Effective temperatures derived using can be also compared with those obtained using the Infrared Flux Method (IRFM hereafter) and spectrophotometric methods. Although the lack of common observations between this paper and the papers quoted in this section prevents us from repeating the same tests done with the photometric calibrations, the similarity between and MD85 found in Fig. 4.2 (click here) allows us to establish comparisons by using MD85 instead of .
Blackwell & Lynas-Gray (1994) (Bl94, hereafter) constitutes one of the most recent works on the determination of effective temperatures using IRFM. They recalculated the values given in Blackwell et al. (1990) (Bl90, hereafter) by using the newest version of the Kurucz models (1990, 1991, 1992) characterized by providing more accurate determinations of blanketing and ultraviolet flux than the older versions.
Figure 5.1 (click here) compares, for the sample of stars given in Bl94, the values of calculated with the MD85 calibration and the IRFM. As suggested by Bl94, an error of has been adopted in the IRFM values, whereas an error of has been adopted for MD85. These errors have to be understood from the point of view that the error associated to the IRFM is ``absolute" in the sense that it is the result of errors in the adopted values of gravity, metallicity and interstellar extinction as well as errors in Kurucz models, especially those concerned with convection inhomogeneities and non-LTE assumptions, whereas the error of MD85 is an internal error caused by indetermination in the photometric indices which do not include errors due to the models or to the physical parameters of the stars considered as standards.
In Fig. 5.1 (click here) we see how the IRFM gives values which are systematically lower than MD85. Some authors (Mégessier 1988; Napiwotzki et al. 1993; Smalley 1993) have found similar results suggesting two possible sources of errors: the infrared calibrations and the flux models employed. To quantify the influence of the calibration as well as the importance of the models in our range of temperatures, we have plotted in Figs. 5.1 (click here)c and 3d the values of those stars in common to Bl94 and Bl90: whereas Bl94 used the new version of Kurucz models (1990-1992) together with Cohen et al. calibration (1992), the MARCS code of Guftasson et al. (1975) and the Dreiling & Bell (1980) calibration were employed in Bl90. The difference of () between Bl90 and Bl94 can be explained in terms of blanketing: if blanketing increases at short wavelengths (as it happens with the new Kurucz models) the total flux conservation will make the infrared flux to be higher and, following the IRFM, the temperature lower.
The series of papers by Malagnini et al. (1982, 1985, 1986) and Morossi & Malagnini (1985) constitute one of the most extensive works in the calculation of using spectrophotometric methods. In Fig. 5.1 (click here) we have compared the values of given in Malagnini et al. (1986) (Mal86, hereafter) and Morossi & Malagnini (1985) (Mor85, hereafter) with those values obtained from the MD85 calibration. The average difference between MD85 and Mal86 is and between MD85 and Mor85 it is . A possible explanation of the differences between the two samples found in Fig. 5.1 (click here)b could be the adopted color excess: in the Mor85 sample, the reported values of depend on E(B-V) since the reddening modifies both the shape and the absolute level of the flux distribution. The color excess was measured according to the values of adopted by Fitzgerald (1970). These have an associated error of 0.02 mag. which corresponds to an error in of . To produce the systematic difference of 132 K found in Fig. 5.1 (click here)b a deviation of would be enough. Hence, we see how a slight difference in the definition of the intrinsic colors could explain the trend found in Fig. 5.1 (click here)b. This problem is not present in Mal86 since in this case the sample is formed of bright stars with .