1 Introduction

The precise relationships which link the physical parameters of stellar atmospheres (i.e. $T_{\rm eff}$, log(g) and [Fe/H]) with measured colours, photometric indices or spectral features of stars play a central role connecting several areas of stellar physics. In particular, the accurate calibration of the scale of stellar effective temperatures is of increasing importance in spectroscopic studies since $T_{\rm eff}$-colour or -spectral type calibrations are usually adopted in deriving chemical abundances. On the one hand, noteworthy differences in the derived ratios between different elements are obtained when analysing the same spectra with different temperature scales (e.g. King 1994), thus implying quite different scenarios for the chemical evolution of the galaxy. On the other hand, non-negligible variations in primordial abundances may be deduced from the adoption of diverse temperature scales in spectroscopic analysis (e.g. Bonifacio & Molaro 1997). Stellar effective temperatures are also a most influential consideration in the observed differences of the giant branch and main sequence slopes when comparing empirical lines of globular clusters with theoretical isochrones (e.g. Bell 1992). Here a part of the discrepancy might be induced by the adoption of an incorrect temperature scale (since this is crucial in the transformation of the HR diagram from the theoretical to the observational plane). Furthermore, the effective temperature scale is necessary for the analysis of the global behaviour of stellar atmosphere models (e.g. Castelli et al 1997), in particular it is a relevant parameter for the modelling of molecular bands in the synthesis of infrared colours (e.g. Bell & Gustafsson 1989). Finally, it is also important for the overall colour synthesis of stellar populations (e.g. Vazdekis et al. 1996).

We have carried out these observations as a part of a continuing programme aimed at the (semi-)empirical calibration of $T_{\rm eff}$ as a function of photometric colours, [Fe/H] and $\log (g)$ for population I and II stars. The part of the programme concerning stars with $\log (g)\, \raisebox{-0.6ex}{$\stackrel{\textstyle \gt}{\sim}$}\, 3.5$ (i.e. the main sequence and the blue part of subgiant branch) has been recently accomplished, and is described in Alonso et al. (1996a,b; Papers III and IV respectively). As for the part of the programme devoted to the giant branch temperature scale, we have selected a sample of $\sim$400 stars with spectral types from F0 to K5 covering a wide range in metal content (+0.5 > [Fe/H] > -3.0). In order to obtain their effective temperatures, we will apply the Infrared Flux Method (Blackwell et al. 1990), which has proved useful for deriving stellar effective temperatures of metal-poor giants of globular clusters (Arribas & Martínez-Roger 1987a; Arribas et al. 1991). Furthermore, this method relies only weakly on theoretical models and its main requirement from the observational side is the measurement of accurate photometry needed to derive near infrared monochromatic fluxes. In this paper, we present and discuss the results of the observational campaign of near-IR photometry for the stars of the sample lacking previously measured JHK(L') magnitudes, and also for stars common to other systems which will be necessary to homogenize the photometric measurements taken from the literature. The infrared magnitudes will be used in a later stage of the work to derive IR monochromatic fluxes for the application of the Infrared Flux Method.