Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~fmassi/papers/6438pap.ps.gz Äàòà èçìåíåíèÿ: Tue Feb 27 18:13:09 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 04:29:30 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: massive stars

Astronomy & Astrophysics manuscript no. 6438pap c
# ESO 2007
February 26, 2007
Star formation in the Vela Molecular Ridge
Large scale mapping of cloud D in the mm continuum #
F. Massi 1 , M. De Luca 2,3 , D. Elia 4 , T. Giannini 3 , D. Lorenzetti 3 , and B. Nisini 3
1 INAF ­ Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, I­50125 Firenze, Italy e­mail: fmassi@arcetri.astro.it
2 Dipartimento di Fisica, Universit‘a degli studi di Roma Tor Vergata, Via della Ricerca Scientifica 1, I­00133 Roma, Italy
3 INAF ­ Osservatorio Astronomico di Roma, Via Frascati 33, I­00040 Monteporzio Catone, Roma, Italy e­mail:
deluca,giannini,dloren,bruni@mporzio.astro.it
4 Dipartimento di Fisica, Universit‘a del Salento, CP 193, I­73100 Lecce, Italy e­mail: eliad@le.infn.it
Received ; accepted
ABSTRACT
Context. The Vela Molecular Ridge is one of the nearest intermediate­mass star forming regions, located within the galactic plane and outside
the solar circle. Cloud D, in particular, hosts a number of small embedded young clusters.
Aims. We present the results of a large­scale map in the dust continuum at 1.2 mm of a # 1 # â 1 # area within cloud D. The main aim of the
observations was to obtain a complete census of cluster­forming cores and isolated (both high­ and low­mass) young stellar objects in early
evolutionary phases.
Methods. The bolometer array SIMBA at SEST was used to map the dust emission in the region with a typical sensitivity of # 20 mJy/beam.
This allows a mass sensitivity of # 0.2 M # . The resolution is 24 ## , corresponding to # 0.08 pc, roughly the radius of a typical young embedded
cluster in the region. The continuum map is also compared to a large scale map of CO(1--0) integrated emission.
Results. Using the CLUMPFIND algorithm, a robust sample of 29 cores has been obtained, spanning the size range 0.03 - 0.25 pc and the
mass range 0.4 - 88 M # . The most massive cores are associated both with red IRAS sources and with embedded young clusters, and coincide
with CO(1--0) integrated emission peaks. The cores are distributed according to a mass spectrum # M -# and a mass­versus­size relation # D x ,
with # # 1.45 - 1.9 and x # 1.1 - 1.7. They appear to originate in the fragmentation of gas filaments seen in CO(1--0) emission and their
formation is probably induced by expanding shells of gas. The core mass spectrum is flatter than the Initial Mass Function of the associated
clusters in the same mass range, suggesting further fragmentation within the most massive cores. A threshold A V # 12 mag seems to be
required for the onset of star formation in the gas.
Key words. ISM: clouds -- ISM: individual objects: Vela Molecular Ridge -- ISM: dust, extinction -- Stars: formation -- Submillimeter
1. Introduction
CO(1--0) is the best tracer of molecular gas and the most ef­
ficient transition to map large sky areas, but it fails to probe
the densest molecular cores. This is due both to its large op­
tical depth and to the fact that CO freezes onto dust grains at
the highest densities. However, the availability of large arrays
of submillimetre/millimetre bolometers has recently provided a
means to carry out large surveys of entire star forming regions
searching for protostellar and starless cores with high sensitiv­
ity (see, e. g., Motte et al. 1998, 2001; Johnstone et al. 2000;
Kerton et al. 2001; Mitchell et al. 2001; Reid & Wilson 2005;
Mookerjea et al. 2004; Hatchell et al. 2005).
Send o#print requests to: F. Massi
# Based on observations collected at the European Southern
Observatory, La Silla, Chile, program 71.C--0088
All these studies testify to the need to obtain as completely
as possible a census of the youngest stellar population down
to the lowest masses, and to perform accurate studies of the
environment where young stars form. This is particularly true
for massive star forming regions, since they usually are farther
away than low­mass star forming regions, with obvious prob­
lems in terms of resolution. In this respect, the Vela Molecular
Ridge (VMR) has proved to be one of the most interesting tar­
gets. It is the nearest giant molecular cloud complex hosting
massive (although mostly intermediate­mass) star formation
after Orion, but, unlike Orion, it is located within the galac­
tic plane, where most of star formation in the Galaxy occurs.
