UC Berkeley: A Very Small HI Cloud
My solution was to find a project that I might be able to do, but one in which I would obviously need the help of my colleagues. Again I was lucky. Observations with the new technique of very long baseline interferometry (VLBI) were yielding new insights because it provided far greater resolution than most other techniques. A group of astronomers from Caltech wanted to use their 130-foot antenna in the Owens Valley together with our 85-foot dish in Hat Creek to observe quasars at 20 cm wavelength. Linking the two systems together to form a single instrument leads to great complexity both in the observations and their reduction. The primary requirement is that observations at the two sites are carried out with a time standard that is as close to identical as possible. Data is accumulated very rapidly and is stored on magnetic tape. The use of two-inch wide tape reduced the number of reels required to record it all Sources are observed throughout the time that they lie above the horizon of both observatories, thus requiring a team of observers at each of them.
The "magic" of this instrument is that the size of its beamwidth approximates that of an antenna with a diameter equal to the separation of the two observatories, resulting here in a resolution on the sky of 0.09" of arc. Of course the sensitivity is determined only by the total area of the two antennas -- one does not quite get something for nothing. As we were planning for the observing time I made a bargain with the Caltech team. We would help with their observations if they would help us make our own. Since their chosen wavelength was so close to that of the hydrogen line, my idea was to use the high resolution to observe interstellar neutral hydrogen in absorption in front of the quasar 3C147. No observations of the gas at such high resolution had ever been attempted, largely because no one expected such small scale features in the neutral gas.
I saw no harm in trying, but did not realize how much study of a spectral line would complicate matters compared to continuum studies, although I should have. In my observations details of the frequency structure of the signal were of prime importance, and the additional variable meant further demands on the system both in observing and handling the data. For example, each receiver used 100 separate channels, each isolating a narrow band of frequencies, allowing determination of the shape of the line profile. In the first observing run in November 1973 the frequency resolution was 10 kHz, but that proved to be inadequate. So we tried again in July 1974 with 5 kHz channels, and that did the trick. After each of these observing runs I took the data to the National Radio Astronomy Observatory in Charlottesville, Virginia to put it through the only existing "processor". This device, clearly still under development, provided the first step in the lengthy process of converting raw data into meaningful results. Also, it was here that the crucial question "Did the system work?" was answered, providing for any observer some tense moments. This processor still required active participation by its attendant engineers. My principal contribution was cheerleading and running up and down stairs (for what purpose I now don't remember).
In order to detect small scale structure in the interstellar hydrogen I needed a background source that was known to have shown 21-cm absorption lines, and one made up of distinct and closely spaced components. I tried several candidates, but only the quasar 3C147 yielded positive results. It is made up of several small components, the major ones 0.16" apart and therefore resolved by our instrument. The hydrogen-line absorption spectrum in the direction of 3C147 observed at low resolution shows two clouds, at velocities of 0 and -8 km/s. Both these clouds appear in our high-resolution observations, but there is a significant difference between the two. The zero velocity cloud covers one of the features within 3C147 but not the other 0.16" away. We may, of course, be seeing the edge of a large feature that happens to lie between the two sources. (This seems unlikely, but until more sources can be observed remains a possibility.) The alternative, that we are seeing a very small cloud having an angular size of only about 0.1" of arc, is the interpretation I explored.
In order to deduce the nature of this cloud in spite of the limitations of the observations, I had to make several (I hope reasonable) assumptions. I found the linear diameter to be at most 3 x 10-4 pc or 70 AU (AU = diameter of the earth's orbit)!! The last paragraph in the paper on this research describes it best.
"The cloud described here is smaller by several orders of magnitude than any neutral hydrogen cloud previously observed; its size and density [105 atoms/cc] are more similar to those associated with dense molecular clouds. Speculations on the origin of such clouds, their place in the interstellar medium, and their influence on star formation depend critically on how large a fraction of the interstellar gas exists in this form."
The paper was published in June 1976, and immediately afterwards I began planning for further observations. This time we were able to use three antennas as our interferometer -- Hat Creek and Owens Valley in California and the National Radio Astronomy Observatory in Green Bank, West Virginia. This latter of course added a 3000 mile baseline. Scheduling the three antennas took some time, but the observations were made in early 1977. By now a new "processor" was operating at Caltech and it was there I scheduled time for examining our new material. Unfortunately, by the time I had finished there the date for my early retirement was imminent and I, therefore, had to leave the material with a colleague for final reduction. Still more unfortunately, he never did anything with it.
That however was not the end of the story. Many years later, 13 in fact, I heard from my old friend and colleague, Miller Goss, that he (and collaborators) had observed again the neutral hydrogen absorption in front of 3C147 with three European VLBI antennas. They essentially verified our conclusions and in fact extended them to include two other sources, 3C138 and 3C380, where they were able to detect still smaller clouds, 25 AU in size. Miller, with other collaborators, in 1998 was able to produce maps of neutral hydrogen in front of three sources and again found very small-scale structure. Other astronomers added further information by using six pulsars as the background sources, and here small-scale HI structure was found in every one, suggesting that this AU-scale structure is common throughout the galaxy. (In the diffuse molecular gas, in particular H2CO, structures of this size and density have now also been found.) Explaining the presence of these very dense, small condensations in the interstellar gas remains a problem. One possibility proposed is that we are seeing not spherical clouds but a cross-section of filaments, but there is no direct evidence to support the idea. The model reminds me of conversation I had at lunch one day long ago with Rudolph Minkowski, who was at the time a senior fellow in the radio astronomy laboratory at Berkeley. As I put cream in my iced coffee (producing the usual swirling pattern), he said, "That's how the interstellar medium really looks." When I had stirred the coffee to a uniform brown he added, "And that's how we see it."