Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.eso.org/~fcomeron/cham.ps Äàòà èçìåíåíèÿ: Fri Feb 19 16:32:27 1999 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 20:42:51 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: photosphere

A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08.03.4, 08.03.5, 08.05.2, 08.12.2, 08.12.3, 08.16.5, 13.09.6
ASTRONOMY
AND
ASTROPHYSICS
1.12.1998
Faint members of the Chamaeleon I cloud
F. Comer'on 1 , G.H. Rieke 2 , R. Neuh¨auser 3
1 European Southern Observatory, Karl­Schwarzschild­Strasse 2, D­85748 Garching bei M¨unchen, Germany; fcomeron@eso.org
2 Steward Observatory, University of Arizona, Tucson, AZ 85721, USA; grieke@as.arizona.edu
3 MPI Extraterrestrische Physik, Giessenbachstrasse 1, D­85740 Garching bei M¨unchen, Germany; rne@mpe.mpg.de
Received; accepted
Abstract. We present a survey of the central ¸ 100
arcmin 2 of the Chamaeleon I star forming cloud, including
objective prism spectroscopy in the Hff region and deep
imaging in the near­infrared. We estimate the expected
number of very low mass objects within the survey, tak­
ing as a reference the higher mass members identified in
previous studies, and assuming different ages and slopes of
the initial mass function of the Chamaeleon I population.
A new approach is introduced to estimate the contribu­
tion of background objects to the counts of low luminosity
sources. This method takes advantage of the fact that the
contribution of Chamaeleon I members should be negligi­
ble at the faintest magnitudes covered by our survey for
any reasonable shape of the initial mass function.
K­band source counts indicate the absence of a signif­
icant population of very low mass stars, implying that the
initial mass function at very low masses, approximated by
a power law, has a form \Phi(M)dM / M \Gamma1 dM or flat­
ter. This conclusion is in qualitative agreement with the
discovery of six new emission line objects in the objective
prism survey, and with the fact that only 2­3 faint objects
are detected in the region of the (J \Gamma H), (H \Gamma K) dia­
gram diagnostic of near infrared excesses of circumstellar
origin. The masses of the new emission line objects, de­
rived from recent pre­main sequence evolutionary tracks,
are found to be near, and possibly below, the hydrogen
burning limit, and their ages to be younger than 3 \Theta 10 6
years. One of them is found to be a bona­fide brown dwarf,
and its detection in a deep ROSAT exposure makes it the
first, and so far the only, brown dwarf known to emit X­
rays (Neuh¨auser & Comer'on 1998, Science, 282, 83). The
near­infrared properties of the Hff emission objects sug­
gest that, unlike at higher masses, strong Hff emission
near the hydrogen­burning limit is not accompanied by
infrared excess detectable in the K band. Comparing the
numbers of very low mass objects expected from K band
counts with the number of new Hff­emitting members, for
which we derive individual masses and ages, we find that
Send offprint requests to: F. Comer'on
the spectroscopic survey samples the initial mass function
completely, or nearly completely, down to the hydrogen­
burning limit.
Key words: Stars: circumstellar matter; coronae; emis­
sion line; low­mass, brown dwarfs; mass function; pre­main
sequence. Infrared: stars
1. Introduction
Does the process of star formation vary with the density
of the natal cloud or with the rate at which its mass
is being converted into stars? The molecular clouds in
Chamaeleon, one of the most nearby star forming regions,
contain a sparse population of young stellar objects and
other signposts of recent star formation (Schwartz 1991).
These clouds are therefore a prime site to probe star for­
mation at the low­density, low­rate extreme.
A considerable observational effort has been devoted
to follow up photometry and spectroscopy of Chamaeleon
objects detected in Hff (Gauvin & Strom 1992, hereafter
GS92; Hartigan 1993) and X­ray surveys (Feigelson et al.
1993, Huenemoerder et al. 1994, Alcal'a et al. 1995, 1997,
Covino et al. 1997). Embedded sources, including possi­
ble protostars, have also been detected in molecular­line
surveys (Reipurth et al. 1996) and in mid­infrared surveys
carried out by IRAS (Baud et al. 1984) and ISO (Nordh
et al. 1996). The near­infrared properties of young stel­
lar objects have been used to identify possible new mem­
bers in large scale surveys like DENIS (Cambr'esy et al.
1998). These studies have probed the star formation his­
tory of the region, the coevality of weak­lined and classi­
cal T Tauri stars, and the correlation of the X­ray activ­
ity with bulk properties of the emitting stars. As a result,
the properties of the stellar population of the Chamaeleon
clouds are well known down to masses of a few tenths of
a solar mass.

2 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
The very low mass stellar population, however, has re­
mained largely elusive. The low surface density of young
stellar objects inferred from observations of the more mas­
sive population implies that broad band surveys in the
visible or the near infrared will detect mostly background,
unrelated sources (Hyland et al. 1982). On the other hand,
X­ray surveys published so far, even with ROSAT pointed
observations, were limited to stars with bolometric lumi­
nosities higher than ¸ 0:15 L fi , corresponding to masses
higher than ¸ 0:3 M fi for pre­main sequence objects with
an age of 3 \Theta 10 6 years (D'Antona & Mazzitelli 1997).
The situation has been similar with available Hff surveys,
where severe incompleteness appeared at masses below
¸ 0:6 M fi (GS92). In both observational approaches, sub­
stantially longer exposure times with the available instru­
mentation are mandatory to probe the stellar initial mass
function at masses near the stellar/substellar borderline.
To constrain the characteristics of the low mass stellar
population of the Chamaeleon I cloud, we have carried out
deep surveys of a 100 arcmin 2 area in its most obscured
region (Cambr'esy et al. 1997), using JHK imaging and
spectroscopy in the Hff region. This area has the high­
est concentration of identified pre­main sequence objects
in the entire Chamaeleon complex, and thus the ratio of
member to background objects can be expected a priori
to be highest. We introduce a new method for deriving the
population of very low mass Chamaeleon I members, even
if the membership of individual objects cannot be estab­
lished. A comparison to the observed K­band magnitude
distribution allows us to constrain the slope of the initial
mass function to be similar to the behavior in other re­
gions with a higher density of young stellar objects -- i.e.,
to be roughly flat if expressed in differential logarithmic
mass units. In addition, we identify new faint young stel­
lar objects by means of the excess near infrared emission
associated with the presence of circumstellar material or
by their Hff emission, and individual masses and ages are
derived in the latter case. Masses for the new members are
found to be near, and in some cases below, the hydrogen­
burning limit. Unpublished ROSAT archive observations
with exposure times much longer than those of published
X­ray surveys of this area are used to determine or to put
stringent limits on the X­ray luminosities of the new mem­
bers. One of them, a 0.03­0.04 M fi brown dwarf, is clearly
detected in X­rays (Neuh¨auser & Comer'on 1998, hereafter
NC98).
Our observations and data reduction are described in
Section 2. The results are presented in Section 3, and their
interpretation is discussed in Section 4. We summarize our
conclusions in Section 5.
2. Observations
2.1. JHK imaging
The near­infrared observations were obtained on 27, 28,
and 29 March 1997, using the 256 \Theta 256 pixel 2 NICMOS3­
based IRAC­2b infrared camera at the ESO­MPI 2.2 m
telescope in La Silla, Chile. Two additional half­nights
were kindly made available to us on 4 and 5 April 1997.
We obtained a mosaic of 25 fields centered at the co­
ordinates ff = 11 h 07 m 26 s , ffi = \Gamma77 ffi 36 0 50 00 (J2000.0),
approximately coincident with the center of the darkest
nebulosity in visible images of Chamaeleon I, as well as
with the highest surface density of X­ray detected sources
(Feigelson et al. 1993). The star CD ­76 ffi 486 is near
the center of our mosaic. The frames cover an area of
\Deltaff \Theta \Deltaffi = 10 0 \Theta 10 0 5. In general, adjacent fields had an
overlapping strip about 10 pixels wide in common. How­
ever, due to a pointing problem of the telescope near the
South celestial pole, a few narrow gaps exist along the
North­South direction, amounting to ¸ 12% of the sur­
veyed area.
Each field consisted of nine frames per filter, taken
with 5 00 offsets over a 10 00 \Theta 10 00 grid. Each frame was in
turn the coadd of 6 individual exposures of 10 sec, pro­
viding a total exposure of 9 min per frame in each filter.
Image reduction was performed under IRAF, using stan­
dard tasks and dedicated scripts. Due to the absence of
nebulosity at infrared wavelengths, frames obtained in a
single night of observation were median­filtered to con­
struct flat fields in each filter. Median averaging of the
frames of each field also yielded the sky frame to subtract
from the object frame prior to shifting and adding. Bad
pixels were rejected in this final step.
Objects were automatically found in each frame using
the DAOFIND task under the NOAO's APPHOT pack­
age layered on IRAF. The fitting parameters were ad­
justed interactively, until the results of the object search
were found to be satisfactory by comparison to a visual
inspection of the frames. Next, digital photometry was
performed on the detected objects using the PHOT task,
also under APPHOT. The photometry was calibrated by
observing the standard stars HD 84090 and HD 106807
(Carter & Meadows 1995). For the Chamaeleon fields, an
aperture of 5'' was chosen in view of the point spread
function and the crowding of the region. The magnitudes
obtained were in general nearly independent of reasonable
changes in the aperture, except for the faintest stars. Dif­
ferences above the 0.02 mag level, which we take as an
appropriate threshold on the internal accuracy, only be­
gan to appear for magnitudes above K = 16:5, H = 17:0,
J = 18:0. We are confident that the sample is complete to
those limits, as objects fainter by more than 0.5 mag were
still detected by the automated finding procedure. How­
ever, these objects have been excluded from our study due
to their poorer photometric accuracy.

