Category Archives: Uncategorized

Przybylski’s Star II: Abundance Anomalies

Last time we talked about Ap stars in general. Now, let’s get to the really weird part.

In cool stars (hotter than type M), most of the lines by number are from the element iron. This is because of two “accidents”. The first is that due to the rules of quantum mechanics, iron’s six(!) outer (“active”) electrons have a lot of ways they can be excited. In fact, the 3d shell alone has 120 substates that form 25 distinct spectroscopic states, each with its own energy level. The second is that due to the rules of nuclear physics, iron is the end product of runaway fusion in stars, and so gets spewed out in great quantities in supernova explosions. The result is a large number of lines with relatively large strengths. In fact, until Cecilia Payne’s PhD thesis showed otherwise (Best. Thesis. Ever, by the way) this was sometimes taken as evidence that the Sun was mostly iron (it’s mostly hydrogen in fact, no matter what “iron sun guy” tells you.  Stellar astronomers know who I mean.)

We might expect, then, that most of the lines in Ap stars would also be iron, though perhaps of the ionized variety (since they’re too hot for neutral iron).  Not so; in fact for a long time Antoni Przybylski wondered if his eponymous star even had any iron; its abundance is down by a factor of at least an order of magnitude from the Sun’s.

Instead, Przybylski found lots of other elements in his weird star: strontium, lanthanum, cerium, praseodymium, neodymium, europium, gadolinium, terbium, dysprosium, holmium, erbium, thulium… stuff you rarely hear mentioned outside of a Tom Lehrer song.

Now, these things should be only present in the tiniest of abundances, not the most easily seen lines in the atmosphere! What’s going on?

The answer seems to be levitation, although it appears we still don’t have a great model for it.

Here’s the idea, as I roughly understand it: The strong fields, high temperatures, and still atmospheres of Ap stars combine to create a situation where the atoms in the atmosphere, being mostly ionized, get stuck to the field lines and sort themselves out by element. This is because they can only move in one direction, along the field lines, and the strong radiation from the hot star provides a sort of upward force on the atoms, depending on the specifics of how they absorb photons (which depends on how many electrons and protons they have—that is to say, what element they are). I’m fuzzy on the details (I think maybe everyone is), but the bottom line is that the ions of a certain element get concentrated in a thin (or at least high) layer in the atmosphere.  If this layer is high enough, we see it in absorption and the element looks much more abundant than it really is (because we usually assume the atmosphere is well mixed, not stratified). And if this layer is low enough, we might not see the element at all!

So that’s apparently what puts the “p” in “Ap,” the bulk star does not have weird abundances, but its upper atmosphere does because the upper layers of the star are differentiated and stratified!

But that’s not what’s so weird about Przybylski’s star.  No, that star doesn’t just have weird abundance patterns; it has apparently impossible abundance patterns.  In 2008 Gopka et al. reported the identification of short-lived actinides in the spectrum. This means radioactive elements with half-lives of order thousands of years (or in the case of actinim, decades) are in the atmosphere. 

What?! The only way that could be true is if these products of nuclear reactions are being replenished on that timescale, which means… what exactly?  What sorts of nuclear reactions could be going on near the surface of this star?

There are three proposed solutions I’m aware of.  The first is about 8 years old; the second is brand new and a “huge if true” sort of exciting idea. The last is quite fanciful and has never, so far as I can tell, gotten past a journal referee (if anyone’s even tried to publish it), but sort of dials the “huge if true” up to 11.

More on these next time.

Przybylski’s Star I: What’s that?

OK, a new slow blogging.  This one in three parts.

Przybylski’s Star is my favorite astrophysical enigma (this coming from the guy notorious for making Tabby’s Star famous!)  It is occasionally mentioned as a SETI target, but usually in private conversations or irreverent asides on social media. I’m not sure when I first heard about it, but it may have been when looking at the Wikipedia page for stars named after people while doing research for that other star.

Przybylski’s Star is famous for having bizarre abundance patterns.  Not like: “oh look, the carbon-to-oxygen ratio is greater than one”; more like this star has more praseodymium than iron. Yeah.  How could that be?!

First off, let’s start with the basics.  Przybylski’s star is an Ap star.  That’s not “app”, that’s “Ay-Pee”, as in spectral type “A”, with a note “p” meaning “peculiar” (which is an astrophysical understatement.)

Normally, A stars are pretty boring, spectrally.  Cool stars, like the sun, a G star, have convecting envelopes (and atmospheres) like a boiling pot of water; this drives a magnetic dynamo that gives the Sun and other cool stars their magnetic fields. Hot stars (spectral types mid-F and hotter, including A stars) have radiative envelopes, meaning that most of the outer layers are very still, like the water in a bathtub left untouched for days. This means no dynamo.

Apparently, this figure from the Wikipedia page for magnetic braking of stars is supposed to illustrate how it works.

As stars form from collapsing clouds of gas, they spin up, going faster and faster.  When they finally “turn on” they can be spinning so fast they they are near break-up speed: the centrifugal force of the spinning is enough to make them oblate.  Cool stars have magnetic fields, and as the star’s ionized outer atmosphere escapes in a wind, the wind particles get stuck on these field lines and fly down them like beads on a spinning wire.  The fields, being anchored to the star’s surface, impart some of the star’s angular momentum onto the particles, and the star slows down ever so slowly. Over billions of years, the star spins down to rotation speeds of more like once per month.

But hot stars lack this field! As a result, they never spin down. The rapid rotation greatly Doppler broadens the lines: the light from the approaching limb is moving towards us very fast, so its light is blueshifted (just as the receding limb is redshifted). The result is that the “missing colors” that characterize the elements in the star’s atmosphere (the absorption lines) get smeared out a lot.  In fact, the only lines broad enough not to be smeared out beyond recognition are usually the hydrogen lines (the Balmer lines), which are particularly strong in A stars which are hot enough to excite hydrogen but not hot enough to completely ionize it.  (In fact, A stars are so named because in Williamina Fleming’s stellar classification scheme they came first, having the strongest Balmer lines.)

But Ap stars break all the rules.  They have intensely strong magnetic fields, and as a result they don’t rotate fast (presumably having slowed down long ago), and as a result they have very narrow lines, and as a result you can see all of the spectral features of the elements in their atmosphere.  Why?

I’ve never seen a good answer as to why Ap stars have strong fields.  They could be primordial or generated from a dynamo, says Wikipedia, which is fine but misses the weird part: regardless of where the field comes from, why do only a minority of A stars have such fields? What’s different about them?

And here’s the even weirder thing: the abundances of the elements that we get to see thanks to the slow rotation are way off of the abundance patterns we see elsewhere in the universe.  Why?

Next time, I’ll discuss likely answers, and then get to the weirdest member of this already weird class: Przybylski’s Star.

Oh, two last notes before moving on.  First, that name. It’s Polish, and despite what a phonetic Polish pronunciation guide might imply, it’s apparently pronounced “shi-BILL-skee”.1  Not as hard as it looks. (And not as hard as some Polish words are to say. Incidentally, first-exoplanet-discover and Penn State Professor Alexander Wolszczan’s name is pronounced “VOLSH-chan”.  Mentally replace the z’s with h’s and you’ll be fine.)

Secondly, the seminal paper on Przybylski’s star gets fewer citations than it should because, as far as I can tell, Nature misspelled his name!  You can find it here in ADS, but not if you search by author! [Edit: The folks at ADS noticed my Tweet about this, and they have fixed it! It is still wrong on Nature’s website, though.]

1Apparently the initial “P” is not completely silent, but it’s so subtle that to untrained US/UK ears it might as well be. I think it’s roughly approximated by having your lips closed when you start the “sh”: the slight plosive as your lips pop open is all it takes, as in the interjection “pshaw“. It’s not “puh-shi-BILL-skee”. Thanks Andrew Przybylski‏ (@ShuhBillSkee) and Jackie Monkiewicz (@jmonkiew) for setting me straight.

Trump and Fascism

To my friends and followers who disagree with or don’t understand the strong reactions Trump gets from across the political spectrum, I offer this:

After WWII, there emerged a liberal consensus in America that has dominated politics and journalism. By “liberal” I don’t mean “Democratic”, I mean the broader understanding that tolerance of differences among Americans was an essential component of the American fabric. Until recently, even the most conservative Republicans understood that speaking out in explicit contradiction of this consensus was political suicide. When liberals warn of the dangers of “normalizing Trump”, they aren’t just talking about his behavior and tone; they also mean his regular, even gleeful violations of this consensus.

These values were imparted to me and many of my generation (and the next) by my parents’ generation, and onto them by the American reaction to the rise and fall of fascism, especially its most evil and visible manifestation in Europe. Certainly some of this came from the inevitable vilification of our wartime enemies, but it was largely justified by the horrors revealed by the post-war occupation of Germany. Family lore is that my maternal grandfather was one of the officers that led American forces through Germany after VE day, and gave ordinary Germans tours of the concentration camps.

This video has been making the rounds, and I encourage you to watch it with this mindset. It is a propaganda video produced by the US government in 1943 in an attempt to inoculate America against the tactics the Nazis used to take control of Germany. Its parallels to recent events are eerie (so eerie I at first suspected that it was an elaborate hoax, but it appears to be genuine, having been used as part of the effort to desegregate the armed forces.)

Today’s left will find much to fault in the video (it is, after all, almost 75 years old) but I’m not offering it as a guide to how you should think. I’m offering it as a good example for the foundational values that many Americans find violated by Trump, and why the analogies to fascist Europe are not made hyperbolically. Trump’s rhetoric and actions are literally, exactly what we were raised to be on guard against.

(A)eon, era, epoch, age

Because I’m a recreational (and occasionally professional) pedant, I was curious if I was using the word “epoch” right in a recent paper. I dug around and got my answer.  For my own and others’ future reference, here’s what I found:

In general or popular writing, the terms era, epoch, and age can be used, without error, synonymously to refer to a particular span of time.

Writers who wish to respect their more precise and historical meanings or who need to respect jargon, can follow these rules:

“Eon” is the preferred spelling; “aeon” is a variant that may appear affected.

In general, an eon is a very long time, comparable to the age of the universe.

An epoch is a fixed point in time (like the zero date of a calendar, or the moment a world-changing event occurred), especially one that marks the beginning of a new era. One can “make an epoch” by doing something that changes things forever.

An era follows an epoch and is defined by it. For instance, the “Christian Era” is the time since AD 1, with Christ’s birth (roughly) making the epoch.

An age is basically an era, but seems to be a bit more general, not necessarily needing an epoch. “The Tudor Age” would then just refer to “the time when the Tudors were in charge,” while “The Victorian Era” carries the connotations of the many things that distinguished the time when Queen Victoria was alive.  This distinction might be too subtle to honor.

In geology jargon, time is divided into eons, then further divided into eras, periods, epochs, and finally stages.

In astronomy jargon, an epoch is the moment of an observation. It most commonly comes up in ephemerides, giving the moment in time that a certain object had or will have certain coordinates or orbital parameters. Not to be confused with “equinox” which specifies the Celestial coordinate system one is using.  “Age” should probably be avoided except to refer to how old something is, to avoid confusion.

And now you know.

AstroWright Science at #AAS229

It’s the “Superbowl of Astronomy” again, this time in Grapevine, Texas, at the 229th meeting of the American Astronomical Society.  You can follow the fun on social media at #AAS229.

