Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~inr/post/nearl.ps Äàòà èçìåíåíèÿ: Fri Jun 8 23:31:48 2007 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 03:43:56 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 87 jet

Four nearby L dwarfs
I. Neill Reid
Dept. of Physics & Astronomy, University of Pennsylvania, 209 S. 33rd Street,
Philadelphia, PA 19104­6396; e­mail: inr@herschel.physics.upenn.edu
J. Davy Kirkpatrick
Infrared Processing and Analysis Center, 100­22, California Institute of Technology,
Pasadena, CA 91125
J. E. Gizis
Department of Physics and Astronomy, University of Massachusetts, Amherst, MA 01003
C. C. Dahn, D. G. Monet
U.S. Naval Observatory, P.O. Box 1149, Flagstaff, AZ 86002
Rik J. Williams
Department of Astronomy, MSC 152, California Institute of Technology, Pasadena, CA
91126­0152
James Liebert
Steward Observatory, University of Arizona, Tucson, AZ 85721
A. J. Burgasser
Dept. of Physics, 103­33, California Institute of Technology, Pasadena, CA 91125
ABSTRACT
We present spectroscopic, photometric and astrometric observations of four
bright L dwarfs identified in the course of the 2MASS near­infrared survey.
Our spectroscopic data extend to wavelengths shortward of 5000 š A in the L0
dwarf 2MASSJ0746+2000 and the L4 dwarf 2MASSJ0036+1840, allowing the
identification of absorption bands due to MgH and CaOH. The atomic resonance
lines Ca I 4227 š A and Na I 5890/5896 š A are extremely strong, with the latter
having an equivalent width of 240 š A in the L4 dwarf. By spectral type L5, the D
lines extend over ¸ 1000 š A and absorb a substantial fraction of the flux emitted
in the V band, with a corresponding effect on the (V­I) broadband colour. The
KI resonance doublet at 7665/7699 š A increases in equivalent width from spectral
type M3 to M7, but decreases in strength from M7 to L0 before broadening
substantially at later types. These variations are likely driven by dust formation
in these cool atmospheres.

-- 2 --
Subject headings: stars: low­mass, brown dwarfs; stars: luminosity function,
mass function; Galaxy: stellar content
1. Introduction
Familiarity with one's immediate neighbours is, in general, good policy. In the case
of the Solar Neighbourhood, our knowledge of the local constituents forms the basis of
the determination of fundamental statistical quantities such as the luminosity function,
the mass function, the local mass density and the star formation history of the disk.
Moreover, as the apparently­brightest members of their respective spectral classes, the
nearest celestial neighbours are most accessible to detailed astrophysical analysis. The latter
consideration is of particular importance for objects of intrinsically low luminosity, such
as old, low­temperature white dwarfs and ultracool, very low­mass (VLM) main­sequence
dwarfs.
Until recently, the main resource for the identification of VLM dwarfs remained
the proper motion catalogues compiled by Luyten from photographic material obtained
in the 1950s and 1960s using the Palomar 48­inch Oschin Schmidt. The development
of higher­sensitivity red and photographic­infrared emulsions in the 1970s permitted
photometric surveys to extend to somewhat larger depths, but this field of study has been
revolutionised through the advent of deep, near­infrared all­sky surveys, such as DENIS
(Epchstein et al, 1994) and 2MASS (Skrutskie et al, 1997). Follow­up observations of
sources with extremely red JHK or optical­to­infrared (RIJHK) colours (Delfosse et al, 1997;
Kirkpatrick et al, 1999a, hereinafter paper I) have resulted in the identification of numerous
ultracool dwarfs. Many are spectroscopically similar to the previously­unique white dwarf
companion, GD 165B, which has been transformed from an anomaly to a prototype. The
far­red optical spectra of these dwarfs are characterised by the disappearance of TiO and
VO absorption bands, the defining signature of spectral class M, and the presence of metal
hydride (CaH, FeH, CrH) bands and neutral alkali (Cs, Rb, sometimes Li) lines. The
progression of those features was ordered in paper I to define a new spectral class, type L.
The initial sample of ultracool L dwarfs discovered by 2MASS (20 objects) and
other surveys (5 dwarfs) includes only two objects with magnitudes brighter than K=12:
Kelu 1 (Ruiz et al, 1997) and 2MASSJ1439284+192915. As a result, apart from Ruiz
et al's observations of Kelu 1, spectroscopy of these sources has been confined largely to
wavelengths longward of 6400 š A. We have since extended the areal coverage of our 2MASS

-- 3 --
analysis by almost a factor of three, concentrating on identifying late­type L dwarfs. Our
current sample includes 74 spectroscopically­confirmed L dwarfs (Kirkpatrick et al, 1999b,
hereinafter Paper II). Four (including 2MASSJ1439284+192915) are of particular interest,
since their properties imply that they lie at distances of no more than 15 parsecs. All
are sufficiently bright that they supply an opportunity of extending high signal­to­noise
observations to bluer wavelengths and to higher spectral resolution. This paper provides a
brief discussion of the properties of these ultracool dwarfs.
2. Observations
The four L dwarfs discussed in this paper were all identified as candidate low­
temperature objects based on analysis of JHK S photometric catalogues derived from the
Two­Micron All­Sky Survey (Skrutskie et al, 1997). 2MASSWJ1439284+192915 forms part
of the original L dwarf sample discussed in Paper I; 2MASSWJ0746425+200032 was selected
amongst a sample of candidate bright, ultracool late­type dwarfs (discussed further by Gizis
et al, 1999); 2MASSWJ0036159+182110 and 2MASSWJ1507476­162738 were identified as
likely to be mid­ to late­type L dwarfs based on their having (J­K S ) colours redder than
1.3 magnitudes. For brevity, we shall refer to these sources as 2M0036, 2M0746, 2M1439
and 2M1507 throughout the rest of this paper. The individual photometric measurements
of each object are listed in table 1: 2M0036 and 2M1507 fall in overlap regions between
separate scans and the JHK s magnitudes are averages of the two observations. A finding
chart for 2M1439 is available in Paper I, and finding charts for the other three dwarfs are
presented in Paper II.
2.1. Spectroscopy
Each L dwarf has been observed using the Low Resolution Imaging Spectrograph
(Oke et al, 1995) on the Keck II telescope. Initial observations were obtained using a
1­arcsecond slit and the 400 l/mm grating blazed at –8500 š A, covering the wavelength range
6300 to 10200 š A at a resolution of 9 š A. An OG570 filter was used to eliminate second­order
flux. This is the standard instrumental set­up used in our L dwarf observations, and data
reduction and calibration followed the procedures described in paper I. The UT dates of
the individual observations were 14 & 16 Dec 1998 (2M0036), 24 Dec 1998 (2M0746), 8
Dec 1997 (2M1439) and 24 Dec, 1998 (2M1507). 2M0746 was also observed on Dec 4, 1998
using the modular spectrograph on the Las Campanas Observatory Du Pont 2.5­metre (see
Gizis et al, 1999 for further details), while the 2M1439 observations are described in Paper

