A CATALOG OF SOFT X-RAY SHADOWS, AND MORE CONTEMPLATION OF THE 1 keV BACKGROUND S. L. SNOWDEN 1'2'3. M. J. FREYBERG _. K. D. KUNTZ 4, W. T. SANDERS s, ABSTRACT This paper presents a catalog of shadows in the _i keV soft X-ray diffuse background (SXRB) that were identified by a comparison between ROSAT All-Sky Survey maps and DIRBE-corrected IRAS 100 #m maps. These "shadows" are the negative correlations between the surface brightness of the SXRB and the column density of the Galactic interstellar medium (ISM) over limited angular regions (a few degrees in extent). We have compiled an exte'nsive but not exhaustive set of 378 shadows in the polar regions of the Galaxy (Ibl _. 20°), and determined their foreground and background X-ray intensities (relative to the absorbing features), and the respective hardness ratios of that emission. The portion of the sky that was examined to find these shadows was restricted in general to regions where the minimum column density is £ 4 x 1020 H cm -2, i.e., relatively high Galactic latitudes, and to regions away from distinct extended features in the SXRB such as supernova remnants and superbubbles. The results for the foreground intensities agree well with the recent results of a general analysis of the local 41-keV emission while the background intensities show additional, but not unexpected scatter. The results also confirm the existence of a gradient in the hardness of the local ¼keV emission along a Galactic center/anticenter axis with a temperature that varies from 10613 K to 10TM h:, respectively. The average temperature of the foreground component from this analysis is 106°s K, compared to 106.06 K in the previous analysis. Likewise, the average temperature for the distant component for the current and previous analyses are 106.00 K and 106"°2 K, respectively. Finally, the results for the ¼ keV halo emission are compared to the observed fluxes at 3 keV. where the lack of correlation suggests that the Galactic halo's ¼ keV and 3 keV fluxes are likely produced by separate emission regions. -] . . ;Code 662. NASA/Goddard Space Flight Center, Greenbelt, MD 20771. :Max-Planck-Institut fiir Extraterrestrische Physik, D-85740 Garching, Germany 3Universities Space Research Association. 4Department of Astronomy, University of Maryland, Collage Park, MD 20742 'Department of Physics, University of Wisconsin-Madison, 1150 University Avenue, _Iadison, WI 53706. 2 Subject headings: galaxies: Milky Way I interstellar: matter- X-rays: galaxies, general 1. INTRODUCTION A negative correlation between the column density distribution of a Galactic H I feature and the surface brightness of the _1keV soft X-ray....diffuse background (SXRB). a "'shadow," provides a mechanism for determining the location along the line of sight of X-ray emitting plasmas relative to the X-ray-absorbing H I. The fundamental result of a shadowing observation is the separation of the observed X-ray flux into foreground and background components relative to the shadowing object. An "object" in this case can be either a local column-density enhancement (i.e., cloud) or a local minimum in the Galactic H I column density. With the additional information of the distance to the shadowing object, which can be determined using interstellar absorption line measurements towards stars with a range of known distances, constraints can be placed on the locations of the emission coml:;onents. If a number of shadowing observations can be combined, then it becomes possible to create a three-dimensional map of the distribution of both the X-ray emitting and absorbing material. This process becomes particularly effective when the results are compared with the extensive mappings of the distribution of neutral material in the solar neighborhood that can be found in the literature (e.g., Frisch _ York 1983; Paresce 1984; Welsh et al. 1994; Sfeir et al. 1999). Shadows in the general SXRB (i.e., away from discrete emission features such as supernova remnants) were first unambiguously detected using the ROSAT observatory (Snowden et al. 1991; Burrows _ .Niendenhall 1991). The continuing study of such shadows has provided crucial information about the soft X-ray emission from the Galactic halo and from the Local Hot Bubble (LHB), an irregularly shaped region of ,-_ 106 K plasma which extends -,- 50 - 1.50 