As if the Kepler star KIC 8462852 weren’t interesting enough, Bradley Schaefer (Louisiana State) added to the controversy when he discovered what appeared to be a steady dimming of the star over the past century. Schaefer called the result “completely unprecedented for any F-type main sequence star,” and given the discussion about KIC 8462852 as a SETI target, this raised the stakes. Something just as odd as the object’s strange lightcurves was going on here, and it seemed natural to think that the dimming and the lightcurves were related.
But Michael Hippke now begs to disagree. An old friend of Centauri Dreams (see, for example, his Exomoons: A Data Search for the Orbital Sampling Effect and the Scatter Peak), Hippke takes a close look at Schaefer’s work and reaches a different conclusion. As he sees it, the ‘dimming’ of up 0.165 ± 0.013 magnitudes per century in this F3 star may actually be the result of imperfect calibration on the Harvard plates. In other words, while the lightcurve anomalies remain, the dimming may well be a data artifact rather than an astrophysical enigma.
Image: KIC 8462852 as photographed from Aguadilla, Puerto Rico by Efraín Morales, of the Astronomical Society of the Caribbean (SAC).
First, though, a word about Bradley Schaefer’s work, about which Hippke says “Schaefer had the excellent idea to look into the old plate archives. To solve this mystery, we need all the information we can get, and Schaefer did very careful and high-quality work.”
This parallels comments I’ve heard from other professionals, who praise the quality of Schaefer’s analysis. Submitted to Astrophysical Journal Letters, the Hippke paper looks to contrast the ‘dimming’ of KIC 8462852 with an analysis of other F-type main sequence stars from the same dataset. Along the way, Hippke double-checks Schaefer and finds sound work:
Although the process of data cleansing and binning involves arbitrary choices, we have reproduced this part of the analysis for all variants with virtually identical results. It is therefore important to note that the method and results in Schaefer (2016) appear to be adequately careful and accurate. In the following, we will thus concentrate solely on the interpretation of his result – whether the dimming is “unprecedented”.
Take away its odd lightcurves and KIC 8462852 appears to be a relatively normal star. Thus Hippke’s criteria for study are F-stars from the Kepler field of view, from which photometry is studied for the 3 most quiet F-dwarfs and 25 bright F-dwarfs in the Harvard DASCH (Digital Access to a Sky Century @Harvard) archive. Trends in the data may not, Hippke believes, be slow drifts but ‘structural breaks’ — in other words, changes caused by abrupt changes in technology or calibration techniques. Evidence for this occurs not only for KIC 8462852 but also for KIC 7180968, indicating we are dealing with a phenomenon not isolated to KIC 8462852.
From the paper:
The significant trends (and/or structural breaks) found in 18 of 28 comparison stars support the interpretation that the dimming of KIC 8462852 is not extraordinary. A careful analysis of each dataset is time-consuming, which is why we have not performed this analysis for hundreds of stars. In case of further doubt on the significance of such trends, the analysis presented could simply be expanded to more stars.
This would make an astrophysical interpretation of the ‘dimming’ unlikely because it would require that a number of main-sequence F-dwarfs fluctuate by 10% or more over the course of a century. “It seems more likely,” writes Hippke, “that the change of emulsions, errors in calibration etc. cause these trends.” In an email just received, Hippke notes of Schaefer’s work that “It might just be that his check stars were unusually stable, which obfuscated existing trends in the data.”
Thus the paper favors the notion that changes in technology and imperfect calibration — quality issues in the dataset itself — explain what otherwise appears to be long-term dimming of KIC 8462852. This leaves us, as the author notes, with the short-term dimmings found in KIC 8462852’s lightcurves, a problem that the question of century-long dimming does not address.
What can be done to investigate the dimming issue further? Hippke’s email suggests that other data, particularly plates from the Sonneberg Observatory in Germany, will be useful for comparison. “Unfortunately, these plates are not available online and have only partially been scanned, so checking these data might take several months.”
The paper is Hippke, “KIC 8462852 did likely not fade during the last 100 years,” submitted to the Astrophysical Journal Letters (preprint). Bradley Schaefer strongly disputes Hippke’s work, so we haven’t heard the end of this.
