A comparison between the 
image of the Gum nebula as taken
from Sivan (1974) with the giant HI disk (Fig. 1 (click here))
shows a good positional coincidence between the neutral gas disk and the
optical emission, as well as a similar angular size. To the northwest, the
HI emission slightly departs from the circular symmetry. The
two optical wings between 
and 
lie
outside the edge of the disk. The eastern wing depicts a good morphological
correspondence with the HI contours. The other wing, however, appears
detached from the giant HI disk.
The
Puppis-Vela region between 
, 
was thoroughly studied at 12, 25, 60 and 
by Sahu
(1992) using IRAS data. At 25, 60 and 
, the
author finds a 
radius shell centered at 
, which is called the "IRAS Vela Shell''. The center of
this structure is coincident with 
Velorum and with the Vela OB2
association. Thus, it is assumed that the IRAS Vela Shell has the
same distance as the Vela OB2 assocation, i.e. 450 pc. We checked the
image presented by Sahu (1992) and found a
remarkably good positional correspondence between the inner thick HI shell
reported by Dubner et al. \
(1992) and the infrared emission. In both cases, the emission
encloses a minimum at 
. Therefore, the northern part
of the IRAS Vela Shell seems to be the infrared counterpart of the thick
HI shell. The IR emission is probably due to the presence of shock heated
gas. The present results are consistent with Braun's (1985)
findings in the SNR IC443. No HI counterpart is found for the IRAS Vela
Shell as a whole.
The Gum nebula has also been studied in X-rays by Leahy et al. \
(1992), using the HEAO-1 data in the range 0.1 to 3 keV. In this
study, the authors adjusted for the X-ray emission a single temperature
model for a hot optically thick gas. The
results were consistent with a uniform X-ray emitting plasma with
. The estimated temperature agrees with that
expected from the interior of a SNR, according to Chevalier's model
(1974). The observed X-ray source region is not completely
coincident with the extent of the Gum nebula in 
. Thus, Leahy et al. \
(1992) suggest that the origin of the X-ray emission may
be ascribed to a SNR located within the Gum nebula but not filling it entirely.
Based on HI data, Dubner et al. (1992) have investigated
the thick HI shell in connection with the Gum nebula. They have found that
the strong stellar winds of 
Puppis, an O4If-type star, and
Velorum, a Wolf-Rayet
spectroscopic binary (WC8+O9I), may have provided enough energy to the ISM
to equal the kinetic energy acquired by the thick HI shell. However, by
applying the model of Chevalier (1974), they also show
that the thick HI shell is consistent with a model of an old SN
explosion occurred 
years ago. They suggest that this explosion
could have originated the Gum nebula. Observations at other
wavelengths also support the hypothesis of a SNR. In conclusion, Dubner et al.
\
(1992) propose that the gas distribution was originally perturbed
by a SN explosion and the hollowed shape is now maintained by the stellar
winds of the hot massive stars 
Puppis and 
Velorum.
In order to find out the origin of the giant HI disk shown in this paper, we first investigate the hypothesis of a wind-driven bubble blown by the massive stars present in the region. For this purpose, we have applied the model of Weaver et al. (1977) using typical parameters for O and WR stars, since it is well known that these type of stars have very powerful stellar winds capable of creating large cavities surrounded by HI shells (Cappa de Nicolau \ & Niemela 1984; Cappa de Nicolau et al. \ 1988; Dubner et al. 1990; Niemela & Cappa de Nicolau \ 1991).
To estimate the original ambient density, it was assumed that the
present swept-up mass was uniformely distributed into a spherical volume
of 150 pc in radius. Thus, an initial density of 
was derived. Combining this parameter with the typical mass loss
rate and terminal velocity of O stars, it follows that a bubble could have
reached that extent after 
years. This value is almost
twice the typical lifetime of 
yrs of an O star. If we
assume that the star blowing the bubble was a WR star, the same estimate
yields an age of 
years old for the bubble, again much
older than the lifetime of a WR star, typically 
yrs (Maeder
1991). In this latter case, the numbers can be reconciled if
we assume that both the WR star and its O-type precursor are responsible
for the creation of the big bubble.