AstroWright group members are there in force.  Here they are:

Wednesday, January 4

Jacob as an undergrad at THE Ohio State University

Jacob as an undergrad at THE Ohio State University

Poster #146.30: NSF graduate research fellow Jacob Luhn presents his work connecting flicker with jitter, a project he’s doing under the supervision of Hubble Fellow (and soon-to-be PSU faculty member) Fabienne Bastien.  Back in 2005, I published a paper trying to quantify the amount of radial velocity noise intrinsic to a star with that star’s properties: evolved stars, quickly spinning stars, and young stars all have more of this noise, which is called “jitter”.  More recently, Fabienne published one showing that photometric noise also correlates with evolution—she dubbed this noise “flicker”. Soon after, I was a co-author on one of her papers showing a good correlation between flicker and jitter.

Now, Jacob is expanding the sample way beyond the 10 stars we used (4 were upper limits!) to make this relation quantitative and explore its relationship with stellar properties.  This will allow us to predict, in advance, which Kepler stars will make good radial velocity follow-up targets.

Thursday, January 5

Kim Cartier

Kimberly M. S. Cartier a.k.a. @AstroKimCartier

Texas A 10:30am 202.04D: Kimberly M. S. Cartier (a.k.a. Astrolady, a.k.a. @AstroKimCartier) presents her ***DISSERTATION TALK!*** on the Exo-Atmosphere of WASP-103b. Kim has nominally been my student, but her primary research advisers have been Ron Gilliland(!), Ming Zhao, and Thomas Beatty. She has worked an a range of projects, the theme being exoplanets with space telescopes, with a brief digression into SETI with me. She’s done a lot of projects, so she’ll have a lot to talk about, probably sticking to the science part of her thesis. Her final science project will be using MINERVA with Tom to do precise differential colors with photometry—”chromometry” Tom’s calling it—you’ll have to go to the talk to learn more! For the other side of her thesis (this mostly with me) you’ll have to visit her poster on Friday (see below).

Posters #245.25, 26, 27: Brendan Miller, Winonah Ojanen, and Spencer Miller present their work on Swift and Chandra data studying M dwarf coronae and high-energy photons’ effects on planetary atmospheres, something I’ve been working on with Brendan for a while, now.

Friday, January 6

Texas D 2:10pm 320.02: Jason Eastman talks about our spectroscopic commissioning results with MINERVA

Kim at her poster last year

Kim at her poster last year

Poster #335.01: The other piece of Kimberly M. S. Cartier‘s thesis: “Multimedia Astronomy Communication”. Kim has been applying best practices in social media outreach and online and print journalism to astronomy research. Together, we’ve written a couple of articles (there’s more to come!) and she’s been working with Penn State’s media relations folks on press releases. Last time she presented her “meta-poster” on good poster design, and this time she’s giving a much broader overview of effective communication of astronomy research to different audiences, and in different media.

Journalists: Kim is pursuing a career in science journalism with her PhD in astrophysics. I encourage you to stop by her poster to chat!

Saturday, January 7


Dr. Bastien as a graduate student at Vanderbilt.

Texas D 10:10am 403.02: Hubble Fellow Fabienne Bastien takes her work on flicker (see Jacob’s poster above) to new domains—in particular those F-ing stars those stubborn F stars and stars observed by TESS and K2.  Those missions will measure flicker, but in a different way than Kepler did, and also target a different typical sort of star from Kepler.

She will also briefly present the recommendations of the RV community on how to reach 10 cm/s RV precision from her recent Aspen Center for Physics workshop. If you couldn’t make the workshop, be sure to stop by. Also, while you’re there, congratulate her on her new tenure-line faculty position at Penn State :)



Texas A 10:50am 401.05: In a talk in a competing session (you’ll have to hustle over after Fabienne’s talk) CEHW postdoctoral fellow Thomas Beatty discusses thermal inversions in hot Jupiters.

Have fun at the meeting!

Metzger, Shen, and Stone

The next round of WTF star papers continues.  Brian Metzger (whom I know from grad school), Ken Shen, and Nicholas Stone have submitted a paper to MNRAS exploring in detail the idea that that Boyajian’s Star is dimming secularly because it recently “ate” a companion, and it’s still processing the energy from the merger, which is slowly “dribbling” out as an excess of luminosity.   

Steinn and I explored this in our rundown of possible explanations.  It appeared as hypothesis 13 in my blog post as “post-merger return to normal” and in our paper in section 11.3.

We had two primary objetctions: it would not really explain the dips, and the timescales are all wrong.  We wrote:

One issue here is that the dimming is too fast.  When confronted with big changes in energy content or flux, stars evolve on the Kelvin-Helmholz timescale, roughly the time it takes for all of the energy in the star at a given moment to finally escape the surface (while being constantly replenished by the fusion in the star’s core). For Boyajian’s Star this timescale is about 1 million years. This means that if the entire star is processing a big change in internal energy or luminosity, it takes around 1 million years to complete the adjustment.  Changing by 15% in 100 years is therefore about 10,000 times too fast.

But, the star’s radiative envelope is not very massive, so perhaps the energy never made it deep into the star? In that case the Kelvin-Helmholz timescale is a bit shorter, so maybe we’re off by only 1,000 times. It’s an order of magnitude argument, so maybe we’re being too pessimistic by a factor of 10, so we’re only off by 100 times. It’s possible that a detailed simulation of such a merger will reveal shorter timescale events, perhaps even things that might produce the dips.

So, I’m intrigued, and I like the idea despite the timescale argument not working out. It’s possible that there are other ways to temporarily brighten a star we haven’t thought of.  I’d like to hear from people who model these things before I commit to a plausibility level, so I’ll say:

Subjective verdict: unclear.

Well, Brian Metzger and company have come through.  In their paper, they look at the same mechanism Neslušan and Budaj explored to put material on highly eccentric, “cometary” orbits around the Boyajian’s Star.  The idea is that the close companion (which is presumably bound to Boyajian’s Star) interacts with material (anything from comets, planets, brown dwarfs to other stars) and slowly perturbs it into a highly eccentric orbit.  Then, if it’s comets, it outgasses when it gets close and you get the big dusty comae that might cause the dips.

Metgter et al. invoke the same mechanism to put a heavier object on an eccentric orbit, then have that object merge with Boyajian’s Star.  They deposit the energy into the envelope of the star, then run a stellar structure and evolution code called MESA to see how the energy is processed.

Their key result is their Figure 2:

screen-shot-2017-01-03-at-8-51-53-amIt shows, on the top, the total brightness of Boyajian’s star after merging with four fiducial objects of very different sizes.  The extra energy here is coming from the object’s orbital kinetic energy, which gets dissipated as heat when the two objects merge and eventually comes out as starlight.  Bigger objects have more energy to deposit, and deposit it at different levels.

The bottom plot shows the fractional change in the star’s luminosity with time (it’s the time derivative of the top plot divided by the top plot).  Zero means the star is not changing brightness, -0.01 means that the star changes its brightness by one e-folding (a factor of 2.7) in about 1/0.01 = 100 years.  The grey bands are the long-term dimmings seen by the DASCH plates over the last 100 years (top) and by Kepler over 4 years(bottom).

Steinn and I argued that the values you get in this scenario are more like -10-6, so way too small to notice.  What Metzger et al. have shown is that most of the energy does indeed end up in the envelope—the top millionth of the star’s mass—so time timescales are correspondingly shorter.  Our order of magnitude estimate was way way off, and so the hypothesis may be plausible after all (we recognized this could be the case in our paper, which is why we declined to give this scenario a plausibility).

So there are regions of the graph where all four curves cross the secular dimming levels seen.  This means that the model does not have to commit to what merged with Boyajian’s Star to explain the dimming.

So where does this scenario rank now? There are still several details to be worked out:

First, there’s the dips.  Metzger et al. point out that the same mechanism that sent the object into the star could also send other material there — a big planet could have lost its moons during a merger, or a planet could have been ripped apart, or something similar. This is essentially Boyajian et al.’s original hypothesis of material on a cometary orbit due to a single disruption event. The big difference is that there was originally a lot more material in the form of something that fell into the star.

Then, there’s the lack of infrared flux.  Again, the highly eccentric orbits save the hypothesis, and Metzger et al. point out that stellar radiation will blow sufficiently small dust out of the system, where it would no longer be warm and radiate.

The next is the details of the Montet & Simon light curve.  It changes slope pretty dramatically, and overall is steeper than the Schaefer dimming. What does this imply? I don’t see similar changes in slope in the Metzger et al. models, but presumably they’re invoking multiple ingestion events. Is this a problem for the model, or does it perhaps tell us the timings and masses of the mergings?

The next is the luminosity. The European Gaia spacecraft will measure the distance to the WTF star very precisely. This, combined with its apparent brightness, will give us the total luminosity of the star quite precisely.  This should give constraints on the merger history of the star.  Combined with the various secular dimmings, this should constrain the model—or prove inexplicable.  It would be nice to know what Gaia weill tell us, if anything, about this model. It would be especially nice if this model turned out to make a falsifiable prediction for the parallax.

Finally—and Metzger et al. acknowledge this is a major flaw in the model—there’s the issue of how likely this is to happen.  Steinn and I have argued that whatever the explanation for Boyajian’s Star, it’s got to be an unlikely one because it’s unique among 200,000+ stars Kepler has observed.  But this scenario turns out to be really unlikely—like Kepler had all but zero chance of seeing such a thing happen.  The effects of these merging events don’t last very long, so you need to stare a long time to have any chance of catching it happen. You would need practically every F star to have planetary material ready to go on eccentric orbits and merge, and even then you need a lot of planetary material.

I’m glad to see this scenario fleshed out so well. I suspect that there are ways to save the model by finding ways to make sort of event occur more frequently—perhaps by making the merging/dips more frequent by getting a chain of material from a single massive object—so I’m optimistic there’s more to this.  I’d say this paper has moved the “post-merger return to normal” scenario from “unclear” plausibility to something like “less plausable,” or even higher.

As I wrote last time:

This is how Tabby’s Star will be solved: a vague and qualitative hypothesis will get turned into a simple, quantitative model like this one, and that model’s success will inspire further work on more complex quantitative models. Eventually, these models will explain all of the data well and make some sort of prediction that will be confirmed by observations. Then we’ll say we have a good model for the system.

Thoughts on Neslušan and Budaj

A lot of folks want to know my opinion about the two new Tabby’s Star papers out this week:

Mohammed A. Sheikh, Richard L. Weaver, and Karin A. Dahmen
Phys. Rev. Lett. 117, 261101
With commentary by Steinn Sigurdsson here:

On this one, my opinion closely matches Steinn’s (because I asked him to explain it to me!). From what I understand of the paper, certain statistics of the dips follow a power law, and so-called “avalanche” models of certain phenomena associated with phase transitions follow a similar power law.  The authors suggest that this means that the processes causing the dips are internal to the star, and represent some sort of transition it is undergoing, like a global magnetic field flip.

That’s interesting, but I don’t know what it really means. It may provide a way for physical models to try to reproduce the data, by asking if the dips they predict follow the same power law.

Neslušan & Budaj  A&A accepted:

In this paper, the authors model four of the deeper dip complexes with a relatively simple but physically motivated model of massive objects (small planets) with very large, extended dust shrouds moving on highly eccentric orbits.

The physical motivation for this model is the same as for Boyajian et al.’s invocation of “comets”: eccentric orbits mean that the material will only spend a brief time near the star where it occults it, and so only be warm briefly.  This concentration in time explains the lack of IR excess at other epochs, and the lack of repetition of the dips.  Like Boyajian et al., Neslušan & Budaj posit that the bodies are the result of a single break-up event, and at least approximately share an orbit.