pc from the Sun in all directions (Cox & Snowden 1986; Cox &: Reynolds 1987; Snowden et al. 1990;1998). For example, the relationship between the ¼ keV SXRB surface brightness and the H I column density has been examined recently for a number of regions of the sky, both large and small. ROSAT All-Sky Survey data were used to study several regions of rather limited extent (Kerp et al. 1996; Kerp et al. 1999). a large region in Ursa Major (Snowden et al. 1994a), the .M complex of high-velocity clouds (Herbstmeier et al. 1995), the Draco Nebula (Snowden et al. 1991; Moritz et al. 1998). the Eridanus enhancement (Snowden et al. 1995a), as well as the whole sky (Wang 1997; Snowden et al. 1998; Pietz et al. 1998). ROSAT pointed observations were used to study the Draco Nebula (Burrows _: Mendenhall 1991), MBM 12 (Snowden, McCammon, _ Verter 1993), several distinct clouds (Wang 8c Yu 1995; Kuntz, Snowden, _ Verter 1997), a region of Ursa Major (Barber _ Warwick 1996), and a region of Eridanus (Guo et al. 1995). With the aim of providing a tool with which to further studies of the small-scale geometry of the local interstellar medium and to provide information on ¼ keV emission from the Galactic halo, this paper presents a catalog of 378 absorption/emission features in the ¼ keV soft X-ray -3- background derived from the ROSAT All-Sky Survey (RASS). This analysis improves upon the general results reported in Snowden et al. (1998, hereafter Paper I) by selecting regions of the sky most likely to provide statistically significant results. These are regions with variations in absorption greater than one optical depth over a solid angle of typically _ 30 deg 2 in directions of relatively low average column densities. Tile derived results are therefore more accurate for the sampled regions of the sky than the smoothed results of Paper I. In addition, we have slightly improved upon the analysis technique of Paper I. The data used in this analysis are discussed in § 2, tile analysis itself is described in § 3, and the results and discussion are presented in § 4. Section 5 presents the conclusions of this paper. 2. DATA For the analysis presented here, we have used the RASS high-resolution maps of the ¼ keV and 43-keV diffuse X-ray background presented in Snowden et al. (1997), and the DIRBE-corrected IRAS 100 pm maps of Schlegel, Finkbeiner, _: Davis (1998) scaled to represent the column density of Galactic neutral hydrogen. These are the same basic data sets (with the addition of the 3 keV data) that were used in Paper I. 2.1. X-ray Data As described in detail in Snowden et al. (1995b;1997), the X-ray data have been cleaned of periods of anomalously high noncosmic background, have had residual noncosmic background contributions subtracted (scattered solar X-rays, Snowden _: Freyberg 1993; particle background, Snowden et al. 1992, Plucinsky et al. 1993; long-term enhancements, Snowden et al. 1994b), are exposure corrected, and have had bright point sources removed. The maps cover ,_ 98% of the sky with roughly 106 12' x 12_pixels, and consist of count rate and count-rate uncertainty pairs in six bands. In this paper we use data from the R1 and R2 bands, and the summed R12 band (¼ keV) and R45 band (3 keV). The R12 band data are formed by summing the R1 and R2 band count rates (which are statistically independent) and adding their uncertainties in quadrature. The R45 band data are formed in the same manner by adding the R4 and R5 band data. Figure 1 displays the band response functions for the four bands. The R1 and R2 bands are clearly not spectrally independent, however the R45 band is reasonably cleanly separated from the R12 band. (The X-ray data used for the all-sky analysis of Paper I consisted of the R1 and R2 band data binned into 24' × 24' pixels.) The individual ROS.4T bands are formed by pulse-height selection on individual events (detection of an X-ray photon), so each event is uniquely assigned to a single band providing statistical independence. The lack of spectral independence of the R1 and R2 bands is due to the poor intrinsic energy resolution of the proportional counters, which is E/AE -_ 1 at E -,_ ¼ keV. -4- Thespectralseparationbetweenthe R12bandandtheR45bandis providedbv thecarbonh:c_ absorptionedgeofthe proportionalcounterentrancewindowat 0.284keV. Aswill beaddressedin §3,theanalysispresentedherehasbeenlimitedto approximately 10_ofthesky.orroughly105pixels.Histogramsofthestatisticalsignificance(countratedivided bytile uncertaintyin thecount rate) of the individual pixels in the three ¼ keV maps used in this analysis are showu in Figure 2. While the significances are not in general as large as one would prefer, they are sufficient for this analysis, and the finer angular binning (than used in Paper I) allows an increased sampling of any fine structure in the column density of the absorbing interstellar medium. 