The data were reduced using Snowden's software for extended objects and X-ray background analysis (cf. Snowden et al. 1994). For each observation, using only the accepted time intervals listed in the raw data files, three images are created. The first contains the total counts, the second is a model of particle background, and the third is an exposure map, which includes the variations in the detector quantum efficiency. These images have 5'' pixels, matching the detector resolution. These maps from individual fields are summed on a pixel-by-pixel basis into three mosaics of the entire Cygnus Loop field with 15'' pixels. Over such large spatial scales each observation is a distinct projection of the sky onto the plane of the detector, and these must be remapped to a common projection. Only the central 17' of each observation is used in order to avoid the significant degradation of the point spread function and decreased efficiency at large off-axis angles. The count-rate map (Fig.2) is the difference of the total counts and the total background, divided by the net exposure. Background subtraction is not a problem for the bright and moderately bright fields. The background does become comparable with the source surface brightness for the lower surface brightness regions. However, we have little difficulty extracting low surface brightness features because of the temporal and spatial stability of the HRI background, and our ability to cross register fields containing predominantly weak emission with adjacent ones containing bright features. To display the high resolution information contained in the mosaic we also show a zoom field which has been binned with 6'' pixels (Fig.3).
Figure: The current status of the Cygnus Loop ROSAT-HRI
mosaic. A total of 391,087 s from 31 pointings of the
ROSAT-HRI in 24 positions are combined into this map
and binned in 15'' pixels.
At first sight the X-ray morphology depicted in the ROSAT-HRI mosaic presents a bewilderingly complex view. The clearest message is that the Cygnus Loop is not an adiabatic blastwave propagating in a uniform medium and cannot be approximated as a Sedov remnant. There is considerable large and small scale azimuthal variation around the perimeter of the Loop and significant structure projected across its face. The X-ray emission is dominated by two prominent areas: one to the west and another to the east/northeast. Each region consists of a complex network of filaments and high surface brightness clumps. The emission is predominantly limb brightened, but the contrast is much greater than for an adiabatic blastwave. Significant emission (e.g. the eastern limb) lies well within the projected edge of the blastwave - closer to the center than expected for an adiabatic blastwave.
Figure: The NE rim of the Cygnus Loop
from the ROSAT-HRI mosaic. The data have been
binned in 6'' pixels. The field of view is
.
Figure: The NE rim of the Cygnus Loop in H 
.
The field has been registered with the X-ray field
shown in Fig.3.
The correlation between X-ray and optical line-emission falls into three broad categories that are distinguished by surface brightness and morphology: 1) regions of high surface brightness X-ray emission distributed in long coherent arcs which are aligned tangentially with, and lie interior to, the bright optical filaments, e.g., NE (NGC 6992, Figs. 3, 4); 2) indentations in the shell where bright X-ray emission occurs at large distances behind the projected edge of the blastwave, e.g., the eastern edge, the SE cloud, and the bright funnel shaped section at the N. These regions tend to be more chaotic and include the most compact and highest surface brightness regions within the Loop; 3) regions of low surface brightness limb-brightened X-rays bounded on the outside by faint, thin Balmer-line filaments, e.g., extreme N in Figs. 3 & 4. Type (3) filaments trace the projected edge of the blastwave since Balmer-line filaments occur where previously undisturbed atomic gas is shocked (Hester, Danielson, & Raymond 1986).
Category (1) and (2) occur where the shock has encountered a greater than average column of gas, which has caused substantial deceleration of the blastwave (Hester & Cox 1986, Hester, Raymond, & Blair 1994, Graham et al. 1995, Levenson et al. 1996). The difference between regions (1) and (2) may be due to viewing geometry, the shape of the cloud which is begin encountered, and the time since enhanced density was encountered by the shock. For example, the morphology of the western edge can be interpreted because this region is viewed almost exactly edge and the complication of projection effects are minimized (Levenson et al.).
It is crucial to decide whether the clouds that have been struck by the blastwave are isolated, or part of a coherent structure, e.g., an H II region cavity or wind blown bubble formed by the SN progenitor. We have argued that the appearance of the eastern edge and the SE cloud can only be explained if it is determined by the shape of a preexisting cavity (Graham et al. 1995). To a large degree the cavity must be spherical because the Cygnus Loop is approximately circular. However, to explain the regions of high optical and X-ray surface brightness there are several locations where the cavity walls protrude into the Cygnus Loop. Columns of gas can protrude into the H II region cavity if they are overdense regions - just like the elephant-trunk structures seen in HST/WFPC-2 images of M 16.
Evidence from filaments in category (3) clinch the argument in favor of
a cavity. The new HRI and H data trace the projected edge of
the blast wave over an unprecedented 70% of the perimeter of the
Loop. Remarkably, the X-ray emission extends right to the edge of the
blastwave in every case where it is bounded by Balmer-line filaments.
This contrary to what is expected for either a Sedov or evaporative
remnant. The fact that bright X-ray emission extends all the way to
the edge of the blastwave suggests that the shock has collided with a
density enhancement. The ubiquity of this morphology indicates that the
Cygnus Loop is located within a cavity.