What Neslušan & Budaj add to this is a rough physical model for the clouds of material.  They assume it appears around 4AU (around the time comets get their comae) and that the dust particles are orbiting the planet (not how I imagined they would behave—I would have guessed like a collisional gas).  They have a variety of models for the initial conditions of these orbits, including some that generate spherical clouds and rings.  They then modeled the gravitational interactions of the star and massless dust particles, and included PR drag (radiation pressure from the star and the dust particle re-emission).  They used MERCURY to do the integrations.

They tried lots of variations of these parameters, and found a few that gave good fits.  Here are some of their efforts:

Neslušan & Budaj


The green lines (very hard to see: don’t use bright green on white, and use heavier weight for your lines, people!) are good qualitative fits to the data.  The right hand side shows the evolution of the dust cloud (lots of colors) and the orbit of the planet (red line) about the star (not shown, but at (0,0) and the focus of the red line).

In all cases what seems to be happening from the right hand side is that a dust cloud is released from the body, and then radiation pressure blows it away from the progenitor body near periapse.  They get similar results and some better fits with other models, including the ring model.

The authors have not really addressed the long term dimming seen (they mention it but have only a hand-wavey explanation that it’s accumulated dust in the orbits), nor the lack if IR excess (qualitatively, these things are only IR-bright when they are close, but the long-term dimming demands significant dust all the time, so, one would think, IR emission all the time).  As Steinn and I wrote, the “comets” explanation is “plausible for the dips, [but] very unlikely for the secular dimming.”  I think that still stands.

Keep in mind that the reason Neslušan & Budaj get good results for the “comets” hypothesis while Bodman & Qullien had more equivocal results is that they have more free parameters to play with from a more sophisticated model.  I don’t think they can do it with much less mass, but they can do it with fewer objects because their objects are bigger.  They still need one object per dip complex.

I’m also surprised that they want to model dust like massless particles in orbit around the planet—I would have guessed that dust cloud would be better modeled with a gas dynamics code than an n-body code. I’d like to hear from someone who studies cometary tails or atmospheric escape about how physically plausible these initial conditions and equations of motion are.

But this is a great next step.  This is how Tabby’s Star will be solved: a vague and qualitative hypothesis will get turned into a simple, quantitative model like this one, and that model’s success will inspire further work on more complex quantitative models. Eventually, these models will explain all of the data well and make some sort of prediction that will be confirmed by observations. Then we’ll say we have a good model for the system, and, if that model includes interesting features with wider applicability, we will use it to understand the Universe better.

Finally, Neslušan & Budaj conclude with “with such physical models at hand, at present, there is no need to invoke alien mega-structures into the explanation of these light-curves.” My thoughts on the propriety of the ETI hypothesis are well documented at this point, but let me say that I don’t think this paper takes the comets hypothesis across any critical threshold that we can say that we now have a good physical model for the system.  They’ve shown that one can get the sorts of complex structures we see in four the dips from a very simple model that is still missing a lot of physics — but a spline is also a very simple model that will fit the data well, which shows that simple models that fit are not by themselves enough to close the book here.  We still have a lot of work to do!

Some notes on the term “comets” here:

  • Boyajian et al. invoked giant comets, meaning bodies on highly eccentric orbits with extended clouds of gas and dust, but much bigger than Solar System comets.
  • Now, Neslušan and Budaj have taken exactly that idea and built a physical model for it.  It seems to be able to generate the dips.
  • Makarov & Goldin invoked interstellar “comets,” but that term usually refers to comets that have been ejected from the Solar System: such objects would not have comae, and so would not be big enough to cause Tabby’s Star to dim. Like Boyajian et al. and now Neslušan & Budaj, they mean things with big clouds of material around them, although it’s unclear why they would emit big clouds in interstellar space.
  • In all cases, we’re not really talking about comets like in the Solar System.  Both Boyajian et al. and Neslušan & Budaj are invoking a single break-up event that generates things that have orbits like comets and, like comets, generate big comae and tails when they get close to the star.  But these are not the primordial “dirty snowballs” of our comets.

Jargon update: “dispositive null”

Back when Sharon Wang published her first paper with me, we struggled to title it. A big result in the paper was that our team had shown that the planet did not transit: that is, we had ruled out transits.

But calling this a “null detection” of a planet seemed to send the wrong message: we had not failed to find a transit, we had succeeded in showing that the planet had failed to transit! I mulled over this in this space here and here.

Eventually, we coined some jargon in the paper: “dispositive null detection”.  Such a result “disposes” of the question of whether an effect exists. You use it when you are ruling out an effect with a known minimum amplitude (like non-grazing transits of a giant planet, or the effects of the ether in the Michelson-Morley experiment) and you do so definitively.  You don’t use it when you don’t find anything, but there could still be something beneath your detection threshold: that’s an ordinary “null detection”.

I wondered if it would catch on.  So far, I count six additional instances of its use, only 3 of which are my own uses of the term!

  1. Greg Henry used the term in text I contributed to his paper on HD 38529.
  2. I used it in our first Ĝ paper.
  3. Natalie Hinkel, at my suggestion, included it in the abstract of our paper on HD 130322.
  4. Natalie later included the term in the body of her paper on HD 6434.
  5. Danielle Piscorz and Heather Knutson defined and used the term in their “Cold Friends of Hot Jupiters” series.
  6. And just last month Andrew Howard and BJ Fulton used the term in their big RV limits paper.

So it hasn’t exactly taken off, but it is getting more use than I thought it would.  We’ll see!

A New Scale for SETI interest

I spent way too much time today trying my hand at another version of the Rio scale, one that does not assume that you understand the signal you see or know where it comes from, exactly.  I think it’s a little more useful down at the bottom (unlikely to be aliens signals) and a bit compressed at the top (very little discrimination among truly interesting signals). I also don’t think I have the weighting quite right yet.

Anyway, here it is.  It applies only to astrophysical anomalies for which the ETI hypothesis has some explanatory power.

It’s a 10-point scale (anything scoring below 1 is uninteresting). I conceived it as a log scale, so it roughly tracks as the product of three probabilities calculated as the sum of three numbers: A+B+C.

The idea is that you find where a given “detection” is in each category (maybe interpolating a bit). I give characteristic examples in brackets for each category, citing some famous (non-) examples of signals (n.b. “first pulsars” means the state of the field when the first pulsars were discovered, and there was not yet any theoretical explanation for their existence).

Note that these three categories are not orthogonal: the routine-ness of detection can influence likelihood of instrumental effects and astrophysical nature of anomaly.

A: Is the detection confirmed?

0: Unclear if detection is real [Poor metadata, possible transcription error, corrupted data]

1: One off: one event, one site [Wow! signal]

2: Verified: simultaneous observations at multiple sites OR multiple events at a single site [FRBs when they were still only seen at Parkes; something that survives strict coincidence detection]

3: Repeated: (multi-site AND multiple events) OR absolutely-no-doubt multiple events at single site [Tabby’s Star, especially with dimming]

4: Routine: Phenomenon regularly detected at different sites by independent groups [long-delayed echoes; FRBs today]

B: Could it be instrumental?

-7. Looks like a known instrumental effect

0. Does not look like a known instrumental effect, but too little metadata or understanding of data to assess signal reality; instrument builders / users not consulted [RATAN-600, SDSS lasers]

1. Data not likely to be instrumental, but instrumental or non-astrophysical cause cannot be ruled out [perytons, cosmic rays that look like lasers, early days of FRBs]

2. Data vetted by instrument builders/users; probably a real signal [Wow! signal]

3. Data thoroughly vetted with control; unambiguously astrophysical [FRBs today, Tabby’s Star’s dips]

C: Could the anomaly detection be natural or terrestrial?

-7: Obvious natural explanations exist [pulsars today, quasars today, RFI]

-4: Rare or somewhat ad-hoc natural explanations exist [J1407 rings, evaporating planets, GHAT III galaxies, FRBs today]

-1: Lack of study makes it unclear if any natural explanations could exist [Przybylski’s Star,SDSS lasers]

0. Only the rarest or most unlikely natural explanations make sense [Tabby’s Star]

1. No known natural cause, but new natural phenomena could be at play [first pulsars, first FRBs]

2. Not obviously intelligent communication or manufacture, but no natural explanations make sense [successful artifact SETI: complete Dyson sphere in the field, short-lived actinides in a spectrum]

3. ETI hypothesis is only plausible explanation [monolith on the Moon; communication: narrow band radio signals, pulsed optical, prime numbers, etc.]

Here’s how I interpret the outcome of a calculation:

Below 1: No interest warranted
1–5: SETI interest potentially warranted; no press interest warranted
6–8: SETI interest definitely warranted; technical popular press interest warranted; fun, off-beat news item for general press, with appropriate caveats. If not aliens, still very interesting.
9: Significant mainstream press interest warranted. Could be aliens
10: Aliens

OK, so how do our favorite candidates compare?

SDSS lasers: 2 + 0 – 1 = 1
GHAT III galaxies: 4 + 3 – 4 = 3
RATAN-600: 1 + 0 + 3 = 4
Przybylski’s Star: 4 + 3 – 1 = 6
FRBs before they were understood or seen at other telescopes: 2 + 3 + 1 = 6
Wow! signal: 1 + 2 + 3 = 6
Tabby’s Star: 3 + 4 + 0 = 7
First quasars (before understanding of their nature): 4 + 3 + 1 = 8
First Pulsars (before understanding of their nature): 4 + 3 + 1 = 8

This roughly (but imperfectly) tracks my impression of the SETI consensus on these “detections”: some signals are really interesting and only explicable as new phenomena (FRB’s, pulsars, quasars, maaayyybe Tabby’s Star, and Wow! signal, if it’s real) and they rightly rank high.

Some anomalies, like Przybylski’s Star are less compelling, but still intriguing and maybe deserve more attention than they get.  Then there’s a gap to things that will almost certainly not pan out on closer inspection.

Anyway, I’m not entirely sure it works well enough to be useful, but that’s my day’s-work contribution to the conversation about improving the Rio scale. Let me know what you think.

When the Press Should Pay Attention to a Possible SETI Detection

How should the media react to news of a claim of detection of alien intelligence? Claims come with various degrees of credit—there is some threshold that needs to be crossed for it to be worth repeating.

With earthquakes, the media can get a quick sense of importance from the Richter scale, which reduces a lot of qualities into a single number that the press has more-or-less learned how to use. Combined with location, even a non-expert can quickly use the scale to make a good guess about whether the earthquake warrants a call to see how loved ones are doing.

The Torino Scale

There is an analogous scale for asteroid threats.

It’s not very often that we have one, but when one is reported there’s generally a range of reactions in the media from totally ignoring it to overhyping how the world will end soon.  Fortunately, responsible journalists can look to the Torino scale to see how newsworthy a potentially hazardous asteroid is.

Visual representation of the Torino scale, from Wikipedia

Visual representation of the Torino scale, from Wikipedia

In short, the worry of an impact is some combination of the likelihood of impact and the damage it would cause if it hit.  A number above 3 is newsworthy, a number above 5 is genuine cause for concern and deliberation, and anything above 7 is call-Bruce-Willis territory.

In the history of the scale, only there have only ever been 2 asteroids that even reached Torino 2, and only one of those was (briefly) upgraded to Torino 4. Both were quickly downgraded to zero upon further measurements.  The media can, if they care to be responsible about this, generally ignore anything that is Torino 0, except maybe as a hook for a think-piece on planetary protection or something.

