4 New V-I and some applications

We have calculated instantaneous (epoch) $(V-I)_{\rm H}$ color indices for 4414 M stars, 50 S stars from the list by van Eck et al. (1998), and 321 carbon stars, which include R, N, and Hd sub-types. A condensed version of this effort is presented in Table 7, which contains HIP number, GCVS name for variable stars, median Hp magnitude (entry H44, ESA 1997, vol. 1), 5-to-95 percentile Hp range or the Hp "amplitude'', coefficients b₀,b₁ (if b₁ has not been determined, it is set equal to zero), median V-I from this study, spectral type (M, S, or C).

We note that about 2% of Hipparcos M, S, and C stars do not have adequate Tycho-2 photometry and, hence, are not given in Table 7. Those include some very bright stars and a number of faint stars. More than a dozen stars of intermediate brightness with 8.0>Hp>5.0failed in the Tycho-2 photometry reductions due to poor astrometry, high background and/or a parasitic signal, which corrupted the signal from the target object.

  
4.1 Remarks on individual carbon stars

We used the derived $(V-I)_{\rm H}$ color index and in some cases individual slopes from the $Hp-V_{\rm T2}$ vs. Hp plot to scrutinize the identity of some Hipparcos carbon stars. If an anonymous field star is measured instead of a real carbon star, it could yield a positive slope in the fit of $Hp-V_{\rm T2}$ vs. Hp. This is because the Hp measures have been overcorrected, using a V-I color index appropriate for an expected carbon star but not for the actual target. On the other hand, the Tycho-2 $V_{\rm T2}$ photometry appears to be insensitive to the color a star really has. The net result is a very small or even positive slope. After identifying such cases, we checked the 2MASS Atlas Images for the true location of a carbon star in question. The offset in position is given in Table 8. If a carbon star has incorrect coordinates in Alksnis et al. (2001), it is coded by "GCGCS:'' in Remarks. If an incorrect identification is already acknowledged in the Hipparcos Catalogue, it is indicated by the "HIP note'' in Remarks. In the case of contradictory spectral classifications, we list only the alternative classification, since in nearly all such cases Hipparcos spectral type is "R...''. None of them can be found in Alksnis et al. (2001); therefore, the true identity of these stars has yet to be confirmed by spectroscopic means. An exception is HIP 94049 = CGCS 4179 which is a genuine carbon star (Houk, private communication; see also Table 1).

 

 

Table 8: Upper part: erroneous Hipparcos pointings of carbon stars or a contradictory spectral classification with Hipparcos indicating "R...'' spectral type. The offsets in coordinates, $\Delta$RA and $\Delta$Dec, are given in (s) and ( $\hbox {$^{\prime \prime }$ }$), respectively, in the sense "true position-Hipparcos''. Lower part contains significant corrections "true position-GCGCS'' required in Alksnis et al. (2001) to the positions of non-Hipparcos R stars from Table 1.

HIP	CGCS	$\Delta$RA	$\Delta$Dec	Remarks
4266				M0 (SAO)
14055				M0 (SAO)
21392				M0 (SAO)
22767	808	-21.0	+9	HIP note
24548	893	0.0	-242
29564				M0 (SAO)
29899	1226	+3.4	+26	GCGCS:
35015	1615	+7.1	-146	GCGCS:
35119	1616	+0.3	+59	HIP note, GCGCS:
37022	1787	-2.6	+32	HIP note, GCGCS:
39337	2007	+16.7	+31
40765				G1V (Houk & Swift 1999)
44235				not C-star? (Stephenson 1989)
75691	3614	+8.38	+94	GCGCS:
83404	3762	-0.4	-197	GCGCS:
85148	3820	-1.6	+58	GCGCS:
88170				M0 (SAO)
94049				C-star, not F5V
95024	4241	+5.3	+10	HIP note, GCGCS:
106599	5371	-7.7	+4	HIP note, GCGCS:
113840				M0 (SAO)
118252	5970	-2.3	-13	HIP note, GCGCS:
	258	-3.0	0
	3765	-0.6	-42
	3810	+10.3	+10
	3813	+0.4	+9
	3864	+0.3	-10
	3939	+0.5	-2
	3966	0.0	+25
	4042	+0.8	-2
	4168	+0.8	-14
	4498	+3.7	-35

  
4.2 Duplicity and V - I color index

Figure 9: Comparison of the V-I color indices used in the preparation of the Hipparcos Catalogue (H75) with those rederived from a color transformation based on $Hp-V_{\rm T2}$. Only red stars with spectral types  M and  S are considered. Single-star solutions are depicted by black dots, whereas the open symbols denote more complex solutions (hexagons - component solutions  C; triangles - acceleration solutions  G; squares - Variability-Induced Mover (VIM) solutions  V; stars - stochastic solutions  X). It is worth noting that nearly all datapoints in the upper right corner of the diagram correspond to complex solutions, thus hinting at problems encountered in the Hipparcos data processing for these stars.