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 3
Saturation was obvious for the brightest stars in the
field. Comparing the magnitudes obtained by us to those
published by Lawson et al. 1996 (hereafter LFH96) shows
nonlinearity to be severe for stars with magnitude brighter
than ¸ 10:5 in each filter. Consequently, we used our
data only for mK ? 10.5, where the photometry gener­
ally agrees between the two surveys to ¸ 5%. Photometric
accuracy to the 0.05 mag level is also suggested by inter­
comparing the observations of the standard stars observed
each night.
In total, we obtained valid photometry for 206
objects 1 . Figure 1 shows the (J \Gamma H); (H \Gamma K) diagram for
the objects detected in all three bands, to which we have
added the brighter members identified in earlier works.
The solid lines indicate the locus occupied by dwarfs and
giants according to the spectral type vs. color calibra­
tion of Bessell & Brett 1988. We have extended the dwarf
branch toward later spectral types, whose colors are taken
from Kirkpatrick et al. 1993. Starting at the position of
the latest type considered by them, M9, we have plotted
the reddening vector corresponding to the extinction law
of Rieke & Lebofsky 1985. Departures from the normal
extinction law in the near infrared have been noticed in
some star forming regions such as R CrA (Wilking et al.
1997). However, in the present case, the upper envelope of
the distribution of stars in the (J \Gamma H); (H \Gamma K) diagram
is parallel to the reddening vector, thus supporting our
choice of extinction law.
Astrometry was performed using as a reference the po­
sitions of Chamaeleon I members listed in Appendix B of
LFH96 (excluding HD 97048, for which their quoted posi­
tion seems to be erroneous, probably due to the associated
nebulosity as pointed out by Feigelson et al. 1993; the Hip­
parcos position was used instead). Because most frames
do not contain stars appearing in that list, most positions
were derived by using stars in the overlapping frame ar­
eas to refer to frames with position references. The overall
astrometric precision is estimated to be better than 5''.
2.2. Spectroscopy
The objective prism observations were carried out in the
nights of 7 and 8 March 1998 using the Danish Faint Ob­
ject Spectrograph and Camera (DFOSC) at the 1.5 m
Danish telescope at the European Southern Observatory.
We used a grism providing a dispersion of 22 nm/mm
on the 2k \Theta 2k pixels 2 , 30.7 mm \Theta 30.7 mm detector.
The spectral resolution is seeing­dependent, as no slit was
used, and is estimated to be –=\Delta– ' 500 from the result­
ing spectra. To eliminate contamination by background
light, we used a Gunn r filter which isolated wavelengths
between approximately 6050 š A and 7200 š A. The com­
bined efficiency curve of the system (filter transmission,
grating efficiency, and detector quantum efficiency) varies
1 The list of positions and JHK photometry can be obtained
from the authors upon request.
Fig. 1. Infrared color­color diagram displaying the position
of the objects detected in our survey and with reliable mag­
nitudes. The open squares correspond to 8 bright confirmed
members in the surveyed area; the excluded object is HM
16, for which no published JHK photometry is available, and
which was saturated in our images. We have used the JHK
photometry provided by Lawson et al. 1996 for 7 of the objects;
for CHXR 76 (= B34: Baud et al. 1984), whose H and K pho­
tometry is quoted as uncertain by Lawson et al., we have used
instead our measured magnitudes, H = 11:32, K = 10:94. The
solid curves represent the locus of unreddened main sequence
dwarfs (lower curve) and giants, and the dashed line is the
normal reddening vector with E(J \Gamma H)=E(H \Gamma K) = 1:7. Its
upper extreme corresponds to an extinction of AV = 20 mag.
smoothly over the useful spectral range, as was confirmed
by inspection of the spectra of the earliest­ type stars in
our images. In addition, two images were obtained in the
V and I filters to enable reliable source identification in
the spectroscopic frames and to perform approximate dig­
ital aperture photometry in the visible. The center of the
field is the same as for the JHK observations described
above.
Eight individual objective prism images, each of 1800
seconds of exposure time, were obtained with small off­
sets of the telescope in between. The sky background in
these frames was high enough to permit the construction
of a flat field by stacking all the images together with­
out correcting for the offsets. This was done by first nor­
malizing the counts in each bias­subtracted frame to the
background level of an empty control region, to correct
for the varying background illumination level. The values
obtained at each pixel position in the normalized frames
were then median­averaged, rejecting the values deviat­
ing significantly from the average. Each individual, bias­
subtracted image was then divided by the normalized flat
field. The flat­fielded images were finally coadded, each
one shifted as determined from the centroids of the Hff
emission of selected stars.
The combined objective prism image was inspected for
traces of Hff emission. The spectra selected in this way

4 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
CHXR 74
CHXR 78C
1
B 34
HM 15
2
Sz 23
3
HM 16
VW Cha
4
5
6
Glass I
HD 97048
CHXR 26
CHXR 73
LkHa 332­17
Fig. 2. Finding chart for the emission line objects discussed in
this paper. The newly detected members are labeled from 1 to
6, while the principal denomination of LFH96 has been used
for previously known members.
were extracted using the NOAO APEXTRACT package
layered on IRAF. The individual spectra were wavelength
calibrated in an approximate way, using as reference wave­
lengths the peak of Hff emission and, for the latest stars,
the peak of the rather sharp feature lying between broad
TiO absorption bands at 7045 š A. For stars earlier than
M0, the TiO bandhead at 6875 š A provided a more accu­
rate reference and was used instead. The positions of the
newly detected Hff sources, labeled 1 to 6, are indicated in
Figure 2 on our I­band image of the area, along with other
known Chamaeleon I members. Spectra are presented in
this paper for all of these sources with the exception of the
two brightest ones, HD 97048 and LkHff 332­17, whose
spectra appear saturated; and of CHXR26, whose spec­
trum is strongly contaminated by that of B 34. Given
the larger field of view of DFOSC compared to the in­
frared survey, spectra were obtained for some Chamaeleon
I members not observed in the infrared.
Digital photometry was performed on the V and I im­
ages in the same way as for the JHK observations. Aper­
tures 9 pixels (=4''3) in diameter were used, although the
results were found to be generally insensitive to the aper­
ture size. The magnitude zeropoint was calibrated by tak­
ing the values published by GS92 and LFH96 for some
of the brighter stars in the field. The zeropoints derived
from each individual star were found to be consistent to
within 0.05 magnitudes. An exception was the star CHXR
78 NE, for which LFH96 give V = 12:88, indicating it was
1.2 magnitudes brighter then than in our data: the quoted
value makes it of similar brightness to Glass I, another star
in the field, while it is clearly much fainter in our image.
To our knowledge, however, the star has not been classi­
fied as variable, and has been considered as unrelated to
the cloud by Huenemoerder et al. 1994.
A comparison with deep X­ray pointed observations
from the ROSAT data archive revealed that the source la­
belled as 1 (hereafter Cha Hff 1) in Figure 2 was the only
clear X­ray detection among the newly identified emission
line objects (see Section 3.5, and also NC98). This is a
very faint object producing a barely visible continuum in
the objective prism image, from which no reliable deter­
mination of the spectral type could be made. At the same
time, its very low luminosity indicated a mass probably
well below that of the other Hff emitting objects, and pos­
sibly substellar. All this made it very desirable to obtain
follow­up observations of this source, aimed at obtaining
better quality spectra from which its position in the H­R
diagram and its physical parameters could be reliably as­
certained. Spectra in the visible and the near infrared of
Cha Hff1 were obtained with the ESO New Technology
Telescope (NTT) on La Silla in May 1998; see NC98 for
technical details on these observations.
2.3. New X­ray observations
Apart from the flux­limited ROSAT All­Sky Survey obser­
vation (cf. Alcal'a et al. 1995), the Chamaeleon I dark cloud
members have been observed twice with the Positional
Sensitive Proportional Counter (PSPC) (Pfeffermann et
al. 1988) onboard ROSAT (Tr¨umper 1983). We have re­
trieved from the ROSAT data archive a 32 ksec ROSAT
pointed observation (# 200207) obtained on 1 February
1991 for H. Zinnecker as Principal Investigator (PI). We
merged these data with a 5 ksec PSPC pointed observa­
tion (# 200046), also centered on Chamaeleon I, which
was obtained in March 1991. These additional data are not
yet in the ROSAT archive, but we can use them by cour­
tesy of Prof. Eric Feigelson, the PI. The ROSAT All­Sky
Survey exposure times are much smaller (below 1 ksec),
so that we do not consider them here, but see Alcal'a et
al. 1995.
We performed standard local and map source detection
algorithms with EXSAS version April 1998 (Zimmermann
et al. 1994) running under ESO­MIDAS version Nov 1997
to reduce the merged data set and found that ¸ 2'' (20'')
can be allowed for the identification of bright (faint) X­
ray sources with optical counterparts (see Figure 3). The
probability P for existence of a source is estimated as max­
imum likelihood ML with ML = \Gamma ln(1 \Gamma P ). A value of,
eg., ML = 14:3 (or 5.9) corresponds to a detection sig­
nificance of 5 (or 3) Gaussian oe. Some results from the
short 5 ksec pointing were already published in Feigelson
et al. 1993, but since we are dealing mainly with very
late­type stars, which are very faint both in the optical

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 5
and X­ray wavelengths, most of them were not detected
in the short pointing, but only in the merged data set. A
preliminary analysis of the long 31 ksec pointing was per­
formed by Braun 1992, whose source list and X­ray data
are in good agreement with our data (Section 3.5). The
merged ROSAT PSPC image of the Chamaeleon I dark
cloud is shown in NC98.
Fig. 3. Summary of the correspondences between objects de­
tected in the infrared, Hff, and X­ray surveys. Indicated are
previously known members of Chamaeleon I (*), the new emis­
sion line objects (+) and the new sources with near­infrared ex­
cess (X). The polygons indicate the ROSAT PSPC error boxes
around sources detected with a confidence level above 3oe.
3. Results
A problem in studying the very low mass population of
Chamaeleon I is separating the cluster members from the
unrelated background objects, which dominate the over­
all source counts. In principle, this may be done if one
can identify cluster members by some of the unique signa­
tures displayed by young stellar objects, such as X­ray or
Hff emission, near­infrared excess, or variability. However,
some objects may not display these phenomena; moreover,
these signatures are a poorly known function of the mass
or the age of the young stellar object. It is therefore de­
sirable to estimate the number of members of the young
stellar aggregate using a procedure that is as independent
as possible of their individual identification through these
signatures.
3.1. A new method for background subtraction
In relatively rich clusters, such an estimate can be achieved
by comparing on­cluster and off­cluster star counts, apply­
ing a reddening model to the off­cluster counts, and sub­
tracting them from the on­cluster values (e.g., Luhman
and Rieke 1998; Luhman et al. 1998). However, in a very
poor aggregate such as Chamaeleon I, we were concerned
about possible subtle differences between an off field and
the true cluster background that might introduce biases
in the subtraction. In addition, unless a huge off­cluster
area is surveyed, the background correction at the lowest
cluster masses has a statistical weight similar to the clus­
ter counts, and the subtraction yields a low significance
for the counts of low mass objects. Therefore, we have de­
veloped a new method in which deep imaging data purely
on the cluster can be used to calibrate and remove the
background counts at high statistical weight.
Our approach is based on the substantial difference in
slope between the K luminosity function in a young clus­
ter and the K counts on the background. For example,
from the D'Antona & Mazzitelli 1997 tracks, the mass­
luminosity relation between 0.1 and 1 M fi at 3 \Theta 10 6 years
is L ¸ M 1:5 , and the mass ­ K luminosity function is
flatter still, LK ¸ M 1:2 . It has been found in a number of
clusters that the low mass IMF is roughly flat in logarith­
mic units. Then, because the exponent in the mass­LK
relation is close to 1, the predicted K luminosity function
(in magnitudes) is virtually flat. In contrast, using the
simplest geometric approximation, the counts in an infi­
nite, extinction­free three dimensional stellar population
should go as 10 fim , where m is the magnitude and fi = 0:6.
So long as they can be approximated by such exponen­
tials, the form of these distribution functions is preserved
for a young population embedded at varying depths in a
clumpy cloud, as well as for a background population seen
through a screen of variable extinction. This is straight­
forward to show for the background population. Let us
assume that k 10 fim dm is the unobscured apparent mag­
nitude distribution of the background: if we call q(A) dA
the fraction of the surveyed area having a foreground ex­
tinction between A and A+ dA, the ''obscured'' distribu­
tion function becomes
N (m) dm =
\Gamma Z 1
A=0
k q(A) 10 fi(m\GammaA) dA
\Delta
dm (1)
which can be written as
N (m) dm = k 0 10 fim dm (2)
with
k 0 =
Z 1
A=0
k q(A) 10 \GammafiA dA ;
Z 1
A=0
q(A) dA = 1
Analogous expressions are found for the embedded
population, if q(A) dA is now understood as the probabil­
ity that the foreground extinction in front of a given source
of the aggregate is comprised between A and A+dA. The

