≡ Menu

KOI-456.04: Earth-like Orbit Highlights New Detection Tools

The planet candidate KOI-456.04 strikes me as significant not so much because of the similarity of its orbit with that of Earth (a 378 day orbital period around a star much like the Sun), but because of the methods used to identify its possible presence. Make no mistake, this is still very much a planet candidate, as co-authors René Heller and Michael Hippke are at pains to explain, noting that systematic measurement errors cannot be ruled out, though they estimate an 85 percent likelihood that it is there.

We don’t have many examples of small planets potentially in the habitable zone of a star like ours, and this is what has received the most media attention. So let’s look at this aspect of the story quickly, because I want to move past it. If this candidate is confirmed, it looks to be less than twice the radius of the Earth, receiving about 93 percent of Earth’s insolation from its star. Make assumptions about its atmosphere and you can arrive at a surface temperature averaging 5℃, 10 degrees lower than Earth’s mean temperature.

Image: Most of the exoplanets from the Kepler mission are the size of Neptune and in relatively close orbits around their host stars, where temperatures on these planets would be far too hot for liquid surface water (third panel from above). Almost all of the Earth-sized planets known to have potentially Earth-like surface temperatures are in orbit around red dwarf stars, which do not emit visible light but infrared radiation instead (bottom panel). The Earth is in the right distance from the Sun to have surface temperatures required for the existence of liquid water. The newly discovered planet candidate KOI-456.04 and its star Kepler-160 (second panel from above) have similarities to Earth and Sun (top panel). MPS / René Heller.

But what I want to dwell on is the methodology used to study this system. Heller (Max Planck Institute for Solar System Research) and Hippke (Sonneberg Observatory, Germany) are joined here by colleagues at the University of Göttingen, UC-Santa Cruz and NASA Ames in a new look at archival data from Kepler on the star Kepler-160 in Lyra, which was observed by the mission between 2009 and 2013. The star is similar to the Sun in mass and radius and previously known to have two confirmed planets.

The new work analyzes transit timing variations in the orbital period of the planet Kepler-160c suggestive of a third planet. They find Kepler-160d, a third world that is disturbing the orbit of Kepler-160c. This is a planet without any transits that is thus only indirectly confirmed.

The intriguing candidate, potentially the fourth planet here, is KOI-456.04, which appears to be 1.9 Earth radii in an orbital period of 378 days. The Max Planck Institute for Solar System Research (MPS) happens to be building the PLATO Data Center, and the suggestion is that the PLATO mission, to be launched in 2026, will have the chance to confirm this interesting object of interest and study it in much greater detail.

Heller and Hippke have been developing their exoplanet detection pipeline in several recent papers, studying twelve detrending algorithms for stellar light curves in detail. ‘Detrending’ refers to eliminating noise within transit data to cull out evidence for a planet. The results pointed to a detrending algorithm available in the open source package called Wōtan, used in combination with a transit search algorithm known as ‘transit least-squares’ as the most accurate choice. Heller and Hippke developed TLS specifically to look for smaller planets by modeling stellar limb darkening (see Dataset Mining Reveals New Planets for more on this).

What emerges is a more precise model of the brightness variations seen in a transit event, one that the duo believe improves upon the more established ‘box-like’ approximation known as the ‘box-fitting least square’ (BLS) algorithm. The latter is somewhat faster in computational terms, but the Wotan/TLS combination is in the authors’ view more sensitive. I talked to both Heller and Hippke about the new paper via email and asked Hippke about the advantages of their method. His response:

I… believe that Wōtan+tls are the leading toolset in finding new transiting exoplanets. You gain about 10% sensitivity going from BLS to TLS. In other words, at the same false alarm rate (e.g., 1%) you get 10% more planets from TLS. Naturally these are at the small end of the size distribution (you find large planets as easily with BLS). Smaller planets are usually more interesting because rocky planets are believed to be < ~ 2 Earth radii.

The dominating noise source in transit observations is in many cases stellar variability, which is why Heller and Hippke tested a dozen detrending methods, all of which are available through Wōtan (TLS is likewise an open source tool). According to Hippke, the Wōtan methods are more important for more active stars — remember that M-dwarfs can be quite active in comparison to G-class stars like the Sun. Young stars just at the end of planet formation are likewise active, making them interesting targets for using the Wōtan tools to achieve optimal detrending for exoplanet detection.

