Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~lt/preprints/maser1/preprint.ps.gz Äàòà èçìåíåíèÿ: Tue Sep 11 16:56:41 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 07:42:05 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: photosphere

A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08(13.09.3; 09.13.2; 08.06.02)
ASTRONOMY
AND
ASTROPHYSICS
22.2.1994
Near infrared images of galactic masers:
I. Association between infrared sources and masers.
L. Testi 1 , M. Felli 2 , P. Persi 3 , and M. Roth 4
1 Dipartimento di Astronomia e Scienza dello Spazio, Universit`a di Firenze, Largo E. Fermi 5, 50125, Firenze, Italy
2 Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125, Firenze, Italy
3 C.N.R., I.A.S., C.P. 67, I­00044 Frascati, Italy
4 Las Campanas Observatory, Casilla 601, La Serena, Chile
Received date; accepted date
Abstract. We present the first results of an extensive
near infrared survey of galactic H 2 O and OH masers in
high--luminosity star forming regions (SFR) aimed to see
if there are near--infrared (NIR) sources directly associ­
ated with the masers. Seventeen fields for which accurate
VLA positions of the masers were available have been im­
aged in the three J, H, and K NIR broad band filters with
pixel resolution of 1:34 arcsec=pixel and a field of view
of roughly 3 0 \Theta 3 0 . All observed fields show a high den­
sity of K--band sources, completely undetected in previ­
ous surveys, probably stellar clusters located in the SFR.
From numerical simulations we find that the distribu­
tions of the observed first--neighbour K--band source to
the maser is very unlikely due to chance coincidence with
uniformly distributed field sources. For this reason, the
infrared source nearest to the maser ( !
¸ 10 arcseconds)
is considered to be associated with the maser. All these
sources have distinctive characteristics: they are weak and
detected only in K, or if the H magnitude is measur­
able, they show an H\GammaK colour index greater than 2.
Although not in all sources there are high sensitivity--
high resolution radio continuum observations, only few
of the K--band/maser sources are closely associated with
known ultracompact (UC) H ii regions. After consider­
ing several plausible alternatives we find that the ob­
served NIR emission is produced by a young stellar ob­
ject (YSO) surrounded by a dusty circumstellar envelope.
In the evolutionary scheme of SFR this result places the
NIR/maser sources in a stage preceding that of UCH ii
regions, in which the radio continuum from ionized gas is
undetectable with present sensitivities either because so
much reduced by self--absorption or by dust absorption of
stellar UV photons in the very dense envelope of the YSO
or intrinsically weak due to low UV photon fluxes.
Send offprint requests to: M. Felli
Key words: Infrared arrays -- H 2 O and OH masers -- Star
formation
1. Introduction
Since their discovery, more than twenty five years ago
(Weaver et al. 1965), H 2 O and OH masers have repre­
sented an interesting problem for astronomers. In partic­
ular the 6 16 \Gamma 5 23 water maser transition (22 GHz) and
the ground state OH masers (at 1612, 1665, 1667, and
1720 MHz) observed in many high mass star forming re­
gions, usually show brightness temperatures (10 11\Xi15 K)
that are difficult to model theoretically (Elitzur 1982). Be­
sides the intrinsic interest in the interstellar masing pro­
cess, the study of masers can be a very powerful method
of probing the evolutionary conditions of high--luminosity
young stellar objects (YSO) and the dynamics of their
interaction with the surrounding medium (Elitzur 1992).
Recently Elitzur et al. (1989) have proposed a model to
explain the high brightness temperatures of H 2 O masers
around YSO. The model involves fast (v s ¸ 100 km s \Gamma1 )
J--type shock propagating in a dense molecular gas (n ¸
10 7 cm \Gamma3 ). In the postshock region the proper density
and temperature conditions should exist to enhance the
H 2 O fractional density and to pump the maser transition.
The geometry is considered to be filamentary in order
to produce the required amplification and beaming an­
gle (but the same results can be obtained in planar ge­
ometry, see Elitzur et al. 1992). A similar model explains
OH masers, with somewhat smaller velocity and density
requirements.
In order to test this model and to understand the rela­
tionship between the masing cloud and the other objects
in the SFR, such as YSO, molecular outflows and UCH ii

2 L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers:
regions, arcsecond resolution observations at various fre­
quencies are required. In particular one would like to know
how the shock forms, how the energy is transferred from
the star to the masers and in which phase of the stellar
evolution does the maser event take place. At the low reso­
lution that can be achieved with single dish observations,
masers and diffuse H ii regions seem to be associated in
many SFR (Genzel & Downes 1977, Codella et al. 1993).
