Sub-Lyman-α observations of supernova remnants

Sub-Lyman-α observations of supernova remnants

0273—1177i~3$6.00 + 0.00 Copynght @ 1993 COSPA1~ Adv. Space Res. Vol. 13, No. 12, pp. (12)67—(12)76, 1993 Printed in GreatBritain. All rights reserve...

916KB Sizes 0 Downloads 57 Views

0273—1177i~3$6.00 + 0.00 Copynght @ 1993 COSPA1~

Adv. Space Res. Vol. 13, No. 12, pp. (12)67—(12)76, 1993 Printed in GreatBritain. All rights reserved.

SUB-LYMAN-a OBSERVATIONS OF SUPERNOVA REMNANTS K. S. Long Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA and Centerfor Astrophysical Sciences, Department of Physics and Astronomy, The Johns Hopkins University, Charles and 34th Streets, Baltimore, MD 21218, USA

ABSTRACT The first map of 0 VI A 1035 emission of the Cygnus Loop has now been obtained with Voyager. The first good spectra of the sub-Lye region in the Cygnus Loop and in the LMC remnant N49 have been obtained with the Hopkins Ultraviolet Telescope. The lines detected below 1200 A include S VI AA933, 945, C III A977, N III A991, and 0 VI AA1032, 1038. The 0 VI luminosities of the Cygnus Loop and of N49 exceed their soft X-ray luminosities. There is 0 VI emission at the primary shock front in the Cygnus Loop and from the bright optical filaments. The HUT spectra of the Cygnus Loop can be interpreted in terms of shocks with velocities of 170—190 km s~ the differences between the two spectra are due to the time since the beginning of the shock-cloud encounter. INTRODUCTION The wavelength range between Lya and the Lyman limit contains many important plasma diagnostics—the Lyman and Werner bands of molecular hydrogen, the higher order Lyman lines of atomic hydrogen, and transitions of ions of many intermediate mass elements. For supernova remnant (SNR) research however, the most important transitions are the ls-2p transitions of 0 VI, which

are one of the few diagnostics we have of plasmas in SNRs at temperatures between i0~K and 106 K. 0 VI AA1032, 1038 probes the boundary between the hot, low density X-ray-emitting gas that is created by the primary shock as it propagates into the ISM and the cooler, higher density clouds that are partially bypassed by the primary shock and are heated by slower secondary shocks producing bright optical and UV emission. The first observations of a SNR in the FUV below Lyc~were made with Copernicus which was used to study absorption lines along the line of sight to two stars behind the Vela SNR /1/. These observations revealed the presence of broad absorption lines of 0 VI AA1032, 1038 (and N V AA1239, 1243), as well as numerous lower ionization lines including CII A1037. The 0 VI column densities that were measured, 4.5 x 1014 cm2 and 2.8 x 1014 cm2, to the two stars behind Vela were much greater than to two comparison stars 10 and 1.5°away from the SNR and also larger than had been observed along other lines of sight /2/. Therefore these observations were the first to reveal substantial amounts of material at temperatures between 10~K and 106 K in SNRs. Unfortunately, these are the only high resolution absorption observations of a SNR below Lyct, although additional UV echelle observations have been made above Lyct with IUE /3,4/. Here I review the observations that have been made of SNRs in the region between Lya and the Lyman limit since Copernicus, concentrating on the observations that my collaborators and I have carried out with the Voyager spacecraft and more recently with the Hopkins Ultraviolet Telescope. (12)67