-- 4 --
I. Spectral types have been derived for each dwarf based on the LRIS spectra plotted in
figure 1 following the precepts given in paper I (see Kirkpatrick et al, in prep. for further
details).
We have supplemented these intermediate­resolution red spectra with a range of other
observations.
LRIS: blue spectra
We also have shorter­wavelength LRIS observations of 2M0036, 2M0746 and 2M1507,
using the 300 l/mm grating blazed at 5000 š A. Those spectra were obtained on 25 Dec,
1998 (2M0036), 5 March, 1999 (2M0746) and 17 July, 1999. The respective exposure
times were 1800, 1800 and 3600 seconds respectively. As with the standard far­red
observations, we used a 1­arcsecond slit, providing a spectral resolution of ¸ 6 š A and
wavelength coverage from ¸ 3900 to 7800 š A. No order­sorting filters were employed. The
data reduction procedures mirror those used in analysing the red data, with flux calibration
provided through observations of Hiltner 600 and LTT 9491(Hamuy et al, 1994). We also
obtained lower­resolution data covering the 5400 to 10400 š A region for 2M1507 using a 158
l/mm grating on the red channel of double spectrograph on the Hale 200­inc (5.08 metre)
telescope. Those data are consistent with the higher signal­to­noise Keck spectrum.
Figure 2 plots the reduced spectra, where we include, for comparison, our observations
of the late­type M dwarf BRI0021­0214 (M9.5, Kirkpatrick et al, 1995) and data for the
L2 dwarf, Kelu 1 (Ruiz et al, 1997). The latter spectrum was kindly made available by
S. Leggett. Figure 3 provides an expanded view of the 4500 to 6600 š A region. The more
prominent molecular and atomic features are labelled in both figures. Note, in particular,
the increasing strength of both the potassium 7665/7699 and sodium 5890/5896 resonance
doublets as one progresses from spectral type M9.5 to L5.
HIRES observations
Finally, we have obtained higher­resolution echelle spectra of all four L dwarfs using
HIRES (Vogt et al, 1994) on the Keck I telescope. The observations were obtained 24 Aug
1998 (2M0036, 2M1439), 6 March 1999 (2M0746) and 14 June, 1999 (2M1507). In each
case, the data provide partial coverage of the wavelength range ––6000 \Gamma 8500 š A, including
important features such as Li I 6708 š A, K I 7665/7699 š A, Rb I 7800 & 7948 š A, and the Na
I 8183/8195 š A doublet. Total exposure times of 6000 seconds were accrued on each source.
As discussed further below, lithium was not detected in any of these four L dwarfs.
The HIRES data were flat­field corrected and the spectra extracted using programmes
written by T. Barlow. The wavelength calibration, based on Th­Ar arc lamp exposures, was

-- 5 --
determined using the iraf routines ECIDENTIFY and DISPCOR. We have not attempted
to set these data on a flux scale. Radial velocities were computed for each star either from
the measured wavelength of the Hff emission line (in 2M0746 and 2M1439) or by measuring
the central wavelengths of atomic lines due to Cs and Rb, adopting heliocentric corrections
given by the IRAF RV package. Our radial velocity measurements for M dwarfs from
the Marcy & Benitz (1989) sample indicate that the latter technique can give velocities
accurate to \Sigma1.5 kms \Gamma1 . However, the atomic lines are relatively broad in the L dwarfs,
and an internal comparison of the individual measurements suggests that the uncertainty is
2­3 kms \Gamma1 . Save for 2M1439, the uncertainties in the derived space motions are dominated
by the parallax measurements.
2.2. Photometry
CCD images in several passbands have been obtained of all four L dwarfs discussed in
this paper. The observations were made using the 40­inch telescope at the Flagstaff station
of the US Naval Observatory. Full details of the data reduction and calibration process are
given by Dahn et al (in prep.). Those data are listed in Table 1.
In addition to these direct measurements, we have used the calibrated spectra
plotted in figures 1 and 2 to synthesise (B­V), (V­R) and (V­I) colours. As in Paper I,
square passbands are adopted for each filter, and the flux zeropoints are those of the
Johnson/Kron­Cousins system (Bessell, 1979). In general, there is reasonable agreement
between the spectroscopic colours and the available direct measurements.
Finally, we have estimated bolometric magnitudes for each dwarf. While none of
these sources, and relatively few late­type M or L dwarfs in general, have observations
at wavelengths longward of 2.2¯m, the available data suggest that m bol can be inferred
with reasonable accuracy from the observed magnitude at the 1.25¯m J band. Leggett et
al (1996) infer BC J = 2:07 magnitudes for the M6.5 dwarf GJ 1111; Tinney et al (1993)
infer BC J = 1:7 mag for the L4 dwarf GD165B; and Leggett et al (1999) have derived
BC J = 2:19 mag for Gl 229B. These results indicate that there is relatively little variation
in BC J over this temperature range (¸ 2700K to ¸ 950K) and we have adopted a uniform
correction of M bol =M J + 1.75 mag for each object in the current sample.