Its distance is probably within a factor 2 of that of Orion (700
pc vs. 450 pc), so the losses in terms of resolution and sensitiv­
ity are not very large and are partly compensated by the bigger
observable area with the same instrumental field of view. Also,

2 F. Massi et al.: Star formation in the Vela Molecular Ridge
one can in this manner study many di#erent star forming sites
with the same observational and environmental biases.
The VMR was first mapped in the 12 CO(1--0) transition
with low resolution (# 30 # ) by Murphy & May (1991). They
subdivided it into four main regions, named A, B, C and D.
Yamaguchi et al. (1999) presented new higher­resolution (# 8 # )
maps of the VMR in the 12 CO(1--0) and 13 CO(1--0) transitions,
better evidencing the complex structure. The issue of distance
was discussed by Liseau et al. (1992), who found that clouds A,
B and D are at 700±200 pc. Liseau et al. (1992) and Lorenzetti
et al. (1993) obtained the first data on young stellar objects in
the region measuring the spectral energy distribution (SED) for
the associated IRAS sources. Massi et al. (1999, 2000, 2003)
subsequently found that the IRAS sources with L bol > 10 3 L
#
are associated with young embedded clusters.
Vela D appears to be the most e#cient of the VMR clouds
in producing young embedded clusters. A look at Fig. 1 of
Yamaguchi et al. (1999) shows that the region consists of more
than one cloud; that located at # = 264 # , b = 0 # contains at
least 5 embedded clusters (see Fig. 1 of Massi et al. 2003),
all studied by Massi et al. (2000, 2003, 2006). With the aim
of investigating the mechanisms leading to this high e#ciency,
and the star formation history in general, in 1999 we decided
to make large scale maps of the cloud at mm wavelengths
with the SEST. Observations in 12 CO(1--0) and 13 CO(2--1) are
presented in Elia et al. (2007). In 2003, the same region was
mapped in the 1.2­mm continuum emission with the bolometer
array SIMBA. The SEST beam, 24 ## , i.e. # 0.08 pc at 700 pc,
matches, or is smaller than, the typical spatial size of cluster­
forming cores and that of the embedded star clusters in the re­
gion (see Massi et al. 2006). A small subset of these data has
already been published in Giannini et al. (2005). This paper
presents the full results of the SIMBA observations. In Sect. 2,
observations and data reduction are described. Sect. 3 reports
the most significant results, that are then discussed in Sect. 4,
and Sect. 5 summarizes our conclusions.
2. Observations and data reduction
A # 1 # â 1 # region of Vela D was observed in the 1.2­mm
continuum between May 23--26, 2003, using the 37­channel
bolometer array SIMBA (Nyman et al. 2001) at the Swedish­
ESO Submillimetre Telescope (SEST) sited in La Silla, Chile.
At this wavelength, the beam HPBW is 24 ## . A total of 17 sky
areas 18 # â 14 # (azimuth â elevation) in size were mapped
in the fast scanning mode, with a scanning speed of 80 ## s -1 .
Each map was repeated 3 to 4 times, except for 3 marginal
regions. We always performed a skydip and pointing check ev­
ery # 2 hrs, while the focus was checked at the beginning of
each observing run and at sunset. The pointing was better than
# 5 ## . The zenith atmospheric opacities remained in the ranges
0.166 - 0.244 (May 26) and 0.263 - 0.317 (May 23).
All data were reduced with MOPSI 1 according to the
SIMBA Observer's Handbook (2003). The steps are summa­
rized in Chini et al. (2003). All the available maps for each of
1 MOPSI is a software package for infrared, millimetre and radio
data reduction developed and regularly upgraded by R. Zylka.
the 17 areas were first coadded yielding 17 images that, once
calibrated (see below) were mosaiced together. Since some of
the images partially overlap, in those cases the reduction steps
were repeated while mosaicing in order to search for faint
sources emerging from the areas with a longer e#ective inte­
gration time. The final large­scale map is shown in Fig. 1. The
r.m.s. is in the range 14--40 mJy/beam, but is # 20 mJy/beam
(or less) over most of the map.
