The velocities and distributions of stars perpendicular to the Galactic plane reflect the amount of mass in the local disk of the Milky Way. Velocities can be used to obtain the stellar velocity dispersion perpendicular to the plane < v2z >, whereas distances provide the density distribution perpendicular to the plane, (z). The gravitational potential due to the disk, (z), is related to these quantities through the Jeans equation:
so that measurements of (z) and vz provide an estimate of the
density of the disk in the solar neighborhood.
Oort (1932)
pioneered attempts to measure the disk density
and found that the local disk
mass exceeded the value which could be accounted for in visible material
(Oort 1960).
The problem was later tackled by Bahcall
(1984a,
b,
c)
who used star counts of both F dwarfs
and K giants, as well as a range of Galactic mass models, to investigate the
presence of DM in the solar neighborhood. He numerically solved the combined
Poisson and Boltzmann equations, given by
and
respectively. Here disk(z) is the total disk density,
haloeff the density contribution due
to the halo, and i(z) is the density of the ith stellar
component with corresponding
velocity dispersion < vz2 >i.
Bahcall (1984b)
assumed that
the disk was made up of a finite number of isothermal components to obtain
his results. His findings supported the
earlier work of Oort and suggested that the density of unseen material in the
Milky Way disk was at least 50% that of the observed material.
From rotation curve constraints, the DM was found
to be distributed in a disk and to have an exponential scale-height
less than 0.7 kpc. Thus at the time of
Trimble's (1987)
review, the evidence
pointed towards a significant component of DM in the Milky Way disk. However,
more recent developments have altered this view.
Bienaymé, Robin,
and Crézé (1987)
adopted a different approach to
estimating the mass density in the Galactic plane. They constructed Galactic
models, including the central bulge, visible disk, dark halo and disk DM,
and incorporated a model of the disk stellar populations. The relative
contributions of the various terms were constrained by the requirement that the
resulting potential should be consistent with observations of the Milky Way
rotation curve. Acceptable models had to provide self-consistent solutions to
the Boltzmann and Poisson equations.
The stellar density laws were translated into star count predictions.
In contrast to earlier studies,
Bienaymé et
al. (1987)
found little evidence for disk DM.
Their technique suggested a dynamical mass density of 0.09 to 0.12
M
pc-3, limiting the local disk DM to a density less than 0.03
M
pc-3, corresponding to a surface density in DM of 24
M
pc-2. Their best fit model had a DM density of just 0.01
M
pc-3, although no DM was also an acceptable solution. Moreover,
Bienaymé et
al. (1987)
found that a moderate flattening of the dark halo
would remove the need for even this small amount of disk DM.
In a subsequent paper,
Crézé,
Robin, and Bienaymé (1989)
addressed the
question of why their technique gave different results from those of
Oort (1960)
and Bahcall (1984b).
Their study essentially reversed the Bahcall models to
calculate predicted star counts, which were then compared to
observed star counts.
Crézé et
al. (1989)
concluded that most Bahcall
models, both with and without disk DM, were consistent with the observed
star counts, and that the extant data simply did not provide strong constraints
on the local mass density. In other words,
Crézé et
al. (1989)
believed that the errors
associated with earlier mass determinations were much larger than
previously claimed.
Thus there was no conflict between the low or zero
disk DM densities inferred by
Bienaymé et
al. (1987)
and the much higher DM fractions found by Bahcall.
Around the same time,
Kuijken and Gilmore
(1989a)
introduced a new technique to estimate the surface
mass density of the local disk. Their method also used a tracer population of
stars. However, unlike the methods of Bahcall and Oort, it employed
the full observed distribution
function of velocities and distances of the stellar tracers, rather than
assuming that the stellar population could be represented by a finite number of
isothermal distributions. This distribution function, along with the
spatial density distribution of the same tracer population, was then used to
determine the local surface mass density.