6 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
complications arising from the structure of the obscuring
clumpy molecular cloud are thus engulfed in k 0 , leaving
fi unaffected. As a result, the observed counts from any
embedded cluster should include an approximately flat
magnitude distribution from the cluster members, with
a rapidly rising distribution of background counts at the
faint end if the data are sufficiently deep to penetrate the
obscuring cloud. Examples of this behavior can be found
in Luhman and Rieke 1998 and Luhman et al. 1998.
The conventional on­cluster versus off­cluster subtrac­
tion is based on modeling k 0 and modifying the off­cluster
counts accordingly to estimate the background. However,
the slope invariance of the background counts lets us in­
stead derive k 0 by normalizing to the faint end of the ob­
served K magnitude distribution, if it is sufficiently be­
low the inflection in the number counts. In addition, if
the data are deep enough so the faint background counts
are large, the subtraction can be made at high statistical
weight even for brighter levels where the counts are small.
3.2. Application to Chamaeleon I
3.2.1. Background population
Because the galaxy is not infinite and due to other effects,
the background stellar counts do not rise as steeply as fi =
0.6 from the simple geometric model. An accurate model­
ing of the galactic structure as traced by different classes of
objects thus becomes necessary to accurately predict the
expected stellar counts in any given direction. Wainscoat
et al. 1992 have produced such a model to predict point
source counts in the mid infrared, finding as a byprod­
uct expected star counts at shorter wavelengths as well. A
comparison with actual star counts, discussed by those au­
thors, finds an excellent match between observations and
model predictions, except at very small angular distances
from the galactic equator where modeling uncertainties
and extinction effects can produce significant departures.
However, these effects are not expected to be important
at the b = \Gamma16 ffi of our survey. The comparison between
model predictions and K­band star counts in Wainscoat
et al.'s work is not extended up to the K = 16:5 limit of
our survey, but the very good match that is achieved in
their comparison with high latitude V ­band data at much
fainter magnitudes gives us confidence in that the model
predictions should also be useable over the whole magni­
tude range of our survey. Wainscoat et al. find that the
apparent magnitude m distribution in the K band away
from the galactic plane is very well represented by an expo­
nential law of the form N (m)dm = k10 fim dm, where k is a
normalization factor, and fi varies slowly with galactic lat­
itude: at l = 90 0 , it changes from 0.30 at b = 10 0 to 0.32 at
b = 30 0 . We therefore adopt fi = 0:31 for the present case.
Since theoretical models predict a rapid decrease of the
K band luminosity with decreasing mass for very young
objects with K ¸ 14 at the distance of Chamaeleon I, and
the survey is complete to K = 16.5, the counts at the faint
end of the distribution will contain a negligible portion of
cluster members.
We should point out that the adoption of the index fi
of the magnitude distribution of background sources as an
external input parameter is not strictly necessary for our
assessment on the contents of the aggregate, as long as a
power law is indeed a good approximation. Model results
are used in order to fix the value of fi in Eq. (1), but the
observations alone would already allow its independent
determination, provided that the condition of a negligible
contribution of aggregate members to the star counts at
faint magnitudes is fulfilled. In this respect, the validity of
the assumption of a power law, and of its adopted index,
as an appropriate representation of the background distri­
bution of magnitudes in our case can be assessed from the
observed magnitude distribution. This will be discussed
again in view of the results obtained at the end of Section
3.2.2.
3.2.2. Expected number of Chamaeleon I sources
To model the number of very low mass members of the
Chamaeleon I aggregate, we assume that the region con­
tains a population of coeval sources embedded at differ­
ent depths in the molecular cloud, with a maximum fore­
ground extinction Amax . The extinction towards a given
source is assumed to have a random value between 0 and
Amax , and the survey is assumed to be limited to, and
complete in, the apparent magnitude interval m 1 to m 2
(m 1 ? m 2 ). If the initial mass function of the population
is \Phi(M), with C \Phi(M)dM being the number of objects
with masses between M and M + dM, then the total
number N of objects expected in the magnitude­limited
sample is
N =
Z Amax
A=0
dA
Z m2 \GammaDM \GammaA
M=m1 \GammaDM \GammaA
C \Phi(M(M )) dM
dM
dM (3)
where M is the absolute magnitude at the reference wave­
length of a star of mass M, and DM is the distance mod­
ulus of the region.
We have applied Eq. (3) to our K­band results, with
m 1 = 16:5 and m 2 = 10:5, as discussed above. We have
adopted DM = 6:0 for Chamaeleon, corresponding to a
distance of 160 pc (Whittet et al. 1997) which is supported
by recent Hipparcos­based determinations (Wichmann et
al. 1998, Knude & HÜg 1998). The M to MK transforma­
tion for a given age has been derived from the theoretical
pre­main sequence tracks of D'Antona & Mazzitelli 1997;
we list in Table 1 the model masses, luminosities, and tem­
peratures as a function of the age for unreddened objects
at the K = 10:5 bright limit of our survey. Our M to MK
transformation ignores the possibility of infrared excesses
of circumstellar origin, which could be noticeable in the
K band. However, as will be discussed in some detail in

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 7
Section 3.5, very few objects in our sample appear to dis­
play such excesses. Table 1 demonstrates that our survey
probes from masses destined to settle near the bottom of
the main sequence downward into the brown dwarf regime.
Although the lower magnitude limit corresponds to masses
of ¸ 0.02 M fi , objects near this limit will contribute neg­
ligibly to the counts due to the small value of dM=dM
predicted by pre­main sequence tracks.
Table 1. Physical parameters at K = 10:5
age (Myr) M (M fi ) L (L fi ) T (K)
1 0.10 0.10 3000
3 0.25 0.12 3350
10 0.55 0.14 3750
To evaluate equation (3), we need to determine two
free parameters, Amax and C. The maximum extinction
Amax is a rather ill­defined quantity, due to the clumpi­
ness of molecular clouds. As most of the sources plotted
in Figure 1 are expected to be background, their scatter
along the reddening line reflects the variable extinction
along different lines of sight. For typical molecular clouds
with a clump mass spectrum characterized by a power
law of slope \Gamma1:6, Comer'on et al. 1996b estimated that
the scatter in extinctions along the line of sight is char­
acterized by a dispersion oe ' 0:5A bg , where A bg is the
average background extinction; i.e., the scatter is compa­
rable to the extinction itself. In view of Figure 1, we have
used AKmax = 1:2 for our calculations, corresponding to
AV = 10:5. This is somewhat higher than the value of
AV = 7 found by Cambr'esy et al. 1997 from star counts,
but the difference can be understood as due to the lower
spatial resolution provided by the method employed by
those authors in regions of higher extinction. Since very
few cluster members are expected near the faint end of
our sample, and since objects intrinsically brighter than
K = 10:5 but obscured above that threshold by the inter­
vening extinction should not be very abundant, the value
of Amax is not critical for our estimates.
The scaling factor C is estimated from the number of
confirmed members in the area of our survey, assuming
that they sample the mass function completely down to
a mass Mmin . Table 2 lists these stars, for which LFH96
provide individual mass and age estimates based on two
different sets of isochrones. We have obtained new deriva­
tions of these quantities using our spectroscopy and new
sets of pre­main sequence evolutionary tracks by Burrows
et al. 1993, 1997 and D'Antona & Mazzitelli 1997. Our
new results are discussed in Section 3.4.
At least eight of these stars are detected X­ray sources.
The ninth one, HM 16, is probably a weak X­ray emitter
too, as can be seen in Figure 3 of Feigelson et al. 1993,
as well as in our Figure 3 and in Figure 1 of NC98: this
Table 2. Known Chamaeleon I members in the surveyed area
Name ff (2000) ffi (2000)
CHXR 73 11:06:28.9 ­77:37:33.0
CHXR 74 11:06:57.4 ­77:42:10.4
LHff 332­17 11:07:20.8 ­77:38:03.3
B 34 11:07:35.4 ­77:34:50.7
CHXR 26 11:07:37.1 ­77:33:32.9
HM 15 11:07:45.0 ­77:39:40.6
HM 16 11:07:59.3 ­77:38:43.9
HD 97048 11:08:04.6 ­77:39:17
Glass I 11:08:15.2 ­77:33:52.7
Note: All the coordinates are from LFH96, with the exception
of HD 97048, for which they were taken from the SIMBAD
database.
elongated source is actually closer to HM 16 than to their
proposed counterpart, HD 97048, which is more likely to
be associated with another faint peak southeast of their
source 29. Taking the new evolutionary tracks and the
relation L x /L \Lambda = 1:6 \Theta 10 \Gamma4 (Feigelson et al. 1993), the
X­ray sample extends on average to ¸ 0:2 M fi . However,
the dispersion in the luminosity ratio implies that some
sources will begin to be missed in the X­ray sample at 0.4
M fi , and that some objects may be included at 0:1M fi
if the X­ray to bolometric luminosity relation extends to
this low mass range (see below). Therefore, for a flat IMF,
we can assume an effective completeness limit of 0.2 M fi .
To show the dependence on this value, we have considered
cases with Mmin = 0:4 M fi and Mmin = 0:2 M fi .
We have modeled the initial mass function \Phi(M) as­
suming it to have the shape given by Miller & Scalo 1979
down to a mass of 0.2 M fi . The only role of the adopted
mass function above this mass is the derivation of the
normalization factor C, which is otherwise little sensitive
to the shape of the mass function in that interval: for
instance, assuming that its slope is constant and equal
to ­2.35 (i.e. the classical Salpeter mass function) all the
way down to 0.2 M fi would change C by less than 30 %.
This is a rather extreme example, as no star forming re­
gion is known to have such a steep mass spectrum at low
masses, and it can be used to illustrate the fact that the
choice of Mmin and the uncertainties of the small num­
ber statistics are expected to have a far greater influence
on our results. For masses below 0.2 M fi , a power law is
assumed, with varying exponents: the values considered,
\Phi(M) / M \Gamma1:5 , / M \Gamma1 , and / M \Gamma0:5 , cover practically
all the range of initial mass functions observationally de­
rived in numerous studies on the field and on clusters.
Table 3 lists the number of low mass Chamaeleon mem­
bers expected to have 10:5 ! K ! 16:5, for different as­
sumptions concerning the age, the shape of the initial mass
function, and the range of masses sampled by the con­
firmed members. The general trends appearing in Table
3 can be understood using qualitative arguments: if the

8 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
Table 3. Expected numbers of faint Chamaeleon I members
\Phi(M) / M \Gammaff
age (Myr) ff = 1:5 ff = 1 ff = 0:5
a) predicted members sampling the IMF down to 0.4 M fi
1 79 29 13
3 86 38 25
10 54 36 32
b) predicted members sampling the IMF down to 0.2 M fi
1 44 16 7
3 47 21 14
known members are distributed over a wider mass range
(i.e., to 0.2 M fi rather than only to 0.4 M fi ), then C
is smaller and fewer members are predicted. Concerning
the dependence on the power law index ff, a steeply rising
initial mass function towards smaller masses is expected
to result in many more faint objects observed than in the
case of a flatter initial mass function, explaining the de­
creasing numbers as we move from left to right in Table
3. On the other hand, the trends seen as we move from
younger to older ages of the population are also the result
of the evolution in luminosity of low mass objects. As the
population ages, bright stars enter the interval of magni­
tudes considered here as their K magnitude increases over
10.5, while initially faint stars leave the interval as they
fade beyond K = 16:5. Whether the total number of stars
in the interval increases or decreases with time depends on
two factors: the slope of the initial mass function, and the
rate of decrease of the luminosity. Thus, although in all the
cases assumed here the faint objects outnumber the bright
ones, the number of stars in the interval can actually grow
with time due to the slower evolution in luminosity of the
less massive stars, provided that the slope of the initial
mass function is shallow enough. This is the case for the
shallowest slope considered here, \Phi(M) / M \Gamma0:5 , while
for an initial mass function \Phi(M) / M \Gamma1 both effects
nearly cancel out over the age interval 1 ­ 10 Myr.
The histogram of K magnitudes is shown in Figure 4.
As explained in Section 2.1, we consider the source detec­
tion and the photometry to be reliable up to K = 16:5; we
will therefore focus our discussion on the 10:5 ! K ! 16:5
interval. Also shown in Figure 4 is the expected magni­
tude distribution of background stars, normalized to the
number of detected stars with 15 ! K ! 16:5, where the
contribution of Chamaeleon I members should be negligi­
ble. This normalization may be affected by a bias due to
the rising shape of the apparent magnitude distribution
combined with the increasing photometric errors as one
goes to fainter magnitudes: the larger number of objects
for which we measure K ? 16:5 implies that more objects
from the 16:5 ! K ! 17:0 bin will be erroneously assigned
to the 16:0 ! K ! 16:5 bin than the opposite, thus leading
to an overestimate of the objects with 16:0 ! K ! 16:5.
Fig. 4. Histogram of observed K magnitudes. The connected
dots represent the values expected from a fit to the Wainscoat
et al. 1992 background source distribution, normalized to the
observed number of objects with 15 ! K ! 16:5. The error
bars are
p
N times the values predicted by the normalized
background model. The faintest bins are included to illustrate
the completeness level of the survey; however, the photometry
becomes inaccurate above K ' 16:5, and sources start to be
missed by the automated detection procedure around K ' 17.
However, the internal consistency of the photometry (0.02
mag; see Section 2.1) at our adopted cutoff suggests that
this bias should not be relevant in practice for the present
case.
The observed distribution of K magnitudes shows a
good overall agreement with the predictions of the back­
ground population model, with hints of an excess of ob­
served objects at the brightest magnitudes that can be
interpreted as the statistical signature of the Chamaeleon
I population. Due to the small numbers of both predicted
and observed objects, this excess is difficult to quantify.
It is clear, however, that Figure 4 argues against the ex­
istence of a large number of Chamaeleon I members in
the interval 10:5 ! K ! 16:5. For example, we can es­
timate the exponential index fi of the background mag­
nitude distribution that would be required if there were
a minimum of 50 members with K ! 15 in the surveyed
region, assuming that the counts are dominated by the
background above K = 15 (which should be the case, un­
less most of the sources that we detect are brown dwarfs
with M ! 0:02 M fi !). We find that such a situation would
require fi ? 0:6, this is, even steeper that in the extreme
case of an infinite Galaxy without extinction.
A closer look at this excess is presented in Figure 5. In
it, we have calculated the excess of sources (i.e., the dif­
ference between the observed number of sources and the
extrapolated background) as a function of the limiting
magnitude. The error bars reflect the
p
N uncertainties
due to the limited­number statistics. Their length equals
q
N observed +N 2
background =N norm , where N norm is the num­

