This is a new update of an interview from August 2014.

Q: Your early papers had nothing to do with young stars. How did you develop your interest in star formation?

A: As a grad student, I had originally been working on the theory of extended stellar atmospheres and winds with Joe Cassinelli at Wisconsin, but decided I needed some observational experience and worked with Chris Anderson on spectra of flare stars. The resulting thesis wasn’t very good but it introduced me to solar-type magnetic activity.

Then I went to CfA as a postdoc and began working with Andrea Dupree on International Ultraviolet Explorer satellite observations of all kinds of objects - globular clusters, X-ray binaries, cataclysmic variables, you name it, along with John Raymond. My major focus eventually became chromospheric and coronal emission of solar-type stars, which then led to my interest in T Tauri stars, which were supposed to be the most magnetically-active late-type stars... but that isn’t the big story.

Q: In the early 1980s you wrote a series of highly influential papers with Keith MacGregor on centrifugally and wave-driven winds.

A: Keith had worked on rotating, magnetized solar-type wind models for his thesis, and so we applied these models to the very massive outflow in Orion-KL, showing that a magnetized protostellar cloud rotating at breakup could drive a massive wind. But we did this for equatorial flow; at roughly the same time the Blandford and Payne and Pudritz and Norman papers came out on disk winds/jets driven by magnetic fields anchored in the rotating disk. The importance of disks only became clear somewhat later with the launch of IRAS. So we missed out on jets/disk winds, though I think our paper did provide some motivation for the X-wind model of Frank Shu and his collaborators.

I tried using Alfvén wave-driven wind models to explain the Balmer line profiles of accreting T Tauri stars, first with Suzan Edwards and Gene Avrett, and then with Nuria Calvet. The winds could predict emission line ratios fairly well, but the line profiles generally didn’t match the observations. Then in 1989 I was visiting Santa Cruz and mentioned the problem of the slow rotation of T Tauri stars to Doug Lin. Doug suggested that the slow rotation was a result of magnetospheric coupling which transferred excess stellar angular momentum to the disk. Cluelessly, I dismissed this idea because I thought making a magnetospheric hole in the disk should produce a gap in the near-IR infrared spectral energy distribution that was not apparent... just before Arieh Königl’s paper came out on magnetospheric braking. (My mistake was to ignore emission from the directly-illuminated, hotter inner “wall” of the dust disk, which fills in the short-wavelength SED, as shown by Antonella Natta and collaborators.) I went back to Nuria and said, we’ve gotten this all wrong; we should be doing models of (magnetospheric) infall, not outflow. Making this change resulted in very good agreement with many observations, as ultimately developed in some detail in James Muzerolle’s thesis work.

Q: Your first venture into observations of young stars came with your 1982 paper on high resolution spectroscopy of T Tauri stars. Another large observational effort was your study of rotational and radial velocities of T Tauri stars. What motivated you to embark on observations?

A: New detectors became available at Mt. Hopkins based on Steve Schectman’s design that provided the ability to go much fainter than before. I figured it would be useful to have a kind of album of T Tauri line profiles using this new opportunity. In particular, I remember an H$\gamma$ profile of DF Tau where it showed simultaneous red- and blue-shifted absorption, and puzzling over it. Of course now we interpet this as magnetospheric infall interior to an expanding wind.

I think my interest in making a survey of rotation in T Tauri stars stemmed from my collaboration with John Stauffer, who sadly passed away a few years ago. John was interested in understanding the rapidly-rotating solar-type stars in the Pleiades. Probably the idea was to see how initial angular momenta mapped into the Pleiades rotation allowing for contraction to the main sequence and seeing how much angular momentum loss via winds one needed. Along with Jerome Bouvier’s simultaneous study, we showed the slow T Tauri rotation more systematically than had been done before. On a personal note, it engendered about the worst referee report I’ve ever gotten, with 70 paragraphs (I counted) of scathing criticism. I was ready to jump off a bridge until I received a very nice letter from George Herbig, who liked the paper. That meant a lot to me.

Q: In 1985 came your paper with Scott Kenyon on the nature of FU Orionis objects, followed by a series of studies of FUors, and culminating in your Annual Reviews article on FUors. What were the key ideas, and how have they held up with time?

A: I became interested in the FUors when, as a finishing grad student at Wisconsin (the official designation was “terminal”), I went to my first AAS meeting and heard George Herbig’s wonderful Russell Lecture on “Eruptive Phenomena in Early Stellar Evolution”. This paper stuck in my mind for years, so when Scott Kenyon came to CfA and gave a talk on novae, I went up to him afterward and asked if FU Ori could also be some kind of young nova. Scott said he didn’t think so, but it did sound like an accretion disk outburst such as were then being invoked to explain the recurrent eruptions of dwarf novae.