Well now, this is interesting.
“The two comparison stars used by Schaefer (2016)
(TYC 3162-1001-1 and TYC 3162-879-1) show a constant
luminosity (within their errors). For our 28 comparison
stars, we find a constant luminosity for 10 of 28
stars (36%). Assuming that all stars have been drawn
randomly from the same sample, the chance of drawing
2 of 2 constant stars is 13%. It might be attributed to
bad luck that these apparent data discontinuities were
not seen in the first place.”
I’m a little shocked that Schaefer only used 2 control stars. Is that correct? If so, maybe I should have read Schaefer’s paper a little more closely. An N of 3 seems very small to draw such broad conclusions.
What about the check stars?
Schaefer chose 2 nearby check stars, saying “I will report here in detail on only the two check stars closest in color and brightness to KIC8462852,” which means that these are presumably also F type stars.
I cannot assess from here whether Schaefer’s check stars were always on the same plate as KIC8462852, but if they are truly angularly close (which makes that likely), I do not see why Hippke and Angerhausen’s finding that for some other stars there are “breaks,” and that
“The structural breaks appear most prominent at the ‘Menzel gap,’ but might also be present at other times. These issues might arise from changes in technology, and imperfect calibration.”
is necessarily relevant. I can easily imagine reasons why different ways of making or processing plates might cause changes from one plate to another, but it is much harder to understand how there would be systematic magnitude changes for one star on a plate, and not for another star of similar color and brightness nearby on the same plate.
Clearly, resolving this issue will require a lot more examination of stars on these plates – for example, by finding check stars for each of the stars that do show breaks in Hippke and Angerhausen.
At first report, it struck me as puzzling why changing tech would affect HIC 8462852 and not the other 2 F stars, but, I don’t know all the details (same plate etc). This is turning out to be a wonderful ride…
I hate to sound so boneheaded but I cant get my head around this statement:
“Hippke notes of Schaefer’s work that “It might just be that his check stars were unusually stable, which obfuscated existing trends in the data.”
If the check stars were shining with unchanging magnitudes and that was evident in the plates then doesn’t that mean the plates portray magnitudes correctly (within error) for ALL of the stars? How can it be both ways?
Coacervate, I would also hope that the Stable F-Types in question would act as benchmarks to normalize, not randomize the data.
What about Schaefer’s two control stars, which did not show dimming? How does Hippke deal with that?
Well, it was a fun ride. Still might be. Keep it up!
This “ride” is far from over.
Getting to be like a TV mystery show.
” Assuming that all stars have been drawn
randomly from the same sample, the chance of drawing
2 of 2 constant stars is 13%.”
But were the 2 calibration stars in the Schaefer paper chosen randomly? Were they not on the same plates?
I kinda emailed Schaefer about the century long dimming, suggesting it would have to be something like a hydrogen build up between us (a cloud moving our way large enough to envelop the system) and kic8462852 which he rejected as hydrogen would have blocked certain wavelengths. I asked him if this phenomena was apparent in close proximity stars (because something like that would be/ could be affecting multiple systems) and he didn’t email back. Oops.
Gas is there but not over the star from what I can see, but it may have affected the other control stars.
http://aladin.u-strasbg.fr/AladinLite/?target=kic%208462852&fov=0.05&survey=P%2FFinkbeiner
Forgot to add you need to zoom out a bit.
If the gas moves with significant velocity it is possible that the ‘veil’ of gas has altered the brightness over time. Do the other controls stars have any identifiers so we can see if they are affected by the gas and dust?
Talking to myself here please ignore, if you look at the absorption spectrum of hydrogen there are many that are close to the spectrum of the star. Perhaps hydrogen from an evaporating planet/moon?
https://www.learner.org/courses/physics/visual/img_lrg/Hydrogen_spectrum.jpg
Page 8
http://arxiv.org/pdf/1509.03622v2.pdf
Here is what WISE sees in IR around the star, you can see tenuous gas clouds around.
http://aladin.u-strasbg.fr/AladinLite/?target=kic%208462852&fov=1.73&survey=P%2FallWISE%2Fcolor
So a constant flow (in excess of a century) of hydrogen in our direction flowing over the star system. The planets act as scoops through ice cream as gravity pulls in the hydrogen. This leaves ‘hydrogen shadows’ where there is no hydrogen. Rotating planets create a pressure shift where hydrogen flowing over the surface of the spinning ‘wing’ is escaping at different velocities, so there are vortices in the hydrogen flow.
