)
As mentioned in the introduction, the VLA observations presented by
Bieging et al. (1991),
have a similar velocity resolution, but have a factor 
higher
angular resolution as compared with that of the present data. There is good
qualitative agreement. The distinction of four velocity intervals by
Bieging et al. (1991), with
characteristic morphology of the HI absorption ("curtain'', "floccules'',
"arc and loops'' and "linear filaments'') are not as obvious in our observations;
this effect arises due to the inferior angular resolution here.
The velocity-right ascension diagram (Fig. 3 (click here)) is adequate to show the large scale velocity
behaviour of the absorbing cool HI, such as gradients and "bends'' (as
at 
, v=-35 km s
). Due to the undersampling of the
contours, details are less visible. For these Figs. 5 (click here) with the
collection of spectra are more suitable, e.g. there is a strong feature seen at
, 
(
), v=-43 km s
.
The physical interpretation of the saturated line (
) at
v=-50 km s
is complex; in spite of the saturation, we can derive some limits.
The absorption line could be caused by a large scale feature of high
density and a spin temperature, 
K; this latter value was estimated
as follows: first we fitted Gaussians to the optical depth profiles, thereby
disregarding the residuals of these velocities, where the line is saturated.
We derive a peak optical depth of about 6, with a velocity width (FWHM) of
km s
. From this width we find an upper limit for the kinetic
temperature of 310 K using
The spin temperature 
is probably considerably
lower because single dish observations give a brightness temperature of about 50 K
(Mebold & Hills 1975); therefore 
is about the same for a cloud of
such high optical depth. The integral of the optical depth over the velocity is 25 km s
,
which results in a value of the column density 
. If we assume that the feature has a diameter
of 5 arcmin (the moving "curtain'' across the source) and that it is the
same in the line of sight, the space density is 8 
400 cm
.
Another interpretation would be that the saturated line is the blend of
many small scale features of very cold HI. The rather regular morphology
of the feature at -50 km s
makes the latter explanation less attractive.
)
The velocity-right ascension diagram (Fig. 4 (click here)) gives the impression of uniformity;
even the spectra (Fig. 6 (click here)) seem to confirm this impression. However in the
channel images several concentrations are visible; one of these is at
, 
. In order to study the small scale
structure, we computed images by
subtracting the average optical depth from the channel images. In Fig. 7 (click here) we
display the spectra based on these images. In these spectra the feature
mentioned above is clearly visible.
Gaussian fitting of this narrow feature gives the following average
parameters: integral 
km s
, v= -1.31 km s
,
observed velocity width (FWHM)
km s
. The velocity width (FWHM) corrected for instrumental
broadening of 0.62 km s
gives an average of
km s
.
The small deconvolved value of the velocity width
corresponds to an very low kinetic temperature 
K (
using Eq. (2).
The large scatter in the corrected velocity width is due to the fact, that the
correction is more uncertain in cases where the fitted velocity widths
are only slightly larger than the
instrumental resolution. For such a narrow feature the velocity resolution is
barely adequate. The scatter is strongly influenced by the most narrow lines, whereas
the average width is more reliably determined as well as the kinetic temperature.
However the question arises,whether this narrow velocity feature is an isolated entity in
space or if it is imbedded in a larger structure. In order to investigate this latter
possibility we have attempted to isolate the narrow velocity feature from the broader feature
at 
km s
in the original spectra by Gaussian fitting.
This fitting gave quite consistent results with the values derived above, mainly in velocity
and velocity width. However characteristic of
the fitting procedure of spectra of this type is that the broad component is affected by
variations of the narrow component. This effect occured here
which resulted in amplitudes with increased scatter. We tried
a different approach to avoid this problem. First we fitted Gaussians to the average
profile; three components gave a good fit (see Fig. 8 (click here)): a narrow component at
-1.5 km s
, the broader component at 0 km s
and the third, a very broad
component of velocity width 12 km s
. The latter two components represents portions of the spectra showing less variation
across the source; then we subtracted these (mean) components from the original spectra.
In the remaining spectra we fitted one component, the -1.5 km s
feature.
The
images of the parameters of this component are shown in Fig. 9 (click here). In the upper two panels
we show the the integral (
) and the central velocity of the component.
The -1.5 km s
feature is present across the the whole source, with a strong
maximum at the position of the narrow feature discussed above. The velocity is quite regular
with a slight gradient. Only in the SE corner there is a region with different velocities.
The halfwidth was corrected for instrumental broadening and then converted to the
kinetic temperature 
, using Eq. (2). A clear minimum is
visible with kinetic temperatures below 10 K at the position of maximum 
. When we make
the reasonable assumption that the spin temperature of the HI can be estimated to be close
to 
, then 
can be computed by 
. 
is shown in the
last panel of Fig. 9 (click here). The strong peak in the image of 
disappears.
The remaining fluctuations
across the source are most unlikely to be significant. This analysis indicates that the narrow
velocity feature is in fact only a cold (
K) part of a larger, warmer (
K)
HI structure. This may well be the first detection of an HI feature with such
a low spin temperature.
The depression in spin temperature could either be explained by the lack of heating (most
likely by dust shielding the radiation field) or by a larger cooling rate (by additional
heavy elements, which cause the cooling). Similar temperatures
are found in molecular clouds, which are thought to be dense clouds and
are shielded from the background uv-radiation (cf.
Burton et al. 1978).
Comparison with results of CO emission with
a spatial resolution of 
(Troland et al. 1985;
Wilson et al. 1993) with our synthesis HI absorption
observations show no corresponding
molecular cloud at the position of the low
temperature HI. However
there is a low velocity feature which seems to consist of both a narrow and broad
component; the narrow component has a velocity of -1.5 km s
, exactly in
agreement with the velocity of the HI cloud. In the observations of 
CO by
Wilson et al. (1993) there is an isolated narrow component at -1.5 km s
;
this spectrum is included in Fig. 8 (click here) as dashed line; the agreement with the HI is striking.
However there is limited spatial information available for a certain identification of the CO
with the HI feature.
The 
column density derived by Wilson et al. (1993) is 
cm
and should now be
compared with our average value of 
cm
for the -1.5 km s
feature,
instead of the single dish HI observations quoted in their paper.
The ratio
is then 0.05 and not two.
The new value agrees roughly with
the ratio observed for the clouds at velocities of the Perseus arm (0.1).
A molecular line at -1.5 km s
was also detected in CH by
Rydbeck et al. (1976).
If the angular size (
arcmin) of the narrow HI feature is also characteristic for
the line of sight and we assume a reasonable distance of 100 pc, the thickness of the feature
is 0.07 pc. From an average column density of 
at cm
, a space density of
100 at cm
follows.
The total mass of the gas (including 
) of the structure is then
Other small scale features are visible in Fig. 2 (click here). But due to the optimum scaling between minimum and maximum optical depth, these features are enhanced and are of much lower significance than the feature discussed above.
)
At the velocities of the interarm gas ( -26 < v < -16 km s
)
there is almost no detectable absorption. However there is a
possible small scale feature at v=-19.9 km s
in the NE of the source.
)
Reynoso et al. (1997), have used the VLA to investigate HI absorption at velocities lower
than the Perseus arm features.
They find several "knots" with velocities
< -68 km s
. We plotted these concentrations on top of our
channel images in Fig. 10 (click here). There is in general good agreement, but not
in detail; this is however not so surprising in view of the
differencies in resolution (spatially higher and in velocity lower
resolution than ours). These high velocity knots are interpreted as
recombined wind driven clumps moving ahead of the SN shock front.