My first thought was to look for the double-peaked absorption line profiles one would expect for a disk, but was having little success taking spectra at 5200 Å. Then Peter Petrov visited CfA and told me about some HIRES red spectra that he and Herbig had taken which showed double absorption lines in FU Ori as if it were a double-lined spectroscopic binary, but the lines didn’t vary in radial velocity. I became very excited, and even had the nerve to cold-call George in Hawaii to tell him he had discovered the signature of an accretion disk. George was noncommittal; as you know, he never liked the disk model, and wrote papers with Peter trying to support alternative explanations. Anyway, we eventually got red spectra with the MMT echelle which showed the line doubling. Scott and I subsequently went to the Kitt Peak 4m FTS and obtained high-resolution 2 $\mu$m spectra which showed slower infrared rotation, as expected for a Keplerian disk.

If I can digress a bit, one of the things I am most proud of from this period is the 1987 paper that Scott and I did on the spectral energy distributions of classical T Tauri stars. It was noticed that the infrared excesses of the disks did not follow the expected power law dependence on wavelength that would be expected for either steady accretion disks or geometrically flat disks irradiated by the star. Frank Shu and collaborators tried to come up with some mechanism by which gravitational heating could produce more heating at larger radii, but the way this would happen didn’t seem clear. What Scott and I showed was that disks would naturally be “flared”, i.e. have photospheres that curved away from the midplane, and this flaring would capture more stellar radiation and produce better comparisons with observations. Although we mostly parameterized the flaring, we actually did perform a numerical integration which turns out to asymptotically match the simple power-law solution of Chiang and Goldreich derived ten years later. I’ll also note as an aside that the prediction of temperature inversions by Nuria Calvet followed closely after in 1991. Today, disk flaring and temperature inversions are of course well-established by a huge set of imaging and spectral observations not available to us at the time.

Q: Your most cited first-author paper, together with Nuria Calvet and collaborators, is the 1998 study $'$Accretion and the Evolution of T Tauri Disks$'$. What were the key new insights?

A: With Erik Gullbring and Cesar Briceño, we had just derived new accretion rates for T Tauri stars from optical-near UV excesses, using the Blue Channel spectograph on the MMT plus the luck of a fabulously transparent sky on one special night at Mt. Hopkins to get spectrophotometry down to about 3200 Å. It seemed to me that we should try to connect the mass flow with the disk masses that had recently begun to be estimated from mm-wave dust emission. We used the similarity solution for viscous accretion disks of Lynden-Bell and Pringle to connect the observed accretion rates with limits on disk masses and radii to try to estimate the angular momentum transport - as parameterized by the “alpha” viscosity. Unfortunately protoplanetary disks are not very viscous. The most popular explanation currently is that magnetic disk winds drive the accretion as originally argued by Ralph Pudritz and Arieh Konigl, but there are significant uncertainties in how well this works because we don’t know enough about magnetic field strengths and the ionization of upper disk layers. If our old models have any current value it is as a benchmark to compare against predictions of disk evolution from more complicated treatments.

Q: Spitzer photometry has been remarkably efficient in characterizing the circumstellar disks of T Tauri stars. You have been involved in a number of disk studies using IRAC, IRS, and MIPS.

A: We were fortunate to have been able to collaborate with the IRAC and IRS GTO teams led by John Stauffer, Lori Allen, Tom Megeath, Dan Watson, and Bill Forrest. I was trying to improve the statistics on disk lifetimes using the IRAC surveys of young clusters in collaboration with the above people, much of it done by Jesus Hernandez. We also used the expanding shells of Cep OB2 to provide independent age-dating at 3 Myr and 10 Myr for Tr 37 and NGC 7160, respectively, in Aurora Sicilia-Aguilar’s thesis. The real advance was in the IRS studies of the pre/transitional disks lead by Nuria, Paola D’Alessio, and Catherine Espaillat showing that T Tauri disks have gaps and holes. Of course, that was only the tip of the iceberg, given the ALMA images of disk rings etc.; and now JWST spectra are exploring disk chemistry in amazing detail.

Q: In 1998 came your widely read book $'$Accretion Processes in Star Formation$'$, which appeared in a second edition in 2008. What motivated you to write this book?

