1 Introduction

Classical Be stars are observed to be rotating close to their break up speeds (Slettebak 1982) with a modal value of $\sim0.7v_{\rm crit}$ (Porter 1996), where $v_{\rm crit}$ is the stars' critical break-up velocity. The distribution of circumstellar material giving rise to the emission features in their spectra (leading to their designation "e'') is concentrated in the equatorial plane (Dougherty & Taylor 1992; Quirrenbach et al. 1994). Emission line profiles, e.g. see Dachs et al. (1986), Hummel & Vrancken (2000) are most easily explained if it is assumed that there are Keplerian motions within a disc. Observations of Be star discs have shown that some can vanish and re-appear in a apparently random fashion with time-scales of 100s of days e.g. Hirata 1995; Hanuschik et al. 1993). The major problem in Be star research is to identify the physical processes which produce and disperse the discs.

Theoretical models which have been presented include wind bi-stability (Lamers & Pauldrach 1991), wind compressed discs (Bjorkman & Cassinelli 1993), and viscous outflowing discs (Lee et al. 1991, also see Porter 1999). Each of these models has problems: the wind models (bi-stability and wind compression) cannot reproduce the Keplerian rotation in the disc as the wind conserves angular momentum leading to a rotation law similar to $v_{\phi}\propto 1/r$ (see Owocki et al. 1994). Viscous outflowing discs successfully reproduce most of the attributes of the observed Be star discs (e.g. continuum excess Porter 1999, V/R variations in the emission lines Okazaki 1997), although the major problem is that the required source of angular momentum to sustain the disc remains obscure (a tentative suggestion has been made that the angular momentum source relates to non-radial pulsations, Osaki 1986).

Consequently our understanding of Be stars is still incomplete. Theoretical models can only be constrained, or indeed ruled out, if their predictions can be confronted with observations of well understood Be star samples. Hitherto, a homogeneous data set involving extensive wavelength coverage across all Be spectral types has been lacking.

In a series of papers we have been addressing that demand by defining and observing in a homogeneous fashion a representative sample of Be stars. The sample contains objects from O9 to B8.5 and of luminosity classes III (giants) to V (dwarfs), as well as three shell stars. It was selected in an attempt to contain several objects that were typical of each spectral and luminosity class in the above range; it therefore does not reflect the spectral and luminosity class space distribution of Be stars, but only the average properties of each subclass in temperature and luminosity. A spectral type and measure of $v \sin (i)$ were derived for each object in the sample and were presented in Steele et al. (1999, hereafter Paper I). In Clark & Steele (2000, hereafter Paper II) we presented K band spectroscopy of the sample, and in Steele & Clark (2000, hereafter Paper III) H band spectroscopy. In a forthcoming paper, Steele & Negueruela (2001, hereafter Paper V) will present spectra in the regions of the H$\alpha$ and Paschen series lines.

This paper presents infrared JHK photometry of the sample and a new technique for separating out the interstellar reddening and circumstellar excess. We then go on to correlate the derived circumstellar excess of the objects in the sample, their emission line strengths, spectral types, and rotational velocities. Throughout this paper we shall be using non-parametric correlation analysis (using the Spearman rank correlation coefficient; e.g. Press et al. 1992) to study these relationships. In this way we impose no assumptions about the form of the dependence between parameters.

In Sect. 2 we describe our observations and how we carried out the preliminary data reduction. Section 3 describes the process we have used to separate the circumstellar excess and interstellar reddening associated with observations of Be stars. In Sect. 4 we present the correlations we have found between the separated excesses, reddenings and other observed quantities. Finally Sect. 5 presents our conclusions.