Unfortunately, the only planet up during the observations
was Jupiter, which was mapped once per day. Since it is re­
solved by the SEST beam, we decided then to calibrate our
1.2­mm data also using observations of Uranus carried out
soon before and after each of our observing runs. We derived
the conversion factor from counts to mJy/beam using di#er­
ent methods, e.g., by fitting a Gaussian to Uranus, by per­
forming aperture photometry of the calibrator using di#erent
apertures, by correcting the photometry of Jupiter for source
coupling and the error beam. We obtained conversion factors
within the range 55­75 mJy/beam counts -1 , depending on the
adopted method, but remarkably, each method provides a value
that remains constant within a 20 % (most often, within a 10
%) throughout the whole observing period (i. e., one week).
Hence, we adopted the canonical value of 65 mJy/beam
counts -1 for all runs but the last, which always appears to ap­
proximate the measured values within # 10 %. Since all our
estimates of the conversion factor exhibit a small drop during
the last run (May 26), in this case we adopted a value of 60
mJy/beam counts -1 . Anyway, this choice does not a#ect the
results, since only few faint sources were detected during this
run. To check our data for consistency, we performed some
aperture photometry on a few sources using MOPSI. An ex­
ample is the region including IRS16 (here we will use the des­
ignation adopted by Liseau et al. 1992 for the IRAS sources
common to their list) which was observed both on May 23 and
on May 24. The flux from the most intense source is found to
be the same within # 3 %, confirming the excellent stability
of the instrumental setting and atmospheric conditions during
the observations. In order to compare our data with the single
pointing measurements reported by Massi et al. (1999, 2003),
we roughly simulated single pointings by selecting apertures
with a diameter equal to the beam HPBW and centred on the
positions given by those authors. For the eight common sources
observed during our run, we find flux densities within 36 % of
those measured by Massi et al. (1999, 2003); in one case, the
flux density value of Massi et al. (1999) is within 39 % of ours.
However, from a check to the records of the observations listed
in Massi et al. (1999) it seems that the single pointings are sig­
nificantly a#ected by the selected throw.
A much more significant test can be performed by consider­
ing di#erent photometry of sources observed in the same mode.
Fa’undez et al. (2004) mapped a sample of fields, which in­
cludes IRS17, with SIMBA in fast scanning mode. They find
a flux of 6.1 Jy for IRS17, while we find a total flux of 8.6 Jy
(see Giannini et al. 2005). However, we achieved an r.m.s. of
# 15 mJy/beam, less than that of Fa’undez et al. (2004), imply­
ing we might have recovered more low­level emission. Fitting
a Gaussian to the source, we obtained a size of 44 ## (same as
Fa’undez et al. 2004) and a flux of 5.5 Jy, within # 10 % of

F. Massi et al.: Star formation in the Vela Molecular Ridge 3
Fig. 1. SIMBA 1.2­mm continuum map of cloud D (greyscale). The scale (in mJy/beam) is indicated through the bar on the right and the contour
at 60 mJy/beam (# 3#) is also drawn. The red IRAS sources coincident with the most intense features are labelled (along with the designation
adopted by Liseau et al. 1992). Six areas including the main mm sources are enclosed within boxes and numbered, and are shown zoomed­in
in the following figures.
that measured by Fa’undez et al. (2004). Beltr’an et al. (2006)
observed four of our sources in their survey of massive proto­
star candidates with SIMBA, namely IRS16, IRS20 and IRS21
(plus IRS18, observed in our run but located outside the target
area, thus not discussed here). They used CLUMPFIND to de­
convolve the emission into clumps as we also did (see Sect. 3).
Since their maps are noisier and less sampled than ours, the
identified clumps are slightly di#erent from the ones we de­
tected. Hence, where the number of clumps per source di#ers,
we summed together the emission such as to compare the same
emitting areas. In two cases (IRS18 and IRS20), the two sets of
fluxes agree within # 15 %. As for IRS16 and IRS21, the fluxes
measured by Beltr’an et al. (2006) are lower and within # 50 %
of ours. We carried out aperture photometry with MOPSI on
both sets of data, and found that the r.m.s. of the four maps by
Beltr’an et al. (2006) is twice as high as ours. Their scanned ar­
eas are much smaller than ours, also, and when using a small
aperture around the most intense peaks, the fluxes from their
frames are found to be within # 25 % of ours both for IRS16
and IRS21, confirming that probably we recover more low level
emission than Beltr’an et al. (2006). We conclude that our flux
calibration is accurate in general to within # 20 %.