The Rio Scale

What about SETI? SETI has the Rio scale, set up by the International Academy of Astronautics at a conference in Rio to help the public understand how excited to get about “signals” from space.  Seth Shostak has a nice formula to help one calculate it here, and there is a calculator here.  It’s not perfect — it assumes that you understand the nature and origin of the signal you’ve detected — but it does help put things in perspective.

The Rio scale is a 10 point scale that multiplies the credibility of a report (a 5-point scale from 0-4) with the sum of scores from the class of phenomenon (from 1-6), type of discovery (from 1-5) and the distance to the source (from 1-4, important since it goes to whether two-way contact might occur).

Let’s take two famous examples to get a sense of scale: the Wow! signal and Tabby’s Star.

The Wow! Signal

The Wow signal is the most famous unexplained, potentially alien radio signal from space, detected 13 days after I was born in 1977 by Jerry Ehman (there have been many such “one-off” signals detected).

The class of phenomenon here is unclear—there’s no reason to think it was intended for Earth or could be interpreted by us.  This gives it 1 or 2/6 on this portion of the scale.

The discovery type is also low — a one-off signal.  This gives it a 1 or 2 /5 on the second portion.

The distance is unknown because we only have a general direction that it came from— so we only get the minimum 1 point out of 4 here (I think).

Finally, there’s the credibility: professional radio astronomers who understand their equipment well believed they found something.  In my mind, that puts it up to a 3 or 4/4 on the reliability scale.

Following the formula, I get somewhere from 1 to 3 on the Rio scale for this signal — not bad, and commensurate with how it’s talked about.  It’s a neat signal!

Tabby’s Star

What about Tabby’s Star? Plugging in the numbers again:

Class of phenomenon is low: there is no communication here, just possible evidence of astroengineering: 1/6

Discovery type is unclear, but since Kepler saw the dips for years and we’ve ruled out instrumental effects, let’s be very generous and give it the full 5/5.

Distance is well-known: 1,500 light years or so.  So we get 2/4 on this one.

Credibility on the report is high, but at some point we also to include our confidence that the source was not natural, which is low. We still have natural explanations to rule out. Let’s say 2/4 (“Possible, but should be verified before being taken seriously”)

Plugging everything together, we optimistically get a 3 on the Rio scale, which is pretty high as these things go (which is why it gets attention from the SETI community). But this optimistic number is somewhat subjective, so it’s clear why many astronomers give this explanation no credence.  A reasonable person might give much lower numbers to these categories and give it a 1 or 0.  The point is, it’s analogous to those near-miss asteroids: interesting enough for a popular article (or blog post) for lay aficionados, but anything more than that is unwarranted hype.

Unconfirmed Signals

Now, let’s imagine a hypothetical discovery of “something weird” in some archival data of some distant stars. The “something weird” matches their pet theory for how aliens might communicate among each other, but the discoverers are not the people who took the data, and have not consulted with such experts to see if it could be an instrumental effect. They have also not gotten verification that the signal is seen with other instruments.  So:

Type of phenomenon: intercepting undecipherable communication: 2/6

Discovery type: from archival data, no verification: 1/5

Distance: in the galaxy 2/4

Credibility: Instrument experts have not weighed in yet: 1/6 (“very uncertain, worthy of verification”).  Also, “something weird” does not rule out natural explanations, so lower it down again to maybe 0.5/6.

Plugging things in, we get 0.4 on the Rio Scale.  Most astronomers can ignore it, the press should definitely ignore it, but someone should probably follow up, just in case.

Another perspective

I’m not a big fan of the Rio scale because, as you can see if you try it yourself, if you have a signal you don’t understand, it’s hard to put on the scale.  The possibility it might be natural or instrumental is not really a parameter, so I’ve forced it into the “credibility” and “discovery type” categories in my examples above.

Rather, I prefer this sort of sequence for “something strange” detections. Every step of the way a detection gets more interesting from a SETI perspective:

  1. Point to a prediction of such a signal in a SETI context.  Not every weird thing is interesting from a SETI perspective.
  2. Rule out instrumental effects. This can be done by using a different instrument to detect the same signal, or thorough analysis of the data by experts in that instrument.
  3. Rule out obvious natural or terrestrial explanations. This means showing that any natural explanation is extremely unlikely or physically impossible.

Most “signals” discovered in SETI top out on this sequence at steps #1 or #2.  Optical and radio searches find “interesting” “signals” all the time, but they expect a few statistical flukes and equipment hiccups from time to time (astronomers like to blame “cosmic rays”, which are real examples of so-called “gremlins” that sometimes cause electronics glitches). Those that are probably real usually don’t repeat, and so it’s hard to rule out natural or terrestrial phenomena that we don’t understand if we can’t see it happen again (like the recent RATAN-600 “detection”).

But if you do get this far— the signal repeats, it’s definitely seen with other instruments— it’s very interesting!  You’re in the same space now as the discovery of pulsars, GRBs, FRBs, and quasars. Note that all four of those discoveries were very interesting natural phenomena, but the SETI community was rightly interested in them until they were solved. There have probably been many other “discoveries” that got this far and turned out to be not so interesting as those four game-changers (Tabby’s Star will probably end up in that category).

At this point, if all you have is “something weird” and not a clear, deliberate signal, you still have work to do before claiming a “probable” discovery of aliens, but you are at the point that you should tell communication SETI people that they should consider pointing their instruments at your target, just in case.  Then:

  1. See what the reaction is from the community of astronomers by getting the discovery published in a peer-reviewed journal. This will get more eyes on the data, and help determine if you’ve missed anything. At this point, the press may be rightfully interested—you need to point out to them that it’s probably not aliens.
  2. Learn more about your targets. Brainstorm less-likely natural explanations. Organize campaigns to follow it up

At this point you might be a 3 on the Rio scale.  Beyond that, you’re in terra incognito, and if the source resists natural explanation you might creep up higher on the Rio scale, slowly.

If you are sure that you have something that looks like communication from alien intelligence, then you can calculate a proper Rio scale number, but you also go farther along this sequence, following the generally accepted protocols for an ETI signal detection:

  1. Rule out all forms of terrestrial and natural origin for your signal, including hoaxes.
  2. Get confirmation from independent researchers of the signal using different equipment, and set up continuous monitoring

In other words, only if you find and confirm extraterrestrial signals of unambiguously intelligent origin—impossibly narrow-band radio signals or impossibly-brief laser pulses, for instance (impossible for a natural source, I mean)—do you claim any level of certainty of the detection of alien life. At this point a researcher can contact national authorities, the international astronomical community, and start planning for how to manage the press.

So, the press should certainly publish when a researcher gets to the sixth step of this sequence, but anything at an earlier stage is a curiosity for those interested about how the process works and what SETI researchers are interested in these days. Anything before step 2 is just the usual background rate of (presumably) false positives, and responsible journalists should generally ignore it.

[Edit: I have another attempt at a scale, so I’ve tweaked the description of my sequence in this post to avoid confusion with that one.]


What Could Be Going on with Boyajian’s Star? Part X: Wrapup and Gaia’s Promise

If you’re looking for a guide to this series, click here.

Last time we finished up our roundup of hypotheses for Boyajian’s Star.  Let’s summarize them:

Plausible Hypotheses

  • #3 Small-Scale ISM structure: This will find support if future dimmings are accompanied by increased reddening and sodium absorption.
  • #4 An Intervening Bok globule: In addition to the above, this would be supported if we discover a cloud of neutral gas near Boyajian’s Star.
  • #7 Comets and other circumstellar material: Plausible for the dips. Would be supported if future dips are accompanied by significant infrared flux and absorption features consistent with cometary or other circumstellar material.

Less Plausible Hypotheses

  • #5 An Intervening Black Hole Disk: Would find support if we can detect the central object or if the dips repeat in reverse order, as with J1407, or if more rigorous analysis shows this to be more likely than we have estimated.

Hypotheses with Unclear Plausibility

  • #2 A Solar System Cloud: Would have plausibility if such a cloud is theoretically realistic, if the cloud were found by other means, or if the dimmings are accompanied by absorption features characteristic of Solar System ices.
  • #9 Alien Megastructures: Would find support if all natural hypotheses are ruled out, we detect signals, or if star suffers significant achromatic extinction.
  • #13 Post-Merger Return to Normal: Would have plausibility if Boyajian’s Star is found by Gaia to be overluminous, and if simulations show this could explain the dips and dimming timescale.

Not Likely Hypotheses

Very Unlikely Hypotheses

And here’s a handy guide for what the Gaia parallax, which will determine the distance to Boyajian’s Star and is due out on September 14, will tell us:

  • Within 400pc: Means that ISM extinction cannot explain current level of dimming, so favors non-ISM and non-dust explanations.  These are #9 ,11, and 12.  Could also imply that the cause is very opaque dust blocking part of the stellar disk, so #5, 6, 7, and 8 might still be OK
  • Around 450pc: Means that all of the secular dimming can explained by the dust we observe along the line of sight.  Favors #2, 3, 4, 5, 6, 7,  and 8.
  • Beyond 500pc: Means that the star is more luminous than we expected from the reddening.  Favors #13: return to normal brightness.
  • No answer: Star does not have a good astrometric solution.  Unclear how this could happen given the lack of RV variation, but could that the 2″ companion is much bluer than we think, which would be annoying and not tell us very much except that it’s not an M dwarf, maybe boosting #5 (if it’s really the central black hole).  Might imply that there is another star nearby that we haven’t accounted for, suggesting new hypotheses.

You can vote on what you think Gaia will find here.

The paper is Wright & Sigurdsson, on the arXiv here.  Thanks for reading, and hopefully we’ll know a lot bit more in 11 days!

Update: Jason Curtis points out that the first Gaia release will not actually have the precision to distinguish among these cases!  So unless the parallax comes out really for from 450, we’ll have to wait longer than Sep. 14.

What Could Be Going on with Boyajian’s Star? Part IX: Intrinsic Variations

If you’re looking for a guide to this series, click here.

Last time we finished up circumstellar scenarios with a discussion of alien megastructures, the reason this star got so much press in the first place.  This installment: intrinsic variations.

I think Robin Ciardullo was the first to mention this possibility to me, though it may have been Richard Wade. The idea is that somehow the star itself is getting slowly dimmer, with occasional dips.

Hypothesis 10) Starspots and Magnetic Cycles

The most natural way stars change brightness over years and days is from stellar magnetic activity.  The Sun at the maximum part of its cycle is about 0.1% brighter than usual, and sunspots and plage coming into and out of view can also modulate its brightness by a few tenths of a percent.  More magnetically active stars might have even greater variations.

But the timescales and amplitudes are all wrong for Boyajian’s Star.  Activity cycles take decades, not centuries, and Boyajian’s Star rotates about once every 0.88 day, not many days.  If the dips were caused by spots, they should come and go every 0.88 day, but instead they last many days or longer.  Finally, in order to cause the big changes in brightness we see, the spots or cycles would have to be 10–100 times larger than the strongest effects known in other stars.

Another problem here is that Boyajian’s Star is decidedly inactive, being a hot star, not a cool star.  Unlike the Sun, it has a radiative outer layer (not a convective one), so it does not have the same dynamos that produce the Sun’s magnetic fields.  Hot stars like this have very little in the way of surface features (“stellar dermatology”) and no activity cycles (that we know of).  There are hot stars with strong magnetic fields and spots, but they are almost always slow rotators (their fields slow them down), and are generally hotter than Boyajian’s Star.

