Catalog of Positions of Infrared Stellar Sources -- Original 1994 information

bar

Readers should understand that the CPIRSS catalog has been updated since the writing of the following document. Hence, some of the information in this document is outdated or repeated elsewhere. However, since much of the initial work on this project was performed in 1994, it was decided to give the project report from that time in its entirety.


THE U.S. NAVAL OBSERVATORY CATALOG
OF POSITIONS OF INFRARED STELLAR SOURCES
1994 VERSION

Robert B. Hindsley and Robert S. Harrington*

*Deceased 23 Jan 1993

Astrometry Deptartment
U.S. Naval Observatory
Washington, DC 20392-5420


ABSTRACT: Star identifications in the IRAS Point Source Catalog have been combined with optical positions from 4 astrometric catalogs to yield a catalog with positional accuracies of about 0.2 arcsecond for 33,678 infrared sources. The identification of IRAS sources with bright optical stars has been checked by requiring the color V-[12] (with [12] being a magnitude derived from the IRAS flux) to be consistent with the optical colors or spectral type. Observed V-[12] colors several magnitudes redder than predicted from optical colors have usually been interpreted as evidence of excess 12 micron flux, but many stars with such colors are shown also to have an overly red V-K color. A K magnitude (2.2 microns) has been estimated using the 12-micron flux and the spectral type or observed optical color for the the 18,525 stars that have no evidence of an unusual 12-micron flux.

1. INTRODUCTION

The Infrared Astronomical Satellite (IRAS) carried out an all-sky survey at 12, 25, 60, and 100 microns. Approximately 250,000 point sources were identified. However, by the standards of optical astrometry, the positions of these sources are poorly determined, with errors of tens of arcseconds. Optical astrometric catalogs such as the ACRS (Corbin & Urban 1990) have errors in position of about 0.2 arc seconds, some two orders of magnitude less than the IRAS position errors.

Many of the sources in the IRAS Point Source Catalog (1987, henceforth PSC) have been identified as stars. SAO numbers are given for approximately 1/8 of the entries in the PSC. Almost all of these stars have positions given in optical astrometric catalogs. Thus it is possible to construct a higher quality astrometric catalog of infrared sources by combining these positions with the IRAS identifications.

Furthermore, if the spectral type or suitable photometry is available for the star, the identification of IRAS sources with particular stars can be confirmed. Because the IRAS positions have such large errors, the identifications provided in the PSC are suspect. The V-[12] color, with [12] being a magnitude determined from the 12-micron flux, is very sensitive to temperature. If the optical identification of an IRAS source is correct, the V-[12] color should be consistent with optical colors or spectral type, with due allowance made for interstellar reddening.

Section 2 of this paper describes the various optical astrometric catalogs that provided positions for the PSC sources. Section 3 deals with the confirmation of identification based on color. In Section 4 the results of the confirmation are discussed. Section 5 describes the estimation of Johnson system K magnitudes for those stars with confirmed identifications. Most of the catalog data needed for this work was taken from the Astronomical Data Center CDROM "Selected Astronomical Catalogs" (ADC 1991; henceforth CDROM).

2. OPTICAL ASTROMETRY

There are 34,438 stars with SAO identifications in the PSC. For purposes of cross-referencing, the Durchmusterung (DM) catalog number was obtained from the SAO/J2000/DM/HD/GC Cross Reference Index (Roman, Warren, & Schofield 1991). For most stars south of -52 degrees, this number is the Cape Photographic Durchmusterung (CPD) number. For stars north of -23 degrees, the number is from the Bonner Durchmusterung or Southern Durchmusterung (henceforth BD for both). In between, the DM number is from the Cordoba Durchmusterung (CD). Stars with a single entry in the SAO and multiple entries in the DM catalogs, or multiple entries in the SAO catalog corresponding to one DM entry, were eliminated from further consideration.