-- 6 --
2.3. Astrometry
All four L dwarfs discussed in this paper have been placed on the US Naval Observatory
(Flagstaff) CCD parallax programme (Monet et al, 1992). Preliminary absolute parallaxes
are available in each case, and those data are listed in Table 2. These observations are
also used to derive absolute proper motions, and the results for 2M0036, 2M0746 and
2M1439 are listed in Table 2. In the case of 2M1507, the USNO observations span a period
of only 107 days, leading to significant uncertainties in ¯ and `. Fortunately, that dwarf
is visible on both the first and second epoch plates taken by the UK Schmidt telescope
as part of the southern sky survey. Most of the L dwarfs listed in paper I are visible on
the POSS II IVN I­band plates, and several are also detected on POSS II IIIaF (R­band)
plate material. 2M1507 is unusual in that it is sufficiently bright to be detected on even
the IIIaJ (blue­green) 1st­epoch UKST plates. The time difference between the two
UKST observations is 11 years, sufficient to allow a more accurate estimate of the proper
motion than provided by current CCD observations. We have used standard profile­fitting
techniques to measure the displacement between the two epochs and derive an annual
proper motion close to 1 arcsecond directed almost due south. Note that three of the four
L dwarfs have motions consistent with their inclusion in the Luyten Half Second catalogue.
3. Discussion
These new observations allow us to investigate further the spectral energy distribution
and atmospheric composition of L dwarfs. In addition, we can determine space velocities
for the four objects in the present sample.
3.1. Molecular features
Far­red optical spectra of L dwarfs show that metal hydride bands, notably CaH,
FeH and CrH, become increasingly prominent with decreasing temperature (later spectral
types). This behaviour is reminiscent of that observed in late­type metal­poor subdwarfs.
In both cases, the greater visibility of the hydride bands reflects decreasing strength of
TiO and VO absorption, albeit governed by two different mechanisms: in the subdwarfs,
the weak oxides are due to an overall scarcity of metals; in the L dwarfs, TiO and VO are
depleted as dust particles, mainly perovskite, CaTiO 3
, and solid­phase VO respectively.
Other minerals, such as enstatite (MgSi 3
) and forsterite (Mg 2
SiO 4
), are also expected to
condense at temperatures between ¸ 2100K and 1500K (Fegley & Lodders, 1996; Burrows

-- 7 --
& Sharp, 1999; Lodders, 1999).
By analogy with cool subdwarfs, other hydrides are expected to be visible at shorter
wavelengths ­ in particular, MgH (Cottrell, 1978). Figure 4 plots our LRIS observations
of three extreme ([m/H]! \Gamma1:5) dwarfs: LHS 489 (esdM0 on the system defined by Gizis,
1997), LHS 453 (esdM3.5) and LHS 375 (esdM5). Those observations were obtained
on 17 July, 1999 using the same instrumental setup and data reduction process as in
our observations of 2M1507. Comparing spectra for the two sets of objects reveals both
significant similarities and differences. The L dwarfs are substantially cooler than the
¸ 3000 to 4000K esdMs, leading to much steeper spectral energy distributions in the former
than the latter. In both cases, however, the most prominent molecular absorption is due to
metal hydrides, with the 5200 š A MgH feature obvious in all of the L dwarfs. The –4788 š A
band is clearly present in the L5, 2M1507, and is barely detected in 2M0036.
TiO bands at 4761, 4954 and 5448 š A are evident in 2M0746, but have disappeared by
spectral type L4. All of the L dwarfs also exhibit strong absorption at ¸ 5500 š A with the
band most prominent in 2M0746. This feature is likely to be calcium hydroxide, CaOH,
originally identified in mid­type M dwarfs by Pesch (1972) and increasingly strong in
later­type M dwarfs. This molecule also contributes a diffuse band at ¸ 6230 š A (Pearse &
Gaydon, 1965), which is blended with the –6158 š A fl 0 TiO band in M dwarfs (Boeshaar,
1976). The latter two bands are likely responsible for the substantial, double­bottomed
absorption feature at ¸ 6200 š A in 2M0746, evident as a shallower depression in the L4,
2M0036. Both CaOH and MgH can be identified in the spectrum of Kelu 1 presented by
Ruiz et al (1997). Finally, the VO –5736 š A band is probably responsible for a relatively
weak absorption feature in 2M0746.
3.2. Atomic lines
Table 3 lists equivalent widths for some of the more prominent atomic lines present in
the spectra of these objects. We list results from measurements of both the LRIS spectra
plotted in figure 1 and of our HIRES data. The latter provide only incomplete wavelength
coverage, but the higher­resolution data allow more accurate measurements of weaker lines.
In particular, the Ca I 6572 š A and 8256 š A absorption lines and Hff emission are barely
detectable in the LRIS spectra, where the measured equivalent widths have a 1oe uncertainty
of \Sigma0:5 š A.
One of the strongest features, either atomic or molecular, in the far­red spectra of L
dwarfs is the K I 7665/7699 š A resonance doublet. Those lines have individual equivalent

