Date: Fri, 14 Jan 2000 06:01:51 -0800 (PST) From: "David L.Meier" To: george@deimos.caltech.edu Subject: PG history Hi, George. The short pg history follows. When I get back to JPL (early Feb), I will try to send you copies of papers that are hard to find (Weyman 1966 preprint; Sunyaev, Tinsley, & Meier 1978; perhaps the Kaufman papers also). The rest should be easy to find. All the best, Dave Meier ------------------------------- Date: Fri, 14 Jan 2000 06:03:00 -0800 (PST) From: "David L.Meier" To: george@deimos.caltech.edu The history of primeval galaxy research starts some time before all those phenomena we associate with galaxy formation and evolution were known (galaxy mergers, dark matter, inflation cosmology, and even dissipative galaxy formation). 1. The earliest papers. The significance of these papers cannot be overestimated, I think. They were written only a couple of years after the Penzias and Wilson paper, which implied that the universe had a beginning. By inference, therefore, galaxies also not only had a beginning, but it also must have occurred over a fairly narrow span of time. A large fraction of galaxy formation must be visible within some limited redshift range. Each of these papers got some of the features correct and each got some of them wrong. But the basic idea that galaxy formation should be observable eventually held up. a. Partridge & Peebles 1967a, ApJ 147, 868 (PPa). This paper envisioned the bulk of star formation in galaxies to occur at fairly high redshifts (10-30) before they collapsed in a dissipationless "violent relaxation". While the authors only used the spectrum of a hot 30,000K black body to represent a galaxy in rapid star formation, they did recognize that there should be a break at the Lyman limit. Primeval galaxies (PGs), therefore, should appear as rather extended, low-surface-brightness objects in the red or near infrared region of the spectrum. b. Partridge & Peebles 1967b, ApJ 148, 377 (PPb). This follow-on paper recognized that such low-surface-brightness objects might not be seen readily with current instruments, but may be detectable in the integrated optical background light. c. Weymann 1966, preprint only. Generated about the same time as PPa, the PG models in this preprint were, in some ways, closer to reality than the Partridge & Peebles model. I still do not understand why Ray Weymann pulled the paper, since it is as viable a model as the other. There were a few main differences. Both used a hot black body spectrum, but Ray's was a bit hotter, didn't consider the Lyman break that might result, but did suggest that the observable objects might be more collapsed, even stellar in appearance. (I don't recall at this time what the suggested redshift range was.) I will make you a copy of this preprint from the one in my files when I back to JPL. Note that this paper was referred to as late as 1974 Partridge (1974) and by Davis & Wilkinson (1974) below, but eventually was lost and not included in Pritchet's large review. d. Field, 1968, "1975". For the infamous Volume 9 of the "Stars & Stellar Systems" series, it was reported that George Field was going to discuss galaxy formation. Bruce Partridge apparently had a preprint of this paper when he wrote the observing paper discussed below, but I did not, and the publication date of this volume was held up for for a long time, and even afterward it was difficult to get a copy! By the time it finally came out in the mid-late 1970s, the content was quite out-of-date, and I confess that I never checked to see what Field had to say on the subject until recently. There is some basic stuff on the formation of galaxies in Big Bang and Steady State cosmologies, but very little on how they might look to observers. Refers a lot to PPa,b and Weymann for these matters. 2. The first searches. On the basis of the above papers, searches were conducted by Bruce Partridge and by Marc Davis & David Wilkinson. Curiously, they were published back-to-back in 1974 (in the same issue as Larson & Tinsley published a paper on new dissipative galaxy evolution models). These are the first papers of which I am aware that used the term "Primeval Galaxies", although I have not done an exhaustive search. (Before they were called "young" galaxies.) The authors of these papers also spent a little time re-estimating the predicted optical properties of PGs, so they also are sources of the thinking at that time on models of how galaxies might form. Note that there are references to a variety of other papers (e.g., Tinsley 1972; Truran & Cameron 1971; etc.) which discuss galaxy EVOLUTION but not FORMATION. In these cases it is the initial model in the evolutionary sequence that was used to determine the PG properties. Note that both papers briefly mention the possibility of a Lyman break in the spectrum, although in both it is due solely to absorption in the atmospheres of the stars and not in the interstellar medium. Neither address the possibility that the entire galaxy might be black shortward of 912 A and that one must observe in the infrared when the redshift is very high. Note that the limiting fluxes/magnitudes of the experiments are consistent with the current detections of Lyman break galaxies. However, as they were both looking for extended objects, they wouldn't have noticed the Lyman break galaxies anyway, as they are relatively compact, appearing almost stellar to ground-based instruments. a. Partridge 1974, ApJ 192, 241. Bruce did both a photoelectric search, using a chopping photometer in the red (6200-7800 A), and a photographic search, using B, V, and R plates. In both cases he looked for objects > 3" in size. However, his limiting magnitudes were bright by today's standards (~19 and ~22, respectively, for the photometer and plates). b. Davis & Wilkinson 1974, ApJ 192, 250. Amazingly similar to the Partridge paper, these authors also used a chopping photometer and also looked for objects ~5"-30" in size in the range 6200- 8900 A (R & I). Their results are expressed by plotting disallowed regions of the PG luminosity-redshift plane for different cosmological models. 