Heller told me that the team is 100% sure that Kepler-160d exists — this is the non-transiting world found through using transit timing variations of Kepler-160c. But what of the planet in the orbit roughly similar to that of the Earth around the Sun?

Our statistical analysis gives us 85% confidence that the signal belongs to a transiting planet. But 99% would be needed to call this a planet. In this case, this object would be called Kepler-160e. For now, it is not. So this object is transiting (I mean, if it is real in the first place), but we are less certain than for the non-transiting planet Kepler-160d that it actually exists. And so KOI-456.04 remains a candidate unless someone can show that it exists with more than 99% certainty.

Thus the tantalizing ‘world’ in the Earth-like orbit remains a Kepler Object of Interest (KOI), an object that cannot be currently validated or falsified, but one that will doubtless be on the target list for the PLATO exoplanet mission. The larger story is that the tools Heller and Hippke have deployed show the promise of pulling 10% more planets (and smaller ones at that) out of the raw data, which makes analysis of ongoing observations as well as reanalysis of older datasets more accurate. It will be fascinating to watch as the computational methods on display in this paper are applied to other known exoplanet systems, with their validity then put to the test by future space- and ground-based observatories.

The paper is Heller et al., “Transit least-squares survey. III. A 1.9 R⊕ transit candidate in the habitable zone of Kepler-160 and a nontransiting planet characterized by transit-timing variations,” Astronomy & Astrophysics, Volume 638, id. A10 (June, 2020). Abstract/preprint.


Comments on this entry are closed.

  • EricSECT June 9, 2020, 9:11

    I say we need to start considering why we do NOT see smaller terrestrial planets in a star’s habitable zone. Jupiter’s Grand Tack cleared out most of the material interior of Mars’ orbit, so that no super Earth(s) could form (as yet, still unknown if they could support life).

    And then Saturn formed just in time, at just the right size and at the proper 3:2 orbital resonance to pull Jupiter back, or it would have ended up “hot”.

    This scenario MUST be exceedingly rare.

    Now….. Toss is The Big Splat to form Luna, which is necessary to stabilize Earths axial tilt, enlarge our iron core and strengthen it’s magnetic field and possibly start plate tectonics…. which yielded a multi billion year stable climate so that complex creatures and intelligence may arise….

    …and you have a Rare Earth indeed. The answer to the Fermi Paradox?

    • Alex Tolley June 9, 2020, 15:36

      Douglas Adams:

      “This is rather as if you imagine a puddle waking up one morning and thinking, ‘This is an interesting world I find myself in — an interesting hole I find myself in — fits me rather neatly, doesn’t it? In fact it fits me staggeringly well, must have been made to have me in it!’

      We really need evidence that these events are required for the emergence of intelligent life.

      • EricSECT June 9, 2020, 21:05

        Seems what is MOST needed for emergence of complex life (and then maybe intelligence…. and then maybe technology that we might converse with) …. is a stable climate for about 4 billion years.

        I’m certain the cosmos, and about 6 worlds in our system…. are rotten with prokaryotic life that can weather most anything.

        But…. Anything more complex than that? Nope.

        • Eniac June 13, 2020, 13:37

          Seems to me the climate wasn’t really that stable, with snowball Earth, great oxidation event, and all that. Plus, most of the time life was confined to the oceans, which naturally provides some measure of stability, regardless of axis stability or any of the other “rare” factors you invoke. There is even life deep underground, where none of that matters.

          There are plenty of cold Jupiters out there, so that part of your theory also strikes me as fanciful.

          • Michelangelo Pantaleoni June 15, 2020, 8:29

            I agree with EricSECT. Also he the snowball Earth and the great oxidation event happened before there was any complex life on Earth. The last 500 Myr have indeed been quite stable climatically

      • Eniac June 13, 2020, 13:53

        Love this!

    • charlie June 9, 2020, 16:34

      “And then Saturn formed just in time, at just the right size and at the proper 3:2 orbital resonance to pull Jupiter back, or it would have ended up “hot”.”

      “hot”?? What does “hot” mean ??

      • Harold Shaw June 9, 2020, 17:52
      • EricSECT June 9, 2020, 20:54

        A HOT Jupiter.

        A near Jupiter mass solo planet, whose orbit around it’s primary lasts but days.

        As it was the first to form and spiraled inward, it tossed all other planets aside.

    • Nicky June 9, 2020, 16:53

      We know exactly why we don’t see smaller terrestrial planets in a star’s HZ, assuming you are talking about solar-like stars.

