We will first derive the age from our CMD by comparison with isochrones. This result will then be compared with the distribution of stars in the theoretical plane, i.e. in the Hertzsprung-Russell diagram (HRD).
The age of the association has been derived by fitting isochrones to the CMD which were calculated from the stellar evolutionary models of Schaerer et al. (1993). These models include moderate convective core overshooting and new opacities (Rogers & Iglesias 1992). We used the tracks for a metal content of Z=0.008, which is appropriate for most young LMC populations (see e.g. Spite et al. 1993; Hill et al. 1995). For the isochrone fit we used a LMC distance modulus of 18.5 mag (Westerlund 1990).
Using both CMDs the best fit to the youngest population
is reached for an isochrone with an age of 5 Myr
(
) and a reddening of 
0.22 mag.
In the V vs. B-V diagram also an age of 6 Myr would give a good fit.
The foreground reddening in this region of the LMC is very low:
from the map of Oestreicher et al. (1995) we find 0.04 - 0.06 mag,
the map of Schwering & Israel (1991) gives 0.06 - 0.08 mag.
Thus the major part of the reddening is due to LMC gas.
Due to the huge gas amount in the region (cf. Fig. 1 (click here))
this is not astonishing and we have to expect a strongly varying reddening
within our field of view.
Oey & Massey (1995) give a lower mean reddening of 
mag.
The CMD with the 5 Myr isochrone is shown in Fig. 7 (click here).
In the 
vs. 
plane the fit of the isochrone is quite good.
In the 
vs. 
plane the fit
near the turnoff region of the main sequence is bad.
The shape of the upper main sequence is strange and can not be fitted with
any isochrone.
There is also an unusual high scatter in the U-B colour.
A wrong U calibration does not seem to be very probable, because
first this would not produce scatter but a common shift of all stars,
and second the agreement with the U-B colour of
Lee (1990, Fig. 3 (click here)) is good.
Figure 7: CMDs for all stars in the field of view,
vs. 
(left panel) and
vs. 
(right panel), with an overplotted isochrone.
The isochrone is calculated from the modes of
Schaerer et al. (1993)
for an age of 5 Myr and a metallicity of Z = 0.008.
A distance modulus of 18.5 mag and a reddening of
mag have been assumed
Figure 8: Distribution of the reddening values derived from conversion
of the photometry to 
and 
.
The distribution peaks near 
mag, which is
the reddening derived from the isochrone fit
Figure 9: Map of the reddening derived from conversion
of the photometry to 
and 
.
Light grey values mean slight reddening, dark grey values
denote strong reddening.
The field is approximately the same as in Fig. 4 (click here).
Only stars with 
mag and Q < -0.4 mag have
been used for this map
The stars for which a spectral type was derived could easily be placed
in the HRD.
For all other stars we used the transformation equations given by
Massey et al. (1995b) to derive 
and 
.
In these equations the reddening for stars with Q < -0.4 mag is
derived from the photometry individually for each star.
Figure 8 (click here) shows the distribution of the derived reddening values.
The distribution shows a clear peak
near 
mag, thus confirming our fit in the observational plane.
There are some stars which show unusual high reddenings.
These are artefacts of an uncertain U magnitude of faint red stars.
Figure 9 (click here) shows a map of the reddening on the position of N44.
Due to the above mentioned uncertain high reddening values,
only stars with 
mag have been used to construct this map.
Also only stars with Q < -0.4 mag have been used because only for those
stars an individual reddening was derived from the photometry.
The map shows that the reddening in the inner part of the shell is not
very strong and near the mean value of 0.22 mag.
South-west of the shell, between N44B and N44C, we find a peak
of the reddening which could originate from the higher gas and dust content
of this region (see Fig. 1 (click here)).