There were 34,407 stars with DM identifications. Four astrometric catalogs were then searched for positions for these stars. Positions taken from the FK5 (Fricke et. al. 1988) and the FK5 Extension (Fricke, Schwan, & Corbin 1991) have errors at 1992 of 0.10 and 0.14 arcsecond respectively, were the most accurate, and were favored over positions from the other catalogs. The International Reference Stars catalog (IRS; Corbin 1978, Corbin & Urban 1990) consists of those stars with sufficient transit circle observations to define proper motions and exhibits errors at 1992 of 0.20 arcseconds. The Astrographic Catalog Reference Stars catalog (ACRS; Corbin & Urban 1990) is the next most accurate astrometric catalog, with an error at 1992 of 0.21 arcseconds. All of the determinations of positions and proper motions in the ACRS include photographic data. The ACRS and FK5 Extension were supplied by T. Corbin and S. Urban of the U.S. Naval Observatory. The FK5 and the IRS were taken from the CDROM. Four FK5 stars have incorrect DM numbers on the CDROM. The FK5 numbers for these stars, with the correct DM number following in parentheses, are: 655 (BD+55 1944); 833 (BD+32 4349); 1387 (BD-15 3965); and 1527 (BD-12 5683). Some identifications of ACRS stars with DM numbers were also revised with data from Roman (1993).

Of the IRAS PSC stars with Durchmusterung identifications, 33,678 have a position in one of the optical catalogs. The accuracy of the catalog positions is the same as the source catalogs, about 0.2 arcseconds.

3. COLOR CONFIRMATION

Waters, Cote, and Aumann (WCA; 1987) compared V-[12] observed colors for those IRAS sources also found in the Bright Star (BS) Catalog (Hoffleit & Jaschek 1982) to more traditional color indices such as B-V. They chose only those stars that had precise photometry available, and were only slightly affected by interstellar reddening. There proved to be a fairly tight correlation between B-V and V-[12] for that color regime in which B-V is correlated with temperature (i.e., all but the M spectral class stars). Figure 3 from WCA is reproduced here as Figure 1. From Table 1 of WCA, it can be calculated that the rms scatter about the mean relationship is about 0.13 mag for 0.2 < (B-V)0 < 1.6. A similar plot, but with more scatter and a less tightly correlated relationship, can be made for spectral type vs. V-[12] color.

Thus, for unreddened stars with available photometry, one can predict the 12 micron flux. If it deviates from observation by significantly more than the rms scatter around the mean relationship (~ 0.13 mag), the association of the star with the IRAS point source must be viewed suspiciously (or the star has an IR excess/deficit and is very interesting if the identification can be confirmed by other means).

If one uses the spectral type to predict the V-[12] color, the acceptable error would be larger, because there is an intrinsic spread in B-V color within a given spectral class. WCA found that this spread is typically 0.1 mag for stars later than spectral type G. Given that the slope of the V-[12] vs. B-V relationship is about 1.75, this intrinsic spread would imply an additional error of 0.2 mag in estimating V-[12]. Added in quadrature, this would increase the expected error in the prediction from 0.13 mag to 0.24 mag.

A very serious problem encountered in predicting the V-[12] color is interstellar reddening. The difficulty is not so serious when a color is used to predict V-[12], because the effect of reddening is to move the star along a reddening line that is somewhat parallel to the relationship. A B-V color excess of 0.1 mag causes a V-[12] excess of 0.31 mag, but the error in estimating V-[12] is only about 0.14 mag. If the R-I color is used in place of B-V, the error is reduced to 0.05 mag or 0.11 mag, depending on the R-I value. But if the spectral type is used in place of B-V as a temperature indicator, the error is the entire amount of the V-[12] excess, 0.31 mag, because only the V-[12] value is changed by the reddening. In other words, the displacement in Figure 1 would be vertical if spectral type is used instead of an observed color.

Various spectral type and photometric catalogs were obtained from the CDROM. Spectral types were obtained from three different catalogs. The Henry Draper catalog (Cannon & Pickering 1918-24; Cannon 1925-36) has spectral types for nearly 28,000 of the stars in the PSC, more than 85%. However the HD catalog has no luminosity class information, and the M spectral types do not have subclasses equivalent to the MK system. The MHD catalog (Houk & Cowley 1975; Houk 1978; Houk 1982; Houk & Smith-Moore 1988) has the most complete spectroscopic data, but covers only the sky south of -12 degrees declination. The MK Extension catalog (Morris-Kennedy 1983) contains very few of the stars in the PSC. The most extensive photometric catalog is that of Mermilliod (1991), which has entries for about 8500 stars. The combined catalog of Lanz (1986) was also included, but contains data for only about 1600 stars. The catalog of Morel and Magnenat (1978) was used as the primary source of the color R-I on the Johnson system.