      Earth-like planet has transit depth of 90-ppm, and a 13th-mag solar-like star has random noise of 40-ppm in 6.5-hr integration. For a transit duration of 12-hr, you get S/N 6.11 after observing four transits.

      Thats where the problem is. The transit recovery rate at S/N 6.1 by Kepler pipeline Transit Planet Search (TPS) algorithm is 5.2% (Christiansen, 2017). In other words, 94.8% of planets are missed at S/N 6.11

      The next step is to pass the TPS detected transit-like signal to Robovetter, which assigns False Positive or Planet Candidate to every detection. But Robovetter uses S/N 7.1 as a threshold to filter false alarm rate, so any planet with S/N < 7.1 is automatically excluded.

      So detecting "smaller terrestrial planets in a star’s habitable zone" with Kepler pipeline is almost zero chance right now. Thats why there's TLS developed by Heller and Hippke, optimized to detect low S/N planets, and the search has just begun.

      • Enzo June 9, 2020, 18:08

        Interesting. Is there a similar argument for K stars ? Because it should be a lot easier for them.

    • Enzo June 9, 2020, 18:05

      Ever since the first Kepler data started rolling in, I noticed a dearth of earths (earth size in the conservative, read realistic, HZ). Ok, maybe the G stars are noisy etc. but for K star it should have been a lot easier, right ? There are a lot more of them, the eclipse peak is deeper etc.. But no, only around red dwarfs they seem to be more common. How earth-like they really are, well, that’s a different matter.

      BTW, KOI-456.04 is nearly 2x earth’s diameter, wasn’t ~1.5x sort of the top to be expecting it to be earth-like ?

      Finally, the latest David Keeping video on Fermi paradox is a must see :

      There’s also a paper I haven’t looked at though :

      • Nicky June 10, 2020, 17:03

        Not exactly. Kepler’s observation strategy was designed to find transits near solar-like stars, so the number of K-dwarfs is subjected to Kepler’s limitations in magnitude and stellar luminosity. The Kepler sample is dominated by early G-dwarfs to F-dwarfs due this selection bias, very few K-dwarfs observed.

        Assuming stellar radii 0.78 solar value, Earth-like transit yields depth around 140-ppm, with typical random noise of 40-ppm in 6.5-hr integration. The S/N is 9.1, with Kepler TPS completeness 66%. Robovetter further removes 20% to false positives, leaving completeness around 50% for early K-dwarfs.

        • Enzo June 10, 2020, 21:42

          Thanks. Does this takes also into account the substantial shorter orbital period of a planet in the HZ for K stars (i.e. more transits are observed for a given time, better SNR) ?

          Also, I was trying to find out the distribution of the Kepler stars by spectral type with little success. The closest thing I found is this :

          that says that ~61% of the stars in the FOV are dwarfs.

          • Nicky June 11, 2020, 17:40

            Yes, I did take the number of transits into account. There is some promising HZ candidates around early K- and late G-dwarfs, but Gaia DR2 revised most stars radii upward and found a lot of stars to be more evolved.

            Most of these candidates cannot be statistically confirmed, which requires 99% confidence in planetary hypothesis. Thats why you rarely see any K-dwarfs and G-dwarf HZ planets, because even if they are detected it is yet statistically impossible to confirm, so they remain as candidate.

    • Harold Shaw June 9, 2020, 18:26

      Earth’ stable axis and magnetic field are clear advantages that make life and space faring life much more likely. Where your reasoning breaks down is the assumption that the way Earth gained both is the only way. A planet that spins faster, is tidally locked, or has a retrograde spin will have a stable axis. Since one planet’s precession is caused by other planets, there might be system configurations that produce little to no precession. As for a megnetic field, a planet could simply form with all the requirements.

      It isn’t enough to understand our path to a cozy planet if there are multiple paths.

    • Ashley Baldwin June 9, 2020, 19:15

      Richard Feyman :
      “Do you know, the most amazing thing happened to me tonight . I saw a car with the license plate ARW 357. Can you imagine ? Of all the millions of license plates in the state what was the chance I would see that particular one tonight ? Amazing !

      Yup. Rare indeed. There will indeed only ever be one Earth in the universe.

      When another inhabited planet is discovered , I would hope there would be another name for it . Even if its as grand tacky as “New Earth” or “Earth II”

      • Michael Fidler June 10, 2020, 6:38

        In psychology would this not be the idea of Archetypes as in the collective unconscious of people. Leaning on the biblical New Earth as the ideal of a secure pure place in heaven.

