To arrive at our goal of detecting PMS stars, we made use of many observations, of which the numerous photometric and spectroscopic datasets are the most important. With the different photometric datasets we were able to sort out the poor quality observations and measurement errors as well as to select observations taken at the same brightness level. This provided us with reliable data for investigating the SED of each programme star. Combined with the fact that we determined the spectral types of the cluster members using different TCDs combined with spectra, we were able to discuss and determine accurate stellar astrophysical parameters. The photometric data was dereddened with the use of these parameters and was fitted to an appropriate theoretical Kurucz (1991) model.
A first result of this fitting procedure was the detection of an individual
extinction law for each star. We analyzed the extinction laws and concluded
that they are not correlated with the stellar spectral type, nor they depend
on the location of the star in the cluster field. Indeed, stars located
near the central area, do show more often a higher 
value,
i.e. anomalous
extinction. However, the normal and anomalous extinction can sometimes be
found in other regions of the cluster as well, and varies from one star to a
neighbouring one. The anomalous extinction can, therefore, be explained as
being a very local effect or patchy in nature. Possibly, the stellar
radiation influences the
circumstellar or nearby intracluster material. The finding of individual
extinction laws rather than an averaged one for the whole cluster, is in
strong contradiction with previous studies. In such studies, the extinction
law was based on TCDs. However, as shown by Steenman & Thé
(1989), an extinction law is not defined by three colours, but it must be
determined by using colours from the blue to the near-IR.
The approach which makes use of the TCDs to determine extinction laws and or
values is therefore only indicative. Our approach affects the
determination of the stellar 
and 
and hence alters the stellar
location in the HRD and conclusions derived from it.
The SEDs are also used to identify peculiarities in the energy distribution of each star. Likewise, we identified weak and strong IR-excesses and even depletions in the dereddened near-IR. The near-IR depletions could be fitted to Kurucz models by taking higher luminosity classes for late type objects, but could also be due to strong absorption bands or observational errors. The weak near-IR excess could be well fitted with anomalous extinction. A strong near-IR excess was found for nine stars. Possible explanations are the detection of circumstellar dust or nearby or background M-type stars, located in the aperture used for the photometry. However, free-free emission, due to the gaseous inner parts of circumstellar material, could not be excluded as the cause of IR-excess.
One of the main PMS characteristics is the occurrence of a near-IR-excess due
to thermal dust emission of (a) circumstellar shell(s). Analyzing the
temperatures of each near-IR-excess, by comparing them to black-bodies, we
concluded that W245, W266, W494 and W605 are the only ones that
could have circumstellar dust. The other five strong near-IR emitters have
either an M-type companion, need very high 
values (very
anomalous extinction) or could have a large amount of free-free emission.
Checking the programme stars for other PMS characteristics, we noted 29 objects
with possible stellar emission lines in their spectra.
Of the stars that possibly have a shell around
them are W245, W339, W374 and W494 emission line stars.
W266, W273 and W299, three other stars
with a strong near-IR excess, probably have emission lines. Another
PMS characteristic, the variability of the stellar brightness, could not be
analyzed due to an insufficient number of data points for most stars. However,
considering the ranges for the stars having a strong near-IR-excess, besides
W339 and W494, all have 
. Because W339
and W494 are B-type stars, this result is not in contradiction with the
results of Bibo & Thé (1991) regarding the variability of PMS stars.
To verify whether we have detected PMS objects, young stars,
we have to place them in an HRD.
After computation of 
, using the individual 
value, and 
for each star, we composed an HRD in which we could identify three stellar
groups from their location in this diagram, see Fig. 5 (click here). Group I, the
early-type stars, are located on or near the MS. Group II consists of
stars having a later spectral type, up to late G, whereas Group III
consist of stars having even a later spectral type or a higher luminosity.
Group III stars are all identified to be foreground objects, due to their low E(B-V), their low probability of membership and their location in the HRD, which cannot be explained by stellar evolution considering the probable youth of the cluster. This confirms only part of the conclusion of Chini & Wargau (1990) that the Walker (1961) PMS candidates are all foreground objects.
More interesting are the Group I and Group II
objects. Group I, the early-type stars, coincides with the
Hillenbrand et al. (1993) sample which
has a cut-off at A0. They are all located near the transformed
ZAMS from Hillenbrand et al. (1993),
using and confirming the cluster distance of 2.6 
0.2 kpc,
as derived in Thé et al. (1990). Furthermore, we do not
find any stars located to the very right of the MS-band in our
pre-MS sample, Fig. 9 (click here)a. But if we include the MS-sample of
Thé et al. (1990), Fig. 9 (click here)b shows three objects located
right of the MS having an age of about 
in case
they are evolving towards the red-giant phase. Adopting an age
of 5 Myr of the cluster (Thé et al. 1990), we must have
located a clear turn-off point at about 25 
in Fig.
9 (click here)b or Fig. 8 (click here) of Hillenbrand et al.