(12)68

K. S. Long

INSTRUMENTATION Voyager The Ultraviolet Spectrometers (UVSs) on the Voyager spacecraft were primarily designed to study planetary atmospheres in the wavelengthrange 500—1700 A. However, between planetary encounters they have been used to study a wide range of astrophysical objects at low spectral resolution. They are objective grating spectrometers with bare microchannel-plate intensified detectors with 126 channel self-scanning array readouts and dispersions of 9.26 A per channel /5, 6/. These spectrometers have a resolution of —.~38A for extended sources and a FOV of 0.10°x 0.87°.Although the UVSs are sensitive to photons between 500 and 1700 A, the detection efficiency is much lower beyond 1250 A. Because of their large FOV, they are suitable for observing large diameter SNRs, and they have been used to observe both Vela and the Cygnus Loop SNRs. The Hopkins Ultraviolet Telescope The Hopkins Ultraviolet Telescope (HUT) was a part of NASA’s Astro-1 mission, which flew in 1990 December. HUT was designed to obtain moderate (3.5 A) resolution spectra in the wavelength range between 830 A and 1860 A, and is optimized for the wavelength range between 900 A and 1200 A. HUT consists of a 0.9 m f/2 iridium-coated primary mirror that feeds a Rowland spectrograph. UV photons pass through one of several spectrograph apertures, diffract off an an osmium-coated concave grating, and are detected with a CsI-coated microchannel-plate detector with a photodiode array readout /7/. On Astro-1, HUT was mounted on the Shuttle’s Instrument Pointing System, along with two other instruments, the Ultraviolet Imaging Telescope /8/ and the Wisconsin Ultraviolet Photopolarimeter Experiment /9/. Because of its low f/# and large spectrograph apertures, HUT is reasonably well optimized for observations of SNRs. Four observations of SNRs, all using the 9~4~l x 116” aperture, were carried out. Two filaments were observed in the Cygnus Loop; one position was observed in the extragalactic SNR N49; and the final pointing was at the Crab Nebula. CYGNUS LOOP The Cygnus Loop is the best studied of all SNRs at most wavelengths and it is certainly the best studied in the region below Lya. It is nearby (770 pc; /10, 11/), large (3°x 4°),and bright. Absorption along this line of sight is relatively small (E(B V) = 0.08; /12, 10/) and so it is easy to observe at soft X-ray and UV wavelengths. In fact, there are more IUE observations of the Cygnus Loop than of any other SNR /13, 14, 15, 16/. It is one of the two SNR which have been detected with the WFC on ROSAT. —

The Cygnus Loop is limb-brightened at X-ray, optical, IR, and radio wavelengths /17, 18, 19/. The SNR has a soft (0.1—4 keV) X-ray luminosity of 1.1 x 1036 ergs s~,an X-ray temperature of 3 x 106 K, and a limb to center surface brightness ratio of about 3:1. Its global X-ray properties are reasonably well characterized in terms of an 18,000 year old SNR. expanding into a medium with a density of 0.2 cm3 at 400 km s~ /17/. On large scales, the emission at all wavelengths is reasonably well correlated, which presumably reflects large scale density variations around the periphery of the SNR. On smaller scales, however, the correlation between X-ray and optical emission is not as tight, a fact which has been used to argue that cloudlet evaporation is less important than shock heating in understanding the soft X-ray emission in the optically bright regions of the Cygnus Loop /20, 21/. Despite the apparent success of Sedov models at X-ray wavelengths, there is considerable controversy concerning the pre-SN environment of the Cygnus Loop. Various authors (see, e.g. /22/) argue that the SNR is evolving in a cavity created by the progenitor star and has only recently encountered the denser material outside of the cavity. Most of the bright filaments in the Cygnus Loop (and other SNRs) have optical spectra which have forbidden lines e. g. [0111],[S II], and [N II] which are comparable in strength to Hct. Since these forbidden lines are produced in recombining plasmas and since [S II] and [Nil] do not appear in shock models until the temperature of the post-shock plasma drops to 10,000—15,000 K, these spectra are interpreted in terms of steady flow or “complete” shocks. Optical and UV observations —



sub-Lyman-re Observations

~

~

(12)69

600

800

1000 1200 Wavelength (A)

1400

1600

600

800

1000 1200 Wavelength (A)

1400

1600

Fig. 1. The spatially averaged 500—1700 A spectrum of the Cygnus Loop as observed with Voyager /29/. The upper panel shows the count rate spectrum; the lower panel shows the fluxed spectrum. The two prominent features at 980 A and 1035 A are due to C III A977 and 0 VI )tA1032, 1038. of these filaments suggest preshock densities of 10 30 cm3 and shock velocities of 100—150 km s~. There are also some filaments in which [0111]is bright but the other lines including Ha are relatively faint /10/. Since [0111]is produced in plasmas which have higher temperatures (50,000 K) than the other lines, these filaments are thought be locations where the shock has encountered a density enhancement relatively recently, and as a result, the the cooler part of the recombination zone is “incomplete”. Finally in the Cygnus Loop, there is an extensive set of faint Ha ifiaments which are most visible right at the X-ray shock front (/23, 24, 11/). These “Balmer-dominated” filaments show little or no forbidden line emission and are believed to be shocks in which the recombination zone is entirely absent /25, 26, 27/. The Balmer line emission that is observed is produced as the plasma is ionized behind the shock. Such shocks are said to be nonradiative since radiation has not had time to affect the structure of the post-shock flow. —

