THE NEW NEARBY STAR SO025300.5+165258

The new nearby star SO025300.5+165258 was recently reported to have a
preliminary trigonometric parallax of 0.43 +/- 0.13 arcsec by
Teegarden et al. 2003.  This parallax would make the star the third
nearest system to the Sun, at a distance of 2.3 (+1.0 / -0.5 pc, one
sigma error). Teegarden et al. also estimate the distance to be 3.6
+/- 0.4 pc based upon a spectral type of M6.5V and a J magnitude of
8.39.  This distance would rank it 17th among the nearest star
systems.

Our own analysis of the object, based upon the BVRI photometry (and
spectroscopy, below) provided in their paper, yields a rather puzzling
set of distance estimates, most of which are significantly larger than
the parallax reported.  For this analysis, we assume that the reported
BVRI magnitudes determined using "broad band UBVRI Bessell filters"
are on a very similar system (if not identical to) as the Johnson BV
and Cousins RI magnitudes determined by Bessell in many published
papers in the 1990s.  A distinct distance estimate is generated for
each two-filter combination and compared to the object's V magnitude.
The colors and magnitudes are compared to stars with high quality
parallaxes within 10 pc using our RECONS grid of available astrometry
and photometry, and the following distances are derived:

    B-V    16.7 pc
    B-R     6.2 pc
    B-I     3.8 pc
    V-R     4.0 pc
    V-I     2.1 pc
    R-I     2.0 pc

MEAN DISTANCE ESTIMATE FROM BVRI:    5.8 pc   64th nearest system

MEAN DISTANCE ESTIMATE FROM VRI:     2.7 pc    7th nearest system

The distance estimates for BVRI are highly inconsistent with one
another.  The problem lies not in the fits, but can be explained if
(1) the object is not a normal M dwarf, (2) the photometry is not in
the standard UBVRI system, or (3) there are errors in the photometry
reported. Eight additional stars within 10 parsecs having spectral
types M6.0V to M7.0V and BVRI photometry (14 individual sets of
photometry) have been run through the same distance estimate code.
None of the eight stars has such a large disparity in the results, and
the final estimated distances are more than 10% different from the
true distances in only 3 cases of 14 (11.0%, 13.0% and 25.0%).

Running the same code again without the B magnitudes (usually the
least reliable distance estimates include the B magnitude) results in
even better agreement between estimates and reality --- the 11.0%
error set drops to 5.7%, the 13.0% set remains the same, and the 25.0%
set drops to 11.9%.  The average error from the VRI code for all 14
sets of data run on the 8 stars with known distance is 5.1%.  We
therefore adopt the VRI distance as the more reliable of the two
photometric estimates.

In addition, we can use RECONS methodology (Henry et al. 1994, 2002)
to estimate a distance from the spectral type and the V magnitude. The
errors in such distance estimates are at best 20%, and are usually
quite a bit larger (40%).  We find the following distances depending
on the spectral type assigned (typically, spectral type assignments
have an error of +/- 0.5 subtypes):

    M6.0V   7.0 pc
    M6.5V   4.6 pc
    M7.0V   2.9 pc

DISTANCE ESTIMATE FROM SPECTUM:      4.6 pc  38th nearest system

BOTTOM LINE: Nine different estimates of the distance yield distances
with a very large range.  The BVRI photometry values do not yield
consistent answers in this case, although they are quite reliable for
stars of similar spectral type.  Eliminating the B magnitude helps,
but there is still a factor of two in the distance estimates from VRI
alone.  Our most reliable estimate, 2.7 pc, is consistent with the
preliminary parallax reported but is less reliable than usual for
stars of this type.  Obviously, only a high quality trigonometric
parallax will determine a reliable distance to the new object, but it
is likely to lie between 2 and 5 pc.  Its rank in proximity changes
radically within this range:

        2.0 pc            3rd nearest system
        3.0 pc            8th nearest system
        4.0 pc           25th nearest system
        5.0 pc           46th nearest system

NOTE: This star has not yet been added to the RECONS list of nearby
stars because its trigonometric parallax does not meet the RECONS
criterion that the error be less than 20 milliarcseconds.  The current
error is 130 milliarcseconds.