(1993). The early-type stars located at the
right of the ZAMS, close to the upper stellar birthline, must
therefore be younger. If they are still in their pre-MS phase,
it is remarkable that they do not show typical HAeBe
characteristics.
HAeBe candidates, however, are found among the lower luminosity early-type stars, four have a very strong near-IR emission. Two of these stars have IR-excess temperatures possibly indicative of dust (W245 and W494), the other two, W262 and W273, have excess temperatures of roughly 2500 K, which is an upper limit. We conclude from these findings that the emission line stars W245 and W494 are indeed HAeBe stars. As we noticed, they have the PMS characteristics as mentioned before. W262, without any emission lines, and W273, do show some PMS properties, but must be studied further for their IR-properties. W262 has an IR-excess due to stellar dust and it would be interesting to verify whether the spectrum is indeed without any emissions as can be disputed from Fig. 1 (click here). The PMS candidates we found are of spectral type B6 or later. Compared to the location right of the ZAMS and the evolutionary tracks of Palla & Stahler (1993) these objects cannot be older than 1 Myr.
The situation is more complex for the Group II objects. Some
of these stars do fulfill all membership criteria, whereas others do not,
see Fig. 5 (click here). All have a peculiar near-IR characteristic, W605 and
W266 have CS dust emission, the other ones have anomalous
extinction laws. Two objects of
Group II have a very low E(B-V), the other four have not. The location
of the Group II objects, very close to the stellar birthline with the
highest accretion rate, can be explained by the hypothesis that these
stars have just been formed. Compared to
the calculations of Palla & Stahler (1993), pre-main sequence objects
of 3.5 to 5.0 
younger than a few 0.1 Myr can occupy
this region of the HR-diagram.
Normally, such stars are hardly visible
due to their very high 
as they are very much embedded in their
post-natal cloud. For the stars of Group II we could not find such values of
, except for W213 (
).
Due to the fast rate of their formation, there must be an
efficient mechanism clearing the post-natal and shell or disk material. If
a shell or disk is formed, as proposed by Hillenbrand et al. (1993), this
must happen within 1 Myr. Such an effective clearing of the circumstellar
environment of the young star supports the hypothesis that at least part of
the Group I objects
are also in their pre-main sequence phase without showing typical HAeBe
characteristics, normally due to circumstellar material.
The mentioned objects of Group I and Group II do fulfill the
stellar birthline of Palla & Stahler (1993) for protostars with an accretion
rate of 
yr
.
As these objects must have ages of about a few times 0.1 Myr
and do not show evidence of any circumstellar material, it could be that
the disk survival time is of an order of 
, if a disk is formed from a shell left over at birth,
or it seems that the NGC 6611 stars are built up at a
faster rate than generally assumed for most known HAeBes.
Objects of different ages are found in NGC 6611, indicating that the sample objects are not coeval. Hence, we conclude that star formation has an age spread. This was also indicated by Hillenbrand et al. (1993), who discovered many embedded objects, probably in formation as they are still optically thick. It would be of high interest to study whether the natal material is enriched and what are its effects on the early stellar evolution compared to the young evolved objects. The effective clearing of the CS environment could then be due to stellar evolution on a shorter time scale due to the presence of an enhanced metallicity.
Note that the ``gap'' between Group I and Group II objects can be explained by the differences in age of the cluster population. As the lower-MS objects are about a few Myr, the ones of Group II are significantly younger. In fact we have an age-gap. However, we have to note the few objects in this part of the HRD as we suffer from an observational cut-off at lower magnitudes and a cut-off in the Hillenbrand et al. (1993) paper at A0. An additional study of objects of A- and F-type stars being in their PMS-phase must be considered as a main follow-up study. Studying these objects, filling the ``gap'', could give more insight in the theory and understanding of stellar formation and especially the way they become optically visible. Note in this respect the discussion in Hillenbrand et al. (1993) that many emission-line objects in NGC 6611 look like classical Ae/Be type stars rather then having typical HAeBe characteristics. This adds to the confusion of the definitions for these different groups of emission-line objects. The efficient clearing of the dust of the young objects in NGC 6611 could explain why we observe young classical-like Ae/Be type stars.
A possible way to clean the CS environment of young stars could be by the radiation of the already evolved massive early type members of the cluster. Especially as the central part of NGC 6611 is crowded with such objects, the UV radiation field will be very destructive. In this respect the discovery of evaporating gaseous globules (EGGs) by Hester & Scowen (1995) could also well explain our findings as such bodies probably do not contain enough angular momentum to distribute the post-natal material from a shell to a disk-like structure. The impact of star formation by EGGs in this cluster on our results will be discussed in more detail in a forthcoming paper.
Although we have reached our ultimate goal by finding at least three PMS Be-candidates, probably typical HAeBe stars, our main conclusion is that we just started to understand the stellar formation and early evolutionary phases and they appeared to be different and more complex than generally assumed.