≡ Menu

Planets Everywhere You Turn

Exactly what kind of planets can form around M-class dwarf stars is a major issue. After all, these stars, comprising 70 percent or more of the stars in the galaxy, are far more common than stars like the G-class Sun. About 5500 of the 160,000 stars the Kepler mission is looking at are M-dwarfs, and of these, 66 had been found to show at least one planetary transit signal at the time a new paper on M-dwarf planets was in preparation. That paper, the work of John Johnson and postdoc Jonathan Swift (Caltech) and team, homes in on the Kepler-32 system, whose five transiting planets offer a chance to study planet formation and frequency around such stars.

Kepler-32 is about half as massive as the Sun and has half its radius, with about 5 percent of its luminosity. The planets here have radii that range from 0.8 to 2.7 times that of the Earth, all of them orbiting within about a tenth of an astronomical unit from the star, a distance that is about a third of the radius of Mercury’s orbit around the Sun. The outermost of the five planets lies in the habitable zone. The Caltech team used Kepler data in correlation with observations from the Keck Observatory and the Palomar 60-inch telescope to characterize the system in detail. Two of the planets had already been confirmed, and the team has now confirmed the other three.


Image: — Depiction of the Kepler-32 planetary system with the star and orbits drawn to scale. The relative sizes of the planets are shown at the bottom of the figure scaled up by a factor of 80 in relation to their orbits. Credit: Jonathan Swift, John Johnson/Caltech.

Although the study is producing interesting material on planet formation, what’s capturing the bulk of press attention is the researchers’ estimate that there are at least 100 billion planets in the galaxy. It’s a conservative figure at that, because the analysis only includes planets in close orbits around M-dwarfs, leaving aside the issue of outer planets in such systems and planets orbiting other classes of star. To arrive at the figure, the team calculated the probability that an M-dwarf system would have the precise edge-on orientation we see with Kepler-32, combining that with the number of systems Kepler is able to detect. Even the conservative figure — essentially one planet for every star in the galaxy — is happy news for planet hunters, for it suggests that planets are the norm around the most common kind of stars in the Milky Way.

From the paper:

We select out 4682 M dwarfs from the Kepler Input Catalog that have been observed with Kepler and use their observational parameters to derive the planet occurrence rate of Kepler M dwarf planet candidates. We confirm that within the completeness limits of the first 6 quarters of Kepler data, the M dwarf planet candidates have an occurrence rate about 3 times that of solar-type stars, while the occurrence rate of all candidates around M dwarfs is 1.0±0.1. We expect the fidelity of our culled sample to be above 90%. Thus the compact systems of planets around the Kepler M dwarf sample are a major population of planets throughout the Galaxy amplifying the significance of the insights gleaned from Kepler-32.

The likelihood is that the five Kepler-32 planets formed much further out from the star and migrated inwards, one line of evidence being that the mass of the disk where the planets are found would be as much as three Jupiters, too much disk material for the space available based on our understanding of other proto-planetary disks. Moreover, the fact that M-dwarfs are brighter and hotter when young would have kept planet-forming dust from existing so close to the star. Yet a third line of reasoning is that the third and fourth planets of the system are not dense, implying they are composed of volatiles like carbon dioxide, methane or other ices and gases. Thus formation further out in the system with subsequent migration is the likely scenario.

The paper is Swift et al., “Characterizing the Cool KOIs IV: Kepler-32 as a prototype for the formation of compact planetary systems throughout the Galaxy,” accepted at The Astrophysical Journal (preprint).

Related: The Planetary Habitability Laboratory (University of Puerto Rico at Arecibo) notes that of the 18,406 ‘planet-like detection events’ released last month by the Kepler team, some 15,847 meet the criteria for study for possible inclusion in the institution’s Habitable Exoplanet Catalog. PHL analyzed and sorted the remaining candidates and was able to identify 262 potentially habitable worlds. These include four Mars-sized objects, 23 of Earth size, and 235 ‘super-Earths.’ The best candidate PHL sees thus far is an Earth-sized planet in a 231-day orbit around the star KIC-6210395, one that receives 70 percent of the light that Earth receives from the Sun. This interesting if preliminary analysis is reported in a news release wonderfully titled (with a nod to ‘Contact’) ‘My God, it’s full of planets! They should have sent a poet.’


Comments on this entry are closed.