2.2. Measure of Absorption Column Density As in Paper I, we use the DIRBE-corre_ted IRAS 100 pm data from Schlegel et al. (1998), cast into the same projections and pixels as the X-ray data, as a measure of absorption column density. However. they are slightly different from those used in Paper I as they are the product of the final processing used for the Schlegel et al. paper. The differences are minor and do not significantly affect the results. The inherent angular resolution of the IRAS data (,-, 5') exceeds by a factor of five the useful resolution of the X-ray data, which is limited by counting statistics rather than b.v the resolution of the detector (the angular resolution of RASS data, which is an average over the field of view. is ,-, 3'). Using the IRAS data provides a major advantage over using the available HI surveys, which have at best 35' resolution with incomplete sky coverage. The disadvantage of the IRAS data is that the3: do not contain velocity information that would allow the straightforward separation of the X-ray-absorbing interstellar matter into distinct components, as can be done with the H I data. However, velocity information can be determined for distinct IRAS features by comparing the angular structure with the coarser resolution H I data (although that is beyond the scope of this paper). Another major advantage for the IRAS data is that they also sample molecular gas. Since the IRAS data are scaled to column densities of Galactic neutral hydrogen at high Galactic latitudes, they are an appropriate measure of X-ray-absorbing interstellar matter for this analysis (see Kuntz & Snowden 1999 for a more extensive discussion of this subject). The Schlegel et al. (1998) data were scaled to the column density of interstellar hydrogen by using the Leiden-Dwingeloo 21-cm survey of Hartmann _ Burton (1997). Data in the Galactic polar regions were binned into 1.6° x 1.6° pixels and linear fits made to the northern and southern data separately. The range in NH was limited to _<4 × 1020 H I cm -2 to avoid the contribution of molecular gas to the Ira0 intensities and therefore contamination of the results. The scatter plots and fitted relations are shown in Figure 3. The fitted lines for the north and south are given by NH -" 0.1S6 + 1.403 X I100 and NH ----0.334 + 1.305 × I1o0, respectively, where .h:H is in units of 1020 H I cm -2 and I10o is in units of MJy sr-1. The turnover at higher values of I100 is due to the presence of molecular gas. The difference between the Iloo to NH scaling between the northern 5 andsouthernhemisphereiss notsignificantforourpurposes.Theyprovidesimilarresultsforlow column densities where the analysis is most sensitive to systematic uncertainties (the cross-over of the curves is at :\'H "_ 3 x 102° cm-2). At higher column densities where the relations diverge tile column densities are optical])' thick. 3. Analysis 3.1. Selection of Target Regions The selection of the locations and sizes of the regions, or shadows, that were analyzed for this paper was subjective, but not arbitrary. The selection criteria were that there be an apparent absorption feature in the _ keV background and/or an emission feature in the NH map, that there be a reasonable range in the column densities of H l in the vicinity of this feature (most of the regions have a range greater than one Ol_tical depth), and that the minimum column density of H I in the vicinity be reasonably low. These criteria select regions where the total column density is low enough for the R12 band to still be sensitive to distant emission and where the range in NH is large enough to provide an acceptable lever arm for the fit. Since one optical depth for X-rays at ¼ keV is ,,- 1 x 1020 H I cm -2, regions with a minimum sampled column density £ 4 × 