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 9
Fig. 5. Distribution of the excess of sources over the extrapo­
lated background as a function of the limiting magnitude. The
curves represent the distribution expected for some cases se­
lected from Table 3:
a) known members sampling the IMF down to 0.2 M fi , age 10 6
yr, ff = 1 (solid line);
b) known members sampling the IMF down to 0.4 M fi , age 10 6
yr, ff = 0:5 (dotted line);
c) known members sampling the IMF down to 0.2 M fi , age
3 \Theta 10 6 yr, ff = 0:5 (dashed line);
d) known members sampling the IMF down to 0.2 M fi , age 10 6
yr, ff = 1:5 (dot­dashed line).
The first three cases are those providing the best fit to the data.
The last one is the best fitting case with ff = 1:5; even so, it
still predicts too many Chamaeleon members at moderately
faint magnitudes.
ber of objects in the magnitude interval used to calculate
the normalization factor of the background distribution
function. The different curves represent selected examples
of expected source counts as a function of limiting mag­
nitude: they are derived from Eq. (3), using m 2 = 10:5
and varying m 1 to different limits. The examples plotted
in Figure 5 are the three providing the best fit to the data
points, plus the best fit that can be obtained imposing an
IMF slope ff = 1:5. Notice that the vertical scale is now
linear, rather than logarithmic as in Figure 4.
3.3. Objects with Hff emission
In the core of IC 348, Luhman et al. 1998 show that 16 of
67 stars with spectral types of F or later have Hff equiv­
alent widths of 10 š A or greater. Over a larger region in
this cluster, Herbig 1998 found 51 stars with Hff equiv­
alent widths above this threshold out of a total of 109.
These studies suggest that, in a low density region such as
Chamaeleon I, about 40% of the very young stars might be
expected to have detectable Hff in our survey. We assume
that the embedded population is all sufficiently young that
the Hff has not begun to fade (Ÿ 3 million years). From
Figure 4, there are 6 embedded objects with 10.5 ! K !
11.5, corresponding to a total of ¸ 7 objects if we allow for
the missing area in the infrared survey. We would predict
that 40%, or 3, of these stars should have detectable Hff.
Figures 6 and 7 show the spectra of 14 of the objects
with detected Hff emission, identified in Figure 2. The
spectra plotted in Figure 6 correspond to previously iden­
tified Chamaeleon I members; in six of the nine cases,
spectral classifications were already available in the lit­
erature. Figure 7 shows the spectra of five of our newly
discovered faint members; Cha Hff1 is separately discussed
at the end of Section 3.1.1. Table 4 summarizes the avail­
able photometry and the equivalent width measured for
the Hff emission. Most photometric data come from the
V and I images obtained as described in Section 2.2 and
from the JHK survey described in Section 2.1. Excep­
tions are the brighter sources that appear saturated in
those images or those lying outside the area of the JHK
survey, for which values found in the literature are given
instead when available. These cases are marked in Table
4. We note that photometry in other bands (U , B, R, L 0 )
can be found as well for most of the brighter objects in
GS92 and LFH96. Three of the new aggregate members
can be found too in the USNO­PMM catalog, providing
low precision B and R magnitudes which are quoted in the
footnotes to Table 4. Where we predict three stars with
detectable Hff in the interval 10.5 ! K ! 11.5, we detect
four, one with Hff equivalent width ? 10 š A, one close to
this value, and two near 5 š A. This result agrees satisfacto­
rily with our prediction and supports the assumption that
the entire population is 3 million years or less in age.
3.3.1. Spectral Classification
Although the wavelength interval covered by our objective
prism spectra spans only about 1000 š A centered around
Hff, it contains several features useful for spectral clas­
sification (e.g. Prosser et al. 1991, Williams et al. 1994,
Mart'in et al. 1996). We will use the narrow synthetic
bands R 2 , R 3 , R 4 , and R 7 of Prosser et al. 1991 for this
purpose; their limits are listed in Table 5. The reader is re­
ferred to Prosser et al. 1991 for a detailed discussion of the
correlation between flux ratios in these passbands and the
spectral type. We note here that the fact that our spectra
are not flux­calibrated does not have a noticeable impact
on the derived ratios, as the system response does not
change appreciably over the R 2 +R 3 or the R 4 +R 7 inter­
vals, thanks to the fact that the passbands from which the
ratios are calculated are narrow and adjacent. The ratios
that we find are in general somewhat outside the limits
of those considered by Prosser et al., thus indicating later
spectral types than those considered in that work. How­
ever, given their proximity to the extreme values of the
ratios found by those authors, and the fairly monotonic
trends exhibited by the ratios between M0 and M5 (see
Figure 4 of Prosser et al.), we believe that the extrapo­
lation to the ranges found here is a safe one. A possible

10 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
Fig. 6. Spectra of the nine sources having been identified as Chamaeleon I members in previous studies. The vertical scale is
in arbitrary units, and no flux calibration has been applied.

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 11
Fig. 7. Spectra of the brightest five of the newly discovered Chamaeleon I members extracted from the objective prism image.
The vertical scale is in arbitrary units, and no flux calibration has been applied.
exception is Cha Hff 1, that we classify using different
criteria, as discussed below.
Table 5. Narrow bands for M star classification (Prosser et al.
1991)
Band limits ( š A)
R2 6507 ­ 6598
R3 6635 ­ 6718
R4 6750 ­ 6844
R7 7000 ­ 7068
Table 6 lists the values of the R 4 =R 7 and the R 3 =R 2
ratios for all the stars whose spectra are presented in Fig­
ures 6 and 7. To calculate the flux in the R 2 band, we cut
out the Hff emission by interpolating the underlying pseu­
docontinuum between adjacent points. The spectral clas­
sification uses the weighting scheme proposed by Prosser
et al. 1991, i.e., giving twice as much weight to R 4 =R 7 as
to R 3 =R 2 for stars later than M0, and equal weight oth­
erwise. This is not a critical choice, as we find that the
spectral types defined by either index are usually within
one spectral subclass of each other. Spectral types found
in the literature for some of our objects are also given,
and the overall agreement is found to be good. The only
possible exceptions are CHXR 74, for which LFH96 find
a rather uncertain, but apparently earlier spectral type
than we do, and CHXR 73, a highly obscured object for
which no good quality spectra have been obtained yet.
The somewhat earlier spectral types given by GS92 to
VW Cha and Glass I can be explained as due to the de­
crease in sensitivity to spectral type of the narrow band
indices for spectral types earlier than M0, which makes
our classifications more uncertain.

12 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
Table 4. Photometry and Hff equivalent width for all the objects observed
Name ff (2000) ffi (2000) V I J H K W(Hff) ( š A) Notes
CHXR 73 11:06:28.9 ­77:37:33 20.2 15.60 12.81 11.32 10.79 ­
CHXR 74 11:06:57.4 ­77:42:10 17.27 13.59 11.55 b 10.57 b 10.23 b 13
B 34 11:07:35.4 ­77:34:51 18.09 14.31 12.29 11.32 10.94 5.5
HM 15 11:07:45.0 ­77:39:41 15.42 12.55 a 10.41 b 9.09 b 8 .35 b 70
Sz23 11:07:59.4 ­77:42:40 18.04 14.40 ­ ­ ­ 45 out of the JHK survey
HM 16 11:07:59.3 ­77:38:44 15.66: ­ ­ ­ ­ 61 I, J , H, K saturated
VW Cha 11:08:01.6 ­77:42:29 13.16 10.61 a 8.91 b 7.75 b 7.0 8 b 64
Glass I 11:08:15.2 ­77:33:53 12.81 a 10.90 a 8.77 b 7.40 b 6.34 b 11
CHXR78C 11:08:54.6 ­77:32:12 18.98 14.76 12.41 b 11.64 b 11.28 b 3.2
Cha Hff1 11:07:17.0 ­77:35:54 21.0 c 16.23 13.55 12.78 12.28 59
Cha Hff2 11:07:43.0 ­77:33:59 20.4 15.11 12.59 11.43 11.15 39
Cha Hff3 11:07:52.9 ­77:36:56 19.42 14.90 12.46 11.64 11.11 4.5
Cha Hff4 11:08:19.6 ­77:39:17 18.67 14.35 12.20 11.42 11.04 4.7
Cha Hff5 11:08:26.9 ­77:41:21 19.68 14.69 12.14 11.21 10.76 7.6
Cha Hff6 11:08:40.2 ­77:34:17 19.75 14.97 ­ ­ ­ 59 out of the JHK survey
Notes on the photometry:
a : photometry from GS92
b : photometry from LFH96
c : V measurement obtained by R. Neuh¨auser with the ESO­NTT telescope on La Silla in July 1998; photometric accuracy is
\Sigma0:23 mag.
Three of the new Chamaeleon I sources are found in the USNO­PMM catalog, from which rough B and R magnitudes (\Sigma0:5
mag) are derived. These are:
Cha Hff1: B = 23:3, R = 19:6.
Cha Hff2: B = 23:4, R = 18:9.
Cha Hff6: B = 23:1, R = 18:3.
Table 6. Spectral index ratios and classification
Name R4=R7 R3=R2 Sp. type notes
CHXR 73 0.59: 0.74: M4.5: K3­M2 in LFH96
CHXR 74 0.74 0.66 M4.5 K2­M2 in LFH96
B 34 0.72 0.61 M5
HM 15 1.26 0.89 M1 M0.5 in GS92
Sz23 ­ 0.79 M2.5 R7 out of frame
HM 16 1.42 0.93 K7
VW Cha 1.35 0.90 M0.5 K5 in GS92
Glass I 1.46 0.90 K7 K4 in GS92
CHXR78C 0.64 0.58 M5.5 M4­M6 in LFH96
Cha Hff1 ­ ­ M7.5­M8 from NTT spectra
Cha Hff2 0.59 0.50 M6
Cha Hff3 0.55 0.53 M6
Cha Hff4 0.55 0.48 M6.5
Cha Hff5 0.54 0.53 M6
Cha Hff6 0.57 0.50 M6
The classification scheme based on the R i ratios de­
scribed above is based on spectral types M5.5 and earlier,
and may therefore be inappropriate to classify a clearly
later­type object like Cha Hff 1, given the saturation that
is often found for similar indexes at very low temperatures.
Moreover, the much wider wavelength coverage available
for this object thanks to the NTT spectra makes it possi­
ble a much more robust classification based on other fea­
tures. Here we discuss in some detail the derived spectral
type for this object on the basis of different classification
schemes that have been proposed for late M dwarfs, ex­
panding the short discussion given in NC98.
A spectral sequence extending to spectral types as late
as M9, documented with a large number of standards, has
been proposed by Kirkpatrick et al. 1991, 1995. A compar­
ison of composite spectra of stars classified in this scheme
with Cha Hff 1 is shown in Figure 8, which shows that the
latter object must be clearly later than spectral type M6.
This is especially well seen by comparing the dependence
on the temperature of the discontinuity in the slope of the
spectrum near 8000 š A. The composite spectrum whose
overall shape matches most closely the spectrum of Cha
Hff 1 seems to be M8V.
Kirkpatrick et al. 1991 also provide a number of ''color
ratios'' measuring the strength of different temperature­
sensitive features that we can measure on the spectrum of
Cha Hff 1. Ratio A, measuring the CaH feature at 6975 š A,
peaks at spectral type M6.5­M7, and the value measured
for Cha Hff 1, 1.38, is found at spectral types M5 and
M7.5­M8; the former possibility is ruled out by the overall
appearance of the spectrum. Ratio B, measuring the TiI