Figure 10: A histogram showing the fraction of red stars with a DMSA C, G, V or X solution in the Hipparcos Catalogue (field H59), as a function of Hp magnitude, for three different sets of $\Delta (V-I)=(V-I)_{\rm H}-(V-I)_{\rm H75}$, where $(V-I)_{\rm H}$ is the newly derived median color index and $(V-I)_{\rm H75}$ is Hipparcos median V-I (entry H75). The unshaded histogram shows the fraction of these DMSA solutions among the stars with $\Delta (V-I)>2$; hatched for $2\ge \Delta (V-I)>1$; and dark-shaded for $1\ge \Delta (V-I)>-1$. The fraction of G, C, V, and X solutions clearly correlates with $\Delta (V-I)$; except for the faintest stars which always have a large fraction of G, C, V, X solutions.

Perhaps, the star HIP 12086 = 15 Tri is a prototype of a very rare but characteristic Hipparcos problem due to the neglected poor input coordinates. The declination of HIP 12086 listed in the Hipparcos Input Catalogue (ESA 1992) is off by $10\hbox{$^{\prime\prime}$ }$, hence in the detector's instantaneous field of view (see ESA 1997, vol. 3, Fig. 5.2) the signal has apparently been affected by the sensitivity attenuation profile. This kind of bias is absent in the star mapper's instrumentation. As a result, there is a very large positive slope in the $Hp-V_{\rm T2}$ vs. Hp plot. Not only is the Hp photometry clearly corrupted but the astrometry is also degraded as indicated by unusually large errors in the astrometric parameters. A similar effect of poor Hipparcos performance is known to be present, if the targets were wide binaries with separations in the range $\sim$ $15\hbox{$^{\prime\prime}$ }{-}20\hbox{$^{\prime\prime}$ }$ (Fabricius & Makarov 2000a). Here we list such binaries among red stars when the epoch Hp photometry is clearly biased: HIP 7762, 13714 & 13716, 17750, 18465, 45343, 57473, 86961, 87820, 108943, 116191, 114994. We note that from this list the revised astrometry is already available for HIP 17750, 86961, 87820, 116191 (Fabricius & Makarov 2000a).

Strictly speaking the V-I index derived in this study for Hipparcos binary and multiple stars could be affected by the component(s) and, hence should be considered with caution. On the other hand, a peculiar V-I value may very well signal a genuine problem, be it of astrophysical or instrumental character. With this in mind, we examined the location of complex astrometric solutions in the plot given in Fig. 7. It turns out that certain areas, as seen in Fig. 9, are heavily populated by such cases. Why is it so? It is helpful to look at the relative fraction of DMSA C,G,V, and X solutions as a function of differences between our median $(V-I)_{\rm H}$ and Hipparcos $(V-I)_{\rm H75}$. Figure 10 shows that the relative fraction of supposedly complex systems, i.e., binary or multiple stars, is abnormally high for red stars. For $\Delta(V-I)>1$ and Hp<10 (see unshaded and hatched areas in Fig. 10), the relative fraction of such systems is 40% and higher as compared to only $\sim$$10\%$ among the stars having correct $(V-I)_{\rm H75}$ index (dark-shaded histogram).

Table 9 lists all red stars with $(V-I)_{\rm H}-(V-I)_{\rm H75}>2$. As indicated from comparisons with an independent ground based V-I color index (see Col. 3 in Table 9), such differences are real. In essence, the stars listed in Table 9 have been processed with the $(V-I)_{\rm H75}$ color index off by more than 2 mag! Among such stars, the fraction of DMSA C,G,V, and X solutions - nearly 75% - is conspicuous in itself. For example, in the case of HIP 19488 and HIP 91703, it is evident that speckle interferometry could not confirm duplicity and, hence the Hipparcos DMSA/C solution must be spurious. This is nearly a watertight result since the limiting angular resolution of speckle interferometry (Mason et al. 1999; Prieur et al. 2002) is 2-3 times higher than the separation given in the Hipparcos Catalogue. The other stars with a DMSA/C solution listed in Table 9 have not been observed so far under similar conditions nor are they listed in the Fourth Catalog of Interferometric Measurements of Binary Stars, so that their possible spurious nature has yet to be established. Nevertheless, the high fraction of failed confirmations of binarity for Hipparcos stars with a DMSA/C solution (e.g., Mason et al. 1999, 2001; Prieur et al. 2002) is indicative that many such solutions might be spurious. We suspect that the phenomenon of such non-existent binaries among the red stars could very well be rooted in the improper chromaticity correction applied to these stars due to the poor knowledge of their true V-I color.

   

Table 9: All M, S spectral type stars with $\Delta (V-I)_{\rm H}=(V-I)_{\rm H}-(V-I)_{\rm H75}>2$ (Col. 2), where $(V-I)_{\rm H}$ is a color index from this study. When available, $\Delta (V-I)_{\rm0}=(V-I)_{\rm H}-(V-I)_{\rm obs}$, where $(V-I)_{\rm obs}$ is the ground-based photoelectric measurement. The stars are ordered by increasing amplitude of variability $\Delta Hp$(Col. 4). The column labelled DMSA provides a type of Hipparcos solution assuming more than one component. Angular separation between components, $\rho$, is given for DMSA/C solutions only.