3. Results
3.1. Dense cores
Figure 1 shows a number of clumps throughout the region,
most of which are close to red IRAS sources. Among the young
star clusters studied by Massi et al. (2000, 2003, 2006), those
that are located within the boundaries of the observed area
clearly coincide with the brightest mm emission sources.
The map has been scanned, searching for dust con­
densations, using a two­dimensional version of the tool
CLUMPFIND, developed by Williams et al. (1994), capable
of recognizing connected regions of emission with arbitrary
shapes (hereafter cores), starting from the peaks of intensity
and going down to the lowest contour levels depending on the
r.m.s. of the data.
In order to reduce the influence of the variations in the r.m.s.
noise over the map, we divided the observed region into ten
smaller sub­regions. Then, we applied CLUMPFIND to each
of these areas using a local value of the r.m.s., as to properly set
the two input parameters: the lowest contour level, I min , which
acts as a detection threshold, and the intensity level increment,
#I, used to separate resolved components located within the
same emission region. We in practise used I min = 3 r.m.s. and

4 F. Massi et al.: Star formation in the Vela Molecular Ridge
#I = 2 r.m.s. in order to be conservative and to obtain a separa­
tion between substructures which seems reasonable on a visual
inspection.
The varying r.m.s., and the di#erent input parameters cho­
sen for each sub­region, do a#ect the sensitivity over the map,
but ultimately they only change the ``global'' completeness
limit (estimated in Sect. 4). Above this completeness limit, we
do not expect our source statistics to be significantly biased.
The CLUMPFIND output consists of a sample of about
50 cores for which we derived the sizes by assuming a dis­
tance of 700 pc. From the original output we selected a robust
sample of 29 cores (listed in Table 1) which fulfil the crite­
rion of having a size (before beam deconvolution) greater than,
or equal to, the SIMBA HPBW. The sample does not include
those detections (although all are above 3 r.m.s.) that are re­
vealed by CLUMPFIND, but ``under­resolved'' because their
size is less than the SIMBA beam, before deconvolution. For
completeness, we also list coordinates and peak intensity of
these detections (estimates of sizes cannot be given) in Table 2.
Most of them are small components ( # < 1 M # ) of large com­
plex structures, but some (including isolated ones) appear in
some way linked to the star formation activity (e.g. coincidence
with IRAS/MSX point sources, CO peaks and/or evidence of
H 2 emission), although we cannot exclude they are artefacts
of the finding algorithm. In particular, umms1 is a relatively
high­mass core (but being located within a quite noisy area)
and coincides with a CO(1--0) peak at the border of the line
map (hence only partially mapped). However, all these detec­
tions deserve dedicated observations with higher sensitivity in
order to confirm them, to clarify their nature and to give rea­
sonable estimates of their masses and sizes. A detailed analysis
of these uncertain detections will be presented in a forthcom­
ing paper (De Luca et al. 2007), while in the following we will
refer only to the sample of Table 1. The adopted approach is
bound to a#ect the statistic of the faintest sources, dropping
objects that would fall in the robust sample if the sensitivity
were higher. But, again, we do not expect major e#ects above
the completeness limit.
The cores' masses have been determined accordingly from
the expression of Fa’undez et al. (2004), by assuming a gas to
dust ratio of 100, a temperature of 30 K and a dust opacity
k 1.3 = 0.5 cm 2 g -1 at 1.3 mm. We adopted the same k 1.3 as in
Motte et al. (1998) and Testi & Sargent (1998); however, values
as high as 1 cm 2 g -1 are possible, typical of very dense regions
(n H > 10 7 cm -3 ; Ossenkopf & Henning 1994), such as, e.g.,
circumstellar envelopes.