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 13
Fig. 8. A comparison of field M dwarf spectra (individual and
composite) with the spectrum of Cha Hff 1. The M6V spectrum
is that of Gl 406; M7V, that of vB 8; M8V is the combination
of the spectra of LHS 2243, LHS 2397 A, LP 412­31, and vB
10; and M9V, that of BRI 1222­1222, LHS 2065, LHS 2924,
and TVLM 868­110639. The spectral classifications are from
Kirkpatrick et al. 1995. Note in particular the stronger NaI
feature at 8190 š A in the field M dwarfs, indicating a lower
surface gravity of Cha Hff 1.
feature at 7358 š A, has a broad peak between M5.5 and M6;
the value we measure, 1.11, is near the peak value. On the
other hand, Ratio C, measuring the NaI features at 8183
and 8195 š A, is very low for our object at 1.05; such low
values are found for spectral type standards M8 and later,
but would be also expected as a surface gravity effect,
as discussed below. Finally, Ratio D measuring the CaII
feature at 8542 š A is also very low at 1.07. This index has a
minimum near this value between M4 and M5, then rises
at lower temperatures, and drops again to that level for
stars later than M8. Therefore, with the possible exception
of Ratio B, all the other color ratios consistently agree in
classifying Cha Hff 1 as M7.5 or later.
Another set of color ratios, aimed at establishing a
spectral sequence for young brown dwarfs, has been de­
fined by Mart'in et al. 1996. These authors propose a
quantification of the spectral subtype as a function of
several pseudo­continuum spectral regions. Using their
expressions for the spectral type as a function of the
pseudo­continuum indexes PC3, PC4, and PC5, we find
M spectral subtypes of 7.82, 8.28, and 7.01, respectively.
The weighted average based on the standard deviation
of the spectral subtype­pseudocontinuum index relation­
ship yields a M7.8 type for Cha Hff 1. We note that the
other two pseudocontinuum indexes, PC1 and PC2, both
lie above the peak values found in the standard stars used
by those authors, both in the proximities of M8. Finally,
we have measured the VO ratio based on the absorption
band produced by that molecule at 7445 š A, whose defini­
tion and advantages have been discussed by Kirkpatrick
et al. 1995. The value obtained, 1.076, places Cha Hff 1
in subclass 8.
In summary, we find a remarkable overall agreement
among all the classification criteria based on the red part
of the visible spectrum, which almost unanimously assign
a spectral type M7.5 ­ M8 to Cha Hff 1. This classifi­
cation is in good agreement with that derived from the
spectrum in the 2 ¯m region, discussed by NC98, which
is dominated by a broad depression between the H and
K bands due to water absorption in the photosphere of
the star. Its depth cannot be accurately quantified due to
strong telluric absorption in that region, but comparison
with synthetic atmosphere models (Allard & Hauschildt
1995a) clearly suggests a temperature of 2800 K or lower.
The NaI feature around 8190 š A, known to decrease
with decreasing surface gravity (Kirkpatrick et al. 1991),
is clearly weaker in the spectrum of Cha Hff 1 than in
those of any of the field M dwarfs shown for compari­
son. This is to be expected, as the surface gravity of a
young stellar object should be intermediate between that
of an evolved M dwarf (where NaI is strong) and that of
a giant (where NaI is weak). A surface gravity effect in
the same direction is also noticed in the NaI feature at
2.21 ¯m, which is not visible in the spectrum of Cha Hff
1 (NC98), while it is a prominent feature in late field M
dwarfs (Jones et al. 1994). Luhman & Rieke 1998, Luhman
et al. 1998, and Wilking et al. 1998 also find a NaI feature
systematically weaker than expected for a given spectral
type, or even absent, in young stellar objects in L1495E
and IC 348. The relative weakness of NaI absorption in
the visible spectrum of Cha Hff 1, and its absence in the
K­band spectrum, can thus be taken as a proof that it
is a true young stellar object, and therefore a member of
the Chamaeleon I aggregate, rather than an evolved fore­
ground dMe star unrelated to the dark cloud population.
The membership of Cha Hff 1 is independently supported
by its noticeable infrared excess at longer wavelengths,
which has been detected recently by ISOCAM on board
of the Infrared Space Observatory at 6.75 and 15 ¯m (An­
laug Amanda Kaas, private communication; Olofsson et
al. 1998).
3.3.2. Luminosities and temperatures
The spectral classification and photometry presented in
Tables 4 and 6 permit the placing of the Chamaeleon I
objects in the H­R diagram, using temperature­spectral
type calibrations plus intrinsic colors and bolometric cor­
rections for late­type atmospheres. In turn, a comparison
to the L­T eff loci predicted by theoretical pre­main se­
quence evolutionary tracks allows an estimate of the age
and the mass of the objects.
We will use the compilation of intrinsic colors and bolo­
metric corrections for main sequence stars by Kenyon &
Hartmann 1995 to derive foreground extinctions and lu­
minosities for our objects, based on the spectral types

14 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
given in Table 6. Allard & Hauschildt 1995b show that
model spectra for solar metallicity M dwarfs change lit­
tle within the range of surface gravities relevant for our
objects. We have adopted the extinction law of Rieke &
Lebofsky 1985; as discussed in Section 2.1, this law ade­
quately reproduces the extinction vector traced by back­
ground sources in Chamaeleon I in the (J \Gamma H); (H \Gamma K)
diagram. At visible wavelengths, departures from the uni­
versal extinction law have been found in Chamaeleon I
(Steenman & The 1989, Whittet et al. 1987). To mini­
mize their effect, we have based the extinction estimates
on the I and H bands, under the assumptions that I is
at a long enough wavelength to be free from anomalous
extinction effects, and that H is short enough not to be
affected by circumstellar emission. At the J band, where
the continuum spectral energy distributions of our newly
found low mass sources peak, the extinction A J is:
A J = 0:919 [(I \Gamma H) \Gamma (I \Gamma H) 0 ] (4)
where (I \Gamma H) and (I \Gamma H) 0 are respectively the observed
and intrinsic color indices. For the cases in which no H
measurements are available, namely Sz23 and our new ob­
ject Cha Hff6, a formula analogous to Eq. (4) has been
used instead, replacing the (I \Gamma H) index by (V \Gamma I), and
the numerical coefficient by 0.544.
The luminosity L is:
log L(L fi ) = 1:86 \Gamma 0:4 [J \Gamma A J \Gamma DM+(V \Gamma J) 0 +BC] (5)
where J is the measured J magnitude, DM = 5:73 is the
distance modulus (Section 3.2.2), and BC = M bol \Gamma MV
is the bolometric correction. Again, for those cases for
which only V and I are available, an equivalent form of
Eq. (5) has been used, in which the term between square
parentheses has been replaced by [V \Gamma3:55A J \GammaDM +BC].
Luhman & Rieke 1998 compare existing spectral type­
temperature calibrations. Following their recommenda­
tion, we have adopted their best fit to the calibration of
Leggett et al. 1996, which in turn is based on fits to syn­
thetic spectra provided by cool model atmospheres. The
temperature scale is in good agreement with that derived
by Jones et al. 1996. Uncertainties in the temperature cal­
ibration of these models may be estimated to be around
200 K (Allard et al. 1997). Then, using the correspon­
dences between spectral types and narrow band flux ratios
of Prosser et al. 1991, we have transformed this calibration
into:
T eff (K) = 1922 + 818:6 (R 4 =R 7 ) + 852:3 (R 3 =R 2 ) (6)
which includes the weighted average of the two flux ratios
described above. Due to the flattening of the spectral type
vs. (R 4 =R 7 ) or (R 3 =R 2 ) relationships at earlier spectral
types, Eq. (6) is applicable only to spectral types M0 or
later. A modified form of Eq. (6) has been used for Sz23,
because (R 4 =R 7 ) is not available in this case. For Cha
Hff1, we have used directly the spectral subtype which
enters the best­fit formula of Luhman & Rieke 1998.
Table 7. Physical parameters, extinctions and infrared ex­
cesses
Name Teff AJ log L(L fi ) \Delta(H \Gamma K)
CHXR 73 3035: 2.0: ­0.79: ­0.23:
CHXR 74 3090 0.7 ­0.82 0.04
B 34 3031 0.6 ­1.18 0.07
HM 15 3712 1.6 0.12 0.55
Sz23 3448 0.7 ­1.18 ­
VW Cha 3794 1.1 0.51 0.49
CHXR78C 2940 0.6 ­1.23 0.02
Cha Hff1 2600 0.4 ­1.79 ­0.06
Cha Hff2 2831 1.0 ­1.16 ­0.09
Cha Hff3 2824 0.6 ­1.27 0.15
Cha Hff4 2781 0.2 ­1.31 ­0.01
Cha Hff5 2816 0.7 ­1.06 0.07
Cha Hff6 2815 0.2 ­1.57 ­
Table 7 gives the resulting physical parameters, plus
the extinction and the intrinsic infrared excess calculated
as \Delta(H \Gamma K) = (H \Gamma K) \Gamma (H \Gamma K) 0 \Gamma 0:22A J . Only
the objects with spectral types M0 or later have been re­
tained. The positions of the objects of Table 7 in the H­R
diagram are plotted in Figure 9, together with recent the­
oretical pre­main sequence evolutionary tracks. The com­
parison between different isochrones allows an estimate of
the uncertainties in the physical parameters due to mod­
eling alone, thus providing a lower limit to the uncertain­
ties in derived mass and ages. As expected, the results
are somewhat model­dependent; nevertheless, they agree
qualitatively that at least five of the new emission line ob­
jects are very young, with ages of 3 \Theta 10 6 years or less,
and masses near the hydrogen­burning limit. From the
preceding Section, it appears that most of the stars in
the infrared survey are of similar age; there is no signif­
icant older population with negligible Hff emission. The
young age for these objects may simply reflect the fact
that Chamaeleon I is of too low a density to be gravi­
tationally bound, so older stars will have wandered far
from their formation sites. Of course, it may be possible
as well that star formation in Chamaeleon I has simply
taken place only in the last 3 Myr. The ages and lumi­
nosities of our objects as determined from both sets of
isochrones are given in Table 8.
Although some of the new emission line objects may be
brown dwarfs according to both sets of pre­main sequence
evolutionary tracks, confirmation awaits a more precise
determination of T eff than presently available and higher
accuracy in the models. The exception is Cha Hff1, for
which uncertainties in the photometric measurements, the

