Your thesis in 1978 dealt with HII regions as extragalactic distance indicators. How did you choose this subject, and who was your advisor?
There is an interesting story behind that. During my first couple of years of graduate school my research projects were assigned by my supervisors, but when it came to a thesis my advisor, Paul Hodge, insisted that I put together my own thesis topic, without suggestions from him; he really wanted me to develop a project from scratch. So for a couple of months, I struggled with various project ideas, and one by one Paul would gently shoot them down, explaining why they were not likely to grow into something interesting.
In between those discussions, Paul would also talk about the latest conferences he had attended, and would often bring up the topic of the extragalactic distance scale. At the time Allan Sandage and Gustav Tammann had completed a monumental series of papers on the Hubble constant, which rested in large part on using the diameters of HII regions as standard yardsticks and it was attracting a great deal of attention everywhere because the values of H0 that they measured were much lower than had ever been derived previously. Paul was an expert on HII regions in galaxies and was intrigued by the results but somewhat skeptical of their methodology. I had been working with Paul on other projects using HII regions, so this was a natural topic for us to discuss, but the conversations were always completely separate from the biweekly deconstructions of my thesis ideas.
Finally, during a long break, while Paul was traveling, I came up with the "original" idea to revisit the work of Sandage and Tammann, by obtaining much deeper narrowband imaging of many of their HII regions and using surface photometry to measure physically-based diameters, in place of Sandage's visually estimated diameters. I developed a detailed proposal, and when I was finally able to present it to Paul he loved the idea and congratulated me on the choice (but with the warning that I would be challenging the work of the almighty Sandage and would need to be prepared for the consequences). Thankfully my proposals to Kitt Peak to make the measurements was approved (with Sandage on the time allocation committee!) and the thesis was completed successfully.
For the next few years, I truly believed that the concept for this project was entirely my own, and Paul did his part to encourage the myth. Over time, however, it dawned on me how Paul had been seeding this idea all along, but in a way so subtle that I took personal ownership of the project. To me, he was the ideal in a supervisor, and I have always tried to live up to his standard when working with students myself.
Among the numerous papers from your early years, one, in particular, stands out, namely a 1983 study on the rate of star formation in normal disk galaxies based on Hα fluxes.
If nothing else, my thesis demonstrated that HII regions were not the best way to measure distances to galaxies, and other new methods such as the Tully-Fisher relation offered a much more robust path to the Hubble constant. My research needed to move on.
Along the way, I had realized that the emission-line properties of HII regions and galaxies as a whole could be powerful probes of star formation. My postdoc (a Carnegie Fellowship) exposed me to conversations with a rich community of astronomers in Pasadena, including Sandage, Leonard Searle, John Huchra, Steve Kent, and Frank Israel, and out of those, I developed a plan for a survey of integrated SFRs in nearby galaxies. The observatory generously allocated me a vast amount of time (>100 nights over two years) on the small 0.5m and 1m telescopes at Palomar and Las Campanas Observatories, to measure global Hα fluxes for about 170 nearby galaxies, including funding and custom equipment for filters, special large aperture-plates for the photometers, and even a special red-sensitive phototube for the Palomar work.
The other key ingredient was developed in the early 1980s at the University of Minnesota, where I was working in my first faculty position. Hα luminosities of galaxies clearly scaled with their massive star formation rates but calibrating the relation robustly required constraints for dust attenuation and constraints on the IMF. This was accomplished by writing a simple evolutionary synthesis computer code (it actually ran on a pocket calculator!) to model the broadband colors and Hα emission equivalent widths with galaxies with different star formation histories and IMFs. It was built on previous work by Beatrice Tinsley and Searle, Sargent, Bagnuolo, and Huchra, but with the addition of modeling the photoionization rates and Hα luminosities together with the continuum luminosities and colors. It all came together in that 1983 paper, and it was the first paper I wrote that really attracted widespread attention in the field.
This was shortly followed by a paper comparing the properties of giant HII regions in nearby galaxies with those in our own Galaxy.
