Celestia Forums

Starnames.dat has a couple of errors in connecting Hip catalogue numbers to star names for some of the Sun's very near neighbours.

1475 should be Groombridge 34, not Lacaille 9352
114046 should be Lacaille 9352, not Ross 128
57548 should be Ross 128 (at present unnamed)

Grant

Thanks, Grant. I'll correct those ASAP.

--Chris

Chris:
I'm just adding a few other familiar (well, to those interested in nearby star systems!) names to starnames.dat, so as to get rid of the HIP and HD numbers that appear in the Star Browser for near space. I can commit the whole lot to the CVS tree when I'm done, if you like.

Grant

I'll let you take care of it, then. It will be nice to see some more familiar names in our stellar neighborhood.

--Chris

But I've just run into a problem in stars.dat, too. The star HD 213209, which Celestia shows at 13.070ly, is actually much further away (that spectral class of B9 is a giveaway!) http://vizier.u-strasbg.fr/viz-bin/VizieR-S?HIP%20110923 gives its parallax as just 1.17mas, rather than the whopping 249.52 listed in stars.dat.

Grant

And another. HD 20936, KIII, http://vizier.u-strasbg.fr/viz-bin/VizieR-S?HIP%2015690, with a parallax of just 2.59mas - it appears in Celestia at 14.338ly.

Grant

That's sort of disturbing. I'll have to look at the HIPPARCOS catalog and the star database builder to figure out what the source of these errors is. My guess is that's it's a HIPPARCOS glitch . . .

--Chris

Quite serious problems with stars.dat - I've run into five or six parallax errors within 25 ly of Earth. Several stars very much too close, and some too far away - most notably Achird, which is 1400 ly away, instead of 19.

Grant

I figured out what is happening with HD 20936 . . .

In star database creation, the parallaxes of stars identified as components of multiple star systems are set be equal. Without this processing, hardly any components of binary or trinary systems would be close to each other due to the significant uncertainties in the parallax measure ments. HD 20936 is HIP 15690. It shares a CCDM identifier with HIP 15689, indicating that these stars together form a binary system. However, the parallax measurements for these stars indicate otherwise: HIP 15689 has a parallax of 227.45mas, compared to 2.59mas for HIP 15690. I'm quite confused . . . either I'm misinterpreting the HIPPARCOS data, or there are some serious inconsistencies in the data set.

--Chris

Note that there's a comment assocated with HIP 15689, which says "Ambiguous double-star solution of HIP 15689 + 15690." So it seems somebody noticed the discrepancy. Maybe records with notes that include the word "Ambiguous" should be discarded?

(added later)
Well, apparently not. The star with the same CCDM identifier as HIP 110923 is HIP 110893 and that one does not have the "Ambiguous" comment.

I'm guessing that perhaps you used just the "goodness of fit" (field H30) to decide between the two distances. I fear that may be too simplistic a decision.

In both examples here, the component of the pair with the larger parallax (apparently the one that's wrong) also has a larger standard error (field H16) than the star with the smaller parallax , but that's probably proportional to the magnitude of the value. It may be a reasonable deciding factor, though, when H30 values are the same.

Also note that sometimes samples have been discarded (field H29), as with HIP 110893. That'd make me suspicious of comparable H30 values.

I hope these thoughts help a little.

Yep, that seems to be what's happening - 36 Oph A&B (Hip 84405) and 36 Oph C (Hip 84478) are way out at the back of beyond, apparently because they've been tied to Hip 84391. Achird has the same problem, though I haven't hunted for the offending star.
I've done some hand-editing of stars.dat to fix local space (out to 25ly), but obviously that's not an option for the vast bulk of stars.

Grant