The first sub-Lya observations of the Cygnus Loop were made with Voyager by Shemansky, Sandel, & Broadfoot /28/. Below Lya, the spectra they obtained showed prominent peaks centered at 977 ±2 A and 1037.2 ±1 A. They identified the 977 A emission as CIII A977 and the 1037 A feature as CII A1037. They selected C II over 0 VI because they felt that the line should have been centered closer to the the stronger member of the 0 VI doublet at 1032 A than they measured. looseness=+2More recently, Blair et al. /29/ described a series of 21 pointings with Voyager, spaced in a regular grid across the Cygnus Loop, designed to characterize the global FUV properties of the Cygnus Loop. Since the Voyager spacecraft are allowed to drift within a 0.3°deadband, these observations covered nearly all of the Cygnus Loop. These observations confirmed that the 900— 1200 A spectrum of the Cygnus Loop is dominated by peaks at 980 A and 1035 A; the 1035 A 980 A ratio varies from —~0.6to 3. The spatially averaged, background subtracted count rate spectrum of the Cygnus Loop and the corresponding fluxed spectrum are shown in the upper and lower panels of Fig. 1. The fluxed spectrum also shows evidence for two lines above Lya, one at 1550 A that is surely CIV AA1548, 1551 and the other at 1650 A that may be due to a combination

(12)70

K. S. Long

of 0111 A 1665 and Hell A1640. (Blair et al. expressed concern that the latter feature was narrower than expected.) Blair et al. used these data to construct crude maps of the Cygnus Loop in the two features below Lya. All of the pointings at the Cygnus Loop were carried out with the long axis of the slit oriented at a position angle of 49°. Because the UVSs have no spatial resolution other than that provided by the spectrograph aperture, the spatial resolution is quite different in the two directions. Blair et al. elected to represent the spatial information contained in their observations by constructing a 77 x 100 point grid on the Cygnus Loop. They then created similar maps from X-ray /30/ and [0111]optical images of the Cygnus Loops. They found a general similarity between in the FUV images and the optical and X-ray maps. Blair et a!. argued that the correspondence between the 1035 A and the X-ray emission is better than the 980 A map. Conversely, the 980 A map appeared to resemble the convolved [0 III] image more closely. It is difficult to reconcile this trend with the idea that the 1035 A feature could be CII A1037 and so Blair et al. concluded that the 1035 A feature was dominated by 0 VI AA1032, 1038. More recently, Vancura /31/ analyzed a new set of observations specifically aimed at distinguishing the X-ray shock from the optically bright filaments. One series of exposures was obtained with Voyager UVS aperture aligned along the NE limb of the Cygnus Loop where the X-ray shock lies ~ arcmin outside the bright optical filaments. These observations show that part of the 0 VI emission rises with the X-ray gas, but also confirm that the brightest CIII and 0 VI (and X-ray) surface brightnesses are associated with the bright radiative filaments in the NE. Blair et a!. /29/ also used their maps to determine the luminosity of the Cygnus Loop in C Ill and in 0 VI. This was the first attempt to determine the luminosity of the Cygnus Loop in the UV, since only individual filaments can be observed with IUE. The luminosity of the Cygnus Loop is 2.7 x 10~ergs s~ and 4.6 x 1036 ergs s~1in CIII and OvI, respectively. By comparison, the X-ray luminosity derived from the Einstein IPC map is 1.1 x 1036 ergs s_i /17/. Assuming pressure equilibrium between the X-ray gas at 3 x 106 K, the 0 VI gas at 3 x i05 K and the C III at io~K, Blair et a!. show that the volume emission measures at all three wavelengths are comparable, although the mass of CIII and 0 VI producing gas is ~~0.2%and 4.5% of the X-ray gas. Recently, Rasmussen & Martin /32/ have constructed a new map of 0 VI in the Cygnus Loop using data obtained from a rocket-borne echelle spectrograph. Although this image contains <200 photons, there is no doubt that the photons which were observed are from 0 VI. It may be, in fact, that the main importance of this image is that it confirms that C II is quite weak everywhere. Rasmussen & Martin derive a somewhat lower luminosity, 1.2 ±0.5 x 1O~ergs s1, than did Blair et al. but basically confirm the global morphology of the emission. Although it is possible to discuss the morphology of the Cygnus Loop in 0 VI with the Voyager and rocket data, it is hard to draw firm conclusions about the shocks that produced the emission because of the limited spectral range and and spatial and spectral resolution of the observations, and the fact that physical conditions are likely to vary tremendously in the regions defined by a single spatial resolution element. The HUT observations of the Cygnus Loop begin to address these problems. Two positions in the Cygnus Loop were observed with HUT. The first observation was of a complex, but optically bright region on the eastern limb of the Cygnus Loop /33/. Cornett et al. /34/ have recently compared the morphology of this emission complex at X-ray, optical, and UV wavelengths, using UV images obtained with the UIT on Astro-1. The HUT slit position appears to lie about 2 arcmin from a bowshock, also seen at X-ray wavelengths, created by the interaction of the SN blast wave with a dense cloud. (See also /20/.) Although there is substantial Ha emission throughout this region, [0111]is exceptionally bright at the HUT slit position. Therefore, optically, this would be regarded as an incomplete shock. The second observation was of one of the brightest Balmerdominated filaments in the NE /35/, a ifiament which had been studied previously with IUE (/23/). About 660 seconds of good observing time were obtained at each position.