By fitting (emissivity # # # # -1 ) the IRAS fluxes at 60 and
100 µm, Liseau et al. (1992) find temperatures in the range 30--
43 K for IRS 17, 18, 19, 20 and 21. Fa’undez et al. (2004) derive
a typical temperature of 32 K for the cold dust component by
a two­components fit to the SED's in their sample of south­
ern high­mass star forming regions. Beltr’an et al. (2006) find
a mean dust temperature of 28 K by fitting the SED's in their
sample of southern massive protostar candidates longward of
60 µm. Hence, we adopted the canonical value of 30 K for
all the cores. Nevertheless, some of the cores may be colder
pre­stellar cores, and decreasing the dust temperature to 15 K
would increase the derived masses by a factor 2.5. Actually, an
external heating could drive the temperature of pre­stellar cores
towards higher values. Evidence of heating resulting from a
strong external radiation field has been found for protostellar
cores in Orion by Jrgensen et al. (2006). However, the pres­
ence of energetic sources in the mapped region of Vela­D is
still controversial (Lorenzetti et al. 1993, Elia et al. 2007).
The deconvolved sizes and the masses of the cores listed
in Table 1 range from 0.03 to 0.25 pc and 0.4 to 88 M # re­
spectively, while their spatial distribution exhibits a high de­
gree of clustering. The derived sizes span the interval from
the observed diameter of pre­stellar cores in cluster­forming
regions to that of cluster­forming cores (or isolated prestellar
cores). The achieved typical sensitivity of 20 mJy/beam trans­
lates into a point source mass sensitivity (at a 1 # level) of
# 0.2 M # , using the above adopted temperature, opacity and
distance. However, see the next section for an assessment of the
mass sensitivity to extended sources.
3.2. Contamination sources of dust emission
While the e#ect of non­uniform dust temperature and opacity
throughout the sample will be discussed in the next section, the
other possible causes of error in the derived masses are contam­
ination by: i) line emission, ii) optically­thin free­free emission,
iii) optically­thick free­free emission and iv) synchrotron radi­
ation. In the following, we discuss each of the contamination
sources.
The line that can mostly a#ect the detected signal is
12 CO(2--1). We estimated its contribution using the relation
given by Braine et al. (1995), adapted to our instrumental pa­
rameters, and our observations of CO(1--0). Since CO(1--0) and
CO(2--1) are optically thick, assuming the same excitation tem­
perature yields a CO(2--1) to CO(1--0) ratio # 1. Due to the
larger beam at the frequency of CO(1--0) with respect to that
at 1.2 mm, by assuming the same main beam temperature for
CO(2--1) and CO(1--0) we still might slightly underestimate the
CO(2--1) main beam temperature. However, we checked that
the line contribution to the integrated flux density is always less
than few percent for the cores with S # # > 10 3 mJy. Conversely,
for the cores with S # # < 10 3 mJy, the estimated line contribu­
tion is # 10-30 % (55 % in one case). This is an overestimate,
since the integrated emission of CO(1--0) is < 60 K km s -1 to­
wards most of these sources and, as can be seen in Fig 8, at
least the integrated emission up to 25 K km s -1 arises from a
distributed structure much larger than the cores' size that has
been e#ectively filtered out by the observing mode of SIMBA
(see also Fig. 3 to Fig. 7).
Hence, line emission is very likely not to contribute more
than 10­20 % of the integrated flux density from the faintest
1.2mm sources. These results agree with what extensively dis­
cussed by Johnstone et al. (2003); they find that at 850 µm,
only occasionally the continuum becomes contaminated by line
emission, including CO(3--2), in strong sources, although this
may be a problem for low continuum flux sources.
Free­free emission from HII regions can also a#ect the
measured flux densities at 1.2 mm. As a first step, we checked
that none of the cores' positions coincide with radio sources

F. Massi et al.: Star formation in the Vela Molecular Ridge 5
Table 1. Dust cores found by CLUMPFIND: the robust sample.