1020 H I cm -2 corresponding to a minimum optical depth g 3 were used (note that optical depth is not a linear function of NH, see Snowden et al. 1994b). These selection criteria limited the analysis to roughly the 407c of the sky with the lowest column density. However, the criteria were occasionally relaxed in order to better sample the local component, i.e., directions of higher column density where all observed emission can reasonably be assumed to be local in origin. For these regions the background emission is not well constrained. Regions affected by known supernova remnants and superbubbles, specifically Loop I, the Eridanus superbubble, and the Monoceros-Gemini ring, were excluded. The studies of emission and absorption variations of such features using ROSAT data are certainly interesting in their own right (e.g., Loop I, Egger 1993; Eridanus, Snowden et al. 1995a, Guo et al. 1995; Monogem, Plucinsky et al. 1996), but are more appropriately left for detailed investigations of the specific object. The locations of the regions selected for analysis here are shown in Figure 4, where rings showing the extent of the regions are overlayed on both ¼ keV X-ray and IRAS maps. The radii of the target regions were determined by the angular extents of the absorption and/or emission features. In practice, all but three of the target regions are _<8° in diameter. Near some of the larger features, several smaller regions were also selected for analysis in order to independently sample different parts of the feature to test for variation in either the foreground or background X-ray intensities. An effort was made to have the selected regions be distributed over the available lower column density parts of the sky, which occasionally required relaxing some of the se]ection criteria above. While the sky coverage is certainly not uniform, because the distribution of suitable targets is not -6- uniform,theskycoveredisreasonablyrepresentativoef thewholeat highlatitudes. 3.2. Fitted Model AphysicalpicturesimilartothatusedinPaperIwasassumedherefortherelativelocationsof theX-ray-emittingplasmasresponsibleforthebulkofthe i keVbackgroundandthecoolerX-ray absorbinggasin the interstellarmedium.Specificallyi,t consistsofanunabsorbedforeground X-rayemissionregion(theLocalHot Bubble,seePaper I), a region of neutral, X-ray absorbing ISM that includes the shadowing cloud and all other Galactic H I along the line of sight, and a region of X-ray emitting plasma in the Galactic halo that is not intermixed in the neutral ISM and that produces most of the observed ¼ keV background of distant origin (from Paper I this component has a temperature of ,-, 106.o K). Paper I assumed an isotropic extragalactic bacl_ground power law with an index taken from Hasinger et al. (1993) of 1.96 and a normalization fixed to be consistent with extragalactic ¼keV shadowing results (e.g., M 101, Snowden & Pietsch 1995; NGC 55, Barber. Roberts, & Warwick 1996; several additional face-on galaxies, Cui eta[. 1996). To make our current picture more physically realistic, we added the following refinement to the distant emission of the Paper I model. The previous isotropic extragalactic background power law has been replaced with two components: 1) an isotropic power law (10.5E -la6 photons cm -2 s-l sr-1 keV -1, the fitted power law of Model A from Chen, Fabian, & Gendreau 1997). which is the extrapolation of the extragalactic background observed above ,,- 1 keV and is consistent with the observed flux in the ROSAT 1 - 2 keV band, and 2) an anisotropic T = 106"a I,: thermal component that accounts for the observed excess of the diffuse X-ray background at 3 keV above the extrapolation of the power law. We used the R45 band data to determine the intensity of the 106"4 h: thermal component for each shadow region individually. The sum of these two distant components, when extrapolated to the ¼ keV band, produces an intensity similar to that of the extragalactic power law assumed for Paper I. The advantageto this revision is that the model now represents the entire 0.1 - 2.0 keV ROSATspectrum in a self-consistent manner, at least in a general sense. The broadband intensities in the 0.5 - 2.0 keV band are well fit with reasonable spectra. These spectra can then be extrapolated down to the 1_keV band to derive reasonable estimates for their contributions to that band. As will be shown 4 below, this more complicated model has little actual effect on the shadow results in the ¼ keV band. The assumed geometry for the LHB, absorbing gas, and cooler (T -,- 106.o keV) region of halo X-ray emission is shown schematically in Figure 5a. Mathematically, the following equation is fit separately to the R1, R2, and R12 band data: Ix = I0 + I1 x exp[-er(NH, 7"6.0)x NH] + 16.4 × exp[-a(NH, T6.4) × -VH]+ leg × exp[--a(NH, 01.46) X NH]. 