Near infrared images are the most suitable tools for in­
vestigating the powering sources of the masing molecules
because of the reduced extinction.
Near and mid infrared observations of maser regions
have been carried out since the late seventies, using
single detector photometers and typical diaphragms of
about 10 arcsecond in diameter (Evans et al. 1979, Mor­
wood & Salinari 1981a,b, Epchtein & L'epine 1981, and
Baz & Epchtein 1982). A very strong correlation between
infrared sources and masers observed with single dishes
was found at this low angular resolution.
After the successful completion of the IRAS mission,
far infrared data are also available for comparison with the
radio measurements. Many authors have found a high cor­
relation between masers and FIR sources from the IRAS
PSC. Moreover, radio and infrared fluxes seem to be corre­
lated (Wouterloot & Walmsley 1986, Palagi et al. 1993).
However, the complex morphology of UCH ii re­
gions found with VLA observations and the un­
clear relationship between these and the maser spots
has opened many fundamental questions that the
old low resolution and low sensitivity NIR observa­
tions could not address. Wood & Churchwell (1989a) and
Wood & Churchwell (1989b), in an extensive study of
UCH ii regions, found that these objects seem to have
peculiar FIR colours. Many of these are observed near
or coincident with OH and, sometimes, H 2 O masers. On
the other hand the studies of Gaume & Mutel (1987) and
Forster & Caswell (1989) give a large catalogue of accu­
rate positions (absolute position better than 0:5 arcsec) of
OH and H 2 O masers in the ``classical'' lines (ground state
OH and the 6 16 --5 23 H 2 O transition). The observations
show that the maser spots tend to be grouped in com­
pact well separated groups: OH main and satellite lines
tend to be more closely related to UCH ii regions, while
H 2 O masers are usually not closely associated with them.
When both masers are associated with an UCH ii region
the H 2 O masers are usually closer to the peak of the radio
continuum emission.
The advent of the modern infrared arrays makes
them the most suitable devices to be used for a new
NIR survey of regions that shows H 2 O and OH maser
effect. The comparison between this new generation
of infrared images and the latest high resolution ra­
dio data will enable us to address most of the fun­
damental questions concerning the maser and their oc­
currence in the life time of a YSO. Studies of this
type have been performed in individual sources such
as W3(OH) (Guilloteau et al. 1985, Wilson et al. 1991,
Baudry et al. 1993 and Wink et al. 1993), GM24
(G'omez et al. 1993), W75N (Hunter et al. 1993). We have
begun a systematic survey in the J, H, and K broad
band filters using a modern NIR camera of all the
masers for which accurate positions are available. In
particular we have chosen the sources in the list of
Forster & Caswell (1989), because these represent the first
homogeneous set of sources observed at high angular res­
olution in both maser lines (OH at 1665 MHz and H 2 O at
22 GHz). The aim is to find if there is any source of NIR
emission close to the maser spots, with a positional accu­
racy of !
¸ 1 00 and to study the nature of these sources and
their role in the creation of the maser effect. In this paper,
we present the results of the observations of a subsample
of 17 regions from the list of Forster & Caswell (1989),
concentrating the analysis on the NIR sources closer to
the maser position. In a forthcoming paper, a study of the
young stellar population associated with these regions will
be given.
2. Observations and data reduction
The observations were carried out during June 1991 at Las
Campanas Observatory, using the Near--Infrared Camera
mounted on the 1 meter telescope (see Persson et al. 1992
for a complete description of the instrument). The detec­
tor was a NICMOS2 128\Theta128 pixel HgCdTe array devel­
oped by Rockwell International. We observed 17 sources
from the list of Forster & Caswell (1989) (the respective
names are given in column 2 of Table 1). Various frames
in five overlapping positions were taken for each field and
for each of the J, H and K filters, in such a way that
the maser position was contained in all the frames. Flux
calibration was achieved observing a number of standard
stars from the catalogue of Elias et al. 1982. We estimate
a mean absolute flux calibration accuracy of about 10% .