So, subjective verdict: very unlikely.

Nonetheless, something weird an unexpected is going on, so maybe we should not dismiss this entire line of reasoning entirely.  This bring us to

Hypothesis 11) Polar Spots

Montet & Simon point out that a polar spot could be doing this.  Some very cool stars are known to have large, dark, polar starspots.  This explanation nicely explains why the spots don’t come and go with a 0.88 day period: since they are at the pole, we always see them, and so at best their irregular shapes get modulated. This would explain why we see a 0.88 day photometric signature at all, on what we might expect would be a featureless star. Then, the dimmings could be due to the ebb and flow of this spot’s size and darkness on days- and century-long timescales.

Reconstruction of the surface features of the T Tauri star V410 Tauri (from Carroll et al., A&A 548, A95). Note the large spot at the north pole of the star. Very young, cool stars have strong magnetic fields that seem to produce spots like this which, if the star is inclined with the spot towards us, we see persistently, with only weak rotational modulation.

Reconstruction of the surface features of the T Tauri star V410 Tauri (from Carroll et al., A&A 548, A95) shown from 4 different vantages. Note the large spot at the top pole of the star. Very young, cool stars have strong magnetic fields that seem to produce spots like this which, if the star is inclined with the spot towards us, we see persistently, with only weak rotational modulation.

This is very clever, but for the reasons I stated above we have good reasons to think this star can’t have spots like this.  (Again, if this were a thing, we should see it all the time!)  Montet & Simon offer the hypothesis very provisionally for this reason.

So my subjective verdict on this one is: not likely.

This hypothesis would find support if Gaia finds the star to be much farther away than we expect from the reddening and sodium absorption.  This would indicate that reddening is not accompanying the dimming we see, so it could be spots.  It would also find support if signs of a polar spot could be identified, such as narrow absorption features from the much cooler atmosphere, perhaps with Zeeman splitting if the spot has strong magnetic fields.

OK, what else could the star be doing?  Well, many forms of variable star are pulsating, so how about:

Hypothesis 12) Stellar Pulsations

Stars have a few characteristic timescales on which they change.  If you poke a star with a regular cadence, there are certain frequencies that will cause it to respond strongly.  Some ways this can happen are:

  1. In the Sun, random convective motions cause it to “ring” with many frequencies, the strongest being around 5 minutes.  These “asteroseismic” modes are generally very, very small: they can never cause variations of order 20%, and they are usually periodic, not random like Boyajian’s Star.  The timescales are wrong for a Main Sequence star, too: these operate on sound-crossing times and their harmonics, not days or centuries.
  2. Many classes of pulsating stars rely on internal instabilities involving the opacity of their constituent gasses.  If these gasses heat up, they ionize, which makes them more opaque, which makes it harder for starlight to emerge from the core to the surface, which causes the star to expand, which makes the gasses cool, which makes them recombine, which lowers their opacity, which makes it easier for starlight to emerge, which makes the star shrink, which heats the gasses… etc. etc.  These timescales can be of order days, but, like asteroseismic modes, they are periodic, not episodic, and don’t cause century-long dimming.  Delta Scuti stars are a kind of star of very similar mass and temperature to Boyajian’s Star, so it’s possible that it has something like this going on, but it’s not clear how that could cause any of the effects that we see.

Basically, all known pulsation modes are regular or semi-regular, not episodic, and none we know of operate over centuries.   If it’s some form of pulsations, it’s one we haven’t seen before or thought of before.

Subjective verdict: not likely

The last one is my favorite, from Steinn:

Hypothesis 13) Post-Merger Return to Normal

What if the star isn’t dimmer than it should be: what if it’s brighter, and we’re seeing it return to normal brightness?  In that case Gaia will reveal that it is much farther away than we think based on its brightness and reddening.  This would imply that the long-term “dimming” we see is the star returning to a normal state after somehow getting way too bright.

But what could that mean?  The best idea we have is some sort of merger: perhaps the star recently coalesced with another star, or a brown dwarf or planet.  This would deposit a lot of orbital and gravitational potential energy into Boyajian’s Star, which would eventually get dissipated as heat and escape as starlight, causing a temporary brightening.

It’s not clear why this would create the dips, but perhaps the merger was quite recent, and there are still hydrodynamical effects going on inside the star as the companion is “digested”, causing the internal structure of Boyajian’s Star to adjust on days-long timescales.

One issue here is that the dimming is too fast.  When confronted with big changes in energy content or flux, stars evolve on the Kelvin-Helmholz timescale, roughly the time it takes for all of the energy in the star at a given moment to finally escape the surface (while being constantly replenished by the fusion in the star’s core). For Boyajian’s Star this timescale is about 1 million years. This means that if the entire star is processing a big change in internal energy or luminosity, it takes around 1 million years to complete the adjustment.  Changing by 15% in 100 years is therefore about 10,000 times too fast.

But, the star’s radiative envelope is not very massive, so perhaps the energy never made it deep into the star? In that case the Kelvin-Helmholz timescale is a bit shorter, so maybe we’re off by only 1,000 times. It’s an order of magnitude argument, so maybe we’re being too pessimistic by a factor of 10, so we’re only off by 100 times. It’s possible that a detailed simulation of such a merger will reveal shorter timescale events, perhaps even things that might produce the dips.

So, I’m intrigued, and I like the idea despite the timescale argument not working out. It’s possible that there are other ways to temporarily brighten a star we haven’t thought of.  I’d like to hear from people who model these things before I commit to a plausibility level, so I’ll say:

Subjective verdict: unclear.

So that’s it!  That’s all I’ve got.  I’ve tried to be comprehensive, but cleverer people than me keep coming up with new ones, so I’m sure I’ve missed something.  I’m still not convinced that the right answer has appeared in the literature, but I’m hopeful that one of the explanations in our paper is correct.  Before wrapping up in the final post, though, I’d like to discuss some nonstarters that keep coming up:

Popular Nonstarters:

  1. Gravity darkening.  This one got a lot of attention for some reason. Basically, stars that spin very fast have equators that are darker than their poles.  A transit or eclipse of a gravity-darkened star can therefore have an unusual shape, potentially being asymmetric.  Since some of the dips are asymmetric, some people declared the system “solved” with this suggestion.  Problem is, Boyajian’s Star is not spinning fast enough to have significant gravity-darkening, and this still doesn’t address the issues of what is causing the dips!
  2. Black hole or other nearby companions.  Boyajian et al. ruled this one out, but just to reiterate their argument: Boyajian’s Star is RV stable, so it does not have any close-orbiting stellar-mass companions.  It’s not any kind of binary system.
  3. Gravitational waves. I’ve heard this one suggested, presumably because they were just discovered by LIGO and so are in the zeitgeist right now.  They have no explanatory power here (i.e. there is no way they could cause what we see).
  4. Asteroids in the Solar System.  We discuss Solar System solutions here.  Anything closer than the outer Solar System would not persist for years, and asteroid occultations have a very different photometric signature than Boyajian’s Star.
  5. It’s an ordinary “dipper”. There are “dipper” stars that have superficially similar light curves to Boyajian’s Star. But these as caused by close-in disks and other circumstellar debris that are revealed by their long-wavelength emission, which Boyajian’s Star lacks. That was the original reason Boyajian’s Star stood out as being so weird: it’s not young, it has no disk, so it’s not a dipper. I don’t understand why this one keeps coming up.
  6. Starkiller Base.  Yes, I saw The Force Awakens, too.  I presume these suggestions were all made facetiously.

OK, next time: summary of hypotheses, and what Gaia will tell us.


What Could Be Going on with Boyajian’s Star? Part VIII: Alien Megastrutures

If you’re looking for a guide to this series, click here.

Last time I tried to wrap up circumstellar solutions, but before moving on to intrinsic variations, it’s clear I need to address the megastruture in the room.

Hypothesis 9) Alien Megastructures

Background on this hypothesis is here, and here:

Now, I can’t give this hypothesis the treatment I’ve given the others, because, even compared to the vague, general families of solutions in the previous hypotheses, this one has very little to hang any physics on. Ancient alien civilizations could be arbitrarily advanced, and so it’s not clear what physics we’re allowed to assume. We don’t know why they would create megastructures (though energy collection seems like a good guess to me) so we have no good reason to expect any particular shape or size for them.

That said, we can build up a straw man model, and see how well it holds up.

But first, let’s dispense with some unnecessarily complicated ideas. Lintott & Simmons made me smile with their article in the Journal of Brief Ideas in which they used the timescale of secular dimming to estimate the construction time for a Dyson sphere. Similarly, Villaroel et al. skeptically mentioned this possibility in their paper.

But we don’t have to imagine building a gigantic sphere in a century to hypothesize how megastructures could explain the dimming.

Imagine that it is advantageous to collect solar energy to be used for some purpose, and that there is enough material to do so, but not an infinite supply.  Imagine that the panels have a range of sizes and orbit the star in a range of orbital periods.  In this case, the cost (in time, energy, and mass) to construct a panel is balanced by the benefit (the energy intercepted, which favors close-in panels, times the efficiency of the panels, which favors far-out panels). If too many such panels are created, the close-in ones will shadow the farther-out ones, reducing their efficiency.

In this toy model, one expects (in a rough order-of-magnitude sense) to end up with a swarm of panels that have an optical depth near 1. That is, they should not capture, say 99.999% of the photons, because the marginal efficiency of the next panel is reduced by a factor of 100,000 compared to the first one.  And they shouldn’t intercept only 0.001% of the light, because 99.999% of the light still free to take with virtually zero efficiency hit. You’d expect somewhere between, say, 10% and 90% of the light to get absorbed.

The smaller panels will appear as a translucent fluid around the star, constantly blocking some fraction between 10-90% of the light, roughly speaking. This fraction will vary as denser parts of the swarm come into and out of view, and as chance alignments of parts of the swarms at different orbital distances align. We might see variations in brightness on scales from hours to centuries. Particularly large panels—even bigger, perhaps, than the star itself—will cause large, discrete dips as they transit, with profiles according to their shape. The timescale for crossing may not be a good indicator of their distance from the star, because they might be so thin that radiation pressure is important (see the appendix here).

This is actually right in line with the observations of Boyajian’s Star: we see a constant dimming of the star, one that erratically has increased by about 15% in the past 115 years, and occasional dips.

OK, so how does the toy model hold up against long-wavelength observations?  Well, not great. The balance of efficiency with orbital distance is trickier, and interacts with the shadowing argument, but you might expect there to be a typical, characteristic temperature that is around 150K (95% maximum efficiency for a star like the Sun—we argued all of this in more detail in our paper here).

The long-wavelength constraints apply almost as well to megastructures as they do to dust. We argued in that paper that it is reasonable to expect nearly all of the energy  collected to be reradiated as waste heat.  As we saw from the long-wavelength limits, it just can’t be that 15% of the starlight is being reprocessed at 150K, or at any other temperature:

Spectral energy distribution of Boyajian's Star, with arrows indicating upper limits. The right-most measurements are from Thompson et al. They rule out big clouds of dust blocking 15% of the star's light in all directions, because the dust would reradiate at levels easy to detect.

Spectral energy distribution of Boyajian’s Star, with arrows indicating upper limits. The right-most measurements are from Thompson et al. They rule out big clouds of megastructures blocking and reradiating 15% of the star’s light in all directions (black line) at 65K, because the heat would be easy to detect.  Other temperatures are actually in worse agreement with the data.

The black line is our toy model swarm blocking 15% of the light, and it is strongly inconsistent with the red upper limits from WISE and Thompson et al.