For those stars with known color and spectral class the intrinsic color was obtained from FitzGerald (1970). Comparing the intrinsic color to the observed color yielded the color excess directly. The effect of reddening was then removed from the estimation of V-[12]. The error resulting from removal of the reddening is equal to the spread in color with spectral type, so that the overall error is 0.24 mag. To confirm an IRAS source as a particular star, it was required that the residual calculated-observed V-[12] color be within 0.5 mag, or almost exactly 2 sigma.

For stars with only a spectral type, the problem was approached in a different manner. The intrinsic color determined from the spectral type was used to calculate a value of V-[12]. The B-V color excess was determined using the empirical relationship of Arenou, Grenon, and Gomez (1992). This relationship estimates the visual absorption from the galactic coordinates and the distance. If a luminosity class was available, the absolute magnitude was determined from the spectral type using the empirical relationships of Schmidt-Kaler (1982), and distance determined from the difference between the apparent and absolute magnitude (this was calculated in an iterative manner, with the absorption used to correct the distance estimate). If no luminosity data were available, the star was assumed to be a dwarf (class V) if earlier than G5; otherwise the star was assumed to be a giant (class III), consistent with the treatment of Arenou, Grenon, and Gomez (1992). It should also be noted that while this scheme is much too smooth to match the rather clumpy extinction within a few degrees of the galactic plane, only a small fraction of the sources are in that region. The value of the color excess E(B-V) calculated this way was used to correct the estimation of V-[12].

For those stars with only an observed color, the reddening was again estimated in the same manner, but the technique was more complicated in practice because the intrinsic color, absolute magnitude, and distance had to be recalculated in every iteration as they depend on the reddening. Again the luminosity was assumed to be class III or V. A color excess was calculated for both classes, and the value that gave the best match between the observed and predicted V-[12] was adopted.

As lack of a good reddening estimate can be expected to increase the error, it was decided to adopt a maximum acceptable difference between the observed and calculated V-[12] of 0.75 mag when only color or spectral type data are available. When using the spectral type to estimate V-[12], this limit corresponds to E(B-V) = 0.75/3.1 = 0.24 mag (that is, an error in color excess of 0.24 mag would produce an error in V-[12] of 0.75 mag). When B-V color is used, E(B-V) = 0.75/1.4 = 0.54 mag corresponds to the limit. Use of the R-I color on the Johnson system yields E(B-V) = 0.75/0.55 = 1.4 mag (R-I<0.72) or 0.75/1.1 = 0.68 (R-I>0.72).

4. DISCUSSION

The numbers of stars with each type of data available, and the numbers of color or questionable identifications, are given in Table 2. For a normal distribution one would expect to find 95% of the residuals within 2 sigma. For those stars with photometric colors and spectral type data that permit accurate determination of the reddening, 91% are confirmed. This suggests that almost all of these stars do match the IRAS sources. For stars with only photometric or only spectral type data, the percentages of confirmed matches are somewhat lower. This is partially explained in terms of the cruder estimate of the reddening. In addition, V magnitudes for stars without an observed color have been determined photographically, and are much less accurate than photometric magnitudes.

There is a striking difference in confirmation rates between those stars with an MK spectral type (from either the MK Extension catalog or the MHD catalog) but no observed color and those stars with an HD spectral type but no observed color. Spectral types from the MK Extension catalog and the MHD catalog are two-dimensional (i.e. have luminosity class data), while the HD catalog provides only one-dimensional spectral types. To determine if the luminosity data might be the reason for the difference in confirmation rates, the analysis was repeated using the MK spectral types, but the luminosity class data were ignored; instead the luminosity class was assumed in the same manner as for the HD spectral types. The confirmation rates were unchanged, which implies that the MK Extension and MHD catalogs are superior for this purpose regardless of the luminosity data. On the other hand, for those stars with both an observed color and a spectral type, the confirmation rates seem practically independent of the spectral type source catalog used.

In an effort to better understand this dependence of confirmation rate on the spectral types catalog, the data in Table 2 were broken down by spectral type, with the results given in Table 3. The confirmation rates are nearly identical when an observed color is available regardless of the spectral type source catalog, except for the M stars. For stars without an observed color, the rates are very different for stars of spectral type F or later. Thus it seems that the spectral types in all the catalogs are consistent for those stars with an observed color, but not for the redder stars without an observed color.