A: Doug Lin suggested writing the book on behalf of Cambridge University Press. I accepted the challenge because I figured it would be a good opportunity to learn about a lot of things I didn’t know. A decade later I thought so much more had been learned that a second edition was needed, but I’m not very satisfied with it. I’ve now made a shorter, cleaner version that I’ll put on the web someday (soon).

Q: In the 2000s you developed theoretical ideas on rapid formation of molecular clouds and rapid star formation, which led to much debate. What are the key issues, and how have these ideas held up?

A: The standard picture in the 70s and 80s was that star formation was slow because the collapse of protostellar cores is regulated by magnetic flux loss via ambipolar diffusion. This would imply that young stellar populations in molecular clouds would have a wide spread in ages due to continuing formation. But it began to be clear, first from studies in Taurus by Herbig, Vrba, and Rydgren, and a bit later by Stauffer, Burt Jones and me that there was no large age spread. As most substantial molecular clouds contain young stars, they have to form quickly after the parent cloud, and on timescales shorter than the cloud crossing time. Then when Javier Ballesteros visited CfA in the late 90s he showed me some old large-scale ISM simulations by Enrique Vázquez-Semadeni and collaborators with stellar energy input driving large-scale flows that formed dense clouds that were far from spherical, but elongated and filamentary; this rendered the crossing time problem moot. This led to the work with Javier, Ted Bergin, and Fabian Heitsch where we showed that initially atomic gas, when swept up, would not be visible in CO until there was sufficient shielding; and by that time, surface densities were large enough that self-gravity would be important and so star formation could ensue rapidly. We then had to postulate that clouds were disrupted quickly.

I think the debate is pretty much over. Almost everyone now accepts that star formation is rapid because of the importance of stellar feedback in rapidly forming and disrupting clouds. The magnetic flux problem has mostly gone away with the understanding that gas collects along magnetic field lines, so that the mass to flux ratio can increase by simply adding mass rather than diffusing field. The “slow” star formation on large scales is really slow depletion of gas; individual formation events are quick but efficiencies are kept low by rapid cloud dispersal.

Q: You and Nuria moved from the Harvard-Smithsonian Center for Astrophysics to the University of Michigan to join the star formation group there. How has that gone?

A: It’s been great. Speaking for myself, I have been blessed to have had a set of fantastic grad students. Zhaohuan Zhu and I wrote a number of papers on protostellar disk outbursts with Charles Gammie, and then on giant planets opening gaps in disks to try to explain the pre-transitional disks (disks with large gaps). Jaehan Bae and I worked on disk outbursts and instabilities, formation of multiple rings and gaps, structure induced by infall to disks - research in various collaborations with Richard Nelson and Zhaohuan. John Tobin started with me studying stellar radial velocities in the Orion Nebula Cluster, leading to the suggestion that infall may still be occurring there. But John then went on to studies of protostars (his first love), including the suggestion that infall to disks occurs along filaments - an early version of the recent observations of infalling “streamers”. Marina Kounkel became one of the leading researchers in observationally characterizing the age distributions and spatial structures of star-forming regions. All of these folks have gone way beyond what they did at Michigan to become world leaders.

Although it hasn’t gotten much recognition, I’m happy with the work with Javier, Fabian, Aleksandra Kuznetsova (and Andi Burkert) on Bondi-Hoyle-type accretion as producing the power law upper stellar IMF (and young cluster IMF). This idea, first suggested by Hans Zinnecker many years ago, had been deprecated for assuming an environment of constant density and velocity dispersion. However, we showed that some dispersion in the environmental properties is actually helpful in producing the power law.

Another nice result is a simulation that Aleksandra and Fabian and I did on protostellar core formation from a global simulation of a molecular cloud. We found that the cores formed with masses higher than the magnetic flux limit by about a factor of three and that the field directions were generally uncorrelated with the accreting material’s angular momentum, which has implications for reducing the amount of magnetic braking that might otherwise prevent disk formation.

Q: What problems are you currently focused on?

A: In my “retirement” I have returned to FU Ori outbursts and related issues. A recent paper with John Tobin uses the new measurements of protostellar masses from disk rotation, along with a guess at “birthline” properties, to argue that the luminosities of many protostars are too low to have evolutionarily-important accretion rates. This reinforces the argument Scott and I made >25 years ago that there truly is a “luminosity problem”. That is, protostellar accretion cannot be steady; whether it is mostly “front-loaded” at very early stages, or mostly in outbursts, or both is still uncertain. I’m working with Zhaohuan and new postdoc Yisheng Tu to see if we can understand the onset and propagation of FU Ori bursts using non-ideal MHD simulations with radiative transfer. Maybe we will finally get some clues as to what is triggering these events...