limit the search to those stars matching Herschel's measurements or description.

4. The reference star was one of Flamsteed's many "missing" stars, as identified either in Dreyer's or Baily's notes or in Morton Wagman's article "Flamsteed's Missing Stars" (2000). In these cases I proceeded as described in (3), above.

5. Herschel's identifying information was incorrect. In a handful of cases, an obvious error has occurred in Herschel's notes (such as "north following" for "south preceding"). Where a simple correction provided an identification that matched Herschel's measurements for the star, I considered it a valid match and note the correction.

6. The identifying information in either the CDSA or WDS was incorrect. I was disappointed by the number of labeling errors that appear in the CDSA — STF 1066 appears as STF 1006, STF 974 appears as STF 947, and so on. In those cases, I proceeded as if the star had no label. The WDS also mislabeled several Herschel stars: the WDS star "H 6 68" in fact referred to H IV 68; Herschel's H III 30 is labeled STF 2390 in WDS, but his H VI 30 is labeled H 3 30! And the Herschel double H VI 108, clearly identified as a double star in both CDSA and MSA, is not found in WDS at all: positional information was calculated using Stellarium. (The WDS has since corrected these few errors.)

When an identification was made based on any of these conjectural or complex methods, or the identification does not match Herschel's description of the system, I list it as a "doubtful attribution".

Where there was really no clear indication of the star that Herschel intended, or no plausible match could be found using his instructions, I list the star as "unidentified".

The identifications in this 10% of stars can only be considered probable. As W.H. van den Bos cautions in his article "Double Stars that Vex the Observer", "two pairs of stars may occur near together in the sky, whose descriptions resemble each other so closely that [an observer] is measuring one pair when he is really measuring the other." Herschel himself adds: "It will sometimes happen, that other stars are very near those which are thus pointed out, that might be mistaken for them. In such cases an additional precaution has been used by mentioning some circumstance either of magnitude or situation, to distinguish the intended star from the rest." But as the examples above demonstrate, Herschel was not rigorous in keeping to his method.

Peculiarities of Herschel's Catalogs

Speaking of method, several peculiarities appear in the Herschel catalogs that reflect instrument limitations and reliance on primitive, soon to be obsolete observing methods. These deserve comment for their historical interest.

Star Measurements – Herschel's measurements of position angle and separation are based variously on visual estimates without a reference, visual estimates based on the apparent diameter of the stars as viewed at different magnifications, visual judgments in comparison to markings on a wall, drift times, or use of two or more different micrometers of his own invention and construction.

Herschel was aware that the telescopic disk images of stars were "in large part" spurious (a diffraction artifact called the Airy disk), a fact he deduced through repeated and careful measurements that showed the disk actually becomes smaller with high magnification (noticeable at exit pupils below ~0.3 mm). However, he erroneously believed that in some cases the actual body of the star might be observed at high magnifications. He seems unaware that the Airy disk becomes larger when a telescope is stopped down with an aperture mask — something that astronomers of Herschel's time did frequently to minimize the effects of poor seeing and worse optics.

A detailed comparison of Herschel's separation estimates with the more precise and reliable measurements made by Struve and others after 1820 suggests that Herschel's metrics are untrustworthy for stars in categories I through III, and for that reason I have omitted them from my transcription of Herschel's notes. His measurements in categories IV through VI tend to be more reliable, and for that reason are included — but enclosed in brackets, to indicate they have been converted to modern units (e.g., 30° north preceding becomes PA = 330°). However, Herschel was primarily interested in very close pairs, for purposes of estimating stellar proper motions and parallax distances: his measurements become increasingly slipshod in the wider classes of 1782/84, and are often omitted in the 1821 entries when he had moved on to galactic investigations and the search for additional planets.

Star Magnitudes – Herschel does not record estimates of a star's visual magnitude, except to refer to 5th to 9th magnitude stars (below the naked eye limit or omitted from the Flamsteed inventory) as "telescopic stars". Instead, he describes the relative difference in magnitude between the binary primary and companion stars. He uses a total of ten descriptive labels for this purpose, and this large number of categories, combined with the difficulty of estimating relative magnitude proportions or differences across different brightness levels, suggests that his judgments will be inaccurate.

To evaluate this issue, the chart (below) shows the actual magnitude difference between 714 pairs of stars, as measured in WDS, within each of Herschel's magnitude difference categories.

click on image for larger format

Although the average magnitude difference within each category (orange dots) increases systematically (from an average difference of 0.65 for "a little unequal" to over 4.9 for "excessively unequal"), there is clearly substantial variance within categories and overlap between categories. (Comparable inaccuracies appear if the differences are expressed as a ratio on the primary star magnitude.) Even assuming that there may be errors in the WDS database or in matching components of multiple star systems, Herschel's magnitude difference estimates are not reliably useful to identify his double stars.