The stars with an observed color and spectral type are different from those with a spectral type but without a color in that the former group is brighter overall. Stars with an observed color have a median visual magnitude of 6.32, with only 31% fainter than 7.0. Those stars without an observed color but with an observed MK spectral type have a median visual magnitude of 7.91, and 88% are fainter than visual magnitude 7.0. Stars with only an HD spectral type are fainter still, with a median visual magnitude of 8.37, and 94% are fainter than 7.0. Because there is little difference in confirmation rates for the brighter stars (those with an observed color), it is tempting to state that the HD spectral types are poor for fainter stars in general. However, because this sample consists of stars with an IRAS 12-micron magnitude brighter than 5.0, a star with V fainter than 7.0 must have V-[12] color redder than 2.0 and is thus spectral type G3 or later. So the difference in confirmation rates suggests that the HD spectral types differ systematically from the MK spectral types for the reddest stars, particularly visually faint stars redder than the Sun.

It can be seen in Table 3 that the confirmation rates vary with spectral type even for those stars with an observed color and an observed spectral type. Large percentages of stars with V-[12] excesses for spectral types B, A, and M have been observed previously. Considering only those stars in Table 3 with an observed color and spectral type, the percentages of stars with a V-[12] excess greater than 0.5 mag (i.e., not confirmed) is given in Table 4, along with the equivalent percentage found by WCA.

The results are very similar. Patten and Willson (1991) considered only stars of types B, A, and F; while the percentages with an excess were not broken down by spectral type, it was shown that an excess is correlated with rapid rotational velocity, which is known to be more common for stars of spectral type earlier than F. Iyengar and Rengarajan (1991) analyzed stars of type A to M in the Catalogue of Nearby Stars of Gliese (1969). Excesses greater than 0.3 mag were found to occur for more than 5% of the stars in class A and possibly M, but less than 5% of F, G, and K stars had such an excess.

Thus it seems that 12-micron excesses greater than 0.5 magnitude are not unusual for stars of spectral types O, B, A, and M. Those star identifications from the PSC that are not confirmed may not be due to misidentifications, but rather due to excess flux at 12 microns. For this reason, no source has been excluded from the catalog because of the results of the color confirmation. Rather, the residual (calculated - observed) value of V-[12] has been included in the catalog for consideration by the user. Those sources considered questionable are flagged, as are those without sufficient data for analysis. Certainly source identifications that are confirmed should be regarded as more reliable.

5. K MAGNITUDES

The technique of WCA to predict the V-[12] color can of course be used to predict other colors involving IRAS magnitudes if sufficient observational data exist. For example, Odenwald (1991) suggested that the Johnson system K filter (2.2 micron) magnitudes should be very similar to the IRAS 12-micron magnitudes. Using those stars with both an observed K magnitude and an IRAS 12-micron flux, the relationship between K-[12] and (B-V)0 can be determined and then used to predict the K magnitudes. The catalog of Morel and Magnenat (1978) contains K magnitudes for 425 stars that have IRAS identifications confirmed by color and 87 stars that do not have confirmed identifications. A two-color plot of these data is shown in Figure 2. As Odenwald predicted, the K magnitudes and 12-micron magnitudes are nearly identical, with the (K-[12])0 color showing only a small trend with (B-V)0. A line was fitted to all the data, then a new fit was derived after deleting all data more than 0.4 mag from the previous line. This iterative approach is very similar to the treatment of WCA. The final result was

(K-[12])0 = 0.012 + 0.104*(B-V)0, (B-V)0 < 1.6 ~ .006 ~ .006
For the 474 points included in the fit (all but 3 of the confirmed stars, plus 52 of the unconfirmed stars), the rms deviation from the line is 0.087 mag.

Because this fitted line is very insensitive to (B-V)0, it is fairly insensitive to errors in (B-V) or to reddening. The color excess E(K-[12]) is assumed to be 0.4*E(B-V), and should be due entirely to absorption in the K filter. The small errors in slope and intercept yield errors in the (K-[12])0 color small relative to the error in the 12-micron magnitude, which is assumed to be 0.10 mag (see Table VII.D.2 of the Explanatory Supplement). Thus the error in the K0 magnitude, derived by adding [12] to (K-[12])0, can be taken to be 0.10 also. If the error in E(B-V) is assumed to be 0.1 mag (due mostly to variation in intrinsic color for a particular spectral type), then the error in the reddened K magnitude K = K0 + 0.4*E(B-V) is about 0.11 mag.