      • EricSECT June 11, 2020, 13:15

        If there is a “Great Filter” it seems that is was well behind us.

        We ended up with a planet, the right size, in the right spot with a strong magnetic field to contain our atmosphere, a stable axial tilt… and vigorous plate tectonics to enable four billion years of a stable climate, which seems what is needed for higher intelligence to flourish.

        • Eniac June 13, 2020, 14:02

          I think the argument here is that these criteria are not that strict and there are plenty of planets that qualify. So, not the Great Filter. A much better candidate is abiogenesis, as nobody really knows how that is even possible.

    • Ronald June 10, 2020, 2:33

      With regard to detection of terrestrial planets in the HZ of solar type stars, Nick has said it right: it is sheer observational bias. We simply do not have the instruments yet for a routine detection of such planets. The detections of such planets that we do have are rare chance events.

      I have been thinking of, and often mentioning, a possible alternative scenario for the formation of small terrestrial planets, i.e. an alternative to a ‘greedy’ giant planet that absorbs most (rocky) planetary material: initial rarity of (rocky) planetary material.
      If stellar metalicity is low to begin with, it is possible that giant and ‘sub-giant’ (Neptune class) planets do not form, or at least not close to the star.
      Something like that seems to be the case with low-metalicity solar type stars like e.g. Tau Ceti and 82 Eridani.

      With regard to the moon as a stabilizing force: this idea keeps popping up like a mantra, but it has been mentioned so many times now: the necessity of a large moon as an axial tilt stabilizer is an antiquated idea, see for instance Lissauer et al., 2011: Obliquity variations of a moonless Earth.

    • Nicky June 10, 2020, 18:39

      With largely unknown disk condition and growth history, we cannot yet constrain whether Jupiter and Saturn would have been migrating inward, outward or have roughly stationary orbits. Grand Tack is a hypothesis, and there is several other competing hypotheses, such as Early Instability, to explain the structure of solar system.

      Rather, from observational constraints, cold Jupiter is far more prevalent than hot Jupiter (Wittenmyer et al., 2016), making your argument “exceedingly rare” invalid.

  • Michael Fidler June 9, 2020, 10:31

    You know I really like detective stories, there not the bang bang shot them up movies that have little depth. This is a prime example of true detective and I think we would all enjoy seeing more of these unique methods that squeeze the last bit of information out of the minuscule shadows of these distant worlds.

  • Alex Tolley June 9, 2020, 13:35

    Interesting to see that Python language is used to do the analysis. I assume that is because of the various scientific analysis libraries that have been developed to make the coding easier. However, watching the demo animation on speed, it is slow. This suggests to me that to do many analyses of the Kepler data will either require a lot of time, or the need for a server farm or cloud computing to allow parallel computing as this is a good example of “embarrassingly parallel” computing. Is no one using compiled languages to do this work? It might take much more time to recreate the tools, but perhaps the orders of magnitude performance improvements would make sense.

    • Keanen June 9, 2020, 15:09

      @Alex- Wotan relies on the optimizations of the SciPy library (parts of it are C/C++), which is common practice in “scientific computing” circles. I would say a better optimization would be GPGPU for getting the most out of those server farms, though I haven’t looked too closely at the code and maybe that’s already in there.

    • andy June 9, 2020, 15:23

      While pure Python is slow, they aren’t actually doing the heavy lifting in Python. Taking a quick look at the source code, it looks like they’re using both numpy (which calls compiled code: according to the GitHub repository, 50.7% of the code is C, with 1.0% C++ and 0.1% Fortran) and numba, which does just-in-time compilation of a subset of Python/numpy code using LLVM.

      In general for this kind of project, rewriting into a compiled language would mainly affect the glue code rather than the areas where performance is actually a concern. Rewriting the entire project into a compiled language is unlikely to be worth the tradeoff in development speed, flexibility (and having to deal with C/C++ footguns).

    • Ron S. June 9, 2020, 16:20

      Just to add to the other comments. I have done lots of signal processing work in Python. The various libraries are compiled and are blindingly efficient (especially numpy as andy notes). For commercial product I have benchmarked Python code against alternatives and there is little difference. Even the interpreted code is quite fast since it is (typically) translated to intermediate code and executed by a virtual machine. Python compilers also exist.