The nominal scale on the detector was 1.34 arcsecond per
pixel and the field covered in a single frame was almost
3\Theta3 arcmin. The part of the sky covered in all the five
positions was approximately 1 square arcmin for almost
all the sources. Here we report only the analysis of those
small portions of the images centered on the maser posi­
tion. All the image reduction and analysis was made using
the IRAF 2.10 and STSDAS software packages. For each
field the single exposures taken in each of the five positions
were averaged, then the sky subtraction was made using
the sky image obtained by median averaging the frames in
the five positions, this image was also normalized to give
an estimate of the flat field. The five images were then
registered and coadded in order to achieve a better signal
to noise ratio in the central part, near the maser position.
1 Photometry was performed using the DAOPHOT rou­
1 The J images of (339:88 \Gamma 1:26) were somewhat bad and we
could not recover any useful information from them, therefore
they have been excluded from the analysis.

L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers: 3
Fig. 1. a) K images for each of the 17 fields observed except (035:20 \Gamma 1:74) which is given in Figure 2. Only a small part of
the coadded frame is shown. The label on top of each image reports the name of the field, and the coordinate grids are correct
for the 2000.0 epoch. Overlayed on the images there are the H2O masers (crosses), the OH masers (open squares), the radio
continuum peak (two concentric circles), and the error box of the IRAS source (ellipse).
tine, with bright isolated stars as point spread function
models, the full width at half maximum of the PSF was
found to be roughly 2.8 arcsec. The mean limiting magni­
tudes found in each of the J, H, and K filters were 16.3,
14.8, and 14.2, respectively, with fluctuations from field to
field of about 0:7 magnitudes.
Table 1 lists the position and magnitudes of the sources
found nearest each maser and the distance between NIR
and maser component positions; In the first two column,
we report the field number (the progressive number in
the list of Forster & Caswell 1989) and the field name,
then the quoted distance of the masers from the Sun (in
Kpc, see Forster & Caswell 1989 and references therein);
for each source detected near a maser component, we re­
port the coordinates (correct for the 1950.0 epoch), a code
that will be used for reference, and the K, H and J mag­
nitudes with their errors. When a symbol ``?'' is present,
this means that the given magnitude represents a lower
limit compatible with the sensitivity of the frame. For
each NIR source, the associated maser components are
then listed: the component identifier is in the c i column
and in the other two columns are the angular and linear

4 L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers:
Fig. 1. b) Same as Figure 1a.
distance from the NIR source in arcsecond and in units of
10 17 cm.
2.1. Astrometry
Because we intend to compare the NIR images with VLA
radio data, which give the absolute positions of the masers
spots with an accuracy of 0:5 00 , careful astrometry on the
NIR images is necessary. To accomplish this task the Hub­
ble Space Telescope Guide Star Catalogue was used in
a first attempt. However, in most cases our maps were
too small and no more than two stars of the catalogue
were in our field. Since we need at least three stars in the
field to do a reliable astrometry, we decided to go directly
to the original digitized plates of the survey, available at
the Space Telescope Science Institute. We identified in the
plates several reference stars. From these we made tables
of the coordinates of ``secondary'' stars that were in the
infrared images and in the optical ones, and we used these
to make the astrometric calibration of our frames. Because
of the two--step procedure the accuracy of the calibration
is not as accurate as that of the Guide Star Catalogue,
but we estimate an astrometric calibration error smaller
than one arcsecond (using IRAF centering algorithms it
is possible to determine star positions with an accuracy of
¸ 0:3 pixel). The procedure is discussed in greater detail
in Testi (1993b).

L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers: 5
Fig. 1. c) Same as Figure 1a.
2.2. Results
On the calibrated images we overlayed the positions of the
radio continuum sources, that of the masers, and the error
box of the peak position of the nearest IRAS point source
(if any was present within 1 arcmin from the water maser
reference). In Figure 1, such overlays between the data at
various frequencies and the K--band images are presented
for all the observed fields (Testi 1993a).