If the (K-[12])0 excess were due to excess flux in the 12-micron filter only, then predicting the (V-K)0 color and determining the K magnitude from that color would be a viable option for those stars whose IRAS identifications were not confirmed by the V-[12] color. For the same sample of 512 stars used previously a two-color plot is shown in Figure 3. Many of the stars whose IRAS identifications cannot be confirmed by color also show unusual (V-K)0 values. One obvious explanation is that both the K and the 12-micron fluxes are larger than expected. Also noteworthy is the extreme sensitivity of V-K to reddening, with E(V-K) = 2.7*E(B-V), but there would have to be huge errors in E(B-V) to account for the V-K excesses seen in Figure 3. Such errors might arise if there were an identification error in the spectral type catalog, for example. It is possible that the absorption as a function of wavelength is atypical, so that the reddening corrections are inappropriate. Multiplicity involving a faint red companion is certainly possible for the bluer stars. Further work, possibly including K magnitude observations, may be needed to explain this correlation of excesses in the K and 12-micron magnitudes. The important fact for this work is that there seems to be no advantage in using the (V-K)0 color to estimate the K magnitudes for stars without confirmed identifications.

6. THE CATALOG

The catalog contains an identifier from one of the Durchmusterung catalogs as well as the IRAS PSC identification. The right ascension, declination, and proper motions are given on both the FK5 (J2000.0) system and the FK4 (B1950.0) system. There is a flag to indicate which of the astrometric catalogs is the source of the position. Besides the IRAS fluxes for each of the bands, the V magnitude is given along with an indicator of the source of the V magnitude value. A spectral type and source is given, along with an estimate of the B-V color excess and a flag to indicate the method by which the excess was determined.

The types of data used for color confirmation of PSC identifications are also indicated. The (calculated - observed) V-[12] is given in magnitudes, and a flag indicates whether this deviation is within the acceptable limits (which are based on the types of data used). If there are not sufficient data to attempt color confirmation, there is an indication as to what data are lacking. Predicted K magnitudes are given for those stars whose identifications are considered to be confirmed.

Finally the catalog contains several useful quantities from the IRAS PSC catalog, such as the flux quality indicator, the point source correlation coefficient, and an estimate of the probability of variability as determined from the IRAS data. These are more fully described in the IRAS Explanatory Supplement.

RBH wishes to thank Sten Odenwald, Tom Corbin, and Gart Westerhout for helpful comments, Sean Urban and Geoff Douglass for help obtaining data, and Stephen Gauss for assistance in arranging the layout of this paper and distribution of the catalog. Also the help and patience of Brenda Corbin and Greg Shelton of the USNO library is greatly appreciated. The conception of and groundwork for this project was done exclusively by Dr. Robert S. Harrington, whose untimely death in January 1993 at age 50 is still mourned by his staff.

REFERENCES

Arenou, F., Grenon, M., and Gomez, A. 1992, Astron. & Astrophys., 258, 104

Cannon, A.J. 1925-36, "The Henry Draper Extension", Ann. Astron. Obs. Harvard College, 100

Cannon, A.J., & Pickering, E.C. 1918-24, "The Henry Draper Catalog", Ann. Astron. Obs. Harvard College, 91-99

Cannon, A.J. 1925-36, "The Henry Draper Extension", Ann. Astron. Obs. Harvard College, 100

Corbin, T.E. 1978, in IAU Colloquium No. 48, "Modern Astrometry", edited by F.V. Prochazka and R.H. Tucker, (University Observatory, Vienna), p.505

Corbin, T.E., & Urban, S.E. 1990, in IAU Symposium No. 141, "Inertial Coordinate System on the Sky", edited by J.H. Lieske and V.K. Abalakin, (Kluwer,Dordrecht), p.433

FitzGerald, M.P. 1970, A&A, 4, 234

Fricke, W., et. al. 1988, Fifth Fundamental Catalogue (FK5) Part I: The Basic Fundamental Stars, Veroeff. Astron., Rechen-Institute Heidelberg, No. 32

Fricke, W., Schwan, H., & Corbin, T.E. 1991, Fifth Fundamental Catalogue (FK5) Part II: The FK5 Extension- New Fundamental Stars, Veroeff. Astron., Rechen-Institute Heidelberg, No. 33

Hoffleit, D., & Jaschek, C. 1982, The Bright Star Catalogue, Yale University Observatory, New Haven, USA