What Schaefer did point out was it didn’t really explain the problem of the daily fluctuations…
I once heard that an area of radiation in space can cause dimming of light passing through it, but that might have been a scifi film rather than a documentary…and the prospect of a radiation cloud headed our way…
I have to agree that if some F-dwarfs show “breaks” in their otherwise flat brightness curves, and others do not, the cause cannot be changes in the sensitivity of the plates for whatever reason. The changes in brightness as seen from Earth must be real.
Perhaps nobody had bothered to monitor large numbers of F-dwarfs for changes in brightness to the needed accuracy before, and the existence of a new type of (fairly common) variable had been overlooked until now?
I am yet to be convinced that the “structural break” during the Menzel gap is solely due to an artefact of the data.
Hippke and Angerhausen state that they show a “structural break” at 12? confidence for KIC8462852 during that time. They mention a 10? confidence “structural break” for KIC7180968.
Finding one other significant structural break in 28 stars during the Menzel gap I would think is insufficient to completely discount the possibility of a real dimming event for KIC8462852.
We need further data from the likes of Palomar POSS and POSS II, etc to resolve this issue.
This is KIC 8462852 and KIC 7180968 gas image, you can flick to IR surveys as well and you will see next to no gas, zoom all the way out, so given the fact that there is gas in the area of Tabbys star (line of sight) it is more likely the dimming events are true but to what extent I don’t know.
http://aladin.u-strasbg.fr/AladinLite/?target=KIC7180968&fov=0.05&survey=P%2FFinkbeiner
What was NOT mentiones in ANY OF THIS was the MAJOR DIPS in bin 3 AND bin 4 of the Schaefer analysis. I believe that BOTH of these are Kepler-LIKE dips. What I need to know is; did ANY of Hippke’s F stars show BIG DIPS LIKE THESE! The proof(or DISPROOF) of Schaefer’s hypothesis will come ONLY from FILLING IN THE MISSING TWENTY TEARS of data BETWEEN the DASCH AND Kepler lightcurves, as I have STRONGLY STRESSED IN TWO PREVIOUS COMMENTS, by checking ALL AVAILABLE CCD OPTICAL IMAGES(which I ASSUME have a MUCH MORE STABLE calibration format) for a 4-5% dimming trend! It is IRONIC that this kind of calibration discrepancy ALSO led to so many FALSE DETECTION OF EXOPLANETS by Van De Kamp and others. It is a BLESSING that CCD technology has ELIMINATED ALL of these discrepancies FOREVER!
Couldn’t agree more. Shorter , reliable CCD data sets on more stars. Trouble with this sort of research too is it takes a lot of dedication to search historical data sets ( especially if not digitalised ) and at best for a negative result ! The dreaded negative publication bias strikes again . But this is falsifiability which is what science should be all about. Not exactly going to be popular with Journal Editors who are under pressure to sell copy via exciting or better still ,controversial positive results.
But it has to be done even if it means we hear the trundle of a Dyson Sphere running way down the street.
Thankyou Ashley for that important reminder. All too often I forget the toil that is personally demanded by researchers mining the data as you point out, especially for a null result. Good to remember.
I wonder if it possible to reconstruct the technology and methods used for the older images and repeat the observations but at the present time and compare the results to modern equipment and methods?
The idea that it is an artefact in the data arising from technological changes etc sounds credible at first sight, but might hopefully be a testable idea…
A fair comment but as stated above even modern photographic plates don’t have anything like the sensitivity of CCDs and it’s only really now with digital sensor arrays that the sort of precision observations required for the Schaefer conclusions can be made . Comparing CCDs to photographic emulsion , oranges and pears. Calibration will always be an issue on top of that but easier to standardise/quantify with digital arrays.