To explain some features common to all the fields,
we shall consider a typical example. The optical, J, H
and K images of the field (035:20 \Gamma 1:74) are presented
in Figure 2. The large ellipse is the error box of the
IRAS point source, the crosses are the H 2 O masers, the
squares the OH masers and the two concentric circles
the position of the peak of the UCHii region found by
Wood & Churchwell (1989b). The four maps in greyscale
are from the upper left to the lower right, the optical, J,
H and K images. Usually the fields are very crowded, and
the NIR maps show many pointlike and a few extended
sources. The unresolved sources are usually stellar objects,
while the extended emissions could be reflection nebulae
or diffuse dusty H ii regions. Most of these infrared sources
are not visible in the optical map, which implies heavy ex­
tinction. The masers are located near a faint source seen
only in the K image. The brightest NIR sources, often
coincident with the IRAS peak and the UCH ii region
peak, are well separated from the masers and are em­

6 L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers:
Fig. 1. d) Same as Figure 1a.
bedded in a large nebulosity that becomes brighter in
the 2.2 ¯m image. Previous observations of these area
(Morwood & Salinari 1981a) collected with a monopixel
photometer and a diaphragm 12 arcsecond in diameter,
showed only a point source (in K and L) coincident with
the bright nebulosity.
The complexity of the morphology of the fields and
the high density of sources detected impose a complete
revision of the previous works on the association between
H 2 O=OH masers and near--infrared sources. In particular,
the task of establishing an association between a maser
and a NIR source is not straightforward. The reliability of
this association will be the main argument of the following
section.
3. Identification of NIR sources near the maser
3.1. Maser morphology
For each field the maser positions were provided by
Forster & Caswell (1989) and Forster (1992), in the form
of offsets with respect to a reference H 2 O maser spot in
the sky. Because of the high spatial and velocity resolution
of the VLA data many maser spots at slightly different po­
sitions (and at different velocities) were detected. We have
grouped in a single component the masers spots located
within an arcsecond in order to match the NIR resolution.
In this way, we derived 22 H 2 O maser components and 18
OH maser components inside the 17 surveyed fields. In
most cases there is only one component per field, otherwise

L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers: 7
H 2 O
OH
NIR
HII
Fig. 2. From left to right and top to bottom: optical, J, H, and K images of the field (035:20 \Gamma 1:74). The symbols are the same
than that of figure 1.
the components are labelled (c i column in Table 1) with
capital letters (A, B, etc.) from increasing distance from
the reference position given by Forster & Caswell (1989).
3.2. NIR sources
A high density of faint sources in the K--band image was
found in each field. The association with the masers is
based only on the criterion of shortest distance, i.e. se­
lecting the ``first neighbour'' within 10 arcsecond from the
component position (the validity of this method will be
justified in the following section). In this way we avoid
any a priori hypothesis on the nature of the NIR sources.
As can be seen from Table 1, almost all the NIR sources
observed near the maser groups are detected only in the K
band and many of them show a H \Gamma K colour index greater
than 2 (all the others have poor signal to noise ratios). A
high value of the H \Gamma K colour index indicates that either
the source is heavily reddened or have a strong NIR excess
due to the circumstellar environment emission.