Sub-Lyman-re Observations

15

I

;T1’ \

o 800

~

1000

I.5_

U.)

2

10

I

I

I

I

-

/

10

~

I

OVI

-

fl

(12)71

I

~

-

800

/

1000

NV

\

CIV

L~

t\0W3

1200

Hell

1400

I

NV

0111]

-

NW]

1600

1800

I

~

1200 1400 Wavelength

CIV

HeIIOffh

1600

1800

(A)

Fig. 2. The two HUT spectra of the Cygnus Loop. The the top panel is the spectrum of the [0111] bright region on the eastern limb of the Cygnus Loop. The bottom panel is of the nonradiative filament in the NE. Although the high ionization lines have similar strengths, the low ionization lines are far weaker in the nonradiative filament because it has a very incomplete recombination zone. The two spectra, flux calibrated and corrected for an assumed reddening of E(B V) = 0.08, are shown in Fig. 2. Both spectra show emission from lines of high ionization, including 0 VI AA1032, 1038, NV AA1239, 1243, CIV AA1548, 1551, 0 V A1371, and Hell )t1640. As was suspected from the Voyager map of the Cygnus Loop, 0 VI does indeed arise both from the bright optical filaments and the fainter Balmer filaments. The two lines of the 0 VI doublet are clearly resolved in the two spectra; the line ratios are close to the ratio of 2:1 expected in the optically thin limit which indicates that resonance scattering does not affect the line strengths too much. The high ionization lines OVI AA1032, 1038, Nv AA1239, 1243, CIV AA1548, 1551, and Hell A1640 have similar line ratios at the two positions. The biggest difference between the two spectra are that the low ionization lines CIII A977, N III A991, CII A1335, Si IV AA1394, 1403, 0 IV A1403, 0 III A1665, and N IV A1750 are much brighter at the optically bright position. One way of illustrating this point, independent of questions of abundances, is to compare the ratios of 0 VI AA1032, 1038: 0 V A1371 : OIV A1403: 0111 A1665. At the optically bright position the ratios are 1.0 : 0.13 : 0.87: 0.66, but at the Balmer-dominated position they are 1.0: 0.05: 0.04: <0.01. There is almost no recombination zone at the Balmer-dominated shock position. —









The spectrum of a radiative shock is basically that of a recombining plasma. The highest ionization lines arise in the highest temperature plasma close to the shock front. The higher the shock speed the higher the ionization potential of the lines in the plasma. In the HUT spectra! range, the three highest ionization lines are the ls-2p transitions of the Li-like lines of CIV, NV, and 0 VI. The turn-on speeds for these lines are 110, 140, 1above 160 km this s~.velocity (See, e.g. line lines rises the /36/.) strengthEach of these sharply over a velocity range of about 30 km s~ saturate because the shocked gas is ionized beyond the appropriate ions in the hotter portion of the

(12)72

K. S. Long

Observed

Ratios vs. Hartigan

12—’



010]:

120

.9

6

140

160

180

~‘:

.8

120

140

160

180

Velocity (km/s)

200



-

-1:

120

200

-~

et al. (1987) Models

140

160

180

200

‘::1~’_’_::~’__’~_~:—

120

140

160

Velocity

180

200

(km/s)

