In his review of the diffuse background Henry (1991) consistently took an extremely skeptical attitude toward claims of the detection of diffuse ultraviolet radiation, and particularly toward works that claimed to understand the physical source of the radiation absent a spectrum. In particular, it was only in the case of the Voyager observation of extended diffuse emission in Ophiuchus by Holberg (1990) that Henry felt that there was a very strong case for the assertion that ultraviolet starlight scattered from dust had been detected. There is now a second very strong case: Murthy, Henry, and Holberg (1993) have detected extremely strong scattered starlight in the direction of the Coalsack nebula (see Figure 7). Detailed modeling shows that this is not light backscattered from the Coalsack, but rather is the forward-scattered light of three very bright ultraviolet stars near the Coalsack. The spectral dependence of this diffuse emission is (as can be seen by our model fit) exactly that of the illuminating early B stars. Unless g is varying with wavelength in such a way as to fortuitously exactly cancel changes in a, we can conclude that the albedo of the grains is as high at 1000 Å as it is at 1350 Å. In fact, exactly the same phenomenon can be seen in Figure 2 of Holberg (1990), where the geometry of the dust relative to the source star is not so clear, but is unlikely to be the same as for the Coalsack observation.
This is an important result, highly relevant to the question of what is happening at high galactic latitudes. Voyager, with its low sensitivity longward of Lyman , could not be expected to detect the background that appears in Figure 1. But it should definitely detect 300 units in the range 1000 Å to 1100 Å if it is there, and furthermore, the data of Figure 7 suggest that if what is being seen at high latitudes at longer wavelengths by many independent observers is starlight scattered from dust, the spectrum should continue strongly down to 912 Å. Voyager shows that it does not.
We have seen, in Figure 1, the Voyager upper limit of 100 units at 1100 Å. The extragalactic radiation field at slightly shorter wavelengths has been measured by Kulkarni and Fall (1993) by applying the proximity effect to Lyman forest lines in the spectra of nearby quasars. They find a intensity of one unit, well below the Voyager upper limit at 1100 Å.
Figure 7. Spectrum of the Coalsack nebula as observed by Murthy, Henry, and Holberg 1993. This is the brightest cosmic diffuse ultraviolet radiation ever reported in the night sky. The radiation is the forward-scattered light of three extremely bright ultraviolet-emitting stars, Cru, Cru, and Cen. A large subtraction has occurred at Lyman (1216 Å). The dark solid line represents our best fit model, after subtraction of inter-planetary lines. Notice that no break occurs in this spectrum of dust-scattered starlight between wavelengths longward of Lyman and wavelengths shortward of Lyman .
Holberg (1986, 1990) and Murthy, Henry, and Holberg (1991) present the evidence for the validity of the Voyager upper limit. There is a great deal more that can be done with the Voyager archive, and Murthy, Henry, Hall, and Holberg have been funded in an archival research program to carry out this project, which is under way. In Figure 12 of his review Henry (1991) indicated that every Voyager diffuse background observation above 20° latitude was only an upper limit. We have subsequently learned (Holberg, private communication) that the Voyager targets selected for analysis included only those that showed no evidence of a signal. That does not change any conclusions: there is still the same considerable number of locations at moderate and high galactic latitudes that show only an upper limit (which does not occur at wavelengths longward of Lyman , Figure 1), and those locations where a signal is present may all contain point sources (the overwhelming majority of Voyager pointings were toward known point sources). The new situation does leave open the possibility, however, that diffuse emission might still be detected at high latitudes near 1100 Å using Voyager. Indeed, if g and a are both large, we would predict a significant signal at high latitudes near the location of bright ultraviolet stars (see Figure 6).
Figure 8 shows the Voyager upper limits of Holberg (1990) and of Murthy, Henry, and Holberg (1991) superposed on a map of the expected scattered light above b = 40° predicted using the model of Onaka and Kodaira (1991). We used = 0.4 csc b in making this plot, with the albedo taken as 0.65 and g = 0.9 (northern hemisphere) and g = 0.8 (southern hemisphere), to illustrate the model. We have evaluated the reduced 2 for these data (treated as detections, each with a standard deviation of 100 units) against this model as a function of a and g. A plot of the reduced 2 appears in Figure 9. An albedo of 0.65 is seen to require that g > 0.8. A sufficiently low albedo will also explain the data.
Figure 8. The shading shows the predicted scattered light as a function of galactic longitude and latitude as predicted using the model of Onaka and Kodaira (1991). We have used albedo a = 0.65, and = 0.4 csc(b), and we have used g = 0.9 for northern galactic latitudes, and g = 0.8 for southern galactic latitudes, simply to exhibit the results of the model.
Figure 9. Reduced 2 for the fit of the Voyager observations of Holberg (1990) and Murthy, Henry, and Holberg (1991) to the sophisticated model of scattered starlight of Onaka and Kodaira (1991). The Henyey-Greenstein (1941) scattering parameter g is plotted against the interstellar grain albedo a. Contours of reduced 2 = 2.0, 1.5, 1.0, and 0.7 are shown. The width of the contour at any point reflects the function's slope.
The new work already done on Voyager data that bears most directly on the present discussion is a new observation by Murthy, Henry, and Holberg of the extended dust patch at high galactic latitudes that was discovered by Sandage (1976). Sandage's Plate 1 shows very clear evidence for dust at b = +38°. Our Voyager observation shows nothing but an upper limit of 100 units, and our measurement is so clean that we have included it as part of our data-reduction template for "no astrophysical signal" (Murthy, Im, Henry, and Holberg 1993). The importance of this observation is that Sandage deduces that AV = 0.3 mag, from the 21 cm observation of Heiles (1975). This translates into an optical depth at 1100 Å of 1.0 mag, using the E1100-V / EB-V of York et al. (1973). Use of the model of Onaka and Kodaira for Sandage's location and value of t shows that for an albedo of 0.65 we require g > 0.9 to explain our Voyager result. (Another possible explanation would be a low albedo for the grains: this of course would also suggest that the high galactic latitude signal at longer wavelengths is extragalactic.) The Sandage region was also scanned at longer ultraviolet wavelengths by Murthy et al. (1989, 1990), and also by Martin, Hurwitz, and Bowyer (1990). No enhancement of background as the line of sight passed over the Sandage region was noted by any of the various UVX spectrometers.