**Table 2:** Adopted temperature scales with validity range in MK spectral type, luminosity class, method of construction, intrinsic errors and validity range of the intrinsic errors. We distinguish between calibrations which are based directly on effective temperature measurements and those only derived from calibrations of other authors. We refer to the former as *primary* and to the latter as *derived* scales. Scales based on older calibrations but that account for new measurements are indicated by "improved''. The origin of the intrinsic errors is explained in Appendix A.
reference	sp. type	lum. class	construction	intrinsic errors	validity range of int. errors
Bessell (1979, Table 2)¹	B7-M 6	dwarfs	primary	220 K²	earlier G2
				$\Delta{\log}T_{\rm eff}=0.035^2$	later than G2
					and earlier than K7
Bessell (1991, Table 2)¹	K 7-M 7.5	dwarfs	primary	290 K²	whole range
Bessell (1979, Table 3)	G7-M 6	giants	primary	220 K²	whole range
Cohen & Kuhi (1979, Table 7)	O9.5-M 6	dwarfs	derived	300 K²	whole range
de Jager & Nieuwenhuijzen (1987, Table 5)	O3-M 9	dwarfs	primary	$\Delta{\log}T_{\rm eff}=0.021^3$	whole range
de Jager & Nieuwenhuijzen (1987, Table 5)	O3-M 9	subgiants	primary	$\Delta{\log}T_{\rm eff}=0.021^3$	whole range
de Jager & Nieuwenhuijzen (1987, Table 5)	O3-M 9	giants	primary	$\Delta{\log}T_{\rm eff}=0.021^3$	whole range
Hartigan et al. (1994, Table 4)	F0-M 6	dwarfs	derived	$\Delta{\log}T_{\rm eff}=0.015^2$	earlier than K7
				290 K²	later than K7
Kenyon & Hartmann (1995, Table 5)	B0-M 6	dwarfs	improved	$\Delta{\log}T_{\rm eff}=0.015^2$	whole range
Perrin et al. (1998, Table 5)	G8-M 8	giants	primary	220 K²	earlier than M 6
				79 K³	M 6
				94 K³	M 7
				42 K³	M 8
Luhman (1999, Table 2)	M 1-M 9	giants	derived	270 K²	earlier than M 7
				100 K²	M 7
				50 K²	M 8
				150 K²	M 9
Luhman et al. (2003, Fig. 8)	M 1-M 9	dwarfs	improved	80 K²	whole range
Luhman et al. (2003, Table 8)	M 1-M 9	intermediate	improved	80 K²	earlier than M 6.5
				100 K²	later than M 6.5
Tokunaga (2000, Table 7.6)	O9-M 6	dwarfs	derived	100 K³	later spectral types

¹We only used a combination, i.e. Bessell (1979) for spectral types earlier than K7 and Bessell (1991) for spectral types later than K7.
²Intrinsic errors were derived by us or taken from another work.
³Intrinsic errors were adopted from the original work.

Source LaTeX | All tables | In the text

	1 Introduction
	2 A new method - extending the approach of White et al. (1999)
	3 Application to known eclipsing binaries
	4 Summary and conclusions
	References
	Online Material
	Appendix A: Derivation of intrinsic scale errors
	Appendix B: Conversion tables