-- 8 --
widths of 10 to 12 š A at spectral type L0, but increase dramatically in strength with
decreasing temperature to the extent that the lines effectively merge at ¸L5, where the
composite feature has a width exceeding 100 š A. Similarly, the Rb I and Cs I lines show
distinctly non­linear behaviour, increasing substantially in strength between the L4 dwarf
2M0036 and 2M1507 (L5).
In paper I we proposed that this behaviour is another consequence of dust formation.
As discussed further in section 3.5, dust initially contributes a scattering layer at late­type
M dwarfs, but in lower temperature atmospheres (later spectral types) the dust particles
either `rain out' to greater depths (below the photosphere) or form larger particles, in either
case reducing scattering at optical wavelengths. The overall atmospheric transparency is
further increased as metals are transformed to solid phase, both by the removal of TiO
and VO molecular absorption, and through the scarcity of free electrons and the resulting
reduced level of H \Gamma continuum opacity.
The Ü = 1 photosphere lies at a large physical depth within the low opacity L­dwarf
atmosphere, with the result that the column density of (relatively) undepleted elements,
such as the alkali metals, can reach very substantial values. In addition, gas pressure
increases with increasing depth leading to substantial van der Waal's broadening, as in
degenerate white dwarfs. Both effects lead to strong atomic lines. As discussed in paper
I, the relative strengths of the resonance lines of those species visible in the far red (K,
Cs, Rb) are consistent with their relative abundances in the Sun. (The Ca I 6572 š A line
and the Na I 8183/8194 doublet are higher­order transitions.) Sodium is not expected to
form grains until temperatures of less than 1200K (the T­dwarf r'egime) and, with a higher
abundance than potassium ([Na] = 6.31 as compared to [K]=5.13 for [H]=12.0, where [m]
is the logarithmic abundance), the D lines at 5890, 5896 š A are predicted to grow in strength
at earlier spectral types than the K I doublet.
This prediction is confirmed by the spectra plotted in figures 2 and 3. The sodium lines,
which already have the substantial equivalent width of ¸ 36 š A in the M9.5 BRI0021 have
doubled in strength to ¸ 80 š A by spectral type L0.5 (2M0746). We measure an equivalent
width of ¸ 170 š A in the L2 dwarf Kelu 1, and our spectrum of 2M0036 yields an equivalent
width of ¸ 240 š A for that L3.5 dwarf, although identifying appropriate pseudo­continuum
points is becoming problematic at these later spectral types.
Initial observations of 2M1507 with the Palomar double spectrograph revealed a
steeply declining spectrum shortward of 6700 š A, with no significant flux detected shortward
of ¸ 6000 š A. Our surmise that this might reflect increased sodium absorption is confirmed
spectacularly by the LRIS data plotted in figure 2. Superimposed on the steeply­rising
underlying spectrum, the D lines produce a smooth, concave feature spanning over 1500 š A,

-- 9 --
with MgH the only identifiable absorption feature between 4500 and 6500 š A. The blue wing
of this atomic doublet extends to ¸ 5000 š A, where the spectral energy distribution reaches
a mild peak before declining towards shorter wavelengths. The red wing of the Ca I 4227 š A
resonance line probably contributes to that smooth decline. Similar behaviour in the K
I 7665/7699 doublet at much cooler temperatures is partly responsible for the steep flux
gradient between 8000 and 9000 š A in the energy distribution of methane­rich T dwarfs such
as Gl 229B (Oppenheimer et al, 1998).
3.3. Chromospheric activity and lithium absorption
Hff emission has long been known as an indicator of chromospheric activity amongst M
dwarfs, and earlier studies suggested that emission became increasingly common amongst
later spectral types. Gizis et al (1999), however, have re­examined the distribution of
chromospheric activity as a function of spectral types, using 2MASS observations to
define a photometrically­selected sample of M dwarfs which includes a significantly larger
number of ultracool (?M7) objects than was previously available. Analysis of that sampel
shows that the frequency of Hff emission peaks at close to 100% at spectral type M7 and
declines thereafter. Only 45% of known early­type (ŸL3) L dwarfs have emission lines with
equivalent widths exceeding ¸ 2 š A while none of the later­type L dwarfs in paper I have
detectable emission, despite the low continuum flux in the latter objects.
The four sources considered here show behaviour similar to the dwarfs in the paper
I sample. Both of the earlier­type dwarfs have weak Hff emission, while no emission is
detectable in the two later­type dwarfs. Figure 5 plots our HIRES data for the Hff region
of the spectrum in three objects ­ the 2M1507 data are of low signal to noise at these
wavelengths and essentially featureless. Both of the Hff profiles, but oparticularly 2M1439,
appear to have a narrow core centred on a broader pedestal. This is morphology is also
found in approximately 10% of the ultracool M dwarfs.
None of these dwarfs is particularly active. We can use our flux­calibrated LRIS
spectra to determine emission line fluxes from our measured equivalent widths. In the
case of 2M0746, we derive F – ¸ 2:6 \Theta 10 \Gamma16 erg cm \Gamma2 sec \Gamma1 , while for 2M1439 we find
F – ¸ 7:8 \Theta 10 \Gamma17 erg cm \Gamma2 sec \Gamma1 . These correspond to activity ratios, L ff /L bol , of 10 \Gamma5:5 and
10 \Gamma5:4 respectively, values which are almost two orders of magnitude lower than the typical
level of activity amongst M dwarfs, h L ff
L bol
i ¸ 10 \Gamma3:8 (Hawley et al, 1996; Gizis et al, 1999)
and an order of magnitude below the quiescent state of the ultracool M9.5e dwarf, 2MASSW
J0149090+295613 h L ff
L bol
i ¸ 10 \Gamma4:6 (Liebert et al, 1999). The upper limits corresponding to
non­detection imply even lower activity ratios for the two later­type dwarfs.