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 15
0.237 M o
0.2 M o
0.1 M o
0.078 M o
0.05 M o
1 Myr
3 Myr
50 Myr
10 Myr
B 34
0.5 Myr
Sz23
CHXR 74
CHXR 73
CHXR 78C
Cha H 5
a
Cha H 2
a
Cha H 3
a
Cha H 4
a
Cha H 6
a
Cha H 1
a
50 Myr
10 Myr
3 Myr
1 Myr
0.5 Myr
0.05 Mo
0.08 M o
0.1 M o
0.5 M o
0.25 M o
a
VW Cha
HM 15
CHXR 73
CHXR 74
CHXR 78C
B 34
Sz23
Cha H 5
a
Cha H 2
a
Cha H 3
a
Cha H 4
a
Cha H 6
a
Cha H 1
Fig. 9. H­R diagram with the position of our objects, su­
perimposed on the pre­main sequence evolutionary tracks of
D'Antona & Mazzitelli 1997 (top panel) and Burrows et al.
1997 (bottom panel). The solid lines represent equal ages, and
the dotted lines equal masses. Note the difference in scales be­
tween both plots.
Table 8. Ages and masses of low mass objects
DAM97 B97
Name age mass age mass
(Myr) (M fi ) (Myr) (M fi )
CHXR 73 0.5 0.14 1.5 0.15
CHXR 74 1 0.15 2 0.17
B 34 2 0.16 4 0.12
HM 15 0.3 0.3 ­ ­
Sz23 10 0.3 ­ ­
VW Cha 0.2 0.25 ­ ­
CHXR78C 2 0.14 3 0.09
Cha Hff1 0.3 0.03 0.5 0.03
Cha Hff2 1 0.10 0.4 0.08
Cha Hff3 1.5 0.09 0.5 0.075
Cha Hff4 1 0.08 0.5 0.065
Cha Hff5 0.5 0.10 0.4 0.09
Cha Hff6 3 0.08 2 0.06
Notes: DAM97 = D'Antona & Mazzitelli 1997 models, B97 =
Burrows et al. 1997 models. The models of Burrows et al. 1997
reach up to 0.237 M fi ; HM 15, Sz23, and VW Cha are out of
their model grid.
spectral classification, the temperature calibration, or the
evolutionary tracks all seem to be insufficient to move it
into the region of hydrogen­burning main sequence stars.
It can be therefore considered as a bona­fide brown dwarf
(NC98).
3.4. Objects with near­infrared excess
Strom et al. 1989, 1993 have studied the incidence of K­
band excesses in Taurus­Auriga. This work should be par­
ticularly relevant for comparison with Chamaeleon I be­
cause of the low density of young stars in both systems.
It is expected that the incidence of disks, or their life­
times, may be affected in very dense environments. These
studies found 22 of 48 sources younger than 3 Myr to
have detectable K excesses (greater than 0.2 magnitudes),
while 10 of 23 sources between 3 and 10 Myr in age had
such excesses. Given the age distribution for objects in the
Chamaeleon I field, we would predict that about 40% of
them would have excesses, based on this work. This pre­
diction is consistent with other observations: 14 sources
with IR excess out of 74 X­ray detections in the entire
Chamaeleon I complex (LFH96), 7 out of 21 in L1495E in
Taurus (Strom & Strom 1994), 19 out of 28 in ae Ophiuchi
(Casanova et al. 1995; the JHK photometry used is from
the compilation of J. Carpenter, private communication
by T. Greene), and 5 out of 19 in NGC 2024 (Freyberg &
Schmitt 1995; Comer'on et al. 1996). Taking all these data,
the overall proportion of excess objects is 35%. Assuming
an IMF slope of 0:5 ! ff ! 1:0, an age of 3 Myr, and a
35 to 40% incidence of excesses, from Table 3b we would
predict 4 to 8 objects in the infrared survey with infrared

16 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
excesses. If ff were as large as 1.5, there should be ¸ 16
to 20 objects with excesses.
We identify excesses on the (J \Gamma H); (H \Gamma K) dia­
gram (Lada & Adams 1992, Meyer et al. 1997). Circum­
stellar disks produce a (H \Gamma K) color redder than that
corresponding to a stellar photosphere obscured by fore­
ground dust for a given (J \Gamma H). If the disk emission is
intense enough in the K band, the position of an object
in a (J \Gamma H); (H \Gamma K) diagram is shifted to the right of
the reddening vector to a region inaccessible to normal
reddened photospheres.
We find only three faint objects with probable excesses,
in rough agreement with the predictions for 0:5 ! ff ! 1:0
but somewhat lower than expected. The detected excesses
are far below the prediction for ff=1.5. The individual data
for these objects are listed in Table 9. The third object
in the list has its J and H magnitudes near the limits
of acceptable precision of our survey. Consequently, its
color indices, especially (J \Gamma H), are rather uncertain, and
its actual location to the right of the limiting reddening
vector is doubtful. The mass, temperature, and extinction
for three possible ages have been estimated following the
procedure described in detail by Comer'on et al. 1996a,
1998.
The unknown ages of these faint Chamaeleon I mem­
bers makes it difficult to assess their true character. If we
assume age estimates from those derived for the brighter
members, they may all be brown dwarfs. However, the age
distribution of the brighter members of Chamaeleon I is
very sensitive to the adopted set of theoretical isochrones,
as illustrated by LFH96 (see also the discussion in Luh­
man & Rieke 1998 on the L1495E stellar population). It
is therefore possible that the two brighter objects are of
stellar mass, while the status of third object is uncer­
tain because of possible photometric errors. Nonetheless,
these objects illustrate the advantage of identifying clus­
ter members through infrared excesses; it allows isolation
of a portion of the very low luminosity cluster population.
The incidence of infrared excess among the brighter
sources can be determined by comparison with the X­ray
detections. It is likely that all nine previously known mem­
bers of Chamaeleon I have been detected by ROSAT (see
discussion in Section 3.2.2), but five of them have JHK
colors compatible with having no near infrared excess. We
could not find JHK photometry for HM 16, which ap­
pears saturated in our images. However, its strong Hff
emission (Henize & Mendoza 1973) suggests that it should
fall to the right of the reddening line, given that all the
other bright objects of Table 2 for which strong Hff emis­
sion has been detected do appear in that region of the
(J \Gamma H); (H \Gamma K) diagram. Therefore, four out of the nine
would be selected as members on the basis of their infrared
excess, in excellent agreement with the prediction that 35
­ 40% of these objects would behave in this way.
Given the ''normal'' portion of excesses among the
more luminous sources, the small number of objects with
Fig. 10. Infrared color­color diagram displaying the position of
the objects in the Chamaeleon I region and a limiting reddening
vector as in Figure 1, plus the faint L1495E sources (open
triangles) and the brown dwarf candidates of ae Ophiuchi (open
pentagons).
K ! 14 and detectable infrared excesses in the (J \Gamma H),
(H \Gamma K) diagram may be a result in part of observational
biases and small number statistics. To assess the expected
detection rate better, we used available JHK photometry
of faint members in other nearby stellar populations with
intrinsic properties similar to those in Chamaeleon. First,
we considered the faint objects of L1495E whose JHK
photometry is given by Strom & Strom 1994. This region
is similar to Chamaeleon I in that it contains a sparse
population of lightly embedded young stellar objects, and
moreover it lies at nearly the same distance from the Sun.
We find in Tables 3 and 6 of Strom & Strom 1994 fourteen
objects with J ? 12 and uncertainties in J , H, and K of
0.1 mag or less. Their position in the (J \Gamma H), (H \Gamma K) dia­
gram is plotted in Figure 10 (open triangles); three of them
lie to the right of the limiting reddening vector. We also
plotted in Figure 10 (open pentagons) the location of the
six candidate brown dwarfs in the ae Ophiuchi cloud core
observed by ISO (Comer'on et al. 1998), whose masses are
relatively well constrained by fits to their spectral energy
distributions. Four of them lie in the region characteristic
of near infrared excesses produced by circumstellar disks.
The combined results for L1495E and ae Ophiuchi thus in­
dicate that circumstellar excesses detectable through the
position of the objects in the (J \Gamma H), (H \Gamma K) diagram
are not substantially less common for very low mass stars
and brown dwarfs than for more massive stars, despite the
low temperatures of the former. This has been recently
confirmed by Wilking et al. 1998, who have identified ex­
cess fluxes in the K band in a number of brown dwarfs in
ae Ophiuchi. The small number of low luminosity sources

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 17
Table 9. Faint sources to the right of the limiting reddening vector
Name R.A. (J2000.0) dec (J2000.0) J J \Gamma H H \Gamma K age (Myr) M (M fi ) T (K) AV
IR­1 11:06:42.7 ­77:35:57 15.44 1.74 1.21 1 0.05 2850 11
3 0.06 2850 11
10 0.3 3500 13
IR­2 11:07:32.9 ­77:37:53 17.67 2.29 1.46 1 0.03 2750 15
3 0.04 2800 16
10 0.12 3150 17
IR­3 11:08:41.2 ­77:37:15 17.92 1.25 0.90 1 ! 0:02 ! 2650 ! 5
3 ! 0:02 ! 2600 ! 5
10 ! 0:02 ! 2600 ! 5
with excesses in Chamaeleon I therefore places an upper
limit of ff ! 1 for the low mass IMF.
3.5. X­ray properties of the very low mass members
As already reported in NC98, the brown dwarf Cha Hff1
is clearly detected and identified with the X­ray source
RXJ 110716\Gamma773552 (source XP 30 in Braun 1992), while
Cha Hff3 and 6 are barely detected (4 oe). None of these
three sources were detected earlier, neither in an Einstein
Observatory X­ray image due to much lower sensitivity,
nor in an earlier analysis of the 5 ksec PSPC pointing.
Cha Hff2 is very close to another, partly elongated, bright
X­ray source, identified here as CHXR 26, unresolvable
with the PSPC (cf. Figure 3). Cha Hff5 is clearly unde­
tected and upper limits are given in NC98.
The modest spectral resolution of the PSPC allows
some spectral analysis. In particular, we can estimate so­
called X­ray hardness ratios (X­ray colors) defined as fol­
lows: if Z s;m;h are the count rates in the bands soft (0.1
to 0.4 keV), medium (0.5 to 0.9 keV), and hard (0.9 to
2.1 keV), respectively, then hardness ratios are given by
HR 1 = Z h + Zm \Gamma Z s
Z h + Zm + Z s
and HR 2 = Z h \Gamma Zm
Z h + Zm
Hence, by definition, hardness ratios can range from
\Gamma1 to +1. If, for any particular source, no soft source
were detected, then one can only give a lower limit to the
hardness ratio HR 1, the upper limit being HR 1 = 1.
The hardness ratios measured for the Chamaeleon
I members (Table 10) are consistent with ¸ 1 keV
Raymond­Smith (Raymond & Smith 1977) spectra of so­
lar abundance absorbed by the foreground column density
as given in Table 7. To obtain X­ray fluxes, we divide the
count rates by the appropriate energy conversion factor,
namely 10 11 cts cm 2 erg \Gamma1 (see Neuh¨auser et al. 1995 for
details).
In Table 10, we compile the X­ray data of the
Chamaeleon I members studied here. We list source desig­
nation, offset between optical and X­ray source position,
maximum likelihood ML of source existence, effective ex­
posure time, background and vignetting corrected counts
in the broad band (0.1 to 2.4 keV), hardness ratios, X­ray
luminosity LX (at 160 pc) and the ratio of the X­ray to
bolometric luminosity, with L bol from either our Table 7
or LFH96.
In addition, we also calculate (and list in Table 10)
the upper limits to X­ray emission of the low­mass (brown
dwarf) candidates found with our IR survey (Table 9); the
second object (IR­2), though, is located too close to LHff
332­17 to be resolved with the PSPC, so that no upper
limit can be given.
4. Discussion
In view of the small­number statistics, and the assump­
tions involved in modeling, the results described in Sec­
tion 3.2 do not firmly establish the slope of the low mass
IMF in Chamaeleon I. However, from the estimates pre­
sented in Table 3, from the comparisons between observed
and predicted numbers presented in Figure 3, and from
the low incidence of infrared excess sources, we conclude
that the slope is rather shallow, in the range ff = 0:5 \Gamma 1
in linear mass units. This behavior is equivalent at low
masses to an IMF rising slowly with increasing mass or be­
ing flat, if expressed in logarithmic mass units. Our result
is consistent with recently determined mass functions of
nearby open, emerged clusters such as the Pleiades (Fes­
tin 1997, and references therein), Praesepe (Williams et
al. 1995), and the Hyades (Leggett et al. 1994). It also
agrees with the results for younger aggregates such as ae
Ophiuchi (Comer'on et al. 1993, Strom et al. 1995), NGC
2024 (Comer'on et al. 1996a), the Taurus clouds (Strom &
Strom 1994), L1495E (Luhman & Rieke 1998) and IC 348
(Luhman et al. 1998). An exception to this behavior may
be the Lupus clouds, where Hughes et al. 1994 find an
excess of low mass stars over the predictions of the Miller
& Scalo 1979 initial mass function. The greater distance
to this complex resulting from a revision using Hipparcos
data (Wichmann et al. 1998) may remove this apparent
discrepancy, although the results of Knude & HÜg 1998,
also based on Hipparcos, cast some doubts on whether the
distance is really higher, or actually lower, than the values

