[vsnet-chat 2606] 'F Type' Carbon Stars

[JG <jgts@jgws.totalserve.co.uk>]

From John Greaves, UK

Recently, Mati Morel noted that SAO 198860 and SAO 251015, two Carbon
Stars, have F type spectra in the SAO.

One fact we've been forgetting, of course, is that there is already a
group of well known Carbon Stars formerly classified as ranging from
around F0 to K0, before Morgan et al introduced the C system.  This
group is of course the R CrB stars, now often quoted with a C
classification.  RY Sgr, for instance, is now C(1,0) in the GCVS4,
whereas it's G0p in the HD (and consequently, also in the SAO).

A relatively chewy but not too difficult introduction into spectra, for
those who may have wondered over the years as to where it all comes
from, exists in the form of "Classification of the Stars" by Jaschek &
Jaschek, Cambridge University Press, 1987/1990, and relatively
inexpensive in paperback [in US English paperback = softcover].

In the section on Carbon Stars therein, comparison trace spectra are
given for XX Cam, RY SGR, R CrB (3 R CrB stars) and the F8 star delta
CMa.  Given the low dispersion of objective prism spectra, and the
probability that the original measures for SAOs 198860 and 251015 were
taken on red insensitive plates, these traces are interesting, for if
the red regions of these 4 spectra were "ignored", ie if they were
invisible, these 4 stars' spectra would be hard to distinguish,
especially with respect to their Calcium H and K lines (somewhat
diagnostic of F stars).

These 4 stars are supergiants, however, and with the Carbon Stars SAO
198860 and SAO 251015 we are probably talking normal giants, and further
there is nothing at all to indicate that they any kind of R CrB or
similar star.

One possible way to test whether these sorts of stars are being
classified F due to incomplete spectra, or due to misidentification with
a nearby field star is to cross compare all known Carbon Stars with
their SAO spectra.  Spectrum F - K is pretty broad, but R CrB stars tend
to cluster mid-group (ie late F and early G) in the old style
classification, so if a small group of Carbon Stars did crop up with
spectra in the F range, the classification would not be that incongruous
(from the above), and more a consequence of the low dispersion and
equipment used etc.

If 'erroneous' SAO spectra for the Carbon Stars had a broader spread,
however, misidentifications would be favoured, especially if the odd A
star crept into the SAO derived spectra!

This ties in to a request Mati Morel made for a cross-check between the
SAO and Carbon Stars.  Stephenson's 2nd catalogue, the CGCS [Cool
Galactic Crbon Stars: of 1989 vintage] was actually compiled with type
cross checks made upon good quality objective prism plates that were
available to that author, so it is the best source for stars highly
likely (_and known_) to be carbon stars, and the number of stars of
reasonable positional precision within it is about 5600.  There are over
a quarter of a million objects in the SAO.  There is no cross reference
between the two in existence to my knowledge, so cross linking will have
to be done via positional data.

Despite the speed of modern home computers, even simple arithmetic tests
take a lot of time when 5600+ objects have to be checked one 250000+
times!  However, I gave it a bash: I converted the B1900 CGCS positions
into B1950, imported it and a copy of the SAO into a relational
database, and set up cross indexing of the precessed B1950 CGCS co-ords
against the catalogue B1950 SAO co-ords, and left it to run overnight.

As expected, by next morning the program had barely made a dent in the
situation!

So I crashed the program, and later tried again with a subset from both
catalogues that _only_ lay within a complete band of RA between
Declinations of -25 and -40 degrees.  This amounts to 500+ objects in
the CGCS and roughly 32000 in the SAO, so only took about an hour to
complete.

These 500x32000 checks result in about only a dozen resultant candidates
that can be unamiguously identified with each other [there were three
times or so that many near misses that had to be rejected, as they were
cases where two or more stars lie equally adjacent to the Carbon Star
position, thus making cross identification difficult.  These stars were
not known variables, so they were not used, as no other criteria for
identification were available.  The remaining stars were highly
coincident in position.]

Most of the dozen or so have appropriate SAO spectra, eg type Nb or R8,
but to some extent this is no great surprise, as these stars were known
to be variable prior to the compilation of the SAO (eg R Scl).  One or
two had no SAO spectrum listed, which left the following 3 CGCS objects
with somewhat incongruous SAO spectra:

CGCS RA     DEC     SAO num   V       SPECTRUM
  777   4  43  -31  37  195274       10.2   G5p
  798   4  47  -31  35  195328         9.8   M
2086   8    7  -36    8  198860         8.6   F5

Where CGCS 2086 = SAO 198860 is FK Pup!  Actually, the positional
agreement between the CGCS and SAO positions was not as good for FK Pup
as the excellent ones for CGCS 777 and 798 were, so it's identification
is aided by prior knowledge.

CGCS 798 is given type M, with no a,b,c or d subclass, and could be said
to be an acceptable mistake for a possible N type star.

So, the only good candidate is CGCS 777, with G5p spectrum in the SAO.
Not quite F, and the CGCS carries no C spectrum, but Stephenson felt it
was a Carbon Star in his critical compilation, and it is somewhat
analogous to the RY Sgr situation noted above.

But, as usual, the data are glorioiusly inconclusive, either one way or
the other.  A 'check' test on a Northern band between arbitrarily chosen
+40 and +50 degrees declination showed no abbarent SAO spectra for the
smattering of Carbon Stars cross identified.


The upshot?  Well, there are two to three instances of Carbon Stars
having spectra such as F0, F5 and G5p in earlier studies/spectroscopic
data.  There is not necessarily any need to mark these as errors,
however, nor even peculiar in any way, for -

There is already a precedent of Carbon Stars with spectral types in this
range.


And there, you'll be glad to hear, I shall leave it, and return VS-Chat
back to the gentle mercies of the CV junkies!


Cheers

John

John Greaves, UK


PS: I would be interested to hear if anybody knows any simple
algorithmic short cuts [ie ways of looking at the problem, rather than
hardwire computer techniques] when it comes to cross comparing large
numbers of objects.  I can think of none.  One way crude way seems to be
via chopping up the data into groups of smaller range, as using
conditionals/booleans seems to be non-linear!  Taking a 15 degree swath
of relatively populous sky took just over an hour to process, whereas
extrapolating the progress of the overnight run suggested it'd take nigh
on a week!

VSNET Home Page

Return to Daisaku Nogami

---

vsnet-adm@kusastro.kyoto-u.ac.jp