-- 10 --
Our HIRES observations also allow us to set limits on the equivalent width of the
Li I 6708 š A absorption line in these four dwarfs. The presence of atmospheric lithium in
late­type dwarfs is now well­recognised as an indicator of substellar mass (Rebolo et al,
1992; Magazzu et al, 1993). Recent models indicate that all dwarfs with masses exceeding
0.06M fi should have depleted lithium by the time that their surface temperature has fallen
to ¸ 2400K, equivalent to spectral type M7 (Baraffe et al, 1998). Based on the scale
derived in paper I, we estimate temperatures between ¸ 2100K (2M0746) and ¸1700K
(2M1507) for the four dwarfs considered here. None has lithium absorption exceeding
200m š A. Approximately one in four of the 80 L dwarfs identified to date from 2MASS data
have detectable lithium absorption, with equivalent widths rising to ¸ 20 š A amongst the
later spectral types (Kirkpatrick et al, in prep). Thus, the absence of detectable lithium in
these dwarfs implies that all of these dwarfs have depleted their primordial store of lithium:
that is, all four have masses which exceed 0.06M fi .
That these four L dwarfs have masses relatviely close to the hydrogen­burning limit
is not surprising. Figure 19 in paper I shows the predicted time evolution of temperature
for models spanning the mass range 0.01 to 0.1 M fi . Both the calculations by Baraffe et
al (1998) and Burrows et al (1997) predict that objects with masses as high as ¸ 0:08M fi
(i.e. very low mass stars) can acheive temperatures of 2100K, the value we associate with
spectral class L0. Similarly, the upper mass limit at T eff ¸ 1700K (L5) is 0.07 to 0.075
M fi . In both cases, higher mass objects spend several Gyrs at those temperatures, while
brown dwarfs with masses below ¸ 0:06M fi have total residence times of no more than
¸ 10 8 years. Those circumstances lead to much higher probabilities of detecting high­mass
brown dwarfs and very low mass stars at early and mid­L spectral types. Lower­mass brown
dwarfs make a larger contribution to samples of late L dwarfs.
3.4. Kinematics
Table 2 shows that all four L dwarfs have substantial velocities relative to the Sun.
The correlation between space motion is statistical rather than direct, but since velocity
dispersion increases with age, there is less ambiguity in interpreting a high velocity as
implying a relatively old age than in taking a low velocity as implying youth. Representative
tracers of the `young disk' population (A stars, active late­type dwarfs, Cepheids) indicate
that a 1­Gyr old population can be modelled as a Schwarzschild ellipsoid with [U = ­10,
V=­10, W=­7; oe U = 38, oe V = 26, oe W = 21 kms \Gamma1 ] (Soderblom, 1990). The average space
velocity for those kinematics is V tot = 34kms \Gamma1 , and even the lowest velocity L dwarf in the
current sample, 2M1507, lies at the 75th percentile of the predicted velocity distribution,

-- 11 --
albeit within 1oe of the mean.
The measured velocities are more characteristic of an older stellar population. Hawley
et al (1996) derive the following ellipsoid from observations of nearby M dwarfs: [U = ­10,
V = ­21 , W = ­8 ; oe U = 38, oe V = 26, oe W = 21kms \Gamma1 ]. Matched against that distribution,
2M0036, 2M0746, 2M1439 and 2M1507 fall at the 48th, 68th, 93rd and 39th percentiles. It
therefore seems unlikely that these dwarfs are younger than ¸ 1 Gyr, further corroborating
their identification as high­mass brown dwarfs or low­mass stars.
3.5. Colour­magnitude diagrams
Our spectrophotometry provides the first opportunity of examining the BV colours of L
dwarfs. It is notable that the (B­V) colour inferred from the spectrophotometry for 2M0036
is ¸ 0:5 magnitudes bluer than that for the L0 dwarf, 2M0746. This counter­intuitive
blueward evolution with decreasing temperature can be ascribed in large part to the
increasing strength of the Na D lines. A complementary effect can be expected in the (V­I)
colour.
Figure 6 plots the (M V , (V­I)) colour magnitude diagram for nearby stars with accurate
parallaxes and reliable photometry (Bessell, 1990; Leggett, 1992) supplemented by our own
own data for the four bright L dwarfs discussed in this paper. 2M1507 has a formal visual
absolute magnitude of M V ¸ 22:9. Among M dwarfs, (V­I) reaches a local maximum at
spectral type úM7: VB8 (M7) has(V­I)=4.56 mag, while VB10 is only slightly redder at
(V­I)¸ 4:7 mag. Later­type M dwarfs, such as LHS 2924 (M9, (V­I)¸ 4:37), have lower
luminosities, but bluer (V­I) colours (Monet et al, 1992). Our new data show that the
march redward resumes amongst the L dwarfs, with the growing strength of the sodium
D lines contributing to the decreased flux in the V band, notably the nearly 1 magnitude
offset in (V­I) between 2M0036 (L4) and 2M1507 (L5).
The cause of the reversal in (V­I) colour amongst the later­type M dwarfs has received
little discussion in the literature. Spectroscopy shows no evidence for increased molecular
absorption in the far red, which might decrease the emergent flux in the I band. Indeed,
the strongest molecular absorber, TiO, peaks between ¸M6 and M8 (the fl 7050 š A band
is strongest at M6.5) and decreases in strength in dwarfs of later spectral type, while
other species, such as VO, have less extensive absorption bands. These same stars show
a near­monotonic trend toward redder colours with decreasing luminosity in optical to
infrared colours, such as (I­J) (figure 4). This suggests that the colour reversal in (V­I)