7 NotethatthedataintheR1,R2,andRI2 bandsarefit independentlyandthatonlytheparameters Io and 11 are allowed to vary. Ix is the observed X-ray intensity, I0 is the fitted foreground component. 11 is the fitted distant (halo) component, which is absorbed by the column density NH. a(NH./'6.o) is the theoretical band-averaged absorption cross section (based on the cross sections of Morrison & McCammon 1983), which is a function both of NH and the temperature of the I1 component (see Snowden eta[. 1994b). The fits were first done using T = 106.02 K, the value from Paper I, but were refit using the value T = 106.00 K based on the initial results. 16.4 is the intensity of the hot Galactic halo component that is determined from the R45 band data and is fixed separately for each region; it is absorbed by the column density NH. o'(NH, T6.4) is the theoretical band-averaged absorption cross section for this component, leg is the fixed isotropic extragalactic power-law contribution, and a(NH, ch.46) is the theoretical band-averaged absorption cross section, which is a function both of NH and power-law index, a. As noted above, the spectrum of the extragalactic power law is taken to be 10.5E -1"46 photons cm -2 s-l sr -1 keV -1, and'is extrapolated to ¼keV and evaluated ,on a band-by-band basis. The normalization for the hot (106.4 K) halo contribution was determined in the following manner. 1) The absorbed power law contribution in the R45 band was subtracted from the average R45 band intensity over each shadow region. 2) This excess, minus a small amount (10 -5 counts s-1 arcmin -2) assumed to arise from the LHB (with an assumed temperature of 106.o K and typical high-latitude normalization set by Paper I), was deabsorbed by the average column density of the region. 3) This deabsorbed value was attributed to R45 band emission from the T = 106.4 K component. 4) Finally, the T = l0 s'4 K spectrum was extrapolated to the R1, R2, and R12 bands to fix 16.4. Following Paper I, the assumed spectra of the I1 and I6.4 components are Raymond _ Smith (1977; Raymond 1992; 1991 computer code update 6) thermal equilibrium plasma models using cosmic abundances with T = 106.0 I,_and T = 106.4 K, respectively. The choice of the spectrum affects the analysis in two ways: 1) The band-averaged absorption cross Sections are spectrally dependent, but only to a limited extent, and 2) the choice of the temperature for the hotter halo component affects the amount of emission and band ratios (to the few percent level) attributed to the ¼keV band. The R12/R45 band ratio decreases by a factor of four for spectra between 106.3 K and 106"s K. However, the predicted unabsorbed R12 band intensity for a typical region, even assuming a 106.3 K spectrum, is only comparable to the intensity from the extragalactic power law. We have chosen T = 10_'4 K following Nousek et al. (1982). but this choice is confirmed bv Kuntz & Snowden (1999). There are a number of alternatives to our simple assumed geometry. The most likely are that the cloud or absorbing feature is located within the foreground X-ray emission region (Fig. 5b) or within the X-ray emission region(s) in the halo (Fig. 5c), i.e., there is some form of intermixture of the X-ray emitting and absorbing gas. These variations, while adding some 8While there are more recent versions of various plasma codes, use of the 1991 vintage allows direct comparisons with previous work by these authors. -8- complexity to their interpretations, do not appreciably affect the results. For example, if the shadowing cloud lies within the local emission region, Figure 5b, the fitted distant component includes the emission between the cloud and the edge of the emission region (wall) as well as any more distant emission beyond the wall. In such a situation, the fitting algorithm incorrectly deabsorbs the emission between the cloud and wall by the HI of the wall and "adds" it to the total distant component. Because of this, the attributions of "'foreground" and "distant" emission may need to be reevaluated once the distances to the absorbing features are known. 