During this period Paul Hodge and I were obtaining a large body of deep (for that time) Hα imaging of nearby galaxies for various projects: the distance scale, spiral structure mapping, my SFR surveys, and to study the properties of the HII regions themselves. Paul's passion was to publish atlases of the HII region identifications and positions, because he realized that they would fuel a generation of follow-up imaging and spectroscopy, for problems ranging from chemical abundance surveys to the star formation and evolutionary properties of galaxies.
The paper you mention arose because it became clear by the early 1980s that many researchers were not aware of the extraordinary diversity in luminosities, sizes, and structures of HII regions outside of the Milky Way. Some colleagues thought they were just slightly larger and brighter than the Orion nebula when often they were thousands of times brighter and massive if not more! So to make the point I assembled a diverse collection of some of the best measured HII regions in our surveys, and measured their basic properties (sizes, luminosities, numbers of ionizing stars, mean electron densities, density profiles, etc) on a consistent basis. What people probably remember most about the paper, however, is a multi-panel figure which displayed Hα images of all of the HII regions, from Orion to 30 Doradus upwards, on the same linear scale, to drive home the enormous range of properties. There was not a whole lot of astrophysical insight in the paper --- it was intended more as a sort of Hubble atlas of HII regions --- but it really had a great deal of impact just the same.
In 1989 came your first study on the star formation law in galactic disks. What were your conclusions?
From my time as a graduate student I was fascinated by Maarten Schmidt's suggestion that the star formation rate could be parameterized as a simple power-law function of the gas density (volume or surface density), and subsequently used to derive a simple analytical model for galaxy evolution. Like scores of other observational astronomers at the time, I began to dabble with ways of testing and quantifying this relation, but was repeatedly frustrated by the lack of good data, either decent SFR measurements or lack of complete information on the atomic and molecular gas densities. So a decade went by until finally there was enough information on spatially-resolved HI, CO, and Hα to look at the relation on kiloparsec scales in a handful of nearby galaxies --- that was the paper you refer to.
There were two main conclusions to the study. Within the main gas-rich star-forming disks of the 15 galaxies for which I had data, the azimuthally-averaged SFRs at any radius were fairly tightly correlated with the total surface densities of atomic and molecular gas, with a power-law index of about 1.4. But in every galaxy, there was also a radius where the SFR declined rather abruptly, while the gas disks (at that radius mostly atomic) continued radially without a noticeable turnover, i.e., a star formation threshold. The paper went on to speculate that the thresholds coincided with the onset of gravitational instability in the gas. That interpretation is still debated today (it may be more connected with atomic/molecular phase thresholds, for example), but the basic observational results have held up.
1998 was your annus mirabilis, with an ApJ paper on the global Schmidt law in star-forming galaxies, and an Annual Reviews paper on star formation in galaxies along the Hubble sequence, with so far more than 3000 and 5000 citations, respectively. Please tell me how this came about.
The two papers actually were unconnected --- they just happened to be published at about the same time --- but the core motivations for both were similar. Most of my work on star formation in galaxies up to that time had focused on optical-wavelength measurements of normal star-forming disks, where dust attenuation was significant but not debilitating (typically a magnitude or less at Hα). However, the results from the Infrared Astronomical Satellite (IRAS) in the mid-1980's revealed the importance of a second, high-density, and heavily dust-obscured mode of star formation, preferentially residing in the central regions of galaxies and in ultra-luminous starburst galaxies. A natural question was whether the starbursts represented a physically distinct regime, or whether the normal and bursty regimes were somehow connected?
To address this I collected global SFRs, HI maps, and CO maps for as many normal galaxies as I could lay hands on first to examine the disk-averaged star formation law and do the same for a handful of starburst galaxies and circumnuclear star-forming regions, which were being mapped in CO for the first time. This required developing new SFR calibrations for dusty galaxies based on infrared luminosities because visible indicators (including Hα) were useless in such dusty regions. The short end to the story is that when I plotted the SFR and gas surface densities for both samples they actually lined up, with a common power-law slope close to the 1.4 value I had found in 1989. This gave the theorists and simulators a simple and single powerful "recipe" to plug into their models-- even if we didn't understand its physics-- hence the major impact of the paper. A recent revisiting of the relation, however, suggests that the conclusions drawn in 1998 were probably somewhat oversimplistic.