Figure 10 (click here) shows the HRD with overlayed tracks (left panel) and isochrones (right panel) from the models of Schaerer et al. (1993). The stars hotter than the main sequence band are faint stars with uncertain, i.e. too blue colours. Along the entire main sequence we recognize the same effect that we have already seen in Fig. 7 (click here). There is no single isochrone which gives a good fit, instead the stars spread over the entire main sequence band. This spread has the same reason than the scatter in the V vs. U-B CMD (Fig. 7 (click here), right panel), i.e. an unusual high scatter of the bright stars in U-B, which is brought into the HRD via the transformation equations. We do not understand this high scatter, but it is also present in the HRD of Oey & Massey (1995, their Fig. 4 (click here)) and thus may be an intrinsic feature of LH47.
From the CMDs we find an age of 5 to 6 Myr for LH47 in N44. Oey & Massey (1995) find 10 Myr for LH47 (inside the central shell of N44), Kontizas et al. (1996) give an upper limit of 3 Myr for the associations LH47, 48 and 49.
Oey & Massey (1995) find a significant difference in the
ages of the stars inside (10 Myr) and outside (5 Myr) the shell.
We do not see such a difference.
From Fig. 4 (click here) it is clear that there are some bright
stars in the very centre of the shell.
Those stars include N44:01 and N44:05, for which we do have spectra.
Our spectral classification of those stars
is in good agreement with that of Oey & Massey (1995).
These two stars are as young as 5 Myr, thus implying that there is no
age difference between the interior and exterior of the shell.
However, they could still be located at the border or outside of the shell,
so that they are only projected in front of the centre.
In Fig. 11 (click here) we compare the CMDs of the stars inside
and outside the shell.
Inside means all stars nearer than 
to the center and outside
means all stars more far than 
from the center of the shell.
In this way we selected only stars which are really inside or outside
the gas ring of N44, omitting stars which fall on the gas ring.
The 5 Myr isochrone is also shown for easier comparison of both plots.
As we concluded already from the HRD, we do not see a difference of
the ages of the stellar population inside and outside the shell.
Due to the calibration problem in the work of Oey & Massey (1995; see Sect. 2.1 (click here)), the age difference found by those authors may be an artefact of the photometric offset between their CCD fields. This seems very plausible because of the fact that a distinction between inside and outside the shell in their photometry may be traced back to a different calibration of their CCD fields which had only a narrow overlap.
The very hot star observed by Stasinska et al. (1986, their #2)
is also included in our photometry.
For this star we find 
mag, 
mag and
mag.
After transformation to the theoretical plane, this leads to a temperature
of 38500 K (i.e. 
) and 
.
Its reddening is 
mag and thus only slightly higher than the
average reddening.
The star lies well in the main sequence band,
confirming the classification as a normal O star by Oey & Massey
(1995; their lh47-338) and Pakull & Motch (1989) but not confirming
the very high temperature given by Stasinska et al. (1986).
Figure 10: HRD of the stars in N44.
The stars for which spectra exist are marked by the larger symbols.
On the left panel tracks with Z =0.008 from
Schaerer et al. (1993)
for masses from 120 down to 2.5 
have been overplotted
(masses = 120, 85, 60, 40, 25, 20,
15, 12, 10, 7, 5, 4, 3, and 2.5 
).
On the right panel isochrones for ages of 4 Myr to 100 Myr
(log age = 6.6, 6.7, 6.8, 6.9, 7.0, 7.1, 7.2, 7.5, and 8.0)
calculated from those tracks have been overplotted
Figure 11: Comparison of the CMDs of stars a) inside and
b) outside the shell of N44.
An isochrone with an age of 5 Myr is overlayed.
A reddening of 
mag has been assumed for both plots
One of our goals of the spectroscopy and the transformation of the
photometry to 
and 
was to refine the age determination
from the pure BV photometry.
This was not as successful as we expected due to the not understood scatter in
the U-B colour, which is transformed also into the HRD.
The inspection of Fig. 10 (click here) however shows that the stars with
spectra suggest a minimum age of the association
of 4 or 5 Myr, thus confirming our result from the photometry.