-- 12 --
stems primarily from increased flux in the V band rather than a deficit at I­band 1 . The
(M J , (I­J)) diagram beautifully illustrates the 'step' in the main sequence at spectral type
úM4, originally highlighted by Reid & Gizis (1997) and probably due to the onset of
convection. The L0 dwarf 2M0746 lies ¸ 0:7 magnitudes above the `main sequence' in this
plane, raising the possibility that it is an equal­mass binary.
We suggest that the behaviour the (V­I) colour is driven by the formation of dust in
the upper atmospheric layers of mid­type M dwarfs, and by the subsequent evolution of
the particle size and/or spatial distribution at lower effective temperatures. Tsuji et al
(1996) originally demonstrated that dust formation has an important effect on the emergent
spectral energy distribution of cool dwarfs, notably a reduction in the strength of the
near­infrared H 2
O bands due to atmospheric heating through dust re­radiation. Allowing
for the latter effect reconciles a long­standing discrepancy between theoretical models and
observations of late­type M dwarfs (cf Reid & Gilmore, 1984). Our hypothesis is that the
colour reversal in (V­I) has the same origin.
Tsuji et al (1996) place the onset of dust formation at T eff ¸ 2600K. Leggett et al
(1996) estimate T eff ¸ 2700K for the M6.5 dwarf GJ 1111, suggesting that dust should
become evident at spectral types of úM7 and later. Supporting evidence for dust formation
at this spectral type comes from variations in the equivalent width of the 7665/7699 KI
doublet in mid­ the late­M dwarfs. Figure 7 plots HIRES data covering this region of the
spectrum for eight dwarfs with spectral types between M3 and L4. While the detailed
profile of the shorter wavelength component is obscured partially by terrestial O 2
absorption
(the A band), it is clear that the overall variation mimics that of the (V­I) colour. The
equivalent widths rise to a maximum at spectral type M6.5/M7, declines noticeably in
strength to spectral type M9.5/L0, before increasing dramatically throughout the L dwarf
sequence, as discussed above and in paper I.
We explain this behaviour as a combination of two effects. First, at spectral types
M7­M9.5, dust is present in the atmosphere in sufficient quantities to act as a scattering
layer, raising the atmospheric opacity and hence reducing the physical depth (and hence
both gas pressure and column density) of the Ü = 1 layer for line formation; second, dust
re­radiation not only reduces the strength of the H 2 O bands, but also increases the flux
emitted at visual wavelengths, resulting in bluer (V­I) colours. Dust formation may also
reduce the overall molecular (mainly TiO) opacity to a greater extent at visual wavelengths
1 Note, however, that the growth in strength of the K I 7665/7699 š A resonance lines amongst the later L
dwarfs is likely to result in an effect on M I analagous to the effect of the D lines on M V between spectral types
L4 and L5. Gl 229B is almost 1 magnitude redder in (I­J) than the L8 dwarf 2MASSWJ1632291+190441.

-- 13 --
than at 0.8¯m. In late M dwarfs, such as LHS 2924, the total flux emitted at visual
wavelengths amounts to less than 0.1% of the bolometric flux, so a small flux redistribution
can have a large effect on F V . Section 3.2 summarises the likely explanations for the
increased equivalent widths in all of the alkali lines at spectral types beyond L0: increased
particle size or rain out. More detailed spectrophotometry of mid­ to late­type M dwarfs at
blue and visual wavelengths can test the overall validity of this hypothesis.
4. Summary and Conclusions
We have presented spectroscopic and photometric data for four bright L dwarfs lying
at distances of less than 15 parsecs from the Sun. Our observations permit the first detailed
examination of the properties of these objects at blue and visual wavelengths, revealing
the presence of MgH and CaOH molecular absorption. In addition, the sodium D lines
are extremely strong, reaching equivalent widths in excess of 240 š A in later­type L dwarfs.
This behaviour likely stems from the low atmospheric opacity in the latter objects and the
consequent substantial pressure broadening. The growth in strength of the Na D lines is
also responsible for the (V­I) colour becoming significantly redder between spectral types
L4 and L5. The KI 7665/7699 doublet probably has a similar effect on the I­band flux
between spectral types L8+ and T.
Dust formation is clearly an important factor governing spectral evolution at these low
temperatures. Theoretical models suggest that dust first forms, primarily as TiO­based
agglomerates, at ¸ 2600K, a prediction which is supported by the behaviour of the KI
lines at 7665/7699 š A at spectral types between M3 and L0. Indeed, we suggest that the
reversal in the (M V , (V­I)) relation at spectral type úM7 may be a consequence of both
lower molecular opacities and dust re­radiation heating the atmosphere, with a consequent
increase in the flux emitted at visual wavelengths.
None of the four L dwarfs considered here has detectable lithium absorption, indicating
masses of at least 0.06M fi . All, however, are also chromospherically inactive, implying
masses close to, if not below, the hydrogen­burning limit, and the relatively high space
motions suggest ages of ¸ 1 Gyr or more. Taken together, these indicators suggest masses
of from 0.07 to 0.09M fi . Further detailed observations of these and other bright L dwarfs
will prove important in determining the general physical characteristics of these objects.
The authors would like to thank Pat Boeshaar for illuminating discussion on the CaOH
molecule and Sandy Legget for providing a copy of the blue spectrum of Kelu 1. We would
also like to thank the staff of the Keck Observatories for their skilled and enthusiastic

-- 14 --
support in acquiring the observations for this project.
JDK, INR and JL acknowledge funding through a NASA/JPL grant to 2MASS Core
Project science. AJB and RJW acknowledge support from this grant.
Much of the V,I photometry reported in Sec. 2.2 was obtained by H. Harris as part of the
USNO parallax efforts and we thank him for allowing the use of it here. The astrometric
data reported in Sec. 2.3 were aquired by a team of observers which includes B. Canzian,
H. Guetter, S. Levine, C. Luginbuhl, A. Monet, R. Stone, and R. Walker. We thank them
for their contributions.
This publication makes use of data from the 2­Micron All­Sky Survey, which is a joint
project of the University of Massachusetts and the Infrared Processing and Analysis Center,
funded by the National Aeronautics and Space Administration and the National Science
Foundation.
The Keck Observatory is operated by the Californian Association for Research in Astronomy,
and was made possible by generous grants from the Keck W. M. Foundation.
This work is based partly on photographic plates obtained at the Palomar Observatory
48­inch Oschin Telescope for the Second Palomar Observatory Sky Survey which was
funded by the Eastman Kodak Company, the National Geographic Society, the Samuel
Oschin Foundation, the Alfred Sloan Foundation, the National Science Foundation
grants AST84­08225, AST87­19465, AST90­23115 and AST93­18984, and the National
Aeronautics and Space Administration grants NGL 05002140 and NAGW 1710. JDK and
AJB acknowledge the support of the Jet Propulsion Laboratory, California Institute of
Technology, which is operated under contract with the National Aeronautics and Space
Administration.
REFERENCES
Baraffe, I., Chabrier, G., Allard, F., Hauschildt, P.H. 1998, A&A, 337, 403
Bessell, M.S. 1979, PASP, 91, 589
Bessell, M.S. 1990, A&AS, 83, 357
Boeshaar, P. C. 1976, Ph. D. thesis, Ohio State Univ.
Burrows, A., Marley, M., Hubbard, W.B., Lunine, J.I., Guillot, T., Saumon, D., Freedman,
R., Sudarsky, D., Sharp, C. 1997, ApJ, 491, 856
Burrows, A., Sharp, C. 1999, ApJ, 512, 843
Cottrell, P.L., 1978, ApJ, 228, 544