4. RESULTS 4.1. Shadows "The results of the fits for the R1, R2, and R12 bands are listed in Table 1. For each target region, the table lists the Galactic coordinates of the region center, the diameter of the region, the fitted values for the R1, R2, and R12 band fits and their associated .\_ values for the fits, and the number of degrees of freedom. Scatter plots of the R12 band data versus the column density of H I along with the fitted curves are shown in Figure 6. Inspection of both the table and figure shows the wide range of both X-ray emission and absorption covered by the shadows. The errors quoted in the table are the la values derived using the Lampton, Margon, & Bowyer (1976) criteria (\ra2in + 2.3 for two-parameter fits). When the 1 a range in uncertainty includes zero, the one-sigma upper limit is listed. Figure 7 compares the foreground and distant R12 band intensities derived in this paper with those from Paper I. The foreground (I0) results are in reasonable agreement, showing a moderately tight correlation, while the background (I1) results show additional scatter. This is not particularly surprising as the Ix results are more dependent on the chosen model, and the Paper I results were significantly smoothed. Since the regions for this analysis were specifically chosen to provide good measurements, the results of this paper are more reliable. Figure 8 shows scatter plots of the hardness ratio (R2/R1 band ratio) for the I0 and 11 components as a function of the R12 band I0 and I1 intensities, respectively. In both cases while there is significant scatter in the data, there is no suggestion of a systematic variation with intensity. Figure 9 shows histograms for the R2/R1 band ratios for the Io and I1 components. The average values for the data are 1.13 and 0.98, respectively. These values imply temperatures of 106.08 I( and 106.00 K. which are consistent with those of Paper I, where the derived values for the local and distant emission were 106.06 K and 106.02 Kr. rThe quoting of temperatures for the thermal emission must always be done with caution as they are model dependent. Different thermal emission codes will attribute different temperatures to the same broad-band ratio or fitted spectrum, and even different versions of a given code will show variations. In addition, current thermal equilibrium codes do not fit the observed data for the diffuse ¼ keV background particularly well (see Sanders et -9- 4.2. Variation of Intensity with Direction Figure i0 presents the variation in the fitted values for I0 and 11 as a function of position on the sky. The relative size of the plotted circles corresponds to the relative intensities of the emission. From the plots it can be seen that the value for Io varies fairly slowly over the sky with higher values generally at higher latitudes. However, this is more consistently true for the northern hemisphere than for the southern. In the south there is an asymmetry in Io such that the longitude range 0° < l < 180° has in general lower intensities than the range 180° < l < 360 °. The It results displayed in Figure 10 have a completely different character. The regions of bright emission are considerably clumpier, and in general the higher intensities are at lower latitudes. The consistently bright regions are in the direction of Draco in the north and in the directions of 1,b ,_ 40°, -30 ° and the "void" in the south (the void, or "Region of Bizarre Emptiness", Cox 1997, at l ,-_2300 is a direction in which there is little H I near the Galactic plane out to distances of a few hundred parsecs, see, e.g., Sfeir et al. 1999 and references therein). Note that the latitudes of the displayed enhancements are relatively high (20 o < [b[ < 4.5°), although in all probability the enhancements would reach down closer to the Galactic plane if the analysis of this paper was able to sample them. As expected from the correlations shown in Figure 7, the results of this Paper agree well with the position dependencies of the Io and I1 values from Paper I. 4.3. Variation of Hardness Ratio with