Now, there are a couple of ways out.  First, the aliens might be doing some non-dissipative work with this starlight. Perhaps they are launching interstellar probes. Perhaps they are sending out powerful laser or radio transmissions. Anything that puts all of that stellar luminosity into energy that leaves the system in a low-entropy way would reduce their waste heat luminosity.  Can that do it?

Well, if the work is all done at 65K, then the maximum (Carnot) efficiency allowed by thermodynamics is about 99%. This means that they could reduce their long-wavelength luminosity down from 15% of the star’s, to only 0.15%.  The upper limit at this temperature, as you can see in the figure above, is 0.2% (the purple line).

So… not completely ruled out, yet, but I’d have to say we’re close enough that I’d put this one down as very unlikely, even after normalizing for the unknown chance that there are megastructures there in the first place.

Another way out is that it’s not a spherical swarm. Make it an edge-on ring, and you can further reduce the waste heat. Make them radiate towards their ecliptic poles, and you can reduce it further. Turn the energy into low entropy radiation at high efficiency and do both of the above, and they’d be nigh undetectable.

So you see, the hypothesis is sufficiently flexible that it can fit almost any observations, but I do think we can say this about the vanilla, simplest, strawman scenarios:

  1. IR/mm observations rule out a spherical swarm doing only dissipative work (like humanity does) being responsible for the long-term dimming
  2. We can further rule out a spherical swarm sending the energy off in another way at maximum efficiency with temperatures above ~90K
  3. We can lower this cut-off temperature with more sensitive measurements

So, subjective verdict: unclear, with an extra multiplier of very unlikely for the simplest spherical swarm strawman.

The megastructure hypothesis would find support if Gaia shows us that the star is actually much more extinguished than the reddening suggests (meaning there is a significant optical depth of geometric absorbers: the swarm of solar panels).  Or if, you know, we detect those radio waves that all those panels are producing!

OK, next time: back to the natural explanations, this time looking at intrinsic variability.

What Could Be Going on with Boyajian’s Star? Part VII: Circumstellar Dust

If you’re looking for a guide to this series, click here.

The central puzzle of Boyajian’s Star is that although the dips look a lot like those of dippers, it has no close-in disk.  That is, most stars that show this sort of behavior are young and have close-in disks, and those disks cause their dips.  Not only does Boyajian’s Star have no close-in disk, it isn’t near a star forming region, so there is no reason to think it is young, anyway.

So, if it can’t have close-in material, what could be causing the dimmings?  Well, if the material is circumstellar, it must be very cold, and/or there can’t be very much of it.  Boyajian et al. also showed that the timings and slopes of the dips also put constraints on the material’s proximity to Boyajian’s Star, meaning it needs to be pretty far-out (10 AU or so).  This brings us to:

Hypothesis 7) Comets or other circumstellar debris

tabby-cometBoyajian’s original hypothesis was that the dips were caused by giant swarms of giant comets.  This remains a plausible hypothesis for the dips: the comets would certainly have been warm when near periastron, but their eccentric orbits could bring them very far from the star quite quickly.  Thus, the infrared excess would only be noticeable during a dip.  If the swarm is on a long-period orbit, then we may never see them again!

On the other hand, there doesn’t seem to be any way the comets can explain the long-term dimming, especially without an infrared excess.

So, my subjective verdict on this is: plausible for the dips, very unlikely for the long-term dimming. Occum’s razor argues that if they aren’t causing the long-term dimming, then they probably aren’t causing the dips, either, but your mileage may vary on that one.

OK, what about:

Hypothesis 8) A cool annulus of material

spitzer_stardebris.jpg.CROP.original-originalThis is basically the hypothesis that there is a disk of material, but there is a very large (10 AU, at least) gap between the star and the inner edge of the disk (annulus).  The disk is almost, but not quite, edge-on: then, we could invoke corrugation or other irregularities that occasionally block part of the star for the dips.   This needs to be a real protoplanetary disk, not just a debris disk, because it needs to be optically thick.

There are lots of problems with this:

  1. The Thompson millimeter results rule out very much mass for the disk—there is not a lot of wiggle room between what you need for the dips and the secular dimming and what would have been detected
  2. I’m not aware of any theoretical justification for such ring, except maybe for young stars.
  3. If the inner edge is at 10 AU, the probability of edge-on alignment is roughly 0.1%, so in order to expect to have seen one, we need to have about 1% of all Kepler stars to host such a disk (despite #2, and the fact that the field has no young stars)
  4. The secular dimming is not naturally explained—if the disk is always occulting part of the star the sodium lines would probably not be symmetric, but they are.

So, while an annulus is not ruled out by the data, it doesn’t have a ton of explanatory power, and it has no theoretical justification, and it requires a very unlikely geometry and high frequency of such disks in the field.

Although I know some others still like this one a lot, these are just a bridge to far for me.  Subjective verdict: not likely.

We’ve already discussed the long-wavelength constraints that show it can’t be embedded in a cloud of material, so I won’t even mention that one.

OK, now that we’re done with all of the circumstellar possibilities, in the next installment we’ll continue our conceptual journey from Earth to the star itself and discuss intrinsic variations of the star…

Huh? What’s that? Megastructures? You want a rundown on where that hypothesis is?

Well, I hadn’t planned to talk about it, but since you asked, I suppose we could do a brief detour into some serious speculation before finishing our list.  As luck would have it I happen to have some remarks prepared right here…

All right then. Next time: alien megastructures.


What Could Be Going on with Boyajian’s Star? Part VI: Black Holes!

If you’re looking for a guide to this series, click here.

Last time we covered SINS and Bok globules as potential sources of the dimmings in Boyajian’s Star.  This time we’ll finish up the interstellar solutions and start to look at circumstellar solutions.

Our topic? BLACK HOLES!

Hypothesis 5) An Interstellar Black Hole Disk

Now, a very popular suggested explanation from laypeople has been “a black hole”.  For a long time, I treated this as a total non-starter: black holes are tiny, tiny, tiny, and don’t have the geometric cross-section to block any noticeable amount of light.  Further, they tend to gravitationally lens light, so if there really were a black hole between us and Boyajian’s Star we’d expect a brightening, not dimming. Finally, if it’s close enough to suck up any material, we’d see Boyajian’s Star get brighter from the accretion, we’d see light from radio to X rays from the interactions, and we’d see RV variations to boot.  So, no black hole, right?

So it’s ironic that we found another way to do it (though not what those who suggested it were thinking).  This solution is more analogous to J1407, the “planetary system under construction” discovered and characterized by Eric Mamajek and Matt Kenworthy.

Here’s the idea: the dips really suggest small-scale structure in the intervening material, but the long-term dimming suggests a smooth distribution.  One solution is a gigantic disk with annular features, like a protoplanetary disk with gaps and waves from planets and protoplanets.  The timescales involved (115 years!) mean that it must be HUGE — like 600 AU across.  What could host such a disk, but have totally escaped notice?  After all, Boyajian et al. used adaptive optics to hunt for anything nearby that could do the trick, and only found an M dwarf 2 arcseconds away.

Well, it must be very massive, and it must be dark.  Stellar remnants do the trick: a cold white dwarf, a quiescent neutron star, or a black hole.  We focus on the black hole solution in our paper.

The black hole disk from Interstellar, with the black hole's gravity distorting the image of the disk. Our scenario is much less photogenic: the close-in disk is gone, and only a very wide, very cold disk of dust and debris remains. Also, our black hole is several solar masses, not several million.

The black hole disk from Interstellar, with the black hole’s gravity distorting the image of the disk. Our scenario is much less photogenic: the close-in disk is gone, and only a very wide, very cold disk of dust and debris remains. Also, our black hole is several solar masses, not several million.

The idea is that after a supernova explosion, there will be material that falls back towards the remnant black hole, where conservation of angular momentum requires it to collapse to a disk, and some amount of the material to move outward while a lot of it spirals in and accretes onto the black hole.  Eventually, the black hole finishes its meal, and goes quiescent, leaving behind a big inner gap within a large ring of debris outside.  This ring can now get very cold (the central object is dark! nothing to heat it), thin, and wide.  Perna et al. (2014) describe the process.

So, does it work?  Could there really be a black hole aligned with Boyajian’s Star? Well, the alignment doesn’t have to be very good: the disk needs to span a big chunk of the sky for the dimming to persist over 115 years (a couple of arcseconds, at least).  The black hole itself is small and would be unlikely to actually go between us and Boyajian’s Star (if it did it would lens it, but its Einstein radius is only ~4 mas).

We calculated the volume of space probed by Kepler for objects with that size (angle squared times typical distance cubed), multiplied by the number of stars Kepler observed, and decided you needed about 10 billion disk-bearing black holes in the Milky Way for one to have had a good chance to wonder in front of ~1 such star in the field.  That’s not too far off from the estimated number of black holes in the galaxy!

So, the numbers aren’t too bad!

I really like this one, but there still isn’t any observational evidence that such disks exist, or that they are common enough for Kepler have found one.  We haven’t done a rigorous calculation of the probabilities, so it could still fall apart upon closer inspection.

Given the uncertainties, I give it a subjective verdict of: less plausible.

This hypothesis would find support if the dips repeat in reverse order while the star starts brightening again, like J1407 did.  Alternatively, that so-called “M dwarf” 2″ away could turn out to be the central object, in an almost-but-not-totally quiescent state.  Someone get a spectrum of that thing!

Hypothesis 6) An Orbiting Black Hole Disk

I had hoped that we could make an alignment more likely by putting the black hole in orbit around Boyajian’s Star, but it turns out that makes things much harder.  In addition to the low probability of such a binary companion in the first place, the chances that it would be in a part of its orbit such that we would see it are very low, like 1 in a million low.  Since Kepler only looked at 100,000 stars, and since every star does not have such a companion, this one doesn’t work.

Subjective verdict: not likely.

OK, enough with the black holes. Next time: Circumstellar material.

Update: Commenter Herp McDerp (obviously their real name) points to this Nature article by Alastair G. W. Cameron (Bethe Prize winner and originator of the Giant Impact Hypothesis for the formation of the Moon).  In it, Cameron tries to explain the eclipses of the ε Aur system with our Hypothesis 6!  I’d write “great minds think alike” but I’m totally out of my league on this one, so I’ll just write that we’re in very good company with this hypothesis!

What Could Be Going on with Boyajian’s Star? Part V: The Interstellar Medium

If you’re looking for a guide to this series, click here.

Last time we covered the first two hypotheses: instrumental issues and a Solar System cloud. This time, let’s move to the interstellar medium.

Hypothesis 3) Absorption from the Interstellar Medium

Between the stars there is lots of gas and dust. The densest, coldest parts of this “interstellar medium” (ISM) form neutral gas and dust, which cause reddening, dimming, and line absorption in stars. For instance, all evidence points to Boyajian’s Star having its light pass through enough dust and neutral gas on the way to Earth to make it about 35% dimmer at visible wavelengths than it would be without that stuff in the way. This is expected for any star in its part of the Galaxy.

But, one asks, maybe there are some especially dense pockets along that our line of sight occasionally sweeps past? I had always rejected this line of thought because if that were a thing, all sorts of stars would do that. After all, Kepler looked at over 100,000 other stars and none of them ever showed this behavior.

Carl Heiles, radio astronomer extraordinaire, and my wife's thesis adviser at UC Berkeley. He has done a lot of work on "tiny scale atomic structure" in the ISM.