The energy provided by O and WR stars during their entire lives is of the
order of 
ergs, as can be computed using typical values for
their mass loss rates, terminal wind velocities and mean lifetimes
(Abbott 1982; Garmany et al. 1981). By applying
the energy conserving model for wind driven bubbles (McCray
1983), it can be assumed that up to 20% of the total energy could
have been deposited into the surrounding medium as kinetic energy. In
both cases, the estimated kinetic energy originating in the star is higher
than the kinetic energy associated with the giant HI disk.
As for the candidate massive early-type stars, together with the two well
known massive stars 
Pup and 
Vel, which lie near the center
of the Gum nebula and at the same distance, there is another O-type star:
CD-474551, at a distance of 460 pc, placed at 
(Cruz-González et al. 1974). Based on the
above calculation, we conclude that the combined action of these stars can
plausibly explain the origin of the giant HI disk. As an alternative
explanation, we will also analyze the disk as being created by repeated
supernova explosions from an OB association.
Multiple explosions can act in concert to compress the ambient interstellar
gas into giant expanding shells called "supershells''. Supershells
have typical radii of 
and ages of the order of
years. McCray (1988) estimates that if a typical OB
cluster has initially 
stars with 
(i.e.,
massive stars capable of exploding), then, after explosions have started, a
supershell will begin to be created with a radius evolving as
where 
is the energy of a SN explosion in units of
ergs, 
is the initial ambient density and 
is the age
of the supershell in units of 
years. Application of McCray's model
to the giant HI disk produces an age of 
yrs. This model
would be compatible with an expansion velocity of about 14 .
According to this model, our giant HI disk is far from
becoming gravitationally unstable, since it should be at least 
years old. On the contrary, it is possible that
hydrodynamical Rayleigh-Taylor instabilities have occurred.
Tenorio-Tagle et al. \
(1990a) have found that in the presence of a hot halo, the upper
part of the remnants will "blowout'' because of Rayleigh-Taylor
instabilities, and the blowout will result in the formation of chimneys
from which ejecta will stream into the halo. As a consequence, the
shell as a whole would decelerate due to the loss of internal pressure.
The age at which the blowout is produced, about 
years
(Tenorio-Tagle et al. 1990b), is consistent with the age we
have estimated for the supershell. A blowout would also explain the fact
that the observed expansion velocity is lower than the 14 predicted by
McCray's (1988) model.
In the maps shown in Fig. 2 (click here), a feature quite similar to a
chimney at 
can be seen between +10 and
+20 . If this feature is actually related to the giant HI disk,
there is an inconsistency with the kinematic velocity, since it appears at
velocities higher than the systemic velocity. However, this apparent
incompatibility can be explained with
the following argument: the gas streaming upwards through an aperture in
the shell to the halo will have a velocity V substantially higher than the
expansion velocity of the supershell. If the apperture is located at a
latitude b from the galactic plane, the observer will only detect
the radial component 
of the escape velocity V, given by
. Since the feature is most clearly seen between
v=+12 and +18 , an LSR velocity 
can be adopted,
from which 
, assuming that the systemic velocity is
exactly known. The location of the apperture is 
. Thus, the escape velocity can be estimated to be 
, consistent with the predictions of Tenorio-Tagle et al. \
(1990a).
The two upper wings in the 
image of the Gum nebula converge to
the feature possibly associated with a chimney. This coincidence
supports the hypothesis that the giant HI disk can be explained as a
supershell, and reinforces the contention of the existence of a physical
association of the giant HI disk with the Gum nebula. The 
emission probably arises from the ionization of the hydrogen contained in
the shell, where the ionization flux would have been provided by 
Pup, 
Vel and the O star CD-474551. Concerning the origin of the
explosions which created the supershell, they can be atributed to the stars
belonging to the relatively aged OB association Vela OB2, located at 
pc, which is on the verge of disintegration (Sahu 1992).