Fig. 3. A comparison of the lines in the HUT spectrum of the optically bright position on the eastern limb to models of complete shocks. These comparisons suggest that the shock velocity is about 170 km s—i. postshock flow. As is shown in Fig. 3, the importance of observing below Lyct can be illustrated by comparing the line ratios from the radiative shock position to line ratios for complete shocks. The highest ionization line in the IUE range is N V. Notice that the N V: CIV ratio has saturated at 160 km s~which means that lifE (and HST) observations cannot be used as a discriminator of shocks with velocities greater than about 160 km s~. On the basis of this type of comparison, as well as a more detailed comparison with the models, Blair et a!. /33/ conclude that the shock velocity of the radiative ifiament observed with HUT had a shock velocity of about 170 km s~. Complete shock models are not appropriate for comparison with spectra of the Balmer-dominated filament. Long et al. /35/ constructed models for the nonradiative shock observed with HUT using a 30% neutral preshock fraction, and a column density cutoff of 1Q17 cm2. They found that models with shock speeds between 175 and 185 km s1 matched the data well. The HUT data are equally well fit by models that are based on rapid equilibration of the electron and ion temperatures behind the shock and models that assume equilibration proceeds via Coulomb processes, although the “best” shock speeds are slightly different, 179 km s1 for instant equilibration and 184 km s1 for Coulomb equilibration. The IJV spectrum is not very sensitive to how the electron and ion temperature equilibrate because the UV lines, especially the high ionization lines, are produced far enough behind the shock front that the electron and ion temperature are very close to one another even for Coulomb equilibration. The shape of the Balmer lines in nouradiative filaments have two components, a narrow component which arises from neutral H atoms that pass through the shock front and emit Balmer photons before they are ionized and a broad component which is produced by fast neutral hydrogen atoms behind the shock /27/. These fast neutrals arise from a charge exchange reaction between atoms entering the shock and protons behind the shock front. The width of the broad component reflects

Sub-Lyman-re Observations

(12)73

the post shock temperature. Hester, Raymond & Blair /37/ have recently remeasured the width of the broad component of Ha near the Balmer-dominated position observed with HUT. They find that the width is 135 ±5 km s1 (FWHM). In a collisionless shock shock, all of the particles emerge from the shock front with the same random velocity which means that the ion temperature is much greater than the electron temperature. Equilibration raises the temperature of the electrons and lowers the temperature of the ions. At a fixed shock velocity, the predicted postshock proton temperature for instant equilibration is a factor of 2 lower than for Coulomb equilibration. As a result, the width of the broad component will be lower by ~ for Coulomb equilibration. Smith et a!. /38/ have recently calculated the expected FWHM for the broad component of Ha assuming instant equilibration and Coulomb equilibration. Their calculations show that an observed FWHM of 135 km s~corresponds to shock speed of 135 km ~ for Coulomb equilibration and 170 km s1 for instant equilibration. Since the HUT data imply an “average” shock velocity of about 180 km s~1,the simplest way to reconcile the HUT data and the Ha line width is to argue that we are looking at a fully equilibrated 175-185 km s~shock. Although this interpretation is appealing, it does not correspond to today’s conventional wisdom because no mechanism has been identified that would cause the plasma to equilibrate rapidly. Instead, theoretical calculations /39/ suggest equilibration is modest (Te 0.22T). Furthermore, comparison of models with observations of other SNRs with higher velocity nonradiative shocks provides fairly strong evidence against full equilibration models at higher velocities 1000 km s~ to 2000 km s~ /40/. —

An alternative possibility, discussed by Long et a!. /35/, is that the shock in the NE Cygnus Loop is actually slowing down, perhaps as a result of the fact that it is encountering denser gas. They estimate that the preshock density at the position of the Balmer-dominated filament is 8—12 cm3. For this density, the 0 VI is generated about 200 years after material encounters the shock. To slow the shock from 180 to 140 km s~1the density must increase by a factor of about 1.65 in that time. Models of decelerating shocks do not exist. In a relatively crude attempt to construct the expected spectrum, they summed the emission from steady flow models, taking the emission from the first 1.25 x 1016 cm2 from a 130 km s~model, the emission from 1.25 x 1016 2.5 x 1016 cm2 from 140 km s1 and so on out to 1.0 x 1017 cm2. Although this model did produce the appropriate O VI: N V and the appropriate Ha line width, it did not fit most of the other lines as well as a single velocity model. Hence, Long et a!. concluded that the HUT data favors rapid equilibration of the shock. More specifically they concluded that to reconcile the HUT data with a non-decelerating shock, T 5 must be at least 0.5 T2. —