Carl Heiles, radio astronomer extraordinaire, and my wife’s thesis adviser at UC Berkeley. He has done a lot of work on “tiny scale atomic structure” in the ISM, among many other topics.

But, it turns out rare dense patches are a thing! I was at Berkeley recently and chatted with Carl Heiles, and he pointed me to SINS — small ionized and neutral structures in the diffuse interstellar medium. He was actually one of the folks who first brought these things to the attention of the broader community and postulated that they are short-lived, overdense, corrugated sheets and/or filaments in the ISM, and that when our line of sight aligns with a tangent point of such a structure we get a temporary jump in absorption.

So it turns out that this is a whole field—Carl pointed me to this conference on the topic 10 years ago. So, could this be the answer? Well, one problem is that the columns and sizes implied by Boyajian’s Star are much different from the “tiny scale atomic structure” that Carl describes. Those structures are typically around 30 AU across and block maybe 0.1% of the visible light that passes through them.  We need about 100 times as much dust, and we need structures maybe 100–1000 times smaller than that.

But, it’s not clear that we would know about such structures if they did exist—you would basically need to launch Kepler to notice them. The ones they know about are pretty rare (most sources don’t show any evidence for them) so regions this dense and small could be so rare that Boyajian’s Star is the only one that sees them.

So, if we hypothesize that there is a spectrum of these SINS down to even smaller scales than have previously been seen, and that these smaller structures have even higher densities, and if these extensions persist across 2-3 orders of magnitude, then we would expect to very occasionally see stars behave like Boyajian’s Star!

This hypothesis is supported by the fact that the existing dimming implied by the ISM reddening and sodium absorption is enough to explain all of the Schaefer and Montet & Simon dimming — they see 15% or so, and the ISM is causing over twice that.  This hypothesis would find support if during a future dip (or if the star’s long-term brightness continues to change a lot, in either direction) we see a corresponding change in the reddening, and in the sodium absorption features.

I like this one.  Subjective verdict: plausible!

Hypothesis 4) Absorption from an Interstellar Molecular Cloud

Barnard 335, a good example of a small, isolated molecular cloud—or "Bok globule"—and the subject of my undergraduate thesis with Prof. Dan Clemens at Boston University. Source:

Barnard 335, a good example of a small, isolated molecular cloud—or “Bok globule”—and the subject of my undergraduate thesis with Prof. Dan Clemens at Boston University. Source:

A more obvious way to make a star dimmer is with an interstellar molecular cloud.  We don’t expect to see any of these up in the direction of Boyajian’s Star, but such high latitude clouds do exist.  Now, normally such clouds are obvious because they are the site of star formation, and young, forming stars light the cloud up at wavelengths from the radio to X rays.  But some are quiescent, and if a star were not forming in the cloud it might not be obvious.

Now, such quiescent clouds are usually obvious because they are big (or part of a bigger cloud complex). There are small, isolated ones, though, called Bok globules, and they are usually found because they block a lot of starlight, so you see what looks like a big hole in the sky where stars should be.

At least, you do when they’re big and in front of a rich star field, like in the Galactic Plane. If they’re small and farther away, they can be pretty small, and if the background star density is low, you might miss them entirely.  If there were a small (0.1 AU) Bok globule in front of Boyajian’s Star about 300 pc away, I don’t think it would have been noticed before.

Bok globules have smoothly varying densities, so a varying line of sight through it would naturally explain the long-term dimming.  The dips would imply that this Bok globule has dense, sub-AU knots within it that our line of sight is probing.  That’s a bit of a stretch, but, hey, if SINS exist in the diffuse ISM, why not in a dense cloud?

This hypothesis would find support if we could find the cloud, perhaps with a gas map done by the VLA or something.  Even a single-dish radio telescope might at least spot some molecular gas in that general direction compared to neighboring direction, which would lend this hypothesis support.

This one isn’t quite as nice as SINS for me, but I still think it’s got a lot going for it.  Subjective verdict: plausible!

OK, enough words.  Next time: our final interstellar hypothesis, and then we’ll get in close and take a look at the possibilities for material in orbit around Boyajian’s Star!

What Could Be Going on with Boyajian’s Star? Part IV: Nearby Stars, Instrumental Effects, and a Solar System Cloud

If you’re looking for a guide to this series, click here.

Last time, we discussed the periodicities (or lack thereof) in the dips of Boyajian’s Star, and the implications of the sodium absorption and reddening we see.  The last piece of observational evidence I’d like to introduce is whether any neighboring stars are variable.  After all, if there is a large cloud between us and Boyajian’s Star, it should block light from other stars, too.

Photometry of Nearby Stars

Well, there is a star, KIC 8462860, which sits 25′′ NNW of Boyajian’s Star.  It doesn’t have the same sort of time-series as Boyajian’s Star, so we can’t look for dips, but Ben Montet did extract its long-term light curve from the Kepler full-frame imagery, and he graciously gave them to me for publication:

Photometry of KIC 8462860, a star 25′′ NNW of Boyajian's Star, kindly provided by Ben Montet, using the same methods he and Josh Simon used to find the long-term dimming of Boyajian's Star. Nothing to see here, apparently.

Photometry of KIC 8462860, a star 25′′ NNW of Boyajian’s Star, kindly provided by Ben Montet, using the same methods he and Josh Simon used to find the long-term dimming of Boyajian’s Star. Nothing to see here, apparently.

Nothing to see here, apparently—this looks like an ordinary constant star, like most of the >200 others Montet & Simon looked at.  Nonetheless, Steinn and I strongly encourage photometry monitoring of all stars within 1′ of Boyajian’s Star, just in case.

OK, so that’s going to help with explanations invoking a cloud of material. Now on to the hypotheses!

Hypothesis 1) It’s Instrumental

That is, there is no dimming to explain, it’s an artifact of the instruments.  Dr. Boyajian ruled this out in her paper regarding the dips, and since then many groups have gone back to the Kepler data to see if they can find a problem, including Kepler team members.  The dips are real.

The long-term dimming Schaefer sees could be instrumental — indeed there was a big food-fight over this issue, but, as I wrote in Part I, the Montet & Simon discovery of similar dimming during the Kepler mission (and careful use of control stars) provides independent confirmation of the phenomenon.

I think we can put this one to bed.  Subjective verdict: very unlikely

Hypothesis 2) A Solar System Cloud

Could there be a cloud of — something — in the Solar System blocking light from Boyajian’s Star?  Let’s set aside where it came from for now—does the hypothesis have explanatory power?

The Earth, Kepler, the Sun, and Boyajian’s Star are all moving.  The motion of the star is called its space motion and the apparent change in position of the star due to its motion and the Sun’s is called the star’s proper motion.  If there is something between us and the star, then proper motion should change our line of sight through it.  Kepler also moves around the Sun, and that results in an apparent, annual elliptical motion across the sky we call parallax.  Nearby stuff seems to move due to parallax much more than background stars, so if there is a Solar System cloud, then we would expect our line of sight to trace an annual ellipse through it, in addition to a slow drift due to proper motion.

For the moment, let’s put the hypothetical cloud out at 10,000 AU.  Parallax would make it appear to move by about 20 arcseconds, and its orbital motion would move it by about the same amount over 100 years.  So if the cloud is 20″ across, it could be responsible for the long-term dimming.  This would also help explain the 1.96 Kepler year gap between the two dips (although not the lack of dips at 0.98 years): that’s the time it takes our line of sight from Kepler to return to about the same place, with ~1% taken off due to the cloud’s own orbital motion.

So, in this scenario the long-term dimming is just a density gradient in the cloud, and the dips are from dense knots in the cloud that we briefly look through when Kepler and Boyajian’s Star line up just so.

So, what are the problems?  Well for one thing, the density gradient doesn’t seem modulated on an annual timescale.  For another, the sodium lines in the spectrum don’t seem to be at zero velocity with respect to the Solar System barycenter.  Finally, why would there be a cloud of dust out there?  Not only is Boyajian’s Star way above the ecliptic (but does that even matter at 10,000 AU?), but a 20″ cloud at 10,000 AU would be 1 AU across.  What could cause it?

Steinn likes the idea of a very low-velocity cometary collision that kicked up ice and dust in a big cloud.  It wouldn’t take much mass to hold the dust in a Hill sphere 1AU across (an asteroid would do it).  I prefer a big KBO-type object with a very gentle geyser that burps out dense plumes occasionally.

If we see another dip, and the spectrum of Boyajian’s Star shows evidence of Solar System ices in absorption then, I think this idea might gain traction.

Nonetheless, the combination of a highly speculative cloud, a missing annual timescale, no obvious dimming of neighbor stars, and no clear sodium absorption feature cause me to give a subjective rating of unclear to this one.  That is, I’ll need some outer Solar System people to weigh in on likelihoods before I give it a grade.

Credit Where Credit Is Due

I should take this opportunity to note that many of these hypotheses are not original to Steinn and me.  Many have been suggested independently to us privately, both in person and over email, at Q&A sessions after talks, in social media, and among our professional colleagues.  If in this or future installments you recognize an idea as one you pitched to me, let me say:

  1. Thanks for your suggestion; it helped me think about the problem
  2. You weren’t the only one to suggest it, and you probably weren’t the first
  3. I’m sorry if you are not in our acknowledgements section; we could not acknowledge everybody or track down every suggestion, but even so our original acknowledgements were so extensive that the editor asked us to scale them back (Letters are supposed to be short).

OK, next time: Interstellar solutions!

What Could Be Going on with Boyajian’s Star? Part III: Periodic(?) Dips and Interstellar(?) Sodium

If you’re looking for a guide to this series, click here.

Last time, I looked at the infrared constraints on solutions to the puzzle of Boyajian’s Star.  This time, let’s look at other constraints.

Boyajian and others have noted that the deepest dips seem to have some patterns to them.

A 2 year period?

For instance, two of the deepest dips occur 2.000 years apart, which is an awfully precise number to be mere coincidence.

But, coincidence it is: remember that Kepler does not orbit the Earth, it orbits the Sun, and it does so in an Earth-trailing orbit, meaning that it has a different orbital period than the Earth!  So the relevant year is not an Earth year, but a Kepler year.  This makes those two dips’ interval less suspiciously precise: they are 1.96 Kepler years apart.  Also, if the Kepler orbit were responsible for those two dips, then one would also expect to have seen dips 0.98 Kepler years before and after the first of those—and we don’t.

A 48.4 day period?

Another suspicious bit of timing is that many of the deepest dips seem to occur with an interval of 24.2 days.  Boyajian noted that this looks like a 48.4 day orbital period, in which something luminous blocks the star every 48.4 days, and is itself blocked half a cycle later as it goes behind Boyajian’s Star.

I think it was Mark Conde who first pointed out to me that this was probably a coincidence.  I redid the calculation myself, and asked: if the dip times were perfectly random, how often would we be able to find a similarly strong coincidence?  I drew six times of dips randomly 10,000 times and hunted for periods that lined up at least as well as the 24.2 day coincidences in the timings of the six actual dips in the Kepler data.  Over 15% of the time, I found better coincidences, and among all of the best coincidences I found, the median period was 20.8 days.

It other words, with only 6 dips, it’s quite common to see phantom periodicities if you check enough periods.  True, these are a bit better lined up than random, but that’s probably why they were noticed and I did this analysis in the first place.

So, I don’t think there is any reason to suspect that the dips are periodic. That doesn’t mean they’re not, just that we should not feel constrained by any periodicities when looking for solutions to the puzzle.