-- 15 --
Delfosse, X., Tinney, C.G., Forveille, T., Epchstein, N., Bertin, E., Borsenberger, J., Copet,
E., de Batz, B., Fouque, P., Kimeswenger, S., Le Bertre, T., Lacombe, F., Rouan,
D., Tiphene, D. 1997, A&A, 327, L25
Epchtein, N., De Batz, B., Copet, E. et al 1994, in Science with Astronomical Near­infrared
Sky Surveys, ed. N. Epchtein, A. Omont, B. Burton, P. Persei, (Kluwer, Dordrecht),
p. 3
Fegley, B., Lodders, K. 1996, ApJ, 472, L37
Gizis, J.E. 1997, AJ, 113, 806
Gizis, J.E., Monet, D.G., Reid, I.N., Williams, R. 1999, AJ, submitted
Hamuy, M., Suntzeff, N.B., Heathcote, S.R., Walker, A.R., Gigoux, P., Phillips, M.M. 1994,
PASP, 106, 566
Hawley, S.L., Gizis, J.E., Reid, I.N. 1996, AJ, 112, 2799
Kirkpatrick, J.D., Reid, I.N., Liebert, J., Cutri, R.M., Nelson, B., Beichmann, C.A., Dahn,
C.C., Monet, D.G., Gizis, J., Skrutskie, M.F. 1999a, ApJ, 519, 802 (paper I)
Kirkpatrick, J.D., Reid, I.N., Gizis, J.E., Burgasser, A.J., Liebert, J., Monet, D.G., Dahn,
C.C., Nelson, B. 1999, ApJ, submitted
Leggett, S.K. 1992, ApJS, 82, 351
Leggett, S.K., Allard, F., Berriman, G., Dahn, C.C., Hauschildt, P. 1996, ApJS, 104, 117
Leggett, S.K., Toomey, D.W., Geballe, T.R., Brown, R.H. 1999, ApJ, 517, L139
Liebert, J., Kirkpatrick, J.D., Reid, I.N., Fisher, M.D 1999, ApJ, 519, 345
Lodders, K. 1999, ApJ, 519, 793
Magazzu, A., Martin, E.L., Rebolo, R. 1993, ApJ, 404, L17
Marcy, G.W., Benitz, K.J. 1989, ApJ, 344, 441
Monet, D.G., Dahn, C.C., Vrba, F.J., Harris, H.C., Pier, J.R., Luginbuhl, C.B., Ables,
H.D., 1992, AJ, 103, 638
Oke, J. B., Cohen, J. G., Carr, M., Cromer, J., Dingizian, A., Harris, F. H., Labreque, S.,
Lucinio, R., Schaal, W., Epps, H., Miller, J. 1995, PASP, 107, 375

-- 16 --
Oppenheimer, B.R., Kulkarni, S.R., Matthews, K., van Kerkwijk, M.H. 1998, ApJ, 502, 932
Pearse, R.W.B., Gaydon, A.G. 1965, The Identification of Molecular Spectra, (Chapman &
Hall, London)
Pesch, P. 1972, ApJ, 174, L155
Rebolo, R., Mart ' in, E.L., Magazzu, A. 1992, ApJ, 389, L83
Reid, I.N., Gilmore G.F. 1984, MNRAS, 206, 19
Reid, I.N., Gizis, J.E. 1997, AJ, 113, 2246
Ruiz, M.T., Leggett, S.K., Allard, F. 1997, ApJ, 491, L107
Skrutskie, M.F. et al 1997, in The Impact of Large­Scale Near­IR Sky Survey, ed. F. Garzon
et al (Kluwer: Dordrecht), p. 187
Soderblom, D.R. 1990, AJ, 100, 204
Tinney, C.G., Mould, J.R., Reid, I.N. 1993, AJ, 105, 1045
Tinney, C.G. 1996, MNRAS, 281, 644
Tsuji, T., Ohnaka, K., Aoki, W. 1996, A&A, 305, L1
Vogt et al, 1994, S.P.I.E., 2198, 362
This preprint was prepared with the AAS L A T E X macros v4.0.

-- 17 --
Table 1
L dwarf photometry
Name Sp. (B­V) SP (V­R) SP V I C J H K
2M0036 L3.5 1.7\Sigma0.2 3.0\Sigma0.1 21.33\Sigma:06 16.10\Sigma:02 12.44 11.58 11.03
2M0746 L0.5 2.1\Sigma0:2 2.3\Sigma0:1 19.87\Sigma:06 15.11\Sigma:02 11.74 11.00 10.49
2M1439 L1 21.04\Sigma:02 16.12\Sigma:02 12.76 12.05 11.58
2M1507 L5 22.9\Sigma:5 16.65\Sigma:02 12.82 11.90 11.30