Just checked Jason wirght’s tweets! He says ” Schaeffer has some issues with Hippke et al’s paper, and will respond today”. You might wand to check with him PERSONALLY before posting thsi comment. ALSO: What if Hippke is right about the data, but NOT about the INTERPRETATION! In other words, what if the other 18 F stars WERE dimming in a similar fashion to KIC8462852, but just happened to NOT show BIG SHORT DIPS in the 4 year Kepler lightcurve timespan? Could this be interpreted as Kepler-LIKE dips being CYCLICAL to these stars, but for only like ONE 5 YEAR PERIOD EVERY CENTURY or so. Bins 4 and 5 around 1900 would TEND TO SUPPORT this conjecture! If this WERE the case, the CAUSE would ALMOST CERTAINLY be of INTERNAL ORIGIN! NO COMET DUST OR ALIEN MEGASTRUCTURES NEEDED!
I would second Schaeffer’s opinion. I decided to check the data on all of the stars that Hippke used in his paper. So far I have checked four of the stars he uses, and two of them are listed as variable stars (including one gamma Dor variable), and one is a multiple star. Only one out of the first four stars I checked is not known to be a multiple or variable star.
TERRIBLE methodology on the part of Hippke, if this continues with the rest of the “comparison” stars.
I’ve completed my check into the published spectral information of the 28 stars that Hippke used in the paper released on ARXIV yesterday.
I was unable to get spectral information on four of the stars. Of the 24 stars with spectral information, nine were variables (three gamma Dor type, two delta Sct type and four rotationally variable types), one was an eclipsing binary of Algol type, and three more were multiple stars. That leaves only 11 single, non-variable stars. Of those 11 stars, one was a G-type star and six were evolved non-main-sequence stars. One is of unknown evolutionary status. That leaves only THREE single, non-variable, main-sequence F stars:
KIC 10010623 F3V C 1.78
KIC 11498538 F2V C 3.02
KIC 12453925 F3V C 3.94
KIC 8462852 F3V C 9.88 (“Tabby’s Star”)
The three stars in the paper that are actually comparable to KIC 8462852 in their basic characteristics are all fairly stable, much more so than KIC 8462852 (where the last value is the significance in sigmas (according to Hippke) of rejecting a flat line for the brightness).
I’m surprised the gamma Dor weren’t caught by their CDDP screen (the delta Sct might have been too quiet). In any event, while I agree that you probably want to get the most comparable comparison stars, the time scales of variation on those shouldn’t have affected Hippke’s structural break analysis.
How could Hippke have missed that? Its almost as if the KIC 8462852 anomaly is causing everyone to become crazily sloppy.
Wow, very thorough check, thanks. Well, while I think ‘Variable’ may have added something to consider re. timescales, I do find the methodology of choosing different types of class and different class stars for ‘long duration’ comparibility as somewhat puzzling.
All of the stars mentioned lie in gas free areas, Tabby’s star lies in a gas rich light of sight, so it is not unreasonable to have dimming due to moving clouds.
I TOTALLY agree with HARRY. ALSO: my keyboard has a BROKEN CAPS LOCK key that seems to BE TRIGGERED randomly when I hit the SPACE BAR. I think Harry might have the SAME ONE, but that’s a matter of INTERPRETATION!
LOL
Much respect to the scientists who keep working on this mystery. They’re not only offering their personal work and expert insights for indiscriminate worldwide scrutiny, but they don’t even have the comforting option of using pseudonyms. :)
The real disappointment with this turn of the discussion is that zombie comets may be back. While interpretations of the Harvard plate data are reasonably uncertain, don’t the lead hypotheses remain narrowed to (a) disintegrating monster comets, (b) a previously unknown natural phenomenon, and (c) ETI? Maybe even in that order.
Of course, science is science, objective and dispassionate — but I’m optimistically looking forward to TESS “the comet slayer’s” launch next year.
I second that – these guys are doing amazing work.
It is not often that we see the “scientific method” in real time as argument and counter argument play out to refine hypotheses.