This method was applied by
Kuijken and Gilmore
(1989b)
to a sample of K dwarfs
in the direction of the south galactic pole. They obtained the surface
mass density of visible material by integrating
the stellar population near the Sun
through the derived gravitational force perpendicular to the disk, and
adding to this the interstellar gas. By removing
the contribution of the Galactic dark halo,
Kuijken and Gilmore
(1989b)
found a total dynamical surface mass density of the disk in the solar
neighborhood of 46 ± 9
M
pc-2. Their value for
the surface mass density in identified material was 48 ± 8
M pc-2,
implying that there was no disk DM in the solar neighborhood. Moreover,
Kuijken and Gilmore
(1989c)
analyzed the earlier F and K star samples used by
Bahcall and, like
Crézé et
al. (1989),
concluded that the available data were either internally
inconsistent, or that they provided no reliable evidence for disk DM.
Yet another approach was taken by Knapp
(1988;
also DMW), who
used gas rather than stars to trace the Galactic potential.
From an analysis of the relation between velocity dispersion and scale-height
of molecular hydrogen in the Milky Way, she obtained an estimate of the
mid-plane mass density. The inferred disk mass could be accounted for
entirely by known luminous material.
The chief protagonists in this debate then embarked on a discussion in the
literature which centered on the correct method of analyzing the data.
Gould (1990a)
applied a maximum likelihood method to the tracer population of
Kuijken and Gilmore
(1989b)
and found that, even if the same physical
assumptions were retained,
the disk surface mass density was higher than claimed by Kuijken and Gilmore,
with a value of
54 ± 8 M
pc-2. On the basis of this finding,
Gould (1990a)
claimed that the
Kuijken and Gilmore
(1989b)
study did not severely constrain the
amount of disk DM, and that it was consistent with the earlier result of
Bahcall.
Kuijken and Gilmore
(1991)
rediscussed their method and argued that their
technique was more objective than Gould's, still claiming no disk DM within
1.1 kpc of the Galactic plane. In a further study,
Kuijken (1991)
investigated the presence of DM within 160 pc of the plane, again
finding no DM.
The details of this debate are beyond the scope of the present review. In any
case, no consensus seems to have been reached on the ``correct'' method of
analysis. As
Gould (1989)
pointed out, different tracers are sensitive to
different hypothetical DM components, which complicates direct comparison
between results derived from different data sets.
Part of the apparent discrepancy between these results was
probably caused by the systematic
uncertainties which dominated the Bahcall calculation of the local
mass density. This problem was noted by Bahcall in his original
analysis and studied further
by Gould (1990b).
The uncertainties
include distance scale errors, non-isothermality of the stellar sample,
smoothing of the star counts, spurious stellar density gradients, and
other problems arising from the use of data drawn from different
samples. In an attempt to remove some of the confusion,
Gould (1990b)
devised a technique to assign statistical significance to the
local disk DM density derived using the Bahcall method.
He also presented a strategy for obtaining a
suitable sample of stars that could be used with the Bahcall method to
measure accurately the DM surface density.
Bahcall, Flynn, and
Gould (1992)
have adopted
Gould's (1990b)
statistical test and observing strategy.
They have used a survey of K giants carried out by
Flynn and Freeman
(1993)
that was specifically designed to
address this question. This sample is expected to be relatively free from
systematic effects. The principal result of
Bahcall et al. (1992)
is that the hypothesis that there is no disk DM
is consistent with the data at a level of 14%. They conclude that the odds are
6-to-1 in favor of the existence of disk
DM. Assuming that the DM is distributed in the same way as the visible disk
material, the best fit model has 53% more DM than visible matter.
Despite the
Bahcall et al. (1992)
result, the presence of DM in the solar neighborhood seems a less likely
proposition than at the time of
Trimble's (1987)
review. While it can be argued that the
Bahcall et al. (1992)
work is the most comprehensive investigation of disk DM, their result
is somewhat at odds with other independent studies.
It is certainly
possible that there is some DM in the Milky Way disk, but there appears to be
no clear need to invoke such a component.
Thus while this issue is not yet closed, the consensus seems to be that
the burden of proof has fallen on the advocates of disk DM.