3.3. Reliability of the associations between masers and
NIR sources
Because of the high source density in the fields (roughly
twenty sources per square arcminute in the K image), it

8 L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers:
Table 1. Sources detected in the near infrared close to the masers.
FIELD INFRARED SOURCE H 2 O MASER OH MASER
# name D code 1 ff (1950) ffi (1950) mK e K m H e H m J e J c i d 2 d17 3 c i d 2 d17 3
(Kpc) ( 00 ) ( 00 )
2 339:88 \Gamma 1:26 3.1 2--1 16 : 48 : 24:44 \Gamma46 03 : 29:7 13.1 0.2 15.0 ? -- 4 -- 4 A 1.8 0.84
B 2.8 1.30
2--2 16 : 48 : 24:27 \Gamma46 03 : 36:7 12.37 0.08 14.0 0.1 -- 4 -- 4 A 4.0 1.84
3 340:06 \Gamma 0:25 4.6 3--1 16 : 44 : 35:90 \Gamma45 16 : 25:7 13.0 0.1 14.6 ? 16.2 ? A 2.4 1.66 A 2.4 1.66
4 341:21 \Gamma 0:21 15.3 4--1 16 : 48 : 41:72 \Gamma44 21 : 52:6 10.98 0.02 14.6 0.2 15.8 ? A 1.5 3.50 A 1.7 3.83
7 344:58 \Gamma 0:02 0.7 7--1 16 : 59 : 26:31 \Gamma41 37 : 43:9 14.1 0.2 15.1 ? 16.3 ? A 7.2 0.76 A 4.7 0.49
8 345:00 \Gamma 0:22 3.2 8--3 17 : 01 : 40:27 \Gamma41 24 : 58:6 12.82 0.05 15.5 ? 16.9 ? A 1.5 0.74 A 2.0 0.96
B 4.8 2.27
9 345:01 + 1:79 3.0 9--ND 15.0 ? 15.0 ? 16.9 ? A
9--1 16 : 53 : 19:09 \Gamma40 09 : 29:3 14.5 0.1 15.0 ? 16.9 ? B 3.6 1.65
9--b2 16 : 53 : 19:83 \Gamma40 09 : 39:0 13.5 0.2 13.8 ? 15.4 ? C 3.9 1.74
9--4 16 : 53 : 19:06 \Gamma40 09 : 45:9 13.58 0.07 15.0 ? 16.9 ? A 5.6 2.52
10 345:40 \Gamma 0:90 2.3 10--ND 13.0 ? 13.0 ? 15.0 ? A A
11 345:51 + 0:35 2.1 11--1 17 : 00 : 53:55 \Gamma40 40 : 15:0 12.79 0.04 15.5 ? 16.3 ? A 3.5 1.11 A 0.6 0.21
12 345:70 \Gamma 0:09 1.0 12--1 17 : 03 : 20:93 \Gamma40 47 : 01:3 11.06 0.05 13.3 0.1 15.2 0.1 A 1.7 0.26 A 2.5 0.37
16 348:89 \Gamma 0:19 1.0 16--1 17 : 13 : 34:76 \Gamma38 16 : 12:7 12.85 0.07 14.1 ? 15.8 ? A 2.1 0.32 A 1.8 0.28
21 350:11 + 0:09 9.8 21--ND 13.8 ? 14.4 ? 16.0 ? A A
33 359:44 \Gamma 0:10 10 33--1 17 : 41 : 29:17 \Gamma29 26 : 55:5 11.63 0.04 14.6 ? 17.2 ? A 7.7 11.6 A 7.3 11.0
33--b2 17 : 41 : 29:86 \Gamma29 26 : 44:3 13.1 0.4 13.4 ? 15.3 ? B 1.7 2.60
37 000:55 \Gamma 0:85 2.0 37--1 17 : 47 : 03:98 \Gamma28 53 : 35:5 12.25 0.05 14.6 ? 16.1 ? A 4.5 1.34 A 6.8 2.02
B 4.4 1.32
47 012:68 \Gamma 0:18 6.4 47--4 18 : 10 : 59:23 \Gamma18 02 : 33:1 14.5 0.2 15.3 ? 16.3 0.3 A 9.4 9.09 A 6.8 6.53
51 016:59 \Gamma 0:06 5.6 51--1 18 : 18 : 18:20 \Gamma14 33 : 14:9 14.2 0.2 14.5 ? 17.0 0.3 A 2.2 1.85 A 0.7 5.88
57 028:87 + 0:06 8.5 57--1 18 : 41 : 08:56 \Gamma03 38 : 36:5 11.75 0.05 14.1 ? 15.2 ? A 4.7 5.93 A 3.1 3.98
66 035:20 \Gamma 1:74 2.9 66--1 18 : 59 : 13:35 +01 09 : 11:7 13.10 0.04 16.4 ? 17.5 ? A 2.5 1.08 A 2.7 1.18
1. ND means that no NIR source has been detected near the corresponding maser component.
2. The error on the distance is about 1 00 .
3. d17 is the linear distance in units of 10 17 cm.
4. The J images of 339:88 \Gamma 1:26 are somewhat bad and have not been considered.
is important to establish the goodness of the association
between maser components and NIR sources. In order to
do this we have simulated random associations from a uni­
form distribution of field objects and we have constructed
the ``first neighbour distribution''. From all the fields we
derived a mean density for the sources detected in the
K maps. For convenience, we divided the sources in 5
colour classes: Dark, Red, Yellow, Green, Blue; D sources
are detected only in the K map, R have a colour index:
H \Gamma K ? 2, Y: 1:5 ! H \Gamma K Ÿ 2, G: 1 ! H \Gamma K Ÿ 1:5, and B:
H \Gamma K Ÿ 1. Table 2 reports the mean densities of each class
over all fields. The first column lists the code of the class,
the second the class boundaries, and the third the source
density (sources per square arcsecond). We simulated nu­
Table 2. Source density for the various colour classes.