Whatever is the correct answer, the shock in the NE must have moved into a region of higher density within the last few hundred years. Otherwise the lower ionization state lines would be brighter. Several hundred years from now, the Balmer-dominated filaments in the north will resemble the bright optical filaments located inside of the Balmer filaments in the NE and on the western limb of the SNR. OTHER SNRs N49, A SNR in the Large Magellanic Cloud HUT was also used to observe one extragalactic SNR N49 /41/. This SNR has the highest optical surface brightness of any SNR in the Large Magellanic Cloud /42/. It is the third brightest X-ray SNR /43/. As a result it is interesting because it is the type of SNR which will be easiest to observe in more distant galaxies. N49 is a relatively young SNR; age estimates between 6,000 and 18,000 years are typical. It is bright because it is a young object interacting with a very dense ISM. N49’s X-ray spectrum is dominated by emission lines and there is no evidence that it contains a pulsar. Vancura et al. (/44/ ) recently completed a very detailed multiwavelength study of N49 using ground based spectroscopy and imaging, IUE spectra, and archival Einstein data. Their basic conclusions were as follows: (a.) Even though there are small scale differences between the appearance at different wavelengths, N49 has a similar overall morphology at all wavelengths that —

JAS 13:12—6

(12)74

K. S. Long

reflects the presence of a molecular cloud on the SE edge of the SNR /45/. (b.) The UV emission is primarily associated with the bright optical filaments and there is no evidence in the IUE data for spectral variations across the face of N49. (c.) Although the X-ray and echelle data indicate that there are shocks whose velocity exceeds 200 km s~1,the bulk of the optical and IUE emission can be explained in terms of shocks with velocities less than 140 km s~. For the HUT observation the 9.4” x 116” arcsec slit was placed along the bright optical emission in the SE portion of the N49. This is the side of the SNR which is interacting with the molecular cloud. The HUT observation of N49 was carried out in orbital day, and as a result, the quality of the HUT data for N49 are not as good as for the other SNR observations. Many of the peaks in the spectrum are clearly airglow, but there are also a number of lines from the SNR. These include the brighter of the two lines of the 0 VI doublet, Si IV AA1394, 1403, 0 IV A1403, CIV )tA1548, 1551, and Hell A1640. Confusion with airglow lines restricted the ability to detect N V AA1239, 1243 and several other lines. Since the HUT aperture covered much of N49, it is naive to expect that the HUT data can be modeled with a single velocity shock. In addition, because reddening varies on a small scale across the face of N49, the effective reddening correction for 0 VI is not easy to establish. Nevertheless, the total luminosity of N49 in 0 VI appears to lie in the range 7—10 x i0~~’ ergs s~1,which is larger than its soft X-ray luminosity, 2 x i0~ergs s~/43/. The X-ray observations suggest blast wave velocities of 730 km s1 and a mean preshock density density of 0.9 cm3 in N49 /44/. It is unlikely that the 0 VI emission in N49 can come from the nonradiative shock associated with the primary blast wave; the expected 0 VI emission from a single shock with this velocity is about 2 orders of magnitude lower than observed. Instead, the 0 VI emission in N49 probably arises from shocks with velocities between 190 and 270 km s1 propagating into plasmas with densities in the range 20 to 40 cm~3/41/. The Crab Nebula Finally, HUT was also used to obtain a spectrum of the Crab Nebula /46/. About 1200 s of good data were obtained at a region near the center of the Crab where the brightest [0111]filaments lie. Unlike the other SNRs that were observed with HUT, the continuum spectrum of the Crab Nebula is dominated by synchrotron emission. (Excellent images of the UV continuum of the Crab were obtained with UIT on Astro-1 /47/). There are no emission lines below Lya although there is evidence of absorption due to molecular hydrogen. Above Lya, CIV AA1548, 1551 and Hell A1640 are detected; C IV shows two resolved components, perhaps reflecting the front and back sides of a shell in the Crab, expanding at velocities of ±1100km s’; Hell A1640 shows only a redshifted component. There are also statistically significant excesses near 1400 and 1750 A, possibly due to Si IV + 0 VI] A1400 and N III A1750. The HUT continuum is reasonably well fit with a power law (f~,cx A°) with an index a = 1.51 and a reddening E(B V) = 0.51. These fits are consistent with the fits to the continuum obtained with IUE and earlier UV experiments /48, 49/. There is no evidence either for a non-standard reddening law or for the steepening of the spectral index which is observed at X-ray wavelengths. The main importance of these observations is, in fact, the variation in CIV AA1548, 1551: Hell A1640, which probably indicates abundance variations in the filaments of the Crab Nebula, but that is a subject for a different review. —

