Description and background for MacConnell IRAS spectral classifications version: 10 Jul 2005 18 Feb 2005 28 Jul 2004 About 10 years ago Jack MacConnell (now at STScI) made available a large file of spectral classifications for IRAS red stars along the southern Milky Way, some 14,000 objects. Essentially all these stars will be variable, though only a small portion of them are designated so far. You can have a look at the source file starting here: http://vizier.u-strasbg.fr/viz-bin/Cat?III/170 As you can see, it is a terse list, showing the IRAS name followed directly by the spectral type, and often other information as cryptic free-form comments. There are no accurate coordinates or magnitudes for the stars, nor any external linkage. To make this information useful, some work needs to be done on it. The principal value in the present list is to provide precise coordinates for the IRAS sources---often for the first time---along with external IDs in the visible and near-IR, and the spectral classifications. MacConnell occasionally classified companion red stars in the near fields of the IRAS sources. In addition, MacConnell has been very kind to re-examine many ambiguous cases on his plate collection, and has provided new or revised types for red stars. The variety of remarkable objects in this list is fascinating. By way of background, the IRAS spacecraft made the first complete sky survey in the mid-infrared about 1983. Among the products was a catalogue of some third of a million sources that seemed unresolved at the rather poor resolution of the detector system. Objects were detected as they passed across a bunch of slits placed over the detectors (the pattern of the slit-mask looked something like a cheese-grater). Thus the positions were determined from when the source passed over successive holes in the mask together with knowledge of the spacecraft pointing and spin motion. The resulting positions have an uncertainty expressed as an oval, typically about 40"-60" long and 10"-20" wide, often aligned approximately east-west. Dominating the detections close to the Milky Way are red stars. These appear bright due to circumstellar dust whose output peaks at around 10 microns wavelength. (The stars themselves peak at 1-2 microns.) One of the four filters on IRAS was a very wide passband centered near 12 microns that (very conveniently) takes in most of this 10-micron emission from the envelopes around cool stars. In most instances IRAS detected the stars in the 12-micron band, sometimes very weakly at 25-microns, and not at all at two longer wavelengths. This pattern is usually enough to separate the red stars from other things detected by IRAS. Now, Jack MacConnell has made a career of doing surveys on objective-prism plates taken mostly with the Curtis Schmidt at Cerro Tololo in Chile. Among these was a large series of plates covering the southern galactic plane in the photographic infrared, what we would now call the 'far-red', spanning most of the Cousins R and I photometric passbands. The broad molecular bands appearing in the spectra of cool stars makes them easy to identify on such plates. So while looking for stars of more immediate interest (carbon and the rare S-type stars), he also inventoried the ubiquitous M-giants appearing at the locations of IRAS sources. To some extent this helped answer the question (at the time) of what all these IRAS sources were that seemed to have no catalogued visible counterpart. (Answer: a plethora of cool late-M giants obscured by interstellar or circumstellar dust, or both.) With this as background, I've started to chip away at the 14,000-object list to make it more usable. The part of the file that I have worked over is copied out to the Lowell ftp area at: ftp://ftp.lowell.edu/pub/bas/starcats/mac.sample3 (main table) ftp://ftp.lowell.edu/pub/bas/starcats/mac.iras.notes (notes) ftp://ftp.lowell.edu/pub/bas/starcats/mac.txt (this file) ...which deals with the sources up to RA 8h00m among the equinox 1950 IRAS names. This is about 2900 objects (only 20 percent of the total), and extends to galactic longitude ~248 deg (about -30 Dec), starting from the Auriga/Gemini Milky Way in the north. All the stars are within about 7 deg of the galactic plane. The main table is 124 columns wide to accomodate the long MSX6C names and other basic information. The columns should be self-explanatory except 'r', which follows the spectral type. MacConnell uses this flag to indicate significant reddening (spectra strongly tapered toward the blue end), and I preserve this information here. The remarks column shows mainly a single commonly-used external ID, and longer comments are in the separate 'notes' file. I proceeded as follows. Taking the IRAS names in batches of 50 or 100, I first used VizieR to look for 'best' coordinates, which came mostly from UCAC2 or 2MASS. UCAC2 is the current state-of-the-art among astrometric star catalogues, and the positions are mostly better than 0".1 accuracy. The 2MASS catalogue resulted from a survey in the three near-IR photometric bands called