The data for KIC 8462852 (whether binned or not), in my opinion, shows a fair degree of stability for all but during the “Menzel gap” and in the first decade or so of the Harvard plates. The long term measurements appear not to support a “structural dimming” as any dimming/brightening observed are within the “error” bounds of the data.
I have reduced confidence in the early data as there are only small numbers of observations in a lot of those years.
1 for 1889 and 1891, 3 for 1890, 6 for 1893 and 1894, 7 for 1892 compared to over 100 observations each for 1944 and 1945.
That leaves us with a “structural break” during the Menzel gap as the only significant “event” for KIC 8462852 in the historical data. This “event” needs to be checked carefully with other data such as POSS and POSS II and Sonneberg.
Did it occur gradually over the decade or so from 1952 onwards, was it a once off event that left a dimming in KIC 8462852 or is it just an artefact of the data? That to me is the important question that remains unresolved.
I don’t think the zombie comets ever were away. They are a reasonable explanation, and the present long-term dimming controversy does not change that. Regardless of the outcome.
Well, a precedented explanation is certainly sorely needed at this point, but as I follow the discussion I’m just not so sure that comets hold up any more. Dr. Schaefer indicated in ScienceNews today that disintegrating monster comets and ESI have both been effectively refuted, leaving only some form of the Unknown Option:
“Schaefer says[:] ‘We’ve effectively refuted every proposal on the table. Either there’s some completely new idea or we’re doing something wrong.'”
https://www.sciencenews.org/article/odd-star’s-dimming-not-aliens’-doing
In partial contrast, Dr. Wright offered a generous and clear explanation here several discussions back (KIC 8462852: A Century Long Fade?) why the “thermodynamic rebuttal” does not necessarily eliminate an ETI explanation. But as quoted in the same ScienceNews article, even Dr. Boyajian seems skeptical of comets now:
“A comet horde, however, doesn’t explain why the star faded through the 1900s. ‘That’s the second piece of evidence that shows this star is really weird,’ says Boyajian. ‘It’s a very frustrating piece of evidence as well…. It doesn’t point to an obvious explanation for what to think of next.'”
It helps me, at least, to try and rank the leading choices that are able to explain all of the observations, and today it really looks like we’re down to (a) a previously unknown natural phenomenon or (b) ETI. It can’t hurt to leave room for interesting and otherwise rational compound-assumption hypotheses like discarded stellar engines. But by most current accounts, “Insane Comet Posse” sounds more like a hip hop cult band (for astronomers).
You cannot refute a reasonable explanation (comets) of a phenomenon (the light dips) by pointing to a different phenomenon (the dimming) and claiming: “It must have the same explanation, but it can’t be this one”. It is a self-contradicting argument.
Rather, the most reasonable explanation for the second phenomenon (the dimming) is that it is a false positive observation, which is what Hippke set out to show. This is the right thing to do. If he did not execute it well, someone else should. And I would not take Schaefer’s word on that particular point.
Bradley Schaefer’s response re Michael Hippke’s paper will be online here in about 30 minutes.
Does anyone know if the full 28 graphs in the Hippke and Angerhausen are available? We only got to see 2 graphs in the paper.
Would love to see the full range of data they used be made publicly available.
It is something I feared in that the plates could had deteriorated and I hope it is sorted. If you see this image of the local line of sight area to Tabby’s star there is gas and dust and dimming could occur over time.
http://aladin.u-strasbg.fr/AladinLite/?target=KIC%208462852&fov=12.12&survey=P%2FDSS2%2Fred
Well done to Centauri dreams and all subscribers an important issue is being played out , and hopefully resolved , in real time. Could have taken months for these ” letters to the editor ” style replies to have been assimilated in the “old days” , but now it’s just hours. I talked about the need for falsifiability and publication of negative results but these are now being addressed much more excitingly in this format but with no less quality than a journal. Anything that does this has got to be good for Science which is what this site is all about.
I’m enraptured despite the nature of this being deeply technical needing expert interpretation.
Well done Paul. It’s been a tremendous last couple of weeks really with hardly a breath drawn between this and Proxima Centauri. Anyone who said science was boring should follow it this way and would soon change their mind. Centauri Dreams is offering a near peer review function .