Sodium absorption and reddening

Boyajian’s Star’s sodium lines are very broad because the star is rotating rapidly. Between us and the star, there are clouds of neutral gas, and this gas contains sodium and dust (and lots of other stuff).  As a result, the dust makes the star a bit too faint in the bluer wavelengths, and the sodium absorbs light at its characteristic wavelengths near 590 nm.  This is not unusual: at the nominal distance of Boyajian’s Star (450 pc) we expect there to be plenty of this interstellar material.

The reddening from dust makes the blue wavelengths about 10% fainter than it should be compared to green-ish wavelengths (we write E(B-V) = 0.1).  This also implies that about 35% of all of the visible light from the star in our direction is attenuated by dust before it reaches Earth.

We can also use the sodium lines to “find” the source of this dust, presuming that the sodium and the dust go together (a reasonable but imprecise assumption).  Here is what the sodium lines look like, (analysis courtesy of (newly-minted-PhD) Jason Curtis, Keck/HIRES spectrum courtesy of B.J. Fulton and Andrew Howard):

Boyajian's Star's sodium line. The big "bowl" is the star's absorption line, broadened by its rotation. The narrow features are due to intervening sodium gas.

Boyajian’s Star’s sodium line. The big “bowl” is the star’s absorption line, broadened by its rotation. The narrow features are due to intervening sodium gas.

Absorption from three clouds of sodium between us and Boyajian's Star

The narrow sodium features, modeled as absorption from three clouds of sodium between us and Boyajian’s Star by Jason Curtis.

The total sodium absorption we see (420 mÅ for the experts) translates to roughly E(B-V)=0.1, which is exactly what we see, so it all hangs together: at the time of these observations (2015) the sodium between us and Boyajian’s Star an the reddening were consistent.

This has implications for the long-term dimming seen by Shaefer and Montet & Simon: if it is due to dust and neutral gas, then these clouds are the ones responsible.  Note that none of the lines are centered on Boyajian’s Star’s line: this means they have significant line-of-sight motion with respect to Boyajian’s Star, making it unlikely (I would argue) that they are in clouds orbiting the star.  The RV of Boyajian’s Star is +4 km/s, and the red-most cloud is offset by ~10 km/s, so it has ~5 km/s motion with respect to the solar system barycenter.  This all argues to me that the clouds are what they appear to be and what we would guess they would be if Boyajian’s Star were not so weird: ordinary gas and dust in the interstellar medium.

Next time: what are the stars near Boyajain’s Star doing? Then, on to the hypotheses!

What Could Be Going on with Boyajian’s Star? Part II: Long-Wavelength Constraints

If you’re looking for a guide to this series, click here.

Last time, I laid out the background for Boyajian’s Star.  Let’s look at some of the observational constraints on families of possible solutions.

First, there’s the lack of infrared excess. This actually is the main reason Boyajian’s Star is weird. Other stars, like this “dipper”, behave similarly to Boyajian’s Star but have whopping infrared excesses. That is, there is a lot of extra infrared light in excess of what you would expect from a “naked” star. This is because there is a lot of circumstellar material around those stars in the form of a disk, and parts of that disk sometimes occult the star, causing big drops in brightness. In those cases there is some puzzle in the exact geometry of the situation, but there is no mystery regarding what’s causing the dips.

With Boyajian’s Star, the lack of infrared excess is very puzzling in the context of its dips.  If something is blocking the starlight, why is there no infrared light coming from it?  Massimo Marengo and Casey Lisse published papers recently showing that the lack of IR excess in archival data is also a lack of IR excess today, using recent Spitzer and IRTF data. This rules out many categories of solutions where all the circumstellar dust formed recently due to a planetary collision (for instance).

More recently, Thompson et al. showed that there is no detectable millimeter flux from Boyajian’s Star.  This upper limit is weaker, but rules out cooler dust.  Specifically, they claim there can be no more than 7.7 Earth masses of dust within 200 AU, and no more than about 10-3 Earth masses of dust at the orbital distances implied by the durations of the dips.  This rules out some origins for the long-term secular dimming seen by Schaefer: there can’t be a cloud of stuff around the star constantly blocking 15% of the light, or it would certainly intercept and reradiate more light than we see.  Here’s what I mean:

Spectral energy distribution of Boyajian's Star, with arrows indicating upper limits. The right-most measurements are from Thompson et al. They rule out big clouds of dust blocking 15% of the star's light in all directions, because the dust would reradiate at levels easy to detect.

Red points showing the spectral energy distribution of Boyajian’s Star, with arrows indicating upper limits. The right-most measurements are from Thompson et al. They rule out big clouds of 65K dust blocking more than 0.2% of the star’s light in all directions, because the dust would reradiate at levels easy to detect.  The two curves are models consistent with the short-wavelength data and differing only in the fraction of starlight being intercepted and re-radiated as heat.

In this figure, you can see the optical spectrum of the star on the left, with the red points being the measured brightnesses.  It follows a normal curve for a star.  Disks of the sort that cause the “dippers” are firmly ruled out: they have hot, close-in dust that have lots of emission at 5-20 microns, which we don’t see in the red points.

The black line is actually a model of the star that matches the detailed spectrum we really see 20 microns and shorter.  Longward of 10 microns, there are two curves: the black curve is what you would see if the star were surrounded by a cloud of dust absorbing 15% of its light.  I chose 15% because that is the total amount of dimming seen by Shaefer and Montet & Simon.  The red arrows are upper limits: the star is dimmer than that at those wavelengths. Clearly, the long-term dimming can’t be created by such a cloud, or the dust would be warm enough to be seen by Thompson at 1 mm ( =1000 microns).  You can’t save the hypothesis by making the dust colder, either, because that just moves the curve to the right.

In fact, if there is warm-ish dust aroud the star, it cannot be blocking more than 0.2% of the starlight (the purple curve), and even then it only squeezes between the existing upper limits if it’s at 65K, which is pretty cold.  This constrains possible solutions to the long-term dimming.

Now, if the dust is in an edge-on ring, then it could absorb only 0.1% (or less) of the starlight in total, but still block 15% along our line of sight. In that case, it would not be seen by the current observations, as long as it was cold enough.  So: disks and clouds are out, but rings (and comets) are still allowed by the IR and mm observations.

Ok, that’s enough words for one post.  Next time: If the long-term dimming is going on right now, shouldn’t we be able to determine its composition from absorption features in the star’s spectrum?  Also, aren’t the dips periodic, and can we predict the next one?

What Could Be Going on with Boyajian’s Star? Part I

If you’re looking for a guide to this series, click here.

I’ve just re-submitted a paper, which I expect to be accepted soon in light of a very favorable and constructive referee’s report, about KIC 8462852.  It covers the puzzle as I see it now, the landscape of solutions—from the plausible to the all-but-impossible—and opportunities for future progress. I’m especially excited for the GAIA parallax, which will really help us to narrow things down!

So I’m going to slow-blog the paper over the next week or so, up until the day it actually appears on the arXiv (after it’s accepted).

OK, let’s go!

The Story So Far

Dr. Tabetha Boyajian (now an assistant professor at LSU) first announced KIC 8462852 to the world in a paper that described how it was discovered, how it is an apparently ordinary F star except for its bizarre series of photometric “dips,” and how hard she and her co-authors, including Prof. Saul Rappaport, worked to solve the puzzle.  I’ve blogged previously about the whole story, including the subsequent media coverage here, and you should read that to catch up if you’re not familiar with the star.  I also gave a talk about it that you can watch here:

Then Bradley Schaefer looked at old DASCH photometry and found that Boyajian’s Star has been fading over the past 100 years, a claim at least as extraordinary as the star’s Kepler light curve.  There was a lot of shouting over this claim, and whether it was right, which Kimberly Cartier and I documented in an article for the Atlantic, linked in this blog post.

Ben Montet, looking up, presumably at Boyajian's Star.

Ben Montet, looking up, presumably at Boyajian’s Star.

The big news recently is that Ben Montet & Josh Simon very cleverly recently used the Kepler full-frame imagery—some calibration data that doesn’t get much attention because you can’t use it to find planets—to get accurate long-term photometry of Boyajian’s Star over the course of the mission.  Amazingly (to everyone but Bradley, I suspect), they found that the star got 4% dimmer over 4 years, in a monotonic but irregular way.  What’s more it is the only star out of > 200 that show this effect.

In my opinion, this independent confirmation of the unprecedented effect Schaefer claimed—even if not covering the same time period—shows that Shaefer’s analysis is correct and the star really has dimmed a lot.  Adding the two effects, the star is now apparently at least 17% dimmer than it was in 1890!

Where We Are Now

So what the heck is going on?  We now have two inexplicable things going on: long-term, secular dimming of 17% in 115 years, and these days-long, deep “dips” of up to 22%.  Both are very hard to explain.

Well, Occum’s Razor points towards a single explanation for both the secular dimming and the dips. Of course, in principle they may be unrelated, in which case Boyajian’s Star is extraordinary for two independent reasons. But I favor explanations that could plausibly cause both.  At any rate, I’ve been engaging in a lot of “clean-sheet” reasoning lately, trying to cover all of the bases, and working with a lot of people to figure out what’s plausible.  Steinn Sigurdsson, my often-time theorist sounding-board, is a co-author.

We decided that the explanation for the dimming must occur either at Boyajian’s Star itself (intrinsic variability), in orbit around it (circumstellar material), between the Sun and Boyajian’s Star (interstellar material) or in the Solar System (either between us and the star or in our instruments).  In future installments, I’ll cover the observational constraints, including dip periodicities, long-wavelength constraints, and the brightnesses of nearby stars, and then go through each possibility and give it a plausibility.

Wait, What Did You Call the Star?

Prof. Boyajian, looking up, presumably at Boyajian's Star.

Prof. Boyajian, looking up, presumably at Boyajian’s Star.

I’ve decided to call it “Boyajian’s Star“.  Prof. Boyajian herself calls it the “WTF” star, ostensibly after the subtitle of her paper (“Where’s the Flux?”, natch).  Early on during the media firestorm, a reporter asked me what I call it, and I admitted I could never remember its phone number and that within my group we called it “Tabby’s Star,”  because she first showed it to us, back before it was published. The name stuck, and now that’s what it’s usually called.

I’m conflicted about that term.  On the one hand, it has ensured that Tabby gets the credit for her work establishing how strange the star is, which is great (I wish I could claim that’s why I first used the term! It’s certainly why I kept using it).  On the other hand, that name is actually the most common among professional astronomers, which is incongruous with other eponymous stars.  You see, there’s a long tradition of naming stars after the people who made them famous—Barnard’s Star, Kapteyn’s Star, Przybylski’s Star—but professionalism and formality dictate that we always use the namesake’s surname.  Not only does “Tabby’s Star” use her given name, it uses the diminutive form.  It reminds me too much of one of the sexist double-standards common in academia (e.g. a man is “Prof. Smith”, but a woman is “Jenny”, especially from undergraduates!)

So, I’ve decided that when it comes to the professional literature, I’ll follow tradition and call it “Boyajian’s Star”.  If other astronomers do the same, perhaps it will stick.

The paper is Wright & Sigurdsson, “Families of Plausible Solutions to the Puzzle of Boyajian’s Star,” and I’ll dribble out more about it in future installments until the arXiv posting.

Part II is here.

Update: By popular demand, Tabby has given us a pronunciation guide for “Boyajian”. She says “boy-AH-zhun”, with the j sounding like that in “Jean-Luc Picard” (that is, say “sh,” but use your voice, don’t whisper).