  • lurscher January 4, 2013, 11:52

    now we just need a bunch of FOCAL missions to get atmospheric and surface images, powered by the nuclear salt rockets proposed by Zubrin: http://path-2.narod.ru/design/base_e/nswr.pdf this rockets could send us to 500 AU in one year trip!! (thanks to Adam Crowl for the reference)

  • Harry R Ray January 4, 2013, 14:07

    WOW! I DID read Abel Mendez’ blog. Earth sized planets in the ir respective habitable zones appear to FINALLY be in the data! This comes as a BIG surprise due to the discovery of higher than expected jitter in most sun like stars, OR, these host stars may simply be the ones with the lowest jitter. ONE candidate stuck out like a sore thumb! KIC11565158! It appears to be just slightly bigger than Mars in a 348 day orbit around what must be a very quiecent G2 star! It’s temperature is about 70F, compaired with 58F for earth, which is the temperature where Greenland’ ice sheet would RAPIDLY start to melt, but; due to the fact that the planet’s atmosphere is probably LESS dense than earth’s, there is probably considerably less of a greenhouse effect here! The only downside is that the planet’s mass may be below the 0.3 bar threshold for a sustainable atmosphere over the lifetime of the planet. THE BOTTOM LINE IS: We apparently do NOT have to wait for 6 transits (as Barucki et al were afraid of having to do) to find earth analogs

  • Wojciech J January 4, 2013, 15:50

    We should get some official Kepler statements soon with the AAS meeting starting this weekend.
    “There are no fewer than 16 more invited presentations throughout the week. Among the highlights will be the HEAD Rossi Prize lecture by Marco Tavani (INAF-IASF/Università di Roma “Tor Vergata”) on behalf of the AGILE mission team, the AIP/AAS Heineman Prize lecture on gamma-ray bursts and magnetars by Chryssa Kouveliotou (NASA Marshall), and an update on NASA’s Kepler mission by Natalie Batalha (San Jose State).”

  • Wojciech J January 4, 2013, 16:02

    PHL also notes that there are 35 Warm Jovians in the sample-if they have moons(very likely) there are also odds of them being habitable(some Warm Neptunians maybe as well).

  • Mike January 4, 2013, 16:11

    The 221st American Astronomical Society meeting begins on Sunday. There will be much new information about exoplanets released through the 5 day long meeting.
    One session in particular seems relevant to this article. That is # 216 on Tuesday at 10:30 PST. It will have two presentations about the number of Earth size planets discovered by Kepler. The second presentation, ” The occurence rate of small planets around cool stars from Kepler” concerns near Earth size exoplanet discoveries at Kepler’s M-dwarf target stars.
    That session and the many other exoplanet sessions at the AAS are going to be very informative and interesting to read about.
    The other astronomy stuff at the AAS meeting is also cool and worth reading but planets is where it’s at!

  • Enzo January 4, 2013, 16:18

    “The outermost of the five planets lies in the habitable zone.”

    Paul, where did you get this ? I can’t find it in the paper.
    At a distance of 0.13 AU from a star which is 5% of sun’s luminosity, a planet should receive ~2.95 x the amount of Earth.
    I now check every single claim of habitable zone as I find them often wrong.

  • An Infinitude of Tortoises January 4, 2013, 16:25

    And a nod to something else to which that was a nod: ALL THESE WORLDS ARE … WHOSE?

  • Astronist January 4, 2013, 18:34

    I recently read “Origins of Existance” by US physicist Fred Adams. He makes the point (which I have not encountered elsewhere) that towards the end of their multi-trillion-year lifetimes, red dwarfs are expected to brighten up considerably, and turn into what he calls blue dwarfs for a while. Thus if one is particularly interested in the surface water or Earth-analog zone around a star (the “habitable” zone, if one believes that purely subterranean life or non-aqueous life are both impossible), it seems that that zone will gradually migrate outwards during the lifespan of the star, before migrating back inwards as the star dies. However, this will not happen for a while yet, as the universe is still extremely young relative to the projected lifespan of these stars.

  • Rob Henry January 4, 2013, 19:06

    Enzo says “At a distance of 0.13 AU from a star which is 5% of sun’s luminosity, a planet should receive ~2.95 x the amount of Earth.”

    I say – it is tidally locked and traditional calculations are for rotating planets. Does anyone have reference to a paper that concentrates on HZs for just those sort of planets?