A signature of the multiple SN explosions occurred in the past within the Gum nebula can be given by the presence of five pulsars with distances between 0.44 and 0.66 kpc (Manchester & Taylor 1981), although two of them are older than the supershell, according to the age obtained using the model of McCray (1988).
In summary, we propose that the Gum nebula may be a supershell created by the multiple explosions undergone by stars belonging to the Vela OB2 association. The thick HI shell fits in this context as the remnant of a late explosion which also contributed to the creation of the Gum nebula. The Vela SN explosion cannot have entered into the formation of the Gum nebula because of its young age, but on the contrary, its evolution may have been influenced by inhomogenities created in the interstellar medium by the past explosions, like the thick HI shell.
It has already been mentioned in Sect. 3 that the southern part of
the giant HI disk has been distorted by the presence of a 
diameter shell, centered at 
. The expansive effects of this shell can be seen between -10
and +14 , and the maximum size is reached at 
. In
Fig. 5 (click here), a map of this structure integrated between v=-10 and
+14 is shown. At 
, the galactic
rotation model of Fich et al. \
(1989) predicts a distance of 240 pc for a systemic velocity
of +4 . However, it should be borne in mind that a 1 difference in
the systemic velocity produces a change of nearly 150 pc in distance at
this longitude. Thus, since the usage of galactic rotation models may not
yield a precise result,
we will try to get an independent estimate for the distance, based on the search
of an agent capable of having evacuated the cavity. The only powerful
nearby star located in the inner region of the HI hole is the O5/7-type star
HD 49798, lying at a distance of 570 pc (Cruz-González et al. \
1974). This star is indicated in Fig. 5 (click here) as a star. We
will suppose that the shell has been created by the star HD 49798 and
therefore lies also at a distance of 570 pc. Under this assumption, the
radius of the shell can be estimated to be 
pc. The
reliability of our hypothesis will be tested below using the model of
Weaver et al. (1977).
Figure 5: Integrated map between -10 and +14 of the shell lying to the south of the
Gum nebula. Contours are at
90, 120, 150, 180, 210, 240 270, 300, 400 and 500 K. The greyscale is
shown in the edge on top of the image. The star indicates the position
of the O-type star HD 49798
The shell is open towards the lower density region, opposite to the galactic disk, while the expansion has obviously been slowed down in the direction of the denser region. The eccentric position of the star agrees with such an anisotropic expansion, though optical measurements should reveal if the eccentricity is rather due to a proper motion of the star. The southern part of the shell extends beyond the area surveyed in this paper, thus the mass estimates will be based on the available data and considered as a lower limit.
The amount of matter present in the shell is 
. Assuming that 
kpc, then 
. Adopting an expansion velocity of 
, the
expansion kinetic energy is above 
ergs. According to the
calculations carried out in the previous section, an O star could have
provided this energy to the interstellar medium through stellar winds.
If it is assumed that the mass contained in the shell was
initially uniformely distributed throughout a sphere of 142 pc radius, the
initial ambient density should have been higher than 0.16 
.
Application of the model of Weaver et al. (1977) implies that
the O star HD 49798 should have spent more than 
yrs blowing
the bubble. This time is compatible with a typical O-star lifetime.
There has not been found any correlation between this HI structure and the
emission. In particular, no optical filaments appear towards the
most dense regions, as would be expected if there were important energy
dissipation due to the interaction between the HI shell and the Gum nebula.
Based on the above results, it can be concluded that the southern HI shell was probably created by the stellar winds of the O-type star HD 49798, and that its age is close to the age of the Gum nebula. The interaction between these two structures can be noticed through the distortion of the isolines at the region of overlap.