Core RA(J2000) DEC(J2000) Deconvolved Peak Integrated Mass
designation Size Flux Flux
(pc) (mJy/beam) (Jy) (M # )
MMS1 8:45:33.4 ­43:50:20.1 0.19 501 3.17 32
MMS2 8:45:34.8 ­43:52:04.1 0.17 668 3.70 37
MMS3 8:45:40.1 ­43:51:32.3 0.19 544 3.59 36
MMS4 8:46:34.6 ­43:54:36.0 0.25 1702 8.79 88
MMS5 8:46:49.4 ­43:53:08.2 0.09 155 0.28 3
MMS6 8:46:52.3 ­43:52:59.9 0.24 121 0.20 2
MMS7 8:47:58.8 ­43:39:47.9 0.08 180 0.43 4
MMS8 8:48:39.4 ­43:31:23.9 0.02 77 0.08 0.8
MMS9 8:48:43.0 ­43:31:48.0 0.08 87 0.15 1.5
MMS10 8:48:43.0 ­43:37:08.0 0.03 83 0.13 1.3
MMS11 8:48:45.1 ­43:37:40.1 0.06 82 0.12 1.2
MMS12 8:48:49.0 ­43:32:28.0 0.13 617 1.77 18
MMS13 8:48:49.7 ­43:33:15.9 0.05 69 0.04 0.4
MMS14 8:48:51.1 ­43:31:08.1 0.08 117 0.12 1.8
MMS15 8:48:52.6 ­43:30:28.1 0.03 91 0.12 1.2
MMS16 8:48:53.3 ­43:31:00.1 0.05 163 0.23 2.3
MMS17 8:48:57.8 ­43:38:28.0 0.15 156 0.76 7.6
MMS18 8:49:03.6 ­43:38:12.1 0.09 105 0.28 2.8
MMS19 8:49:08.9 ­43:35:48.1 0.11 126 0.38 3.8
MMS20 8:49:11.8 ­43:35:24.0 0.09 126 0.19 1.9
MMS21 8:49:13.2 ­43:36:28.1 0.17 345 1.76 18
MMS22 8:49:27.4 ­43:17:08.2 0.09 567 1.32 13
MMS23 8:49:29.5 ­44:04:36.1 0.06 71 0.14 1.4
MMS24 8:49:31.0 ­43:17:08.2 0.06 421 0.76 7.6
MMS25 8:49:31.0 ­44:10:44.1 0.21 258 1.70 17
MMS26 8:49:33.8 ­44:10:59.9 0.20 298 1.55 15
MMS27 8:49:34.6 ­44:11:56.0 0.18 710 2.85 28
MMS28 8:50:09.4 ­43:16:27.8 0.06 118 0.20 2.0
MMS29 8:50:12.2 ­43:17:16.1 0.10 177 0.51 5.1
from the Parkes­MIT­NRAO survey at 4.85 GHz (Gri#th &
Wright 1993). Hence, their radio fluxes at 4.85 GHz are lower
than the quoted limit of 48 mJy. For optically thin free­free
sources this translates into 32 mJy at 1.2 mm. Only two of the
cores listed in Table 1 exhibit an integrated flux density < 100
mJy, whereas for all those with integrated flux density # 10 2
mJy the value extrapolated from the radio upper limit amount
to < 25 % of that measured at 1.2 mm. As shown by Massi et
al. (2003), at a distance of 700 pc this upper limit imply no HII
regions excited by stars earlier than B1. Although we assumed
optically thin emission, this agrees with the limit on the earliest
star spectral type set by the observed bolometric luminosities
(B1­B2).
The Parkes­MIT­NRAO catalogue lists extended radio
emission that, generally, lies a few arcmin from the cores.
However, there are some exceptions. Towards IRS 16: MMS1,
MMS2 and MMS3 are at a distance of # 1 # from the centre of
a radio source (quoted radius 1.1 # ) at RA(J2000)=08:45:36.3,
DEC(J2000)=-43:51:01. This is the HII region 263.619--0.533
(Caswell & Haynes 1987); the extrapolated flux density (as­
suming optically thin emission) at 1.2 mm is 1.6 Jy, 15 %
of the total integrated flux density of MMS1, MMS2 and
MMS3. Assuming the radio emission is uniform over a cir­
cle of radius 1.1 # , the expected flux density at 1.2 mm is 52
mJy/beam(SEST), i. e., at a 2 - 3# level. Clearly, as can be
seen in Fig. 2, the three cores are located at the boundary of
the HII region (whose centre almost coincides with the IRAS
location and the centre of the radio emission), confirming that
they are mostly due to dust thermal radiation. Towards IRS19:
MMS19, MMS20 and MMS21 lie # 2 # from the centre of a ra­
dio source (PMNJ0849--4335; semi­major by semi­minor axes
3. # 8â1. # 4) at RA(J2000)=08:49:22.6, DEC(J2000)=-43:35:56.