    Also, this is clearly bolometric luminosity. Visual luminosity should be about 1.5% of the sun’s for that type of star (M1), so visual light levels should be similar. How does that effect HZ calculations?

  • Ronald January 4, 2013, 21:44

    Ok, this is the fascinating stuff that we are all eagerly waiting for all the time, particularly referring here to the second mentioned article on the analysis of all Kepler detections by the Planetary Habitability Laboratory.
    Although, as the article mentions, only those with more than three transit events were selected for the analysis, which: “Unfortunately, also eliminated many interesting objects but more analysis will be required to sort out longer period planets”.
    So, there is still considerable observational bias with regard to longer orbital periods.

    However, there is something in those data, which puzzles me. I checked the numbers in that very useful ”Periodic Table of Exoplanets for the 15,847 NASA Kepler Threshold Crossing Events” and did my own little analysis. I don’t mean the little detail that I come to a total of 15,795 planets instead of 15,847, which is 52 less, but that is minor.

    No, what really bothers me is the following: although, as we all know by now and mentioned above, there (still) is considerable observational bias (i.e. fewer detections) with regard to longer orbital periods, this bias should not be the case with regard to planet size. This is the obvious advantage of the transit method as used by Kepler (contrary to the RV method).
    In other words, the planet size distribution as detected should reflect the real situation reasonably well.

    And yet, we see in the data when we compare the planet size distributions of the ‘Hot Zone’ with that of the Habitable Zone, that, whereas the fractions of gas giants and Neptunes remain virtually the same going from the former to the latter zone (resp. 9 versus 10 and 19 versus 18 %), the fraction of Super-Earths goes up spectacularly, from 38 to 65 %, while the fraction of Earthlike planets goes down dramatically from 22 to 6 %, and that of the smallest two planet types (combined) even more so, from 13 to 1 %.
    This striking difference can not just be a matter of the small number statistics.

    Concluding, it means that toward wider orbits the earthlike (and smaller) planets become scarcer and the super-earths become more abundant, while the (sub)giants remain the same.

  • coolstar January 4, 2013, 22:59

    People shouldn’t jump to conclusions about what this paper says about the possibility of “habitable” planets orbiting M dwarfs. The authors’ major conclusion are that Kepler 32 is EXTREMELY representative of a quite substantial sample of M dwarf planets and that ALL of Kepler 32’s planets formed BEYOND the frost line on timescales totally inconsistent with the formation of terrestrial planets. ALL of these planets formed, essentially, as ice giants and are MUCH too small to have moons large enough to hold an atmosphere within their star’s HZ. They point out that true gas giants around M dwarfs are rare enough to be considered anomalies (one in a sample of 100). None of this, of course, rules out terrestrial planets (with earth like densities, not just earth like radii) in the HZs of M dwarfs, but it sure seems to make them highly unlikely if this system is truly representative. (if any terrestrial planets formed in this system, for example, they would have been destroyed by the migration of the ice giants inward). This cannot be considered good news for those of us interested in ETI.
    I did find the phrase “the outermost planet lies within a region where
    the stellar insolation is similar to Earth’s” to be very misleading, as their calculated equilibrium temperature for this planet is 340K!
    They also grossly underestimated the errors in their derived (not measured) parallax by a factor of several. Their error is about 5% while even a 0.001 arc-second error in a measured parallax would give a range of +/- about 100 pc or an error of about 33%. This would be quite a good ground based parallax for such a faint star and there’s no way that a modeled parallax can be anywhere near as good. (I once measured parallaxes for a living, so I’m sorta sensitive on this issue). Kepler 32 is bright enough that a good parallax should be obtainable by the Naval Observatory team.

  • Enzo January 5, 2013, 0:22

    @Rob Henry,

    Tidal locking should only make things worse because the heat would not even be distributed by the rotation but by the atmosphere only.
    Regarding the different type of radiation, I’m not sure but I don’t think it’s any better than visible light. I don’t think oceans reflect infrared well.
    The fact remains that the star throws nearly 3x the amount of energy at the planet than the sun does with Earth.

    Earth is on the edge of the habitable zone and it is has a fairly thin atmosphere compared to what it is possible for an earth size planet (see Venus). Its reasonable to expect that any super earth would have a thicker atmosphere and, hence, more potential to be warmer.