Houk, N. 1978, University of Michigan Catalogue of Two-Dimensional Spectral Types for the HD Stars, vol. 2, University of Michigan

Houk, N. 1982, University of Michigan Catalogue of Two-Dimensional Spectral Types for the HD Stars, vol. 3, University of Michigan

Houk, N., & Cowley, A.P. 1975, University of Michigan Catalogue of Two-Dimensional Spectral Types for the HD Stars, vol. 1, University of Michigan

Houk, N., & Smith-Moore, M. 1988, University of Michigan Catalogue of Two-Dimensional Spectral Types for the HD Stars, vol. 4, University of Michigan

IRAS Catalogues and Atlases: The Explanatory Supplement 1987, edited by C.A. Bechman, G. Neugebauer, H.J. Habing, P.E. Clegg, T.J. Chester, NASA RP-1190, vol. 1

IRAS Catalogues and Atlases: The Point Source Catalog 1987, edited by C.A. Bechman, G. Neugebauer, H.J. Habing, P.E. Clegg, T.J. Chester, NASA RP-1190, vol. 2-6

Iyengar, K.V.K., and Rengarajan, T.N. 1991, Astron. Astrophys., 250, 420

Lanz, T. 1986, A&A Suppl., 65, 195 (from CDROM)

Mermilliod, J.-C. 1991, unpublished (from CDROM)

Morel, M., & Magnenat, P. 1978, Astron. Astrophys. Suppl, 34, 477

Morris-Kennedy, P. 1983, unpublished (from CDROM)

Odenwald, S. 1991 (private communication)

Patten, B.M., & Willson, L. A. 1991, A.J., 102, 323

Roman, N.G. 1993 (private communication)

Roman, N.G., Warren, W.H., and Schofield, N.J. 1991, unpublished (from CDROM)

Schmidt-Kaler, T. 1982, Landolt-Bornstein, Neue Ser., Gruppe VI, 2b, 14

Selected Astronomical Catalogs, vol. 1, No. 1 (USA_NASA_ADC_CAT_0011), 1991, Astronomical Data Center, Goddard Space Flight Center, NASA (CDROM)

Waters, L.B.F.M., Cote, J., and Aumann, H.H. 1987, A&A, 172, 225 (WCA)

Worley, C.E., and Douglass, G.G. 1984, unpublished (from CDROM).

Table 1: 
Catalog            Number         %
=======================================
FK5                 1352          4
FK5 Extension       1229          4
IRS                 8110         24
ACRS               22987         68
=====================================
Total              33678


Table 2                                                                 
                     Confirmation              Number Color     Number
   Data Available      Criterion    Total       Confirmed    Questionable
-----------------------------------------------------------------------

    Color and MK       within
     Spectral Type     0.5 mag      3408      3140 = 92 %     268 =  8%

    Color and HD       within
     Spectral Type     0.5 mag      3564      3206 = 90 %     358 = 10%

    MK Spectral        within
     Type Only        0.75 mag      8571      7374 = 86 %    1197 = 14%

    HD Spectral        within
     Type Only        0.75 mag     10234      4647 = 45 %    5587 = 55%

    Color Only         within
                      0.75 mag       202       158 = 78 %      44 = 22%
------------------------------------------------------------------------
    Total with Data                25979     18525 = 71 %    7454 = 29%
    No Data Available               7699         0 =  0 %    7699 = 100%
    Total                          33678     18525 = 55 %   15153 = 45%


Table 3

   Data Available         O      B      A      F      G      K      M
-----------------------------------------------------------------------
   Color and MK
    Spectral Type
     % confirmed          0     75     93     97     99     98     50
     Total number         2    195    209    434    648   1618    302
 
   Color and HD
    Spectral Type
     % confirmed          -     67     92     95     97     97     40
     Total number         0     92    214    254    450   2199    355
 
   MK Spectral
    Type Only
     % confirmed          -     37     62     78     84     91     72
     Total number         0     37     53    147    854   5985   1495
 
   HD Spectral
    Type Only
     % confirmed          -     31     35     64     52     47     19
     Total number         0     35     57    120    997   8079    946
=======================================================================


Table 4

     Source                 B      A      F      G      K      M
-----------------------------------------------------------------------
     This work(a)          28      7      4      3      4     56
     WCA (1987)            40     13      5      3      1     29


                          Notes to Table 4
(a) includes all data with an observed color and spectra type
 


bar