We have detected 300 out of 500 sources in either
SiO J=1-0, v=1 or 2 transitions. The detection rate is
above 50 %. This value is comparable with the detection rate in the
previous bulge SiO maser
survey with similar color-selection criteria ([Jiang et al. 1995]).
Since the data reduction has been finished for the sources in the
galactic longitude range between 15 and 25
, we will discuss
the kinematic features only in this range in the present paper.
In order to obtain the bolometric correction at the near infrared bands,
the interstellar extinction is required to be corrected for the
flux densities. Since the
interstellar extinction depends on the distance,
the distance and the bolometric correction must be solved
simultaneously. In addition, we must specify the luminosity of the
source. For simplicity, we adopt 
in this
paper; this corresponds to the luminosity of a star of mass 
near to the tip of the AGB in the models of Vassiliadis and
Wood (1993). With these assumptions, the distances to each source
were calculated.
The objects which we have observed are in a relatively narrow color range
of 
; it is expected that the true luminosities of these
objects do not deviate substantially from the average value of 
. Therefore, the luminosity distances we have
obtained from this method are not expected to suffer from large
errors as a result of an incorrect luminosity assignment. However, there
is still some uncertainty originating from the interstellar
extinction correction due to the patchy structure of the dust
distribution in the Galaxy.
Figure 1 shows a plot of the radial velocities corrected for the solar
motion, 
km s
sin
,
against the
distances. Here, the velocities were obtained from SiO maser
observations. The broken (
) and dot-dash (
)
curves show the radial velocities expected
from the galactic rotation curve [which is approximated by equation
(B2) of Izumiura et al. 1998]. The open circles indicate the
identified sources with a well-determined distance from the near-infrared
photometry, and the filled circles indicate the sources
without near-infrared identification (distances determined only from
and 
). Around 
kpc, several sources exhibit a
large deviation of radial velocities from the curve expected from the
galactic rotation (8 sources in the ellipses in figure 8;
also they are noted by ``y'' in the last column in table 3).
It is curious that these sources are concentrated in the region with
kpc.
The average color 
of deviant group
of stars at around 
5.5 kpc
is 
) mag (the number after 
is a standard
deviation), which is not very different from the average color,
) mag, for the rest of the stars. The average
extinction, 
, for the deviant group of stars
is close to the extinction, 
, for the rest.
These stars are not special in terms of their colors.
We also notice that there is a source-deficient region at
around 
(8 kpc, 0 km s
).
The location corresponds to the relatively empty region
of sources around the peak of the rotation curve in figure 8.
This could be
interpreted as being a vacant region behind the near end of the bar.
Of course, the presence of the empty region
is somewhat dubious; these may be caused by
errors in the distance estimation.
It may occur due to the presence
of dark clouds in a spiral arm, which causes more extinction
than that given under a uniform reddening correction.
Inferred positions of the groups and the hole
are plotte2 in Figure 2 as a face-on view of the galaxy.
In Figure 2, the major axis of the bar is drawn in the direction
by 30 from the Sun -- galactic-center line. In fact,
the direction of the major axis of the galactic bar has been
quite controversial.
Nikolaev and Weinberg (1997) gave the angle, 30
2
from the analysis of the color-selected IRAS variable sources
in the disk (|b| < 3), which
probably have a large overlap of sources
in the present sample. The most recent analysis of the COBE
data of the J, H, K, L, and M bands ([Freudenreich 1998]) derived
the semi-major axis and the orientation of the bar
as 2.5 kpc and 25, respectively.
The analysis of the red clump
stars in the bulge ([]) gave a slightly smaller angle,
19---24.
fig7
Weinberg (1994)
investigated the kinematics of stars near a resonance
and concluded that a large increase in the velocity dispersion
(
80 )
with little change in net radial motion could be observable
in the vicinity of Outer Lindblad Resonance (
5 kpc).
In the present observational data described in the previous
subsection, no such large increase is found except the
group of stars, D3. However, if the D3 stars are created
by the large dispersion at the
Outer Lindblad Resonance (OLR) with the velocity centroid
(due to Galactic rotation) of about 50 , the dispersion must be
approximately 80 , which is somewhat larger than the dispersion
of 60 from the model. In addition,
it is difficult to interpret the presence of the hole , H1,
which appears in the v-l diagram.
The D1 stars also exhibit a large deviation
from the galactic rotational motion.
An interpretation on this deviant group of stars is
that they belong to the bulge. An appreciable number of stars
in the bulge must contaminate the sample at l<20 .
The large velocity dispersion of the
bulge SiO masers stars (
82 ; [Izumiura et al. 1995b]) and the
increase of the dispersion with decreasing |b|
for the bulge stars seem to fit to
the above interpretation. However, since the deviant
group of stars seems to appear in a wide range of the galactic longitude
(seen from the OH 1612 MHz data), more refined analysis of the data
is necessary for such interpretation.
More realistic 3D N-body calculations of the barred model of
the Galaxy were made recently ([Fux 1997]).
Figure 16a in the Fux's
paper gave the longitudinal dependence of the average radial
velocity and velocity dispersion and compared them
with the data of OH/IR stars. The mean radial velocity
of about 67.2 in 15<l<25
in the present observation seems to be consistent with the value
for the disk stars (|b|<5) in the Fux's model
at l=20 [after applying the correction by -220
for the motion of the local standard of rest to figure 16a].
The velocity dispersion
in 15<l<25 in the present observation
(the deviation from the linear fit with l) is 52.9 ,
which seems to fit well to the value of about 55 at l=20
in the Fux's model. As far as average quantities are concerned,
the 3D N-body simulation seems to explain the present radial
velocity data well.
However, effects on the velocity field by the bar-potential are much harder to detect in the stellar component than in the gas ([Vauterin & Dejonghe 1998]), and more refined method for comparing discrete kinematic data with N-body simulations must be required ([Saha 1998]) for further investigations.