6 Comparison with other determinations

In this section, we provide the comparison of our temperatures with those derived by other authors for common stars in the sample. We show differences found in Figs. 7 and 8. Furthermore, a detailed analysis of the scale of temperatures derived from the present work will be given in a subsequent paper by considering the mean relations $T_{\rm eff}$: [Fe/H] and the UBVRIJHK and uvby photometric colours.

Direct methods

$\;$

There is a group of 18 giant stars in the sample whose diameters have been measured by optical methods (Code et al. (1976; C76), Ridgway et al. (1980; R80), Di Benedetto & Rabbia (1987; BR87), Hutter et al. (1989; H89), Mozurkewich et al. (1991; M 91), and White & Feiermann (1987; WF87)). In Table 8, we provide the comparison between IRFM temperatures and those obtained considering their measured angular diameters and bolometric fluxes. If we consider the typical errors of both the direct method and the IRFM the differences observed are within the error-bars. Excluding HR 7635 and HR 5301, the average difference $T_{\rm direct}-T_{\rm IRFM}$ amounts to $0.06 \pm 1.25$%. Moreover, no apparent trend of the differences with temperature is observed.

Arribas & Martínez-Roger (1987, AM87)

$\;$

There are 23 stars of M 3 from the article of Arribas & Martínez-Roger (1987, AM87), based on the application of the IRFM and using the empirical absolute flux calibration of the IR flux of Vega derived by Mountain et al. (1985), and the models of Kurucz (1979) and Gustaffson et al. (1975), which are common with our work (Table 9). Temperatures derived by AM87 are, on average, 55 K hotter, with a standard deviation amounting to 59 K. No obvious trend with temperature can be seen in the differences (Fig. 7). The shift in temperatures, amounting to 1.3%, might be explained by taking into account the different absolute flux calibration, and the improvement of Kurucz's models.

Bell & Gustafsson (1989, BG89)

$\;$

This work contains 25 stars in common with the present sample. BG89 temperatures are based on the IRFM corrected with IR synthetic colours. The differences listed in Table 10 are compatible with a zero-point shift of 65 K (T_BG89 are hotter) or 1.3 $\pm$ 1.1%. These differences are probably connected with the differences in the bolometric fluxes adopted in both works (Sect. 4.2).

 

Table 10: Comparison between the temperatures derived in the present work (Col. 2) and those derived by Bell & Gustafsson (1989) (Col. 3). The mean difference $T_{\rm IRFM}-T_{\rm BG89}$ is -65 $\pm$ 56 K discarding HR 4247 and HR 5889

Blackwell & Lynas-Gray (1998; BL98)

$\;$

This extensive study provides temperatures for 420 stars (A0-K3) with luminosity classes between II and V, incorporating previous results of similar accuracy (Blackwell et al. 1990 and Blackwell & Lynas-Gray 1994). It is a good source for comparison, given that it is based on the IRFM with the same models adopted in the present work, the basic differences being the calibration of the IR flux of Vega, and the scale of bolometric fluxes. In Fig. 7, we show the differences in temperature for the 50 Population I giants common to the present sample. On average, $T_{\rm BL98}$ are 36 $\pm$ 67 K hotter than ours; no apparent trend of the differences versus $T_{\rm eff}$ is seen. This shift in temperature (0.70% $\pm$ 1.25%) is compatible with the difference in the bolometric fluxes (around 1.5%) found in Sect. 4.2.

Di Benedetto (1998; DB98)

$\;$

This work reports on the implementation of the empirical surface brightness technique using Jonhson (V-K) colours for 537 dwarf and giant stars A-K. There are 70 giant and subgiant stars of DB98 common both to Paper I and the present work. Below 7000 K, the agreement is fairly good. The DB98 temperatures are 12 K hotter with a dispersion of 42 K (0.23 $\pm$ 0.93%). However, for the four stars with $T_{\rm IRFM}\gt 7000$ K a difference of 400 $\pm$ 300 K is observed (the DB98 temperatures are hotter). This fact is probably related to the difficulty in estimating bolometric fluxes for early-type stars.

Cohenetal.(1978),Frogel etal.(1979,1981,1983); CFP

$\;$

The scale of temperatures defined by CFP has been adopted as standard in many studies devoted to the analysis of chemical abundances of giant stars and the calibration of stellar evolution models of the RGB. In the series of papers above mentioned, the authors provide effective temperatures for an extended sample of Red Giant Branch stars of globular clusters. The effective temperatures derived by CFP are based on atmosphere models and multicolour photometry (a brief description of the method used to derive temperatures is detailed in Frogel et al. 1981). In Fig. 8, we show the differences observed between $T_{\rm IRFM}$ and $T_{\rm CFP}$. In average, $T_{\rm CFP}$ are 56 K hotter than $T_{\rm IRFM}$ with a dispersion of 46 K. No apparent trend of the differences with $T_{\rm eff}$ or [Fe/H] is appreciated.

In summary, temperatures derived here are comparable with those derived by other authors. On the one hand, the differences and the dispersion of the differences found with other works based on the IRFM are within the error-bars of both accidental errors (i.e. uncertainty on the bolometric and monochromatic fluxes and other input parameters of the IRFM) and systematic errors (i.e. uncertainty in the absolute flux calibration as commented in Sect. 4.1 and different grids of model fluxes). On the other hand, the differences and the dispersion of the differences found with direct methods and surface brightness method are consistent with the combination of our internal error estimates and theirs.