Class Colour Sources=arcsec 2
D Only K 1:67 \Theta 10 \Gamma3
R H \Gamma K ? 2 4:68 \Theta 10 \Gamma4
Y 1:5 ! H \Gamma K Ÿ 2 4:68 \Theta 10 \Gamma4
G 1 ! H \Gamma K Ÿ 1:5 5:57 \Theta 10 \Gamma4
B H \Gamma K Ÿ 1 1:98 \Theta 10 \Gamma3
merically random associations from uniform distribution
of sources for three cases:
a) D+ R = 2:14 \Theta 10 \Gamma3 sources per square arcsec
b) D+ R+ Y = 2:61 \Theta 10 \Gamma3 sources per square arcsec
c) D+ R+ Y+ G = 3:16 \Theta 10 \Gamma3 sources per square arcsec
In Figure 3, the distributions of ``first neighbour'' ob­
tained for the models are plotted together with the ob­
served distribution of the distance NIR--H 2 O maser. Even
in the most unfavourable of the three cases considered here
(c), the observed distribution differs considerably from
the random simulations; the observed distances peak be­
tween 2 00 and 5 00 , the simulated ones peak around 8 00 --10 00 .
Because the observed distributions peak at shorter dis­
tances than the simulated ones, we believe that the as­
sociation between masers and NIR sources is statistically
significative and not due to chance coincidence caused by
projection of sources on the line of sight. It is remark­
able that, in spite of the high source density observed, no
class--B, class--G or class--Y source was found as the first
neighbour of a maser group, but only objects of the classes
D and R.

L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers: 9
Fig. 3. Distributions for the observed NIR--H2O maser groups
distance (shaded) and for the three numerical random simula­
tions (see text for codes meanings).
3.4. (H \Gamma K; J \Gamma K) colour--colour diagram
In Figure 4 the sources detected are plotted in the (H \Gamma
K; J \Gamma K) colour--colour diagram. We decided to use such
plot instead of the standard (H \Gamma K; J \Gamma H), because in
most cases the source is detected in the K map and only
upper limits are available for the J and H magnitudes, and
it is impossible to give even a lower/upper limit for the
(J \Gamma K) colour. Of course this choice does not alter in any
sense the informative contents of the graph, but simply
makes it more easily readable. As discussed in section 5,
the emission in K band is probably of stellar in nature, we
will assume this in the following discussion. The solid line
labelled MS represents the colours of main sequence stars.
A reddened main sequence star should lie inside the band
delimited by the two dashed lines. As can be easily seen
many (more than 53%) of the sources in Table 1 have near
infrared colours that are incompatible with those of main
sequence stars, and the fact that these objects are very
likely to be associated with the masers (as discussed in
the previous section) suggests that a heavy reddening due
to distant cold material is not plausible. More probably,
the sources have a strong NIR excess.
In Figure 4(b) the colour--colour diagram of all the
sources detected in the K band images (in all fields) ex­
cluding those of Figure 4(a) is presented. More than 300
sources are plotted and only a few of them (less than 15%)
have colours similar to those found near the masers, im­
plying that the sources in Figure 4(a) represent, in a sta­
tistical sense, a distinct class of objects.
3.5. Relationship between masers and radio/NIR extended
sources
From low resolution (¸ arcmin) radio studies, a close re­
lationship was found between masers and diffuse H ii re­
gions. This simply tells us that masers are present within
SFR, but is insufficient to identify the sources of excita­
tion of the masers. Also previous low resolution NIR ob­
servations proved to be more sensitive to extended than to
pointlike sources. In our images we find extended infrared
emission in less than half of the regions, and only in one
case (345:40\Gamma0:90) the masers are seen inside the extended
emission. The NIR extended emission, probably evolved
H ii regions or bright reflection nebulae around well de­
veloped luminous young stars, do not seem to be closely
related to the maser emission. Usually the maser spots
tend to be well separated from the diffuse regions. The
diffuse nebulae are often observed together with bright
point sources which are likely to be the exciting stars. On
the other hand the masers seem to be associated with faint
point sources, well separated from the brightest members
of the stellar population and the surrounding diffuse emis­
sion. This would imply that the maser effect is related to
the youngest phases of the stellar evolution, when the YSO
is still embedded in a dense cocoon.