J, H, and K (about 1.2, 1.6 and 2.2 microns). Red stars, even if very faint in the visible are booming bright, even overexposed in the K band. This is because the dimming caused by interstellar extinction is reduced by a factor of ten at K compared to the visual. The key element was to find the brightest 'very red' 2MASS source within or near the IRAS position error-ellipse. Recall that these are 95-percent confidence-intervals for the source position. I usually used a 45 arcsec search radius in VizieR. Note that an unreddened M0 giant has J-K ~1.0, so the stars always were very much redder in the 2MASS data and generally very much brighter than anything else in the field. Nearly always this search yielded a single unambiguous red star consistent with MacConnell's classification. If the star was not present in UCAC2 I usually reverted to the 2MASS coordinates (mostly good to < 0".2), but sometimes used other sources if the 2MASS detection was overexposed. Bill Gray's GSC-ACT still comes in handy. Where MacConnell found a 'blank field' or only faint non-banded stars I tried to ascertain why. I sought red stars in 2MASS well outside the nominal position error limits, and also looked in the MSX catalogue. MSX is another near/mid-infrared survey done as part of a military project to map the background sky at wavelengths relevant to detection of ballistic missiles ('Star Wars' stuff). It included several wavelength bands, but the only really sensitive one was centered at about 8-microns, again with a wide passband taking in that 10-micron glow from the cool stars as does IRAS. MSX is about a factor of five more sensitive than IRAS at this wavelength, and the coordinates for sources have errors of typically only a few arcsec (versus a few tens of arcsec for IRAS). As with the other catalogues mentioned, it is on-line and very easily searchable using the Strasbourg VizieR utility. In many instances the red star was simply outside the IRAS error-ellipse, and Jack said he kept pretty strictly within that boundary for the classifications. In other cases it was clear that the red star was simply too faint. I presumed the limit of his far-red plates was in the magnitude range 11 < Ic < maybe 13 depending on things like crowding or overlapping of the spectral images, plate flaws, and perhaps star color. Rough I-band magnitudes could be found from the USNO-B1.0 catalogue or from DENIS, both also available through VizieR searches. A few obscured objects were below this limit in the 2MASS J band (but bright at K and the MSX 8mu band), so obviously these were just too faint in the visible, even at I. Another class of 'blank fields' is for IRAS sources that are not red stars. MacConnell's selection was simply that the IRAS 12mu flux > 25mu flux. But this is insufficient to eliminate things like deeply-embedded hot stars, various sorts of nebulae (planetary, circumstellar shells around hot stars, HII regions), or even galaxies. A look at the complete IRAS fluxes in the SIMBAD headers was usually enough to decide this. Where neither 2MASS nor the MSX catalogue showed a red star, I revert to the original IRAS coordinates (sometimes improved from MSX) and leave the rest of the entry blank. Whatever solution was found to these special cases is given in the notes. Rather than try to determine the type of object (it is often indeterminate), I simply remark "not a red star". I might note that I often used DSS and 2MASS images from the Goddard SkyView utility to help sort out identifications in crowded fields. While getting coordinates I also got at least a rough estimate of the V magnitude for each object. Some of these are good means from ASAS-3 or TASS MkIV, and even Tycho-2, but others are just averages of the GSC-2.2 and USNO-B1.0 red and blue magnitudes. Every star of course is at least somewhat variable, so any of these are merely indicative. As an aside, it is quite interesting too look up the ASAS lightcurves for a bunch of these to see the variety of forms they exhibit. Once accurate coordinates were determined, I ran the corrected positions (again in batches) into VizieR looking at the GSC and MSX6C catalogue and copy those names as available. The MSX survey covers only the narrow strip within a few degrees of the galactic plane (usually out to +/- 3 deg, but out to 7 deg in a few places). I decided to show one "common" external identifier if this was available, and thus also searched the HD/HDE, BD/CD/CPD, Dearborn red stars, GCVS v4.1 (and now v4.2), and CGCS3. I used 3' search radii in VizieR for all these except the Dearborn catalogue, whose soft coordinates require a 5' search radius. Among the carbon stars I show the CGCS name in preference to the GCVS, but retain HD/BD names over all others. Finally, I searched every star in SIMBAD using a 3' search radius. The main result of all this was many dozens of IDs for fusion in SIMBAD, which I have bombarded upon Gerard Jasniewicz (the official SIMBAD database fixer) in the form of a long string of e-mail lists---it'll take him awhile to get all that cleaned up. \Brian