The Magellanic Clouds show a large variety of bubbles, superbubbles and supergiant shells. All those prominent HII regions contain OB associations which allow us to study in detail very young stellar populations in an only moderate reddened surrounding and at a fairly well known distance. Besides the derivation of the ages of those associations, it is easy to derive the shapes of luminosity and mass functions from colour magnitude diagrams (CMDs). Due to the small ages of the associations the mass functions found from the main sequence stars are identical to their initial mass functions (IMF). In studying Large Magellanic Cloud (LMC) associations we have therefore a powerful tool for deriving the shape of the IMF over a large mass range. Furthermore, those shells are suitable laboratories to search for events of triggered star formation.
The OB association LH47 (Lucke & Hodge 1970)
is embedded in the superbubble N44 (Henize 1956).
N44
is also known as Shapley Constellation I (McKibben Nail & Shapley 1953).
The central shell of N44 is identical to DEM 152 (Davies et al. 1976)
and shell 1 of Meaburn & Laspias (1991).
Meaburn & Laspias (1991) studied the kinematics of N44 and showed that
the shell is expanding with 46 km s
towards the observer at the front side
and away from the observer with 33 km s
at the far side.
Kontizas et al. (1996) examined the distribution of the OB stars in
Shapley Constellation I and found that this complex
may be embedded in a larger aggregate of stars,
where they use Efremov's (1988) definition of complex and aggregate.
Stasinska et al. (1986) identified a very hot O star in the HII
knot N44C with a suspected temperature of about 70000 K.
Oey & Massey (1995) give a chart with the identifications of the different designations.
N44 is not only a conspicuous object in the optical and in the light
of the H
line, but also a prominent object in nearly all other
wavelength ranges.
N44 can be recognized in HI (Luks & Rohlfs 1992),
in CO (Cohen et al. 1988), in 
CO (Chin 1995)
and also in the CII line at 158 
m (Mochizuki et al. 1994).
Chin (1995) detected emission from N44 from the molecular transitions
around 
3 mm of HC
, HCN, HCO
, HNC, CS, and CN.
Elliot et al. (1978) published some narrow band images of N44. From the absence of non-thermal radio emission they concluded that supernova explosions do not play an important role in the formation of the shell. However, diffuse X-ray emission has been detected in N44 (Chu & MacLow 1990; Wang & Helfand 1991) with the Einstein Observatory. Chu & MacLow (1990) proposed that this emission is produced by supernovae when their shock fronts hit the walls of the shell. Chu et al. (1993) confirmed this scenario with pointed ROSAT observations.
The most recent CCD study of N44 is that of Oey & Massey (1995). They analyze the stellar population of the shell and find an age difference between the populations inside and outside the shell. They interpret this as a sign for triggered star formation. A comparison with a numerical model for the expansion of the shell, however, fails. This discrepancy can be understood, if the southern part of the X-ray bright part of N44 is the result of a blowout structure, as proposed recently by Magnier et al. (1996).
Massey (1985) and Massey et al. (1995a,b) pointed out that for bright blue stars the B-V colour is not a good temperature indicator any more and age determinations may suffer strongly from that. It is therefore essential that spectroscopic observations are performed besides wideband photometry. We thus follow here the same method as described already in Will et al. (1996) and took spectra at optical and UV wavelengths for some of the bright blue stars (see Sect. 2.2 (click here)).
The structure of this paper is as follows. After the description of the data in Sect. 2 (click here), we turn to the analysis of the optical spectra (Sect. 3.1 (click here)) and the UV spectra (Sect. 3.2 (click here)). In Sect. 3.3 (click here) we combine these results for a final classification of those stars which have been investigated spectroscopically. The age determination of the association is described in Sect. 4 (click here). Sections 5 (click here) and 6 (click here) are devoted to the luminosity and mass functions, respectively. In Sect. 7 (click here) we finally discuss our results.