  • Raffaele Antonio Tavani January 5, 2013, 7:41

    Nell’ultimo articolo pubblicato nel “blog” di Abel Mendez, vi è scritto che negli ultimi dati della “Missione Kepler” si è rilevata la presenza di 23 “Terre” nella “zona abitabile”.
    (Una percentuale molto bassa, pari allo 0,1% tra tutti i pianeti attualmente rilevati).
    Mi chiedo(e chiedo ai lettori di questo “blog”)se davvero il numero di pianeti nella nostra Galassia è pari(per difetto)a circa 100 miliardi di pianeti, se le “Terre” abitabili sono solo lo 0,1%(sempre che gli ultimi dati di “Kepler” abbiano un valore statisticamente rilevante)quanti sono i pianeti identici al nostro, nella intera nostra Galassia?

    Saluti a voi tutti, da Antonio Tavani

  • Paul Gilster January 5, 2013, 7:50

    Translation of Antonio Tavani’s post re Google Translate:

    In the last article published in the “blog” of Abel Mendez, there is written that the latest figures from the “Kepler Mission” was found to contain 23 “Terre” in the “habitable zone”.
    (A very small percentage, 0.1% of all the planets currently detected).
    I wonder (and I ask the readers of this “blog”) if indeed the number of planets in our galaxy is equal (by default) to about 100 billion planets, if the “Lands” are habitable only 0.1% (again the last data of “Kepler” have a value statistically significant) how many planets identical to ours, in our entire galaxy?

    Greetings to you all, by Antonio Tavani

  • Arecibo January 5, 2013, 8:15

    @Rob Henry

    Well, at distance 0.13 AU from 1.5% sun–visual-luminosity star, a planet will receive 0.88 times of “visual” light that earth received .

    But we talk about M-dwarf here, a type of star where most of its output is in infrared. Infrared light warming a planet better than visual light. So the bolometric luminosity is better for the calculation instead the visual one in this case.

  • Paul Gilster January 5, 2013, 13:52

    Enzo writes:

    “The outermost of the five planets lies in the habitable zone.”

    Paul, where did you get this ? I can’t find it in the paper.

    Enzo, I pulled it from the Caltech news release:


    I’ll need to go back to the paper to check this, though. You’re right about the need to be cautious!

  • coolstar January 5, 2013, 17:35

    Here’s a link to a nice habitable zone simulator done by the astronomy education group at the U. of Nebraska-Lincoln. Along with the Project CLEA group at Gettysburg College, these folks produce some of the best interactive labs and simulations available.
    This simulation is quantized in terms of stellar mass, but it’s unique (to my knowledge) in that it lets you evolve the stars to see how the HZ changes with time. Checking a few points, it seems to be quite accurate (and, of course, it doesn’t include atmospheric effects).

  • Enzo January 6, 2013, 5:59


    La percentuale dello 0.01 % per le Terre nella zona abitabile e’ in realta’ molto piu’ alta.
    La ragione e’ che la probabilita’ di osservare un sistema di “taglio” e di conseguenza di osservare un’eclisse con Kepler e’ pari al diametro della stella diviso il dametro dell’orbita.
    Segue che e’ molto meno probabile osservare un terra intorno ad una stella come il solo a 1 AU (p=1.4/150 = 0.00466, 1 su 217) che una a .1 AU (p = 1.4/15 = 0.0466 o 1 su 21).
    Dunque i risultati di KEpelr devono essere corretti per la probabilita’ di ossevazione.
    Le terre sono ancora piuttosto rare ma non quanto i dati “grezzi” suggerirebbero.

    English (my translation) :
    The % of 0.01% for earth like planets in the habitable zone is in reality much higher.
    The reason is that the probability of observing a system “edge” on and, consequently, being able to observe an eclipse with Kepler, is equal to the diameter of the star divided by the diameter of the orbit.
    It follows that it is much less likely observing a earth around a sun like star at 1 AU (p=1.4/150 = 0.00466, 1 over 217) than one at .1 AU (p = 1.4/15 = 0.0466 o 1 over 21).
    Therefore Kepler results must be corrected according the probability of observation.
    Earths are still fairly rare but not as much as the raw data suggests.

