In a previous paper, Taylor has found that the solar value of (R - I)C is 0.337 +/- 0.0024 mag. This result can be converted to a value of (B - V). by combining it with blanketing corrections and a Hyades color-color relation. The latter are derived and discussed in this paper. The resulting value of (B - V). turns out to be 0.633 +/- 0.009 mag. Annoyingly, this datum differs from the mean of a number of previous results at better than 99.5% confidence. A survey of the problem reveals a wavelength boundary: Hbeta and Halpha yield the result quoted above, while spectroscopy at wavelengths shortward of Hbeta yields [(B - V).] = 0.665 +/- 0.003 mag. The Balmer-line derivations are largely blanketing insensitive, so for the present, it seems defensible to assume that the Balmer-line result is correct and that its short-wavelength counterparts are affected by some unknown property of the solar blanketing.

A calibration presented in a previous paper is used in this paper to derive temperatures for FGK stars near the main sequence. The calibration is checked against published counterparts, and it is found that previous calibrations have not established K-dwarf temperatures in particular beyond reasonable doubt. The database assembled to derive the temperatures is described, and the problems posed by close binaries are evaluated. The newly derived temperatures are used to check a line-depth ratio proposed as a thermometer by Gray and Johanson (1991, PASP, 103, 439), and it is found that the ratio is metallicity-sensitive. Temperatures are given for a total of 417 stars.

This paper presents "primary" [Fe/H] averages for 373 evolved K stars (see section 4). These stars are of luminosity classes II-IV, and their values of [Fe/H] lie between -0.9 and +0.21 dex. The contributing data are from 58 high-dispersion papers and three spectrum-synthesis papers. The data have been selected (and in many cases, corrected) to be as free as possible from systematic error. References are given with the averages to encourage users to cite original sources. For the stars considered, the data define a "consensus" zero point with a precision of +/- 0.018 dex. In addition, analysis yields rms errors per datum which are typically 0.08-0.16 dex. There is no dependable reduction in the sizes of these errors when sensitive detectors such as Reticons are used. The averaged values of [Fe/H] reflect the consensus zero point, and they are based on weights which follow from the rms errors. The primary data base makes recalibration possible for two large [Fe/H] catalogs. Temperature corrections are necessary for both catalogs, and, while the corrections are 0.12 dex or less for the data of Hansen & Kjaergaard, they are as large as 0.4 dex for a recent survey of Brown et al. The corrected data are included in a "secondary" data base for 669 stars without primary data. Smaller data sets from two additional papers are also included in this data base. Weighting for the secondary data is again based on rms errors. Also given in section 4 are a set of [Fe/H] standard stars and a new DDO calibration. The rms errors for the standard-star data are 0.07 dex or less. These data and the DDO calibration reflect the consensus zero point. For normal K giants, CN-based values of [Fe/H] turn out to be more precise than many high-dispersion results. Some zero-point errors in the latter are also found, and new examples of continuum-placement problems appear. It therefore appears that, at present, high-dispersion results are not invariably superior to photometric metallicities. In section 5 a review is given of high-dispersion and related work on supermetallicity in K III-IV stars. The super-metal-rich (SMR) status of one subgiant (31 Aql) is found to be well established. Despite some assertions in the literature, however, no decisive high-dispersion case can be made at present for the existence of SMR K giants. Such stars may exist, but deciding this question will require further work.