-- 18 --
Table 2
L dwarf astrometry and kinematics
Parameter 2M0036 2M0746 2M1439 2M1507
¯ arcsec yr \Gamma1 0.833\Sigma0:073 0.464\Sigma0:091 1.2943\Sigma0:0012 0.99\Sigma:05
` degrees 80\Sigma15 250\Sigma2 288.3\Sigma0:1 174\Sigma5
ú milliarcsec 92.2\Sigma16.3 69.4\Sigma16 69.5\Sigma0:6 117.5\Sigma25:2
V rad kms \Gamma1 21.7\Sigma3:0 56.6\Sigma2:0 ­23.9\Sigma2:0 ­36.4\Sigma3:0
MK 10.86\Sigma0:35 9.70\Sigma0:45 10.70 \Sigma0:02 11.65\Sigma0:42
M bol 14.00 12.70 13.65 14.90
U kms \Gamma1 ­46.3 ­60.6 ­80.7 ­15.8
V kms \Gamma1 ­4.1 ­21.5 ­39.1 ­17.5
W kms \Gamma1 ­11.9 ­8.8 17.9 ­48.6
V tot kms \Gamma1 48.0\Sigma8:5 64.9\Sigma13:9 91.4\Sigma2:5 54.0\Sigma9:5

-- 19 --
Table 3
Equivalent widths for atomic lines
2M0746 2M1439 2M0036 2M1507
L0.5 L1 L3.5 L5
LRIS
K I 7665 9.6\Sigma0:5 š A 19\Sigma0:5 š A !85š A ?150 š A
K I 7699 8.8 9.7
Rb I 7800 2.0 3.0 3.3 6.7
Rb I 7947 3.1 3.0 4.0 7.6
Na I 8183/94 9.1 9.9 7.5 5.1
Cs I 8521 2.7 2.1 2.5 4.6
Cs I 8943 2.2 1.7: 3.6 3.5
HIRES
Hff 1.38\Sigma0:05 1.13\Sigma0:05
Ca I 6572 0.69 0.55 0.86
Li I 6708 !0.2 ! 0:05 !0.1 ! 0:1
Rb I 7800 4.32
Rb I 7947 2.0 3.59 4.3
Na I 8183 1.58 2.04 1.39
Na I 8194 3.26 3.76 1.82
8256 0.57 0.57
Cs I 8521 1.14 2.15 3.44

-- 20 --
FIGURE CAPTIONS
Fig. 1.--- Far­red optical spectra of the four bright L dwarfs discussed in this paper.
Fig. 2.--- Blue/visual LRIS spectra of the 4000 to 7800 š A region in late­type dwarfs. In
addition to three of the L dwrafs discussed is this paper, we plot spectra for the M9.5 dwarf
BRI0021­0214 and for 2MASSWJ150654.4+131206, a bright early­type L dwarf. The more
prominent features are identified.
Fig. 3.--- Expanded reproductions of the 4500 to 6600 š A region of the spectrum for the
late­type dwarfs plotted in figure 2, highlighting the strong MgH bands and the substantial
increase in strength of the Na D lines.
Fig. 4.--- Blue­green spectra of three extreme subdwarfs.
Fig. 5.--- The Hff region in 2M0746, 2M1439 and 2M0036 from our HIRES observations.
The Ca I 6572 absorption is also evident in these spectra, as are TiO bands in the earlier­type
dwarfs.
Fig. 6.--- The (M V ), (V­I)) and (M J , (I­J)) diagrams defined by nearby stars with
accurate parallax measurements. Crosses mark objects with Hipparcos astrometry; open
circles are stars in the 8­parsec sample (Reid & Gizis, 1997) or with ground­based parallax
measurements by Monet et al (1992) or Tinney (1996). Note the reversal in (V­I) colour at
M V ? 18. The four L dwarfs discussed in the present paper are plotted as solid points.
Fig. 7.--- High­resolution spectra of the K I 7665/7699 doublet in dwarfs with spectral
types between M3 and L4. VB 8 and, to a lesser extent, VB 10 both exhibit chromospheric
reversals in the core of both lines.

-- 21 --
7000
8000
9000
10000
0
.2
.4
.6
.8
1
2M0746
L0.5
2M1439
L1
2M0036
L3.5
2M1507
L5
Wavelength
Fig. 1.---

-- 22 --
4000
5000
6000
7000
0
.2
.4
.6
.8
1
BRI0021
M9.5
2M0746
L0.5
2M0036
L3.5
Kelu
1
L2
2M1507
L5
K
I
CaH
CaH
CaH
CaH
CaOH
VO
CaOH
Na
D
Na
D
Na
D
Na
D
Na
D
Wavelength
Fig. 2.---

-- 23 --
4500
5000
5500
6000
6500
0
.2
.4
.6
.8
1
Wavelength
BRI0021
2M0746
Kelu
1
2M0036
2M1507
TiO
TiO
TiO
MgH
MgH
MgH
CaOH CaOH CaOH
CaH
CaH
Na
D
TiO/CaOH
Na
D
VO
Fig. 3.---

-- 24 --
4000
5000
6000
7000
0
.2
.4
.6
.8
1
Wavelength
(Angstroms)
LHS
489
esdM0
LHS
453
esdM3.5
LHS
375
esdM5
MgH
MgH
Na
D
CaH
CaH
CaI
Fig. 4.---

-- 25 --
6540 6550 6560 6570 6580 6590 6600
0
.2
.4
.6
.8
1
Wavelength
2M0746
2M0036
2M1439
Fig. 5.---

-- 26 --
0 2 4 6
20
15
10
5
0
(V­I)
0 1 2 3 4
14
12
10
8
6
4
2
0
(I­J)
Fig. 6.---

-- 27 --
7660
7680
7700
7720
0
.2
.4
.6
.8
1
Wavelength
(Angstroms)
relative
flux
K
I
7665
/
7699
Gl
643:
M3.5
Gl
83.1:M4.5
LHS
523:M6.5
VB
8:M7
VB
10:M8
LHS
2924:
M9
2MJ0345:
L0
2M0036:
L4
Fig. 7.---