From high resolution radio studies (see for example
Gaume & Mutel 1987 and Forster & Caswell 1989) a
closer relationship seems to exist between masers (par­
ticularly OH) and UCH ii regions. But in many cases
the maser components are close to, but not coincident
with, the radio continuum source (see the distribution
of distance between masers and UCH ii regions given
by Forster & Caswell 1989). In five of the 17 sources,
Forster & Caswell (1989) report the detection of a con­
tinuum source at 22 GHz. One of them (35:20 \Gamma 1:74) was
also found by Wood & Churchwell (1989b) and another
(345:00 \Gamma 0:22) by Gaume & Mutel (1987). In all (5) cases
we do find a NIR source, either pointlike or barely re­
solved, which coincides with the radio continuum emis­
sion. However, only four of them are associated with OH
maser components, and only two with H 2 O maser compo­
nents.
4. IRAS properties
For all seventeen fields observed, we searched the IRAS
Point Source Catalogue to see if there is any source within
60 arcsecond from the water maser reference position. In
this way we found a mid and far infrared pointlike emis­
sion in 13 cases. In Table 3 we summarize the charac­
teristics of these sources and the associated fields. For
each source we list, the IRAS name, the fluxes in the
four IRAS bands (upper limits are indicated with the !

10 L. Testi, M. Felli, P. Persi, and M. Roth: Near infrared images of galactic masers:
Fig. 4. (H \Gamma K;J \Gamma K) colour--colour diagram. (a) Sources associated with the masers. The codes of the sources are the same
used in Table 2. (b) All the sources detected in the K--band images except those associated with the masers. Lower limits are
indicated by arrows. The mean error bar is shown in the lower right corner. In both figures the continuous line labelled MS is
the region of the graph in which Main--Sequence stars should be found. The band delimited by the two parallel lines define the
effect of reddening on MS stars.
Table 3. IRAS--PSC sources associated with the observed fields.
NO UPPER LIMITS
# IRAS name F 12 F 25 F 60 F 100 F FIR L FIR [25 \Gamma 12] [60 \Gamma 12] type 1
(Jy) (Jy) (Jy) (Jy) (W m \Gamma2 ) (L fi )
3 16445­4516 .2619E+02 .3183E+03 .4756E+04 .8412E+04 3.13E­10 9.46E+04 1.08 2.26 i
7 16594­4137 .1143E+01 .3288E+02 .6615E+03 .1126E+04 4.10E­11 6.33E+02 1.46 2.76 i
8 17016­4124 .3117E+01 .2410E+03 .3664E+04 .7001E+04 2.45E­10 7.91E+04 1.89 3.07 i
10 17059­4132 .5427E+03 .3879E+04 .1271E+05 .2576E+05 1.32E­09 2.21E+05 0.85 1.37 o
21 17160­3707 .4523E+02 .3355E+03 .4246E+04 .1023E+05 3.32E­10 1.00E+06 0.87 1.97 o
66 18592+0108 .1145E+03 .1023E+04 .1015E+05 .1392E+05 6.56E­10 1.74E+05 0.95 1.95 o
UPPER LIMIT AT 100¯m
11 17008­4040 .6850E+02 .5377E+03 .2701E+04 !.1825E+05 0.89 1.60 o
51 18182­1433 .2501E+01 .3532E+02 .4194E+03 !.1071E+04 1.15 2.22 i
57 18411­0338 .8636E+01 .1052E+03 .1703E+04 !.3146E+04 1.09 2.29 i
UPPER LIMIT AT 12¯m
2 16484­4603 !.9993E+01 .1920E+03 .3440E+04 .5324E+04 ?1.28 ?2.54 i
16 17136­3816 !.4118E+01 .3982E+02 .5546E+03 .1059E+04 ?0.99 ?2.13 o
UPPER LIMIT AT 25¯m AND/OR 60¯m
9 16533­4009 .2645E+02 !.4701E+03 .7136E+04 .1324E+05 !1.25 2.43 o
37 17470­2853 !.4250E+02 !.2805E+03 .5477E+04 .1296E+05 ¸0.82 ?2.11 o
1. type i: NIR source inside error ellipse; type o: NIR source outside error ellipse.
symbol), the integrated flux and luminosity, [25 \Gamma 12] and
[60 \Gamma 12] colours and a code, (i) or (o), indicating whether
the NIR source associated with the maser is inside or out­
side the error ellipse of the IRAS position. In the [60 \Gamma 12]
vs. [25 \Gamma 12] colou