1 Introduction

Open clusters are natural laboratories for stellar evolution, because stars in clusters share a common origin from the same progenitor molecular cloud. Young open clusters are particularly suited to study the processes of star formation and early evolution of stars over a wide mass spectrum. Many questions related to young clusters and/or associations remain to be answered, e.g., the constancy of IMF (at higher masses), the coevality of star formation, the evolution of massive stars and its influence on the environment, and the presence of various peculiar stars (emission-line stars, for example) and their evolutionary status, to just cite a few. The study of clusters and associations deeply relies upon the accumulation of various kinds of precise data, be it astrometric, photometric or spectroscopic. While large efforts have been spent on obtaining magnitudes and colours in various systems such as UBV and uvby for stars in clusters, the situation about spectroscopic data is still far from satisfactory. Reliable MKK spectral types, for example, are often available in published literature only to few brightest stars of a cluster (Mermilliod 1986).

Accurate two-dimensional spectral classification (temperature type and luminosity class) of stars is essential to the study of young open clusters. Differential reddenings often affect member stars in young clusters, causing scatters in color-magnitude diagrams. Spectroscopy provides the means of precisely detecting any differential reddening. By combination with photometric data (magnitudes and colors), spectral classification may be used to construct individually-dereddened high-quality H-R diagrams, from which the distances and evolutionary ages of clusters can be reliably determined and even the effects of binarity (usually indistinguishable from small diffential reddenings) may be examined. Spectral classification can also be used to investigate the nature of reddening and extinction, determining if any abnormality exists and if the differential reddening is caused by interstellar, intracluster or circumstellar extinction material (Sagar 1987; van den Ancker et al. 1997). Peculiar stars (e.g., emission-line stars) may also be discovered from spectral classification.

Traditionally spectral classification is made photographically in the blue spectral region of 3900 Å through 4900 Å. Reliable two-dimensional spectral types can be gained only for brighter stars due to the low sensitivity of photographic plates. With the use of CCDs spectroscopy can be done on much fainter objects and on a broader spectral coverage. The peak of the quantum efficiency of CCDs is usually in the yellow-red spectral region, which is advantageous to the observation of late type stars or highly-reddened early type stars such as those in young open clusters. With the spectral coverage extended from the blue into the yellow-red and even the near infrared, additional spectral features suitable for stellar classification also appear in these regions of longer wavelength (Jacoby et al. 1984; Turnshek et al. 1985; Torres-Dodgen & Weaver 1993; Danks & Dennefeld 1994; Shi & Hu 1999).

We have begun a project of systematically investigating the properties and stellar contents of young open clusters and/or associations, especially those with severe differential reddening. The study is to be based upon more complete and homogeneous sets of combined data. As a part of this program and the first step, we engage ourselves in making accurate two-dimensional spectral classification for large samples of stars in young clusters. The first three clusters selected by us are IC 1805 (C0228+612, $\alpha$$_{1950}=02^{\rm h}28\hbox{$.\!\!^{\rm m}$}9$, $\delta$$_{1950}=61^\circ$54'), NGC 654 (C0140+616, $\alpha$$_{1950}=01^{\rm h}40\hbox{$.\!\!^{\rm m}$}6$, $\delta$$_{1950}=61^\circ$38') and NGC 6823 (C1941+231, $\alpha$$_{1950}=19^{\rm h}41\hbox{$.\!\!^{\rm m}$}0$, $\delta$$_{1950}=23^\circ$11'). They are of very young ages and all showed differential reddenings according to earlier investigators (Moffat 1972; Turner 1979; Stone 1980; Sagar & Joshi 1981; Joshi & Sagar 1983a,b). Most of the cluster stars lack MKK types. In this work a total of 263 stars from these three clusters were spectroscopically observed by slit spectrograh and CCD detectors in the blue and yellow-red spectral regions. Results of the MKK classification are given in this paper. The presence of emission-line stars is discussed. Further studies upon these clusters are underway.