CONCLUSIONS The idealized picture of the ISM first proposed by McKee & Ostriker /50/ in which the primary shock of a SNR propagates into the tenuous volume-extensive phase of the ISM while driving secondary shocks into the denser clouds or cloudlets of the ISM has in some sense allowed optical/UV astronomers and X-ray astronomers to proceed independently. The Voyager, HUT, and rocket observations of the Cygnus Loop and a few other SNRs and the earlier observations with Copernicus are important, in part, because they force us bridge this gap. Observations of SNRs with the WFC on ROSAT and the EUVE will reduce the gap from the other (short wavelength) direction. The Astro UV telescopes are scheduled to fly again in 1994.

Sub-Lyman-re Observations

(12)75

ACKNOWLEDGEMENTS I am grateful to my colleagues Bill Blair and Olaf Vancura who did much, if not most, of the work I have described in this review. HUT is supported by NASA contract NAS 5-27000 to the Johns Hopkins University. REFERENCES 1. E.B. Jenkins, J. Silk, and G. Wallerstein, Astrophys. J. Supp., 32, 681 (1976). 2. E.B. Jenkins, Astrophys. J., 220, 107 (1978). 3. E.B. Jenkins, G. Wallerstein, and J. Silk, Astrophys. J., 278, 649 (1984). 4. J.C. Raymond, G. Wallerstein, and B. Ba!ick, Astrophys. J., 383, 226 (1991). 5. A.L. Broadfoot, B.R. Sandel, D.E. Shemansky, S.K. Atreya, T.M. Donahue, H.W. Moos, J.L. Bertaux, J.E. Blamont, J.M. Ajello, D.F. Strobe!, J.C. McConnel, A. Dalgarno, R. Goody, M.B. McElroy, and Y.L. Yung, Space Sci. Rev., 21, 183 (1977). 6. A.L. Broadfoot, B.R. Sandel, D.E. Shemansky, J. McConnell, G.R. Smith, J.B. Holberg, S.K. Atreya, T.M. Donahue, D.F. Strobe!, and J.L. Bertaux, J. Geophys. Res., 86, 8259 (1981). 7. A.F. Davidsen, K.S. Long, S.T. Durrance, W.P. Blair, C.W. Bowers, S.J. Conard, P.D. Feldman, H.C. Ferguson, G.H. Fountain, R.A. Kimble, G.A. Kriss, H.W. Moos, and K.A. Potocki, Astrophys. J., 392, 264 (1992). 8. T. Stecher et a!., Astrophys. J., 395, Li (1992). 9. K. Nordsieck et a!., in preparation (1992). 10. R.A. Fesen, W.P. Blair, and R.P. Kirshner, Astrophys. J., 262, 171 (1982). 11. J.J. Hester, J.C. Raymond, and G.A. Danielson, Astrophys. J., 303, L17 (1986). 12. J.M. Miller, Astrophys. J., 189, 239 (1974). 13. J.C. Raymond, J.H. Black, A.K. Dupree, L. Hartmann, and R.S. Wolff, Astrophys. J., 238, 881 (1980). 14. J.C. Raymond, J.H. Black, A.K. Dupree, L. Hartmann, and R.S. Wolff, Astrophys. J., 246, 100 (1981). 15. J.C. Raymond, J.J. Hester, D. Cox, W.P. Blair, R.A. Fesen, and T.R. Gull, Astrophys. J., 324, 869 (1988). 16. P. Benvenuti, M. Dopita, and S. D’Odorico, Astrophys. J., 238, 601 (1980). 17. W.H.-M. Ku, M. Kahn, R. Pisarski, and K.S. Long, Astrophys. J., 278, 615 (1984). 18. R. Braun and R.G. Strom, Astron. Astrophys., 164, 208 (1986). 19. N.J. Keen, W.E. Wilson, C.G.T. Haslam, D.A. Graham, and P. Thommasson, A8tron. Astrophys., 28, 197 (1973). 20. J.J. Hester and C.P. Cox, Astrophys. J., 300, 675 (1986). 21. R.G. Teske, Astrophys. J., 365, 256 (1990). 22. P. Shu!! and H. Hippelein, Astrophys. J., 383, 714 (1991). 23. J.C. Raymond, W.P. Blair, R.A. Fesen, and T.R. Gull, Astrophys. J., 275, 636 (1983). 24. R.A. Fesen and H. Itoh, Astrophys. J., 295, 43 (1985).