HIP	$\Delta (V-I)_{\rm H}$	$\Delta(V-I)_{\rm0}$	$\Delta Hp$	Hp	DMSA	$\rho (\hbox{$^{\prime\prime}$ })$	Remarks
19488	2.41		0.13	9.535	C	0.18	unresolved (Mason et al. 1999)
78501	2.19		0.14	10.285	C	0.17
24661	2.31		0.15	10.170
87221	2.61		0.19	8.763	C	0.17
87433	2.27	-0.44	0.28	8.537
76296	2.26		0.33	8.878	C	0.16
42068	2.33	-0.49	0.41	8.511	C	0.18
91703	2.65		0.46	8.799	C,V	0.21	unresolved (Prieur et al. 2002)
7762	2.03		0.48	8.615	X		companion star at $20\hbox{$^{\prime\prime}$ }$
84346	2.05		0.61	8.454	V		unresolved (Prieur et al. 2002)
100404	2.14	-0.76	0.61	8.464	V		unresolved (Mason et al. 2001)
37433	2.17		0.64	8.984
56533	2.64		0.65	8.581	C	0.24
84004	2.40		0.78	7.499	X
80259	2.19		0.91	9.017	V		unresolved (Prieur et al. 2002)
16328	2.10		0.98	9.612	C	0.30
90850	2.32		1.38	11.001
78872	2.06	-0.47	1.70	9.841	G
703	2.43	-0.59	2.09	11.112	V
9767	2.19		2.10	9.773	V
11093	2.52	0.16	2.10	9.756	V
89886	3.27		2.26	10.883
96031	2.29		2.64	10.512
75393	3.06	0.32	2.73	8.554	V
16647	3.23		2.87	10.376	V
81026	2.66		2.89	11.538
1901	2.61	1.27	2.97	10.705	V		unresolved (Prieur et al. 2002)
86836	3.43		3.15	11.196	V
47066	2.61	0.87	3.49	10.073	V
57642	2.32	0.78	3.60	9.968	V
60106	2.04		3.81	9.854
110451	2.01		3.90	11.460
94706	2.81	0.67	3.97	10.826	V		T Sgr: composite spectrum
25412	2.39	0.29	4.00	9.974	V
33824	2.02	0.97	4.05	9.922

4.3 Empirical effective temperatures of red stars

Due to very complex spectra the red stars are cumbersome objects for getting their effective temperature - one of the fundamental stellar parameters. From different vantage points this has been investigated, e.g., by Bessell et al. (1998), Bergeat et al. (2001), Houdashelt et al. (2000). Although the Cousins V-I color index may not be the optimal color to calibrate effective temperature due to the strong influence by molecular absorption bands and possible reddening, nevertheless we attempted to derive an empirical calibration of effective temperatures for carbon and M giants. We used median $(V-I)_{\rm H}$ for Hipparcos stars having interferometric angular diameter measurements in K ( $\lambda=2.2$ $\mu$m) bandpass (Dyck et al. 1996; van Belle et al. 1997, 1999) and corresponding effective temperature estimates. It is expected that the interstellar reddening is low for the chosen Hipparcos stars because of their relative proximity to the Sun. In total, from these sources of effective temperature determinations, we selected 27 small amplitude ( $\Delta Hp < 0.5$) M giants in the range 3.6>V-I>1.5 and 16 carbon stars ( 3.8>V-I>2.4) with no restriction on variability. Similarly to Dumm & Schild (1998) we adopted a linear relationship

$$\log T_{\rm eff}=d_0+d_1(V-I).$$ (9)

For M giants, a least squares fit using Eq. (9) yields $d_0=3.749\pm0.014$, $d_1=-0.087\pm0.007$, and the standard error $\sigma_{\rm T}=110$ K. For carbon stars the coefficients from the fit are: $d_0=3.86\pm0.06$, $d_1=-0.153\pm0.021$, and the standard error $\sigma_{\rm T}=210$ K. Apparently, the effective temperature scale is not satisfactory for carbon stars in terms of its precision. The color mismatch between our median $(V-I)_{\rm H}$ and the color of a variable star at the time of interferometric observation can only partly explain the noted large scatter. Another reason might include an unaccounted for circumstellar extinction, carbon abundance and metallicity effects on the color, and rather large errors in the effective temperature determination. The latter is discussed in detail by Dyck et al. (1996). An alternative scale of effective temperatures for carbon stars is given by Bergeat et al. (2001), although it may have the same kind of inherent problems. We note that the slope d₁ for M giants is 2.5 times larger than in Dumm & Schild (1998). The main reason for that is a stretched color scale of Hipparcos V-I (see Fig. 7). It is felt that the empirical effective temperature scale based on V-I color has a limited use, in particular for carbon stars. Near infrared observations in JHKL bandpasses should be used to obtain better estimates of effective temperature for the coolest stars. With the advent of large optical interferometers the number of precise angular diameters for cool and red stars undoubtedly will increase substantially. However, an equal effort should be invested in deriving reliable bolometric fluxes, which are equally important in establishing a precise scale of effective temperatures.