Direction Figure 11 shows scatter plots of theaverage hardness ratios for the data in Figure 8 binned into 10° latitude bins. As in Figure 8, there is relatively little apparent variation in the ratios. On the other hand, when the data are binned into longitude bins they do show significant systematic variation (Figure 12), at least for the I0 ratio. The I0 ratio varies from a high of R2/R1 ,,, 1.25 averaged over longitudes within 40° of the Galactic center to R2/R1 ,., 1.04 averaged over longitudes within 400 of the Galactic anticenter. The range in the I0 ratio as a function of longitude is centered on the average value, and implies a temperature range of 108"°4 K to 106"13 I{. The magnitude of the variation is not as great as that shown in Snowden, Schmitt, & Edwards (1990), which reported a dipole gradient in the ¼ keV hardness ratio in the Wisconsin all-sky survey data (McCammon et al. 1983) with a range of 105.9 I,: to 106.2 K with the low end of the dipole axis pointing at 1,b = 168?7, 11?2. When the data are analyzed with respect to the orientation of the Snowden, Schmitt, & Edwards (1990) dipole (Figure 13), the extrapolated range in the ratio is 1.02 - 1.24 implying only a slightly broader temperature range of 106.02 I( al. 1998; Sanders et al. 1999). However, the use of the temperature does provide a scale (admittedly imperfect) with which to measure various models. We have much to learn about the ionization states and abundances of the X-ray-emitting plasmas. - 10- to 106"13 K. The fitted line in Figure 13 is acceptable at the 30% level with \_ = 1.15 with 18 degrees of freedom. The best fit for a constant value has \_ = 2.50 with 19 degrees of freedom, which can be ruled out at the > 99.9% confidence level, demonstrating the significance of the dipole variation. The FWH.M range of the local component ratio is 0.91 - 1.30 (Fig. 9a), which is dominated by the systematic variation of the fitted values across the sky. Figure 14 displays the variation on the hardness ratio of the I0 data versus position. The southern hemisphere data clearly show the Galactic center/anticenter gradient. The gradient is less obvious in the northern hemisphere data where the ratio appears to be more mixed with only a slight trend of hardness ratio versus position. Because of the statistics of the results for the distant component, it is much less clear whether there is a significant variation of the hardness of the emission across the sky. The FWHM range of the Il ratio distribution is 0.67- 1.22 (Fig. 9b), implying a temperature range of 105.83 -- 106"12 K. Altl_ough the width of the distribution is probably enhanced due to the poorer statistics of the fits for the distant-emission parameters, it is likely that there is some true variation of the temperature in the halo. The emission appears clumpy suggesting both spatially separate emission regions and that the emission is distributed over large distances (when compared to the size of the LHB). 4.4. The LHB: I0 and the Local HI Cavity The premise of tile Local Hot Bubble is that there is a cavity in the H I of the Galactic disk that contains the Sun and is filled with an X-ray-emitting plasma. This plasma produces half to all of the observed intensity at _1 keV in all directions. The existence of the cavity is required bv interstellar absorption-line measurements. The existence of the plasma within the cavity is absolutely required in low-latitude directions because there are nearby, optically-thick walls of H I (the edge of the cavity), yet a non-zero ¼ keV flux is observed. If the plasma is isothermal and is distributed uniformly throughout the cavity, then the fitted value for Io, when properly scaled, should provide a measure of the distance to the boundary of the cavity. A comparison between the I0 intensities of Paper I with the shape of the local cavity is made by Sfeir et al. (1999), who presented a mapping of the local ISM based on an extensive optical absorption-line study. The results were in reasonably good agreement over most of the sky but there are directions where the cavity extends well beyond the required path length of plasma (e.g., the RBE). The dipole gradient in the temperature determined above produces only a -t-10% variation in the emissivity of the plasma, with greater emissivity in directions of lower temperatures.