3. The first revolution: dissipative galaxy formation. The first real revolution in galaxy formation ideas was that star formation was recognized to occur at ALL stages in the early formation of a galaxy, not just in one almost instantaneous burst before galaxy collapse. This was due to the discovery of metallicity gradients in the Galaxy and other galaxies, such that the outer regions appeared to have lower abundances than the inner regions. This had two major effects on the modeling of PGs: 1) the trigger for star formation must be associated with the galaxy collapse somehow (the prime candidate being collisions of molecular clouds and gravitational collapse of stars in the cooled shocks); and 2) as the galaxy collapsed, the rate of star formation actually increased to a peak, making the brightest phase a rather compact (nearly stellar-appearing) phase. Using the collapse to drive the star formation also gave a natural time scale for the bright phase (roughly the collapse time) --- something that was rather ad hoc in the 1960s models. Two series of papers, one driven mainly by Richard Larson and Beatrice Tinsley and the other by Michele Kaufman, addressed these issues. Most people know about the Larson/Tinsley series only (of which I was a part): a. Larson 1974, MNRAS 145, 405. Richard was the champion of hydrodynamical collapse models with star formation driven by cloud-cloud collisions. He of course had to assume a rate and efficiency of star formation as a function of collision rate, and this drew some criticism from the community. Nonetheless, his basic approach is still in use today in combined stellar/SPH codes, but of course the details and numerics are quite a bit more sophisticated now than back then. It is these models that Larson & Tinsley (1974, ApJ 192, 293) used to look at present-day evolutionary properties of model and observable galaxies. b. Meier 1976, ApJ 207, 343. This was the first "modern" primeval galaxy paper. Computers had advanced enough that populations could be synthesized, spectra calculated, detailed K-corrections and magnitudes be computed as a function of redshift, and color-color diagrams be generated. I did this work under the supervision of my advisor Beatrice Tinsley and therefore should acknowledge her contribution both in computing the stellar populations and integrated galaxy spectrum and also in guiding the early aspects of the research. The paper should have been Meier & Tinsley, but she was very student-friendly, and encouraged me to publish as a single author. The main emphasis of the paper is on searching for PGs using 2-color diagrams and looking for the Lyman break ("faint stellar objects in the lower left-hand part of the two-color diagram ... essentially selecting objects whose Lyman limits are redshifted to a point somewhere between the two shorter wavelength filters"). Other important results that are still relevant are 1) the UV continuum should be fairly flat longward of the Lyman limit; 2) for \Omega=1 cosmology and slow collapse (Model F, which is more representative of "merging") the redshift of formation should be about 4, red magnitude about 25, half-light diameter of about 0".6, and sky density of ~3000 per square degree (Table 2). Note that the star formation rate for Model F peaks at about 100 M_\sun per year, so it is about a factor of 3 brighter still than the Lyman break galaxies. The only feature missing from this paper that is in present-day PG models is the eating away of the continuum by intervening Lyman-\alpha absorption (the Gunn-Petersen effect). This paper came out before the Sargent & Young papers on quasar absorption lines in the early 1980s, so the effect was unknown at the time. One result of the paper is now known to be in error, or at least over-estimated --- the strength of the Lyman \alpha emission line. Loss of Lyman continuum photons should produce an equivalent number of Lyman \alpha photons. However, resonance scattering plus even small amounts of dust will reduce or even eliminate this line (see Meier & Terlevich 1981, ApJ Letters, 246, L109). c. Meier, 1976, ApJ 203, L103. This is a paper I would just as soon forget, but it does have some significance here. It suggests the identification of two quasars with Lyman breaks as possible primeval galaxies. Clearly the objects are not classical PGs because the linewidths are very broad (~10**4 km/sec), indicating typical quasar broad line regions. But after our work predicting a flat PG spectrum with a Lyman break, when I showed Beatrice these two Lyman break quasars, she exclaimed "they've GOT to be PGs!" So, even if they really weren't, the paper was submitted in order to emphasize the Lyman break technique. Michele Kaufman and Trinh Xuan Thuan (1977, ApJ 215, 11) also had some (negative) comments on my suggestion that these quasars could be PGs. d. Sunyaev, Tinsley, & Meier 1978, Comments on Astrophysics, 7, 183. This paper should have been submitted to a more visible journal, but Beatrice wanted to get some money to Rashid Sunyaev who was cash-starved in the Soviet Union so he could buy a calculator; as "Comments..." paid a little bit to their