The increased use in recent months of spectral-fivepoints has revealed deficiencies (errors ?) in the coordinates of some of the sources designated for this purpose. The first couple of sources for which this happened were treated as special cases (eg CIT6, V636Mon, V370Aur), but, recognizing eventually that a systemic problem may exist, it is worthwhile examining the entire dataset for its overall accuracy as well as its specific problems.

There are 97 sources in this section of the catalog. I adopted the SIMBAD coordinates in preparing the catalog in 1999, and it was unnerving to discover that these coordinates may not all be accurate to an arcsecond or so. The sources are of 2 main types, each with their own potential sources of positional error :

• Evolved stars, usually faint, red, dust-enshrouded, often masers, and relatively nearby. Inaccuracies in their J2000 positions may be caused by poorly-known proper motions.
• Compact molecular clouds, often not compact enough to have the same location - as defined by peak flux - at each wavelength.

The plots below show the differences in the RA & DEC coordinates for all 97 data (each box is 10" square) :

and for each accuracy grading :

grade 1 (32/97)	grade 2 (41/97)	grade 3 (24/97)

Notes :

• Our listed declination for (AF)GL5102 was in error by 20 arcminutes ! and will be changed following these comparisons.
• The coordinates for IRC+00365 and (AF)GL2316 (both Loup quality 2) differ by approx. 30" and 10" respectively, and are not shown on the relevant figure above (centre).

With a handful of exceptions the grade 1 objects identified by Loup et al have coordinates that agree with those listed by SIMBAD to better than about 1.5" vector - as the Loup grading might suggest. Of course, this would be totally expected if the same source is used by both SIMBAD and Loup et al for any object ! Twenty-two of the 32 grade-1 coordinates listed by Loup et al derive from (different) single original sources. This agreement is worse for 'grade 2' coordinates - especially when the two deviants above are included - but, surprisingly, rather binomial for the 'grade 3' coordinates - with the majority agreeing very well and only a minority showing very large disagreement. This tends to support the theory that Loup and SIMBAD access the same original source material.

It is is not possible, as was initially surmised, to choose between the totality of coordinates provided by SIMBAD or Loup et al. However, with use of spectral-fivepoints revealing problems with the SIMBAD coordinates, and with the larger, obvious, errors supporting the original Loup coordinates, I feel it is worth re-establishing the totality of Loup et al (1993) as the basis for this section of our catalog. I make no reference to Loup's quality codes, since the above analysis sheds no additional light on the intrinsic accuracy of any coordinate pair, although such a coding in such a catalog would be of great utility.

This was accomplished on 04 January 2002.