18 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
Table 10. X­ray data for Chamaeleon I cloud members studied here
Designation pos. detection exp. X­ray counts hardness ratios log LX log
offset level (ML) [ksec] (broad band) HR 1 HR 2 [erg/s] LX=L bol
CHXR 73 15.9'' 91 37.8 61:9 \Sigma 10:0 0:48 \Sigma 0:21 0:67 \Sigma 0:13 28:70 \Sigma 0:06 \Gamma4:10
CHXR 74 1.9'' 257 36.7 118:9 \Sigma 12:3 – 0:86 0:37 \Sigma 0:09 28:99 \Sigma 0:05 \Gamma3:78
B 34 5.2'' 130 37.8 73:3 \Sigma 10:1 – 0:69 0:31 \Sigma 0:13 28:77 \Sigma 0:05 \Gamma3:64
HM 15 7.3'' 460 37.1 178:1 \Sigma 15:1 0:96 \Sigma 0:06 0:68 \Sigma 0:06 29:16 \Sigma 0:04 \Gamma4:55
Sz 23, VW Cha (1) 6856 35.9 1576:5 \Sigma 40:7 0:98 \Sigma 0:01 0:43 \Sigma 0:02 30:12 \Sigma 0:02 (1)
HM 16, HD 97048 (2) 161 37.1 89:7 \Sigma 11:7 0:76 \Sigma 0:13 0:74 \Sigma 0:09 28:87 \Sigma 0:05 (2)
Glass I (3) 5.3'' 4086 37.5 947:4 \Sigma 31:5 0:95 \Sigma 0:02 0:37 \Sigma 0:03 29:88 \Sigma 0:02 \Gamma4:75
CHXR 78 C 3.5'' 61 37.6 59:4 \Sigma 10:0 0:57 \Sigma 0:22 0:12 \Sigma 0:17 28:68 \Sigma 0:07 \Gamma3:68
LHff 332­17 7.1'' 5634 37.4 1221:7 \Sigma 35:7 – 0:98 0:49 \Sigma 0:03 30:00 \Sigma 0:01 \Gamma4:65
CHXR 26 4.7'' 519 37.8 194:9 \Sigma 15:2 0:85 \Sigma 0:07 0:66 \Sigma 0:06 29:19 \Sigma 0:04 \Gamma4:34
Cha Hff 1 4'' 33.7 37.8 31:4 \Sigma 7:7 – 0:40 0:15 \Sigma 0:22 28:40 \Sigma 0:10 \Gamma3:44
Cha Hff 2 not resolved, 29 00 SE of X­ray source identified as CHXR 26
Cha Hff 3 16.8'' 11.8 37.6 11:9 \Sigma 5:7 – 0:13 0:04 \Sigma 0:35 27:98 \Sigma 0:17 \Gamma4:31
Cha Hff 4 not detected 34.8 Ÿ 22:9 Ÿ 28:30 Ÿ \Gamma4:91
Cha Hff 5 not detected 33.9 Ÿ 1:98 Ÿ 27:25 Ÿ \Gamma5:24
Cha Hff 6 10'' 8.1 31.8 8:2 \Sigma 3:4 – 0:26 0:08 \Sigma 0:40 27:89 \Sigma 0:15 \Gamma4:09
IR­1 (3) not detected 30.7 Ÿ 4:7 Ÿ 27:67 Ÿ \Gamma4:49
IR­2 not resolved, located too close to bright X­ray source LHff 332­17
IR­3 not detected 34.0 Ÿ 5:3 Ÿ 27:67
Remarks: (1) X­ray source confusion: Both VW Cha and Sz 23 are very close (3:0 and 11:4 arc sec, respectively) to this X­ray
source, which is too close to be resolved spatially with the ROSAT PSPC. (2) X­ray source confusion: HM 16 and HD 97048
are very close to this elongated, spatially unresolved X­ray source. (3) We use L bol = 8:5 \Delta L fi for Glass I, and L bol = 0:029 \Delta L fi
for IR­1, both from our photometry.
adopted so far. The field initial mass function is more diffi­
cult to determine, due to factors such as the non­coevality,
the uncertainties in the mass­luminosity transformation
(Tinney 1993), or the observational effects of binarity at
different distances (Kroupa 1995a, 1995b), but a shallow
mass function also seems to reproduce the observations
(Jarrett et al. 1994).
The very low mass population in the surveyed region
seems to be younger than 3 \Theta 10 6 years. This is in overall
agreement with the results of LFH96, who found that over
60 % of the Chamaeleon I members in their list had ages
of 5 \Theta 10 6 years or less when derived from the D'Antona
& Mazzitelli 1998 tracks; however, for their five objects
in common with our Hff observations (CHXR 73, CHXR
74, CHXR 78C, VW Cha, and HM 15) we tend to find
somewhat younger ages and lower masses. The differences
are mostly due to the somewhat later spectral types that
we derive and different temperature­spectral type calibra­
tion used, rather than to differences in the two sets of
D'Antona & Mazzitelli tracks, which are not noticeable at
the level of accuracy attainable with our observations.
Although we have detected few low luminosity sources
with near infrared excesses in Chamaeleon I, we show that
this result is probably largely due to small­number statis­
tics. Taken together, the incidence of excesses for low lu­
minosity members of low density clusters is similar to that
for their higher luminosity members. One would expect a
selection effect against detection of excesses in low mass
objects. With cool spectral energy distributions, these ob­
jects will exhibit a relatively high Hff equivalent width
for a given amount of ionized gas, and their photospheric
emission will be relatively bright at K, tending to domi­
nate any excess. Thus, one predicts a decreasing incidence
of excesses as a function of Hff equivalent width, as well as
in general. The detection of excesses in significant numbers
over the full sample of low luminosity members of nearby,
low density clusters therefore is an interesting conclusion.
Models of near infrared emission of circumstellar disks
(Lada & Adams 1992, Calvet et al. 1991, 1992, Meyer et
al. 1997) successfully reproduce the observational locus of
T Tauri stars in the (J \Gamma H), (H \Gamma K) diagram, but do
not extend to the temperatures and luminosities of the
faintest sources in our study. However, some features in
models for higher mass objects suggest that circumstellar
disks around brown dwarfs should not be conspicuous in
the (J \Gamma H), (H \Gamma K) diagram. For their fiducial central
star of M = 0.5 M fi , R = 1:8 R fi , and T = 4000 K, Meyer
et al. 1997 found that accretion rates below 10 \Gamma8 M fi yr \Gamma1
do not contribute to the disk luminosity via viscous heat­
ing. The accretion rates to be expected for massive brown
dwarfs are of this order, unless we are witnessing a tran­
sient stage of rapid accretion; otherwise, steady accretion
in excess of 10 \Gamma8 M fi yr \Gamma1 sustained over ¸ 10 7 yr would
end up by giving a low mass star, rather than a brown

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 19
dwarf. Moreover, the rate of viscous heating depends on
the potential well created by the central object, which is
much smaller for a brown dwarf than for a M = 0:5 M fi
star: the radii of both objects are comparable, whereas
the mass differs by one order of magnitude. Therefore,
only irradiation by the central object is left as a source
of luminosity for the circumstellar disk. This may be in­
sufficient to produce a significant flux excess at K if the
disk has a central hole, a feature inferred by Meyer et al.
1997 to be common among young stellar objects. Thus,
further observational and theoretical study of the behav­
ior of disks around very low mass young objects would be
of great interest.
Concerning the high energy emission properties of the
M­type stars studied here (Table 10), we find X­ray to
bolometric luminosity ratios ranging from \Gamma4:75 to \Gamma3:64,
which is quite typical for X­ray active M­type stars. In
this regard, the Herbig Ae/Be star LHff 332­17 appears
to be quite typical for X­ray emission of Herbig Ae/Be
stars (cf. Zinnecker & Preibisch 1994). The detection of
an X­ray emitting brown dwarf in Chamaeleon I shows
that the luminosity ratios quoted above extend beyond
the end of the main sequence. It is doubtful however to
what extent this applies to young brown dwarfs only, as
an extensive search for X­ray emission from known brown
dwarfs has failed to provide any detections among more
evolved objects (Neuh¨auser et al. 1998)
5. Conclusions
We have presented the results of a deep objective prism
survey in the Hff region and a near­infrared imaging sur­
vey, aimed at the study of the initial mass function for
very low mass stars and brown dwarfs in the densest re­
gion (10 0 \Theta 10 0 5) of the Chamaeleon I star forming cloud.
We have also analyzed very deep pointed X­ray observa­
tions which include the area surveyed in Hff and the in­
frared. We introduce a new approach to separate the cloud
members from the background, a critical issue in studying
this sparse cluster. We find that:
1.) The initial mass function below M ¸ 0.2 M fi can be
constrained to have a rather shallow slope. When the ini­
tial mass function \Phi(M) is approximated by a power law
of the form \Phi(M)dM / M \Gammaff dM, the index ff ' 0:5\Gamma1:0.
This behavior is equivalent to an IMF that is flat or slowly
rising to 0.2 M fi in logarithmic mass units.
2.) The low mass IMF in Chamaeleon I is therefore similar
to that observed in most other young clusters, regardless
of their density.
3.) Six new objects are found in our Hff survey. A compar­
ison to recent theoretical pre­main sequence tracks shows
that they are very young (ages near or below 3 \Theta 10 6 years)
with masses near the hydrogen burning limit, and even be­
low in at least one case.
4.) In addition, two, or possibly three, fainter objects are
detected by means of their position in the (J \GammaH ), (H \GammaK )
diagram, indicative of the existence of circumstellar ma­
terial. They have luminosities and colors consistent with
being brown dwarfs if their ages are below 10 7 yr.
5.) The relatively high incidence of circumstellar excesses
around very low luminosity objects may be in conflict with
simple extrapolation of theoretical models of higher lu­
minosity objects. To explore this behavior, modeling is
needed for accretion rates around 10 \Gamma8 M fi yr \Gamma1 irradi­
ated by a low luminosity central object with a temperature
under 3000 K.
6.) X­ray emission properties of the young late type stars
in the surveyed area are similar to those of the overall
population of X­ray active M­type stars. Moreover, the
detection of a X­ray emitting brown dwarf with a derived
mass of only 0.05 M fi incorporates brown dwarfs as a new
class of X­ray emitters, with a X­ray to bolometric lumi­
nosity ratio similar to that of late M­type stars.
Acknowledgements. It is a pleasure to thank the 2p2 Team
at ESO­La Silla for their assistance during the observations,
particularly the support astronomer, Dr. Jesper Storm, and
the telescope operators, Sres. Luis Ram'irez and Rolando Vega.
The ESO Director's Discretionary Time Committee is acknowl­
edged for its allocation of time to this project. The valuable
introduction of Dr. James Brewer to the use of DFOSC and the
1.5 Danish telescope did much to improve the efficiency of the
spectroscopic observations. Dr. Giovanni Carraro is thanked
for having allowed us to use some non­photometric minutes
of his time at the NTT to obtain the spectrum of Cha Hff1
presented here. Dr. Chris Lidman kindly allowed us to include
that same object in the list of targets to be observed in the
commissioning of SOFI. Drs. Adam Burrows and Francesca
D'Antona kindly provided us with results of their recent mod­
els in machine­readable form. We also thank Dr. Eric Feigel­
son, the PI of the shorter X­ray observation, for allowing us to
use his 2nd processing data. Drs. G¨oran Olofsson and Anlaug
Amanda Kaas are thanked for having provided near infrared
and ISO data prior to publication, Dr. Warrick A. Lawson
for having kindly made available published data in machine­
readable form, and Dr. Tom Greene for making available to us
the compilation of ae Ophiuchi photometry of John Carpenter,
and Dr. Kevin Luhman for providing the comparison spec­
tra shown in Figure 8. Comments from the referee, Dr. Jens
Knude, helped us to significantly improve the contents of this
paper. This work was supported by the NATO Collaborative
Research Grant CRG 940671 and NASA grant NAGW­4083.
RN wishes to thank the Deutsche Forschungs Gemeinschaft for
financial support in their Star Formation program. ROSAT is
supported by the Max­Planck­Society and the German Gov­
ernment (BMBF/DLR).
References
Alcal'a, J.M., Krautter, J., Schmitt, J.H.M.M., Covino, E.,
Wichmann, R., Mundt, R., 1995, A&AS, 114, 109.
Alcal'a, J.M., Covino, E., Neuh¨auser, R., Schmitt, J.H.M.M.,
Wichmann, R., 1997, A&A, 319, 184.
Allard, F., Hauschildt, P.H., 1995a, ApJ, 445, 433.
Allard, F., Hauschildt, P.H., 1995b, in ''The bottom of the main
sequence ­and beyond'', ed. C. Tinney, Springer­Verlag.