The flux density extrapolated at 1.2 mm (assuming optically
thin free­free radiation) is 294 mJy, with a mean flux den­
sity of 2 mJy/beam(SEST), well below our r.m.s. This ex­
plains why no emission is detected at 1.2 mm from this radio
source. Anyway, the extrapolated total flux density is # 13%
of that measured at 1.2 mm for the three cores. Finally, to­
wards IRS21: MMS25, MMS26 and MMS27 are located 2-3 #
from the centre of a radio emission (PMNJ0849--4413; semi­
major by semi­minor axes 2. # 6â1. # 5) at RA(J2000)=08:49:31.2,
DEC(J2000)=-44:13:47. The flux density extrapolated at 1.2
mm (assuming optically thin free­free radiation) is 94 mJy,
with a mean flux density of 1 mJy/beam(SEST), again well be­
low our r.m.s.. In fact, no mm emission is detected from this
radio source, as well. The total extrapolated flux density is only
< 2 % of that measured at 1.2 mm for the three cores.

6 F. Massi et al.: Star formation in the Vela Molecular Ridge
Table 2. Possible dust cores found by CLUMPFIND, but with size less than the SEST beam (before deconvolution).
Core RA(J2000) DEC(J2000) Peak Peak Integrated Mass
Designation Flux Flux Flux
(mJy/beam) (â local r.m.s.) (Jy) (M # )
umms1 8:46:25.8 ­43:42:26.6 310 6 2.78 28
umms2 8:46:37.2 ­43:18:34.9 86 4 0.05 0.5
umms3 8:46:37.2 ­43:19:55.6 175 8 0.12 1.2
umms4 8:46:49.0 ­43:20:27.1 111 5 0.09 0.9
umms5 8:46:50.4 ­43:21:15.1 90 4 0.05 0.5
umms6 8:46:56.6 ­43:53:07.1 97 6 0.16 1.6
umms7 8:47:28.5 ­43:27:00.4 75 4 0.04 0.4
umms8 8:47:37.3 ­43:43:40.1 71 4 0.03 0.3
umms9 8:47:39.5 ­43:43:48.0 65 4 0.02 0.2
umms10 8:47:41.0 ­43:26:28.3 84 5 0.12 1.2
umms11 8:47:42.5 ­43:43:39.1 139 8 0.10 1.0
umms12 8:47:46.8 ­43:25:55.4 79 5 0.03 0.3
umms13 8:47:55.7 ­43:39:41.5 81 5 0.07 0.7
umms14 8:47:57.9 ­43:38:59.6 110 7 0.17 1.7
umms15 8:48:02.4 ­43:39:15.5 107 7 0.12 1.2
umms16 8:48:15.7 ­43:47:07.8 92 5 0.06 0.6
umms17 8:48:23.0 ­43:31:31.1 63 4 0.07 0.7
umms18 8:48:26.7 ­43:31:39.7 66 4 0.04 0.4
umms19 8:48:33.1 ­43:30:43.9 83 6 0.11 1.1
umms20 8:48:35.5 ­43:30:59.8 83 6 0.07 0.7
umms21 8:48:36.8 ­43:31:13.8 76 5 0.07 0.7
umms22 8:48:36.6 ­43:16:51.2 84 4 0.05 0.5
umms23 8:49:24.2 ­43:13:13.4 95 6 0.12 1.2
umms24 8:49:27.1 ­43:12:33.5 86 6 0.07 0.7
umms25 8:49:27.8 ­43:12:17.3 83 6 0.05 0.5
umms26 8:49:59.0 ­43:22:55.6 91 6 0.09 0.9
Clearly, on the one hand the contamination to the cores'
flux densities from optically thin free­free emission is negligi­
ble. Synchrotron radiation has an even steeper spectral index
than free­free (S # # # -0.6 vs. S # # # -0.1 ), so its contribution at
1.2 mm is expected to be much lower. On the other hand, we
cannot exclude some contamination from optically thick free­
free emission (S # # # 2 up to a turnover frequency depending
on the emission measure) arising from ultracompact (or hyper­
compact) HII regions towards some of the cores. Indeed, as
discussed by Massi et al. (2003), some of the IRAS sources
towards the cores have the colours of UCHII regions, but the
fraction of fields exhibiting other signposts of UCHII regions
appears to be very low.
3.3. Star forming regions
Clues about the star formation history of the cloud can be de­
rived from the filamentary morphology of the gas (see Elia et<