This is the third Paper of a series containing near infrared photometry of a
sample of IRAS sources with far infrared colours similar to those shown by
planetary nebulae (PNe, hereafter). In the previous ones, Manchado et al.
(1989a; Paper I) and
García-Lario et al. (1990; Paper II) showed
preliminary results obtained on a subsample of these IRAS sources. The
selection criteria were previously described in Papers I and II. They are
based on the fact that PNe and objects in the transition phase from the
AGB to the PN stage are strong infrared emitters with peculiar
infrared colours. Thus, it is possible to separate a region in the IRAS
two-colour diagram 
vs. 
where most of
the objects found are PNe or objects connected with this evolutionary stage.
Only a small overlap exists with other objects, like Seyfert galaxies,
compact H II regions and T-Tauri stars (Pottasch et al. 1988). The complete
sample satisfying our selection criteria contains 1084 sources (García-Lario 1992), all them well detected in at least the 12, 25 and
60 
m bands. About 50% of them are previously unidentified objects or
poorly known in the
literature. Among the well identified ones, we find a majority of PNe
(49%) and other post-main sequence stars (27 small fraction of young stellar objects (17%) and active galactic nuclei
(7%).
The number of objects known in the short transition phase which precedes the formation of a PN is very small. The final aim of this work is to increase this number through the detection of new candidates among the infrared sources included in the IRAS Point Source Catalogue (PSC) with no previous identification and the adequate colours. Many of them are expected to be heavily obscured by the thick circumstellar envelopes formed during the AGB phase. The determination of the near infrared counterpart is, thus, the natural extension toward bluer wavelengths in the study of these sources and essential for subsequent studies in this or other spectral ranges.
The near infrared photometry, as we proved in Papers I and II, can be used
to determine whether the main origin of the emission observed is stellar,
nebular or due to the dust present in the circumstellar envelope. According
to this, we can try to identify the nature and evolutionary stage of the
sources observed since each type of object in the sample shows characteristic
near infrared properties which can be used to recognize them.
Many PNe, for instance, are known to
exhibit a characteristic J band excess due to the presence of an emission line
of He I at 1.083 m (Whitelock 1985). Also very late-AGB stars and heavily
obscured post-AGB stars can be easily recognized because of their extremely
reddened near infrared colours (Le Bertre 1988;
van der Veen et al. 1989).
In many cases, however, the identification of a given source is not
possible based on near infrared data alone. Compact H II regions, T-Tauri stars
and active galaxies which, as we have refered above, are also present in
our sample, sometimes show near infrared colours very similar to
those observed in late-AGB and post-AGB stars. In this case, the combination
with information obtained in other spectral ranges or the use of
additional criteria is needed. Sometimes the problem is easy to solve, as
for OH/IR stars, which are characterized by the presence of a
double-peaked OH maser emission at 1612 MHz. Unfortunately, this emission
is usually not observed in more evolved stars. In addition, optically bright post-AGB
stars show no or very little near infrared excess and, thus, are
very difficult to distinguish from foreground sources. Confusion is
a major problem specially when observing towards the galactic bulge,
where fields are frequently crowded.
Apart from a few exceptions, observations have been made only for sources satisfying our selection criteria with no previous near infrared measurements. They are described in Sect.2. (click here). In Sect.3.1 (click here) we analyse the near infrared properties of the various types of stellar objects found among very well identified IRAS sources included in our sample for which data are available in this spectral range in the literature. These data will be used for comparison in our analysis of the unidentified objects. The same kind of analysis is done with their far infrared emission in Sect.3.2 (click here) , as derived from IRAS data. The results obtained are shown in Sect.4. (click here) . Combining the characteristics of the near infrared emission with the IRAS properties and a few additional criteria we classify all the sources observed so far, including those for which the near infrared photometry was previously reported in Papers I and II. The conclusions derived are presented in Sect.5. (click here) .
This work is part of a wide multiwavelength survey which combines infrared data with optical spectroscopy (whenever an optical counterpart is found) and radio continuum observations. The analysis of the results obtained in other wavelength ranges will be published elsewhere (van de Steene et al. 1996a,b; García-Lario et al. 1997a). Optical and near infrared finding charts together with the accurate coordinates of the counterparts found will be presented in forthcoming papers.