authors, we agreed to publish there and the money was sent by Jerry Ostriker to Rashid, if I remember correctly, in the form of an already-purchased machine. The paper expands on Meier (1976) by considering ALL wavelengths, not just the UV. A full synthetic spectrum was generated from radio to X-ray. Of course this exercise is not as reliable as the UV-only one, because relative numbers of supernova remnants, binary X-ray sources, etc. depend on what one assumes for their abundance vs. the star formation rate. e. Meier & Sunyaev 1979, Scientific American 241 (November), 130. Because the "Comments..." article was so obscurely published, I wrote the editors of Sci. Am. to see if they were interested in an article about PGs. They agreed, and Rashid and I wrote this by telephone. The phone charges to Russia (which totaled about $250 in those days) were rather steep for Caltech (where I was a postdoc at the time). I appealed to Sci. Am. for another $250 in addition to the $1000 they paid for the article (most of which was sent to Rashid) and they happily accommodated me. So, Caltech was fully repaid for the phone charges, but my reputation as a spendthrift postdoc remained. One regret I have with this article is that we did not put enough emphasis on the Lyman break and the search techniques I had suggested before, even though it is clearly visible in the figures. By this time some doubt had arisen as to whether a lot of dust might heavily redden the UV, perhaps rendering the break largely undetectable. The second series of papers was driven mainly by Michele Kaufman who spent time at Strasbourg, Austin (Texas), and Notre Dame. Last I heard she was at Ohio State. This was an interesting line of investigation and it is a shame that it wasn't pursued more, and published in a more available journal. f. Kaufman, M. 1975, Astrophysics & Space Science, 33, 265. Independent of Richard Larson, Michele developed her own models of collapsing, forming galaxies and their chemical evolution. Her results were similar to Richard's in that the chemical evolution time scale is found to be the collapse time, significantly longer than the supernova time scale. Therefore, she predicted, there should be a significant amount of heavy elements at the galaxy's peak luminosity, and, hence, a significant amount of dust. g. Kaufman, M. 1976, Astrophysics & Space Science, 40, 369. Michele then used her models to predict primeval galaxy properties. Her paper came out about the same time as mine and didn't receive as much attention. Nevertheless, her conclusions still bear listening to today: the relative large amount of dust expected in PGs, might seriously affect the ultraviolet spectrum in many objects. Instead of appearing as Lyman break objects, some PGs might shine only in their (redshifted) thermal dust emission in the sub-millimeter. In the rest frame, these objects would be ULIGs (ultraluminous infrared galaxies), with star formation luminosities comparable to quasars. This paper was written while she was at The University of Texas at the same time Beatrice and I were there! However, we are not mentioned, nor are either of us referenced. There were some strange dynamics going on at that time that, as a starry-eyed grad student, I didn't understand. 4. Second search attempt. Numerous searches for fainter and fainter galaxies were conducted by people like Koo, Kron, and Tyson. However, before the late 1980s, when PG searches increases in frequency, only one other attempt was made to find such objects. a. Koo & Kron 1980, PASP 92, 537. David and Richard looked for both strong Lyman-\alpha line emission AND Lyman continuum breaks using CCDs and slitless spectra, but found neither, even though their sensitivity was almost 2 orders of magnitude better than that of Davis & Wilkinson (1974). For a 25th magnitude object the continuum is, unfortunately, well below their limits and the reason they didn't detect Lyman-\alpha was found out only after their observations (see Meier & Terlevich below). 5. Observational tests of the models at low redshift. Observations of H II regions (both galactic and extragalactic) are a good way to test the results of PG models. We undertook two approaches, one with IUE and one with Voyager UV observations of 30 Doradus in the LMC. Both were done, but the Voyager team never sent me the data for the LMC. (I don't know if it ever was published.) a. Meier & Terlevich 1981, ApJ Letters 246, L109. While I had several Markarian galaxies as candidate extragalactic H II regions, Roberto Terlevich had the best candidates that he had personally discovered. Both the high-metallicity Markarian objects, and Roberto's low-metallicity object 1543+091 were instrumental in deciphering how Lyman-\alpha is produced in such regions: high metallicity produces so much dust that L-\alpha is destroyed by resonance scattering; even low metallicity produces enough dust to significantly reduce the emission line strength from the Case B predictions. Searching for Lyman-\alpha emission is therefore not the method of choice for PG searches. This concludes my short review of the earliest years in primeval galaxy research. Christopher Pritchet (1994, PASP 106, 1052) wrote a good review article on the subject, to which I refer you for more recent information. In that article, he treated Partridge & Peebles (1967) and Meier (1976) very well, but left out much of the other work discussed here (Weymann; Field; Sunyaev, Tinsley, & Meier; Meier & Sunyaev; and the Kaufman papers). I hope reviews published in the near future can rectify some of those omissions. David L. Meier Pasadena, California April, 1999