Science Backstage

Two of my closest collaborators are having what passes for a “fight” in astronomy:  my erstwhie postdoctoral adviser (and contemporary in graduate school at Berkeley) Jamie Lloyd at Cornell, and my good friend and colleague (and contemporary student under Geoff Marcy) John Johnson.  (I’ll point out that there are plenty of spats in our field that would pass for a fight anywhere, maybe even on cable TV, but this sort is more common).

john-johnson-sm1.jpg

John’s graduate thesis was about the frequency of planets orbiting intermediate mass stars, those about 1-2 solar masses.  Such stars are very hard to study on the Main Sequence, where they are called “A stars”, rotate rapidly, and have few spectral lines for us to study for Doppler shifts.  There is a brief period at the end of such a star’s life, however, when it is running out of hydrogen, that its core becomes more luminous and the star begins to reorganize itself internally in response to the new energy trying to get out.  Its surface cools, it develops a convective envelope, a magnetic field develops, it spins down, and it starts to look like a very bright version of the Sun.  In other words, a nearly ideal target for Doppler planet searches.  Astronomers call these stars “subgiants”; in a stroke of marketing genius John dubbed them “retired A stars”.
For the first year of his survey, he was sorely disappointed.  The stars had no hot Jupiters that would make for quick papers, and he dispaired that his thesis would be a statistical analysis of the robustness of his null result:  massive stars have few detectable planets.  But then planets started popping up everywhere!  And big ones!  It turns out that the planets orbiting these massive stars have typical orbits longer than one year but just short enough on the timescale of a graduate student’s career to be useful.  John published his gobs of planets, showing that massive stars have even more planets than Sun-like stars, and from there it was a quick hop, skip, and a jump from a fellowship at Hawaii, to a faculty job at Caltech, to the Pierce Prize (given annually by the AAS to the most outstanding young astronomer;  I guess I should point out that John has a lot of other research notches on his belt, too). 

jamie.jpgWhile all of this was happening, I was looking for work.  Fortunately for me, Jamie Lloyd, a professor at Cornell and someone I had known at Berkeley when I was a younger student, needed someone with planet hunting experience to design the survey and planet-fitting software for TEDI, an experimental Doppler instrument at Palomar.  I spent a year and a half at Cornell working with his group including three people I keep in touch with still:  Angie Wolfgang (now a graduate student at Santa Cruz), Barbara Rojas-Ayala, (now at AMNH), and Phil Muirhead, who is now John’s postdoc! Astronomy is a small, small world;  there is a lesson here about getting to know your coworkers.

Anyway, at Cornell I learned a lot from Jaime about science, funding, advising and being a successful astronomer, which is a big part of the reason I got my job here at Penn State.
A while back, Jaime sent me a draft of a paper he was working on about John’s subgiants.  He was puzzled by a few things that didn’t make sense to him:  why were all of the subgiant-hosted planets orbiting stars with masses greater than 1.3 solar masses?  This seemed odd, since 1.2 solar mass stars should be much more common than 1.5 solar mass stars.  Also, why were some of their rotation periods so anomalous?  Also, didn’t these two stars have obviously miscalculated masses?  Finally, 1.5 solar mass stars are really rare, and such stars that just happen to be near the end of their lives (but not at the end, that is, subgiants, but not giants) should be much, much rarer.  We should be surprised if we find many at all.
Jaime wondered if it was possible that the masses were being incorrectly calculated from the isochrones (basically, an isochrone is a list of properties of stars of varying masses but the same age; we use them to determine the age and mass of a star of given composition).  There could be several reasons for this: the models could be wrong (and, indeed, even the modelers admit that the subgiant and giant tracks for solar mass stars could easily be off by a lot), our measurements of the stars’ temperatures could be wrong (but they would have to be VERY wrong), or something else.  The implication was that John’s stars weren’t actually as massive as he thought.  Jaime speculated about why they might show so many planets:  perhaps those planets were actually false positives caused by the stars’ atmospheres, and the lack of short period planets was due to tidal capture (the star “ate” the planet).  He called his paper “‘Retired’ Planet Hosts: Not So Massive, Maybe Just Portly After Lunch,” and after some back and forth with me and sending a courtesy copy to John for comment, he published it.  Since then, he’s given a about this talk at several places, putting John’s conclusions into doubt throughout the community.
John was never convinced by the paper, but eventually enough people kept asking him about it that he sat down and thought very hard about how the masses of the stars might be wrong.  With a student of his, Tim Morton, they decided to focus on two questions, rather than attempt a point-by-point rebuttal of every claim and assertion in Jaime’s paper (and subsequent talks).  The two questions were:  First, based on everything we know about stellar evolution and the Galaxy, do we really expect to have enough bright 1.3-1.5 solar mass stars in the sky for John’s thesis work to have been possible?  If not, then it’s possible that he was actually looking at some other kind of star.  Second, is there any way that we could mistake a massive star for a less massive star, assuming the models and measurements were correct?
Tim used a program called TRILEGAL to synthesize a Galaxy full of stars and then replicated John’s thesis’s target selection, picking out apparent subgiants using John’s actual search criteria and then looking at their actual masses.  They then performed a Bayesian analysis to determine what the best estimate of the star’s mass should be given measured temperature and abundances, taking into account the fact that massive stars are rare and don’t spend much time on the subgiant branch.  They determined that there was no way to be fooled:  less massive stars are too different from massive stars to be confused for them, and there are plenty of bona fide 1.5 solar mass subgiants bright enough for John to have studied in his thesis.  This careful analysis will lead to a downward revision of the masses of John’s stars, but only by about 5%.
John sent me a copy, and after several readings and back-and-forths about the statistics and interpretation and structure he added me as an author.  We sent a copy to Jaime, and we had a few back-and-forths with him that helped us properly characterize his claims and focus our critique.  
The paper, “Retired A Stars: Truly Massive, Against All Odds”, is available here.  The bottom line is that we suspect that part of Jaime’s error was not accounting for the Malmquist Bias (which John insightfully compares to the Infield Fly Rule in a must-read post on his blog, here).  Also important to note, but not in this paper, because it’s coming out later, is that stars in the 1.1-1.3 solar mass range tend to be too hot as subgiants to make good Doppler targets (which explains the strange distribution of planet host masses, which seem to have too many 1.5 solar mass stars compared to 1.2 solar mass stars).  

It’s been fascinating to be backstage on both sides of a scientific dispute, and to move from “umpire” to “participant”.  I encourage you to read John’s blog post to contrast it with the perspective of being on only one side.  About the “fundamental truth” of the masses of the stars from his thesis, he writes:
That fundamental truth didn’t care about my career or my pride or my dreams. As a scientist I had to step outside of myself, set my pride aside, and seek out the truth.
While you’re there, subscribe to Mahalo.ne.Trash!

One thought on “Science Backstage

  1. hPfcU

    255664 527065Hey I was just searching at your web site in Firefox and the image at the top with the link cant show up properly. Just thought I would let you know. 477820

Comments are closed.