Star Colors – The most intriguing peculiarity in Herschel's observing notes concerns his star color descriptions. He never refers to a star color as "yellow", he sees all stars as shades of red or blue (the primary of beta Cygni is "pale red"), he frequently describes red stars as "dark", and he mentions elsewhere a confusion between green and blue in the judgment of colors. These points all suggest to me that he may have had protanopia (a form of red/green colorblindness) — and this would have affected both his magnitude and color estimates.

Star Locations – Herschel began his work in an era when celestial coordinates were poorly standardized and difficult to use with his altazimuth reflecting telescopes (the most common telescope objective and mounting before the 19th century). Astronomers had to copy by hand the Bayer/Flamsteed designations of stars into the ornamental star atlases available; celestial coordinates were only printed along the margins, not shown as curved coordinate lines across the charts, making estimates of star locations unreliable.

For these reasons, in his first two catalogs Herschel relied on "star hopping" directions, the observational method of using stars as stepping stones or directional guides to arrive at a specific sky location starting from an easily identified reference star. In some cases these instructions are quite intricate:

In pectore Lyncis [In the chest of the Lynx]. Double. Not easy to be found. A line from the 19th Lyncis to upsilon Geminorum crossed by one from theta Ursae majoris to epsilon Aurigae, points out a star but just visible in a fine evening; it is perhaps about three degrees from the 19th Lyncis; then that star is found, we have the double star about 1 degree n. following the same, in a line parallel to tau Geminorum and the 19th Lyncis. (H I 76)

His 1821 catalog combines star hopping directions, celestial coordinates in right ascension and polar declination, and celestial distances measured as drift times or micrometer measurements from stars with catalogued positions. Ironically, this catalog contains the highest number of duplicate and unidentified entries — by then Herschel was searching for new planets and studying nebulae and Galaxy structure, and double stars were of residual interest as he chanced upon them.

Notation Errors – Most double star entries in the catalogs are carefully described and entirely reliable. But other entries are clearly hasty, no more than jottings in the dark or comments made from memory.

Perhaps for that reason, there are 21 duplicate "discoveries" across the three catalogs. (I was able to identify more than had been noticed previously, thanks to modern spreadsheet tools.) And there are 22 stars notated so confusingly or cryptically that my efforts failed to identify them — indeed, the 1821 catalog documents several instances where Herschel himself returned to look for a star but could not find it.

These cumulative inadequacies and anachronisms may explain why Herschel's groundbreaking efforts were superseded within a few decades by the double star catalogs of John Herschel (HJ), James South and Herschel (SHJ), Friedrich Wilhelm Struve (STF, STFA), and others across the 19th century.

The Reconstructed Catalogs

Based on my research, Herschel's original tally of 848 double star entries breaks down as follows:

Class I ["most difficult", 97 total]: 96 identified, 1 unidentified, 0 duplicate.
Class II [less than 5" separation, 102 total]: 102 identified, 0 unidentified, 0 duplicate.
Class III [5" to 15" separation, 114 total]: 107 identified, 3 unidentified, 4 duplicate.
Class IV [15" to 30" separation, 132 total]: 128 identified, 1 unidentified, 3 duplicate.
Class V [30" to 60" separation, 137 total]: 131 identified, 3 unidentified, 3 duplicate.
Class VI [greater than 60" separation, 121 total]: 112 identified, 6 unidentified, 3 duplicate.

Catalogue N [all classes, 145 total]: 129 identified, 8 unidentified, 8 duplicate.

This brings Herschel's confirmed double star entries to 805, or more than 97% of the 827 total (excluding duplicate entries). All are available, with annotations, in the files linked below.

The WDS web site now provides a William Herschel Cross Reference that lists my Herschel identifications with their WDS and modern catalog labels and Bayer/Flamsteed designations.

Web Pages

The Complete William Herschel Double Star Catalog - The entire Herschel catalog of double stars, sorted by right ascension, with full information and notes on each system. (Version of 5.1.2011; 575Kb .html file)

The Herschel 500 - An observational checklist of 500 Herschel double star systems, drawn from his 1782 and 1784 double star catalogs: with star names, modern catalog designations, current celestial coordinates and positional measurements.

The Herschel Checklist - A compact observational checklist of all confirmed Herschel double stars.


William Herschel's Double Star Discoveries - His 805 confirmed discoveries displayed as an all sky star chart.

Spreadsheet Files

The Complete William Herschel Double Star Catalog - The combined Herschel catalogs in one file, sorted by right ascension. (Version of 5.1.2011, 657kb .xlxs file)

The Herschel 500 - (191kb .xlxs file)


All contents ©2011 Bruce MacEvoy.