(12)76

K.S. Long

25. R.A. Chevalier and J.C. Raymond, Astrophys. J., 225, L27 (1978). 26. C.F. McKee and D.J. Hollenbach, Ann. Rev. Astron. Astrophys., 18, 219 (1980). 27. R.A. Chevalier, R.P. Kirshner, and J.C. Raymond, Astrophys. J., 235, 186 (1980). 28. D.E. Shemansky, B.R. Sandel, and A.L. Broadfoot, Astrophys. J., 231, 35 (1979). 29. W.P. Blair, K.S. Long, 0. Vancura, and J.B. Holberg, Astrophys. J., 374, 202 (1991). 30. F.D. Seward, Astrophys. J. Supp., 73, 781 (1990). 31. 0. Vancura, PhD thesis, Johns Hopkins University (1992). 32. A. Rasmussen and C. Martin, Astrophys. J., 396, L103 (1992). 33. W.P. Blair, K.S. Long, 0. Vancura, C.W. Bowers, A.F. Davidsen, W.V.D. Dixon, S.T. Durrance, P.D. Feldman, H.C. Ferguson, R.C. Henry, R.A. Kimble, G.A. Kriss, J.W. Kruk, H.W. Moos, and T.R. Gull, Astrophys. J., 379, L33 (1991). 34. R.H. Cornett, E.B. Jenkins, R.C. Bohlin, K.-P. Cheng, T.R. Gull, P.M. Hintzen, R.W. O’Connell, R.A.R. Parker, M.S. Roberts, A.M. Smith, E.P. Smith, and T.P. Stecher, Astrophys. J., 395, L9 (1992). 35. K.S. Long, W.P. Blair, 0. Vancura, C.W. Bowers, A.F. Davidsen, and J. Raymond, Astrophys. .1., 400, 214 (1992). 36. P. Hartigan, J. Raymond, and L. Hartmann, Astrophys. J., 316, 323 (1987). 37. J.J. Hester, J.C. Raymond, and W.P. Blair, in preparation (1992). 38. R.C. Smith, R.P. Kirshner, W.B. Blair, and P.F. Winkler, Astrophys. J., 375, 652 (1991). 39. P.J. Cargill and K. Papadopoulos, Astrophys. J., 329, L29 (1988). 40. R.C. Smith, M.L. Laming, and J.C. Raymond, in preparation (1992). 41. 0. Vancura, W.P. Blair, K.S. Long, C.W. Bowers, A.F. Davidsen, W.V.D. Dixon, S.T. Durrance, P.D. Feldman, 1I.C. Ferguson, R.C. Henry, R.A. Kimble, G.A. Kriss, J.W. Kruk, and H.W. Moos, Astrophys. J., 401, 220 (1992). 42. M.A. Dopita, Astrophys. J. Supp., 40, 455 (1979). 43. K.S. Long, D.J. Helfand, and D.A. Grabelsky, Astrophys. J., 248, 925 (1981). 44. 0. Vancura, W.P. Blair, K.S. Long, and J.C. Raymond, Astrophys. J., 394, 158 (1992). 45. J.P. Hughes, L. Bronfman, and L. Nyman, in: Supernovae, ed. S.E. Woosley, Springer-Verlag, New York, p. 679 (1989). 46. W.P. Blair, K.S. Long, 0. Vancura, C.W. Bowers, S. Conger, A.F. Davidsen, G.A. Kriss, and R.B.C. Henry, Astrophys. J., 399, 611 (1992). 47. G.S. Hennessy, R.W. O’Connell, K.P. Cheng, R.C. Bohlin, N.R. Collins, T.R. Gull, P. Hintzen, J.E. Isensee, W.B. Landsman, M.S. Roberts, A.M. Smith, E.P. Smith, and T.P. Stecher, Astrophys. J., 395, L13 (1992). 48. C.-C. Wu, Astrophys. J., 245, 581 (1981). 49. K. Davidson, T.R. Gull, S.P. Maran, T.P. Stecher, R.A. Fesen, R..A. Parise, C.A. Harvel, M. Kafatos, and V.L. Trimble, Astrophys. J., 253, 696 (1982). 50. C.F. McKee and J.P. Ostriker, Astrophys. J., 218, 148 (1977).