      I find there are many old hands who refuse to use “high level” languages for this kind of work. They work in C or massage C code exported from MATLAB and its ilk. Even when that’s done the heavy lifting is from standard libraries just as for Python. If the difference in execution time is a problem buy a faster processor! That’s cheaper than the additional people time.

      You can exploit multi-core processors, too, with suitable set up of threading.

    • Alex Tolley June 9, 2020, 20:17

      Thanks all. I think you have knocked my argument on the head. I was laboring under a misapprehension about the libraries, thinking they were just very efficiently coded in Python, rather than C/C++.

      Ron, interesting that you have done performance comparisons and find little performance loss. That strengthens the case for using Python as a glue to call compiled libraries. (Reminds me when Tcl was similarly used to call C code.)

      I have [re]learned something useful today.

      • Ron S. June 10, 2020, 8:49

        When you look at the % of time spent in “glue” code vs number crunching code the latter takes the majority, so it is no surprise that low glue code has little impact on the total. For the number crunching, regardless of language, standard libraries are used, and over the years they’ve become very good. You won’t find anyone reinventing FFT, convolutions etc!

        The availability of Python libraries is quite surprising. Think of an application and search the public libraries and you’ll likely find a Python module that already does it. The breadth of modules for routine and advanced astronomy and other sciences are legion.

  • Brian Schmidt June 9, 2020, 15:00

    So if this KOI is transiting, as they appear to say, why not use a few earth-based telescopes and watch for it during the next transit? I’ve wondered about this in general for other candidate planets.

    I understand that ground telescopes can’t do the same focus on a large number of stars for weeks/months at a time, but here we’re talking about this star (or a few other candidates) and where you approximately know when transits occur, then you only need a few hours of telescope time during the transit.

  • Ivar Brouwer June 10, 2020, 5:12

    I’m not a specialist, but: as the Star gives the same energy as our sun (or thereabouts) but the planet has X amount of mass more. Would that planet not retain it’s own heat better? Making the planet all that much more hotter then Earth would be in this position. (I might have misread it, but the calculations seem to be made for a planet of earth’s properties) (a bit like the inverse thing that happens with smaller planets: Mars in Earth’s position would probably still loose it’s atmosphere & therefore heat)

    • Ronald June 11, 2020, 5:45

      Of course also depending on age and on abundance of radioactive elements (uranium, thorium), the internal heat of the planet is only a very small fraction of total heat at the surface.
      In Earth’s case, the internal heat flow, consisting of primordial heat and radioactive decay, is only 0.03 % of the total heat at the surface, nearly all of the remainder being solar radiation.

      • ivar Brouwer June 11, 2020, 7:41

        @Ronald – Thnx for your reply! – that brings a lot of uniformity as what to expect from the habitable zone, regardless of mass.

        Then there might be an added feature that comes with added mass, which is the added capability to hold on to it’s atmosphere. But I guess that needs to be handled in a case by case fashion.

  • Sukurjan June 11, 2020, 21:04

    Wow that’s so interesting. Will our next step be leaving earth and inhabiting the new planet. It would be so much fun in starting everything from the beginning right?

    • Eniac June 13, 2020, 14:32

      This is a very good point. Much noise is made about the importance of a magnetic field to retain atmosphere. Yet, Venus has no magnetic field and a thick atmosphere. More importantly, as you mention, a larger planet holds atmosphere MUCH better than any magnetic field, exponentially so. Venus lost most of its water (if indeed it had a lot to begin with, which is not clear), but a planet just two times the size of the Earth already retains not only water, but even molecular hydrogen.

      • Nicky June 14, 2020, 22:27

        Recent observations and global simulations are reaching a consensus that magnetic field protection is not effective and possibly detrimental to atmospheric retention.

        Earth’s polar outflow of oxygen ions (10^25-10^26/s) at open magnetic field outpaces total escape rate at Venus through ion pick and spattering (<10^25).

        The explanation is probably that planetary magnetic field directs and concentrates solar energy to the small polar region where reconnection to interplanetary field occurs, resulting very high density energy and thus high escape rate.

        See recent reviews by Brain et (2013), Gunell et (2018), Gronoff et (2020)

  • Michael Fidler June 12, 2020, 6:43

    The First Habitable Zone Earth-Sized Planet From TESS II: Spitzer Confirms TOI-700 d.

    “TOI-700 d is located well within the conservative habitable zone for its host star, and is the first habitable-zone Earth-sized planet discovered from NASA’s TESS mission.”

