So… it seems Sunday’s blog post on Tabby’s Star (KIC 8462852) was a bit premature- the argument over whether or not the star ITSELF is doing crazy things is far from over.
To quickly recap: Bradley Schaefer of Louisiana State University used the DASCH collection of digitized photographic plates and found that the star KIC 8462852 itself had dimmed by 0.16 magnitudes (almost 20%) over the past 100 years. It was an incredible discovery, because stars like Tabby’s Star shouldn’t do that.
Then Michael Hippke (not sure who he’s with, though he lists the Institute for Data analysis in Neukirchen-Vluyn, Germany), and Daniel Angerhausen (a NASA postdoctoral fellow at Goddard) responded with a paper that showed this was all probably problems with the data. Good so far?
Then Bradley Schaefer fired back with a blog post, (granted, this came out in January, but I missed it) pointing out that Hippke and Angerhausen had made bad data selections and poor decisions about which data were considered good… and they were unfairly maligning the accuracy of DASCH’s data as a result.
Well, now Hippke and Angerhausen have responded to THAT. On Monday, they uploaded a new revision of their paper (actually, it’s almost a completely new paper – 9 pages long instead of 3, AND it references their first draft as if it was a separate item) with many, many more details on their process.
The end result is the same: They find that the steady drop is more likely to be a single jump right after the Menzel gap, when Harvard started observing that area of sky with a single telescope and (apparently) photographic plates that were sensitive to slightly different colors of light. There’s a great plot in the paper (Figure 3 on page 5) that explains exactly what they’re talking about. And they still find similar trends in other F-type stars for the same reason.
More than that, the paper addresses all the items in the blog post almost point-by-point, with facts and figures. They defend their “beginner’s mistakes” by citing DASCH publications that did the same things, explain that they can’t reproduce Schaefer’s data selection using his own selection criteria (they don’t retrieve exactly the same number of data points), question the criteria he used to begin with, and point out that DASCH itself does not claim its data is precise enough to have seen the trend Schaefer had found. They also promise to try this with the world’s second-largest catalog of astronomical plates, the Sonneberg Observatory collection.
Hippke’s paper is a masterful display of how to improve a paper and respond to critics. We’re basically watching peer review in real time, so I’ll update this with Bradley Schaefer’s response if and when it happens.
Update (2016.0223): It’s not Bradley Schaefer, but it’s an update. Michael Hippke, Daniel Angerhausen, and three new collaborators – Michael Lund, Joshua Pepper, and Keivan Stassun – have published a THIRD revision of their paper rebutting the dimming of Tabby’s Star. It no longer cites previous versions of itself, and is only 8 pages long.
This new version is missing most of the part where they checked other F-type stars in DASCH and found that they also showed dimming or brightening trends – that seems to have been split off into another paper (Lund et al. in prep).
The new version gains an analysis of Tabby’s star from another exoplanet transit survey. SuperWASP (the Super Wide Angle Search for Planets) is an exoplanet transit survey like Kepler, except it’s conducted from two sets of eight computer-controlled 5″ ground-based telescopes at sites in the southern and northern hemispheres. It’s a lot less sensitive than Kepler, but unlike Kepler it covers the entire sky, and it’s been running for a lot longer – SuperWASP started in 2003, Kepler’s main mission ran from December 2009-May 2013. What the paper calculates is that the dimming of Tabby’s Star should be visible over the timespan of SuperWASP data but… it isn’t.
Interestingly, Lund, Pepper, and Stassun are all involved in the KELT (Kilodegree Extremely Little Telescope) transiting exoplanet survey – they have experience dealing with this kind of data, but KELT itself has very small telescopes (1.5″ diameter. They did promise “extremely little”) and doesn’t collect data on stars as faint as Tabby’s Star.