20 F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I
Allard, F., Hauschildt, P.H., Alexander, D.R., Starrfield, S.,
1997, ARA&A, 35, 137.
Baud, B., Young, E., Beichman, C.A., Beintema, D.A., Emer­
son, J.P., Habing, H.J., Harris, S., Jennings, R.E., Marsden,
P.L., Wesselius, P.R., 1984, ApJ, 278, L53.
Bessell, M.S., Brett, J.M., 1988, PASP, 100, 1134.
Braun M., 1992, Diploma thesis, Friedrich­Schiller Universit¨at
Jena.
Burrows, A., Hubbard, W.B., Saumon, D., Lunine, J.I., 1993,
ApJ, 406, 158.
Burrows, A., Marley, M., Hubbard, W.B., Lunine, J.I., Guillot,
T., Saumon, D., Freedman, R., Sudarsky, D., Sharp, C.,
1997, ApJ, 491, 856.
Calvet, N., Pati~no, A., Magris, G., D'Alessio, P., 1991, ApJ,
380, 617.
Calvet, N., Magris, G., Pati~no, A., D'Alessio, P., 1992,
RMxAA, 24, 27.
Cambr'esy, L., Epchtein, N., Copet, E., de Batz, B.,
Kimeswenger, S., Le Bertre, T., Rouan, D., Tiph`ene, D.,
1997, A&A, 324, L5.
Cambr'esy, L., Copet, E., Epchtein, N., de Batz, B., Borsen­
berger, J., Fouqu'e, P., Kimeswenger, S., Tiph`ene, D., 1998,
A&A, 338, 977
Carter, B.S., Meadows, V.S., 1995, MNRAS, 276, 734.
Casanova, S., Montmerle, T., Feigelson, E.D., Andr'e, P., 1995,
ApJ, 439, 752.
Comer'on, F., Rieke, G.H., Burrows, A., Rieke, M.J., 1993, ApJ,
416, 185
Comer'on, F., Rieke, G.H., Rieke, M.J., 1996a, ApJ, 473, 294.
Comer'on, F., Torra, J., Rieke, G.H., 1996b, A&A, 308, 565.
Comer'on, F., Rieke, G.H., Claes, P., Torra, J., Laureijs, R.J.,
1998, A&A, 335, 522.
Covino, E., Alcala, J. M., Allain, S., Bouvier, J., Terranegra,
L., Krautter, J., 1997 A&A 328, 187.
D'Antona, F., Mazzitelli, I., 1994, ApJS, 90, 467.
D'Antona, F., Mazzitelli, I., 1997, Mem. S. A. It., 68, 807.
Feigelson, E.D., Casanova, S., Montmerle, T., Guibert, J.,
1993, ApJ, 416, 623.
Festin, L., 1997, A&A, 322, 455.
Freyberg, M.J., Schmitt, J.H.M.M., 1995, A&A, 296, L1.
Gauvin, L.S., Strom, K.M., 1992, ApJ, 385, 217 (GS92).
Hartigan, P., 1993, AJ, 105, 1511.
Henize, K.G., Mendoza, E.E., 1973, ApJ, 180, 115.
Herbig, G. H., 1998, ApJ, 497, 736.
Huenemoerder, D.P., Lawson, W.A., Feigelson, E.D., 1994,
MNRAS, 271, 967.
Hughes, J., Hartigan, P., Krautter, J., Kelemen, J., 1994, AJ,
108, 1071.
Hyland, A.R., Jones, T.J., Mitchell, R.M., 1982, MNRAS, 201,
1095.
Kenyon, S.J., Hartmann, L., 1995, ApJS, 101, 117.
Kirkpatrick, J.D., Henry, T.J., McCarthy, D.W., 1991, ApJS,
77, 417.
Kirkpatrick, J.D., Kelly, D.M., Rieke, G.H., Liebert, J., Allard,
F., Wehrse, R., 1993, ApJ, 402, 643.
Kirkpatrick, J.D., Henry, T.J., Simons, D.A., 1995, AJ, 109,
797.
Knude, J., HÜg, E., 1998, A&A, 338, 897.
Kroupa, P., 1995a, ApJ, 453, 350.
Kroupa, P., 1995b, ApJ, 453, 358.
Jarrett, T.H., Dickman, R.L., Herbst, W., 1994, ApJ, 424, 852.
Jones, H.R.A., Longmore, A.J., Jameson, R.F., Mountain,
C.M., 1994, MNRAS, 267, 413.
Jones, H.R.A., Longmore, A.J., Allard, F., Hauschildt, P.H.,
1996, MNRAS, 280, 77.
Lada, C.J., Adams, F.C., 1992, ApJ, 393, 278.
Lawson, W.A., Feigelson, E.D., Huenemoerder, D.P., 1996,
MNRAS, 280, 1071 (LFH96).
Leggett, S.K., Allard, F., Berriman, G., Dahn, C.C.,
Hauschildt, P.H., 1996, ApJS, 104, 117.
Leggett, S.K., Harris, H.C., Dahn, C.C., 1994, AJ, 108, 944.
Luhman, K.L., Rieke, G.H., 1998, ApJ, 497, 354.
Luhman, K.L., Liebert, J., Rieke, G.H., 1997, ApJ, 489, L165.
Luhman, K.L., Rieke, G.H., Lada, C., & Lada, E. 1998, ApJ,
in press.
Mart'in, E.L., Zapatero­Osorio, M.R., Rebolo, R., 1996, ApJ,
469, 706.
Miller, G.E., Scalo, J. M., 1979, ApJS, 41, 513.
Meyer, M.R., Calvet, N., Hillenbrand, L.A., 1997, AJ, 114, 288.
Neuh¨auser R., Comer'on F., 1998, Science, 282, 83.
Neuh¨auser R., Sterzik M.F., Schmitt J.H.M.M., Wichmann R.,
Krautter J., 1995, A&A 297, 391.
Neuh¨auser R., Brice~no, C., Comer'on, F., Hearty, T., Mart'in,
E.L., Schmitt, J.H.M.M., Stelzer, B., Supper, R., Voges, W.,
Zinnecker, H., 1998, A&A, submitted.
Nordh, L., et al., 1996, A&A 315, L185.
Olofsson, G., et al., 1998, in Brown dwarfs and extrasolar plan­
ets, ASP Conf. Ser. 134.
Pfeffermann E., et al., 1988, The focal plane instrumentation
of the ROSAT telescope, in Proc. SPIE 733, 519.
Prosser, C.F., Stauffer, J., Kraft, R.P., 1991, AJ, 101, 1361.
Raymond J.C. & Smith B.S., 1977, ApJS, 35, 419.
Reipurth, B., Nyman, L.­ š A., Chini, R., 1996, A&A, 314, 258.
Rieke, G.H., Lebofsky, M.J., 1985, ApJ, 288, 618.
Schwartz, R.D., 1991, in ''Low Mass Star Formation in South­
ern Molecular Clouds'', Ed. B. Reipurth, ESO Sci. Rep. 11.
Steenman, H., The, P.S., 1989, Ap&SS, 161, 99.
Strom, K.M., Strom, S.E., Edwards, S., Cabrit, S., & Skrutski,
M. F., 1989, AJ, 97, 1451.
Strom, S.E., Edwards, S., & Skrutski, M. F., 1993, in Proto­
stars and Planets III, ed. E. H. levy, J. I. Lunine, & M. S.
Mathews (University of Arizona Press, Tucson), 837.
Strom, K.M., Strom, S.E., 1994, ApJ, 424, 237.
Strom, K.M., Kepner, J., Strom, S.E., 1995, ApJ, 438, 813.
Tinney, C.G., 1993, ApJ, 414, 279.
Tr¨umper J.E., 1983, Adv. Space Res., 2, 241.
Wainscoat, R.J., Cohen, M., Volk, K., Walker, H.J., Schwartz,
D.E., 1992, ApJS, 83, 111.
Whittet, D.C.B., Kirrane, T.M., Kilkenny, D., Oates, A.P.,
Watson, F.G., King, D.J., 1987, MNRAS, 224, 497.
Wichmann, R., et al., 1998, MNRAS, in press.
Wilking, B.A., McCaughrean, M.J., Burton, M.G., Giblin, T.,
Rayner, J.T., Zinnecker, H., 1997, AJ, 114, 2029.
Wilking, B.A., Greene, T.P., Meyer, M.R., 1998, AJ, in press.
Williams, S.D., Stauffer, J.R., Prosser, C.F., Herter, T., 1994,
PASP, 106, 817.
Williams, D.M., Rieke, G.H., Stauffer, J.R., 1995, ApJ, 445,
359.
Zimmermann H.U., Becker W., Belloni T., et al., 1994, EXSAS
Users's Guide, MPE Report 244, ROSAT Scientific Data
Center, Garching.
Zinnecker H. & Preibisch Th., 1994, A&A 292, 152.

F. Comer'on, G.H. Rieke, R. Neuh¨auser: Faint members of Chamaeleon I 21
This article was processed by the author using Springer­Verlag
L A T E X A&A style file L­AA version 3.