    “We present Spitzer 4.5μm observations of the transit of TOI-700 d, a habitable zone Earth-sized planet in a multiplanet system transiting a nearby M-dwarf star (TIC 150428135, 2MASS J06282325-6534456). TOI-700 d has a radius of 1.144+0.062−0.061R⊕ and orbits within its host star’s conservative habitable zone with a period of 37.42 days (Teq∼269K). TOI-700 also hosts two small inner planets (Rb=1.037+0.065−0.064R⊕ & Rc=2.65+0.16−0.15R⊕) with periods of 9.98 and 16.05 days, respectively. Our Spitzer observations confirm the TESS detection of TOI-700 d and remove any remaining doubt that it is a genuine planet. We analyze the Spitzer light curve combined with the 11 sectors of TESS observations and a transit of TOI-700 c from the LCOGT network to determine the full system parameters. Although studying the atmosphere of TOI-700 d is not likely feasible with upcoming facilities, it may be possible to measure the mass of TOI-700 d using state-of-the-art radial velocity instruments (expected RV semi-amplitude of ∼70 cm/s).”

    “The TESS mission was recently selected for its first
    extended mission, which will begin in the summer of 2020.
    According to the draft observing schedule, TOI-700 will be
    observed in 11 more sectors during TESS cycle 36 providing
    a great opportunity to refine the ephemerides and parameters
    of the three known planets, and possibly detect additional
    planets in the system that would enhance our understanding
    of TOI-700’s architecture.”

  • Michael Fidler June 12, 2020, 6:45
  • Ronald June 18, 2020, 5:13

    Good news with regard to earth analogs: Searching the Entirety of Kepler Data. II. Occurrence Rate Estimates for FGK Stars;


    Quote: “For planets with sizes 0.75–1.5 Re orbiting in a conservatively defined habitable zone (0.99–1.70 AU) around G-type stars, we place an upper limit (84.1th percentile) of <0.18 planets per star."

    Ok, I find the upper limits of 1.5 Re and 1.7 AU a bit high, but all the same, almost 0.2 earth analog per G star is amazing.

  • ljk June 24, 2020, 20:18

    Western Space team theorizes rare exomoon discovery

    JUNE 23, 2020


    Western astronomers may have spotted six new moons orbiting planets in solar systems far from our own – an otherworldly discovery so rare it must wait on future technologies to confirm. Until then, however, the mere possibility of the find sparks excitement over our biggest questions about the universe.

    “Our own solar system contains hundreds of moons. If moons are prolific around other stars, too, it greatly increases the potential places where life might be supported, and where humankind might one day venture,” explained Chris Fox, a Physics and Astronomy PhD candidate who made the discoveries.

    The findings of the Institute for Earth & Space Exploration team were recently submitted to the scientific journal Monthly Notices of the Royal Astronomical Society.


    To quote:

    “These exomoon candidates are so small that they can’t be seen from their own transits. Rather, their presence is given away by their gravitational influence on their parent planet,” Wiegert said.

    If an exoplanet orbits its star undisturbed, the transits it produces occur precisely at fixed intervals.

    But for some exoplanets, the timing of the transits is variable, sometimes occurring several minutes early or late. Such transit timing variations – known as TTVs – indicate the gravity of another body. That could mean an exomoon or another planet in the system is? affecting the transiting planet.

    “Because exoplanets are more massive than exomoons, most TTVs observed to date have been linked to the influence of other exoplanets. But now we’ve uncovered six Kepler exoplanet systems whose TTVs are equally well explained by exomoons as by exoplanets,” Fox explained. “That’s why we’re calling them exomoon ‘candidates’ at this point as they still need follow-up confirmation.”

    Unfortunately, the telescopes needed to confirm these or any of the world’s exomoon candidates don’t exist – yet.

    “We can say these six new systems are completely consistent with exomoons: their masses and orbits are such that they would be stable; they would be small enough that their own transits wouldn’t be seen; and they reproduce the pattern of TTVs seen throughout the entire Kepler data set,” Fox said. “But we don’t have the technology to confirm them by imaging them directly. That will have to wait for further advancements.”

    The six exomoon candidates are in the star systems known as Kepler Object of Interest (KOI) 268.01, Kepler 517b (KOI-303.01), Kepler 1000b (KOI-1888.01), Kepler 409b (KOI-1925.01), Kepler 1326b (KOI-2728.01) and Kepler 1442b (KOI-3220.01).