I was invited to write the Annual Review paper long before the just mentioned work came out. I tried to review just about everything that was known at the time about star formation in normal galaxies and starbursts. Along the way, I compiled a set of recipes for measuring SFRs from Hα, infrared, and ultraviolet measurements, and the convenience of this cookbook probably accounts for a majority of the citations. The paper also came along at the time when one could review the full diversity of star-forming galaxies from the normal ones studied largely in the visible and UV, to the starbursts largely studied up to that point in the infrared and millimeter, presaging today's multi-wavelength approach to studying galaxies. That broad perspective added to its readership and impact as well, or at least so I am told. And showing the first results on the global Schmidt law didn't hurt either.
In 2012 you wrote another Annual Reviews article on star formation in the Milky Way and nearby galaxies.
When I was approached by the ARAA editors about writing this second review, they suggested that I enlist as co-author an expert on star formation on smaller scales in the Milky Way, so we could write a review that would bridge the sub-disciplines, educate extragalactic astronomers about all of the wonderful work that was being done on the Galaxy, and vice versa for the Galactic community. I needed no convincing and immediately recruited Neal Evans for the project. It was a wonderful collaboration, where Neal and I educated, debated, and challenged each other with our views of star formation from different perspectives, and I hope that the fruits of those efforts were borne out in the paper itself.
Just a few months ago you published a paper, together with Mithi de los Reyes, on the integrated star formation law. What is the current status on this subject?
The amazing impact of the 1998 Schmidt law paper was very gratifying in many ways, of course, but at times also a bit scary, because the data on which the paper rests were not the best (by today's standards), and the results tend to be overinterpreted at times. As a result, I always planned to revisit the project when much better data and the right student or postdoc came along to work on it. The data came with the GALEX, Spitzer, and Herschel space missions, along with a series of new HI and CO surveys on the ground, and Mithi (Mia), arrived in Cambridge as a MPhil student just when the data were ripe for analysis. We have published our results so far in two papers. For normal galaxies (Mia's thesis) a clear N = 1.4 Schmidt power law is confirmed. The interpretation of this relation is quite different from 1998, however, as the star formation laws for atomic and molecular gas are entirely different, much along the lines seen in earlier spatially-resolved measurements by Adam Leroy and Frank Bigiel and collaborators. The starburst galaxies still lie on average on an N = 1.5 power-law extrapolation of the law for normal galaxies, but the slope of the relation for starbursts alone is completely different (linear), with a systematic offset to higher star formation efficiencies overall. So nature still holds more mysteries about this law that remain to be solved.
You have had some unique perches from which to survey the advance in astronomy, including being Editor-in-Chief of the Astrophysical Journal and most recently as co-chair of the 2020 Decadal Report on Astronomy and Astrophysics. In terms of galactic and extragalactic star formation, where do you see the new opportunities and challenges?
It has been more than 20 years since I began my term as ApJ editor, and I guess what stands out for me, apart from the enormous growth of the subject over that time, is the gradual coming together of the Galactic and extragalactic sub-disciplines. For decades in the past, the two communities largely worked separately from each other, but now that is changing. A key reason is a gradual recognition that by only studying either what happens within the boundaries of a star-forming cloud or alternatively by only studying the ensemble properties of clouds and galaxies we risk overlooking the most important missing physical piece of the puzzle: how those clouds come into being and how the formation of clouds and the formation and evolution of the stars within and around them are entwined in a mutual feedback cycle. The problem is multi-scale, and our only hope of solving it is to understand the physics connecting all of them. This perspective (not a new one, actually appreciated by Oort as early as the 1950s) has been eloquently expressed in the Cosmic Ecosystems theme of the Pathways to Discovery Decadal Survey.
A second driver for this unifying approach is our newly acquired capability to measure star formation in other galaxies with fidelity until recently only attainable in Galactic surveys, for example with the ALMA/ESO/IRAM/HST PHANGS survey, led by Eva Schinnerer and collaborators. Such observations should make it easier to connect the observed patterns and scaling properties of the young stars and gas to the underlying physics. Finally, computational advances are making it possible to construct numerical simulations of the complex physical processes on these same critical scales. As far as we have come over the past 20 years, I can't wait to see what we learn from these kinds of studies in the coming years.