  • Ronald January 6, 2013, 12:13

    @Raffaele Antonio Tavani:
    Good question. Well, 0.1% of 100 billion is 100 million. But that is a low estimate, because the number of stars in ouw Milky Way galaxy is probably considerably higher than 100 billion, rather 200 billion or more, and the average number of planets per star is also a minimum estimate and is possibly higher.
    So, a guesstimate of a few hundred million habitable planets in our MW galaxy seems reasonable.
    Remarkably, that number corresponds well with older estimates, see for instance the post here on CD: https://www.centauri-dreams.org/?p=11625&cpage=1#comments
    Also see the comments.

  • Harry R Ray January 6, 2013, 17:32

    BACK AGAIN! Bad news and goog news about KIC11565158! First the bad news. Contrary to my belief, this candidate is NOT one of the candidates Abel Mendez is working on RIGHT NOW, because its orbital period is close to one earth year, and; there have ONLY been three transits observes as of now. Now the good news, I googled KIC11565158, clicked “Relative Flux: The NASA Exoplanet Archive, waited about 5 minutes for the entire page to fill in, and saw a BEAUTIFUL TCE in the third box! Looks like about a 15 hour transit to me last October(i.e., this must be the THIRD TCE)! CHECK IT OUT FOUR YOURSELVES! It ALSO appears to me that Kepler MAY be able to detect planets SMALLER than Mars around sun like stars.

  • Ronald January 6, 2013, 21:17

    Further to coolstar’s comment on habitable planets orbiting M dwarfs, I wondered how Kepler 32 can be ´EXTREMELY representative´ of the entire M dwarf sample, if it has 5 planets, whereas the norm for M dwarfs is just about 1 per star.
    I checked the paper by Swift et al. and found that they meant by representative that: 1) Kepler-32 as a star is very representative of M dwarf stars, and 2) the Kepler-32 planets are very representative of M dwarf star planets in their mass-orbital radius (semimajor axis in AU) relationship, as well as their inward migration.

    A few notable conclusions from this paper (partic. Ch. 5, Discussion and summary) are:

    – The average number of planets per M dwarf star is about 1.0 (+/- 0.1). This is corrected for observational bias.

    – However, because only just over half of all these M dwarfs have planets, the average number of planets per planetary system is about 2.

    – The great majority (almost 3/4) of M dwarf planetary systems are single planet systems.

    – Giant planets are very rare: only about 1% of all planets around M dwarfs are giant planets, and about 2% of all M dwarf stars with planets have them.
    This despite the fact that the entire metallicity range from (dex) -0.4 to 0.4 was represented.

    – Other than that, very few planets are above about 3 Re or about 10 Me, hardly Uranus/Neptune size. By far most are between about 0.5 and 10 Me. There is a tendency of increasing planet size with increasing orbit.

    – By far most of these planets occur between 0.01 and about 0.2 AU from their host M star, none beyond about 0.3 AU.

    The last two points come from the very relevant graph 6 (page 9).
    This graph is very telling in another way: it shows that by far most planets are well above the Kepler detection limit in mass and well below the Kepler observing baseline in AU. In other words, this batch of planets seems to be rather complete for the Kepler sample used and not a result of observational bias (detection limit or observation time). And the relative paucity of planets for M dwarfs is also real then. As are their close orbits.

    We are now really beginning to learn some important facts about M dwarf star planets: they are relatively scarce, small to medium sized and in close orbit.

    The fact that most of these planets are in close orbits seems a good thing, because M dwarfs have narrow and close-in Habitable Zones.
    It would be very interesting to learn how many of the planets in this study are indeed in their host star’s HZ. Tidal locking is a real issue in such close orbit.

  • Rob Henry January 6, 2013, 21:40

    In a tidally locked planet, most of the lit surface is far closer to the terminator than the sub solar point, so surely it is the temperature near that terminator that is all important, not that near the sub solar point. Actually I imagine that the lit side of a water world, might even typically become very Earth-like, with the vast bulk of its water locked up in the dark-side ice cap.

    In such a planet, water and frozen atmosphere will trickle back from the dark side forming very different biogeochemical cycles than for Earth. Thus I find it hard to believe that their HZ model happens to be identical.

  • Ronald January 7, 2013, 11:26

    Further to my previous comment, these dim M dwarfs also have very narrow habitable zones (HZ). A typical M dwarf with 1% solar luminosity will have a HZ only about 10% of ours, and the chance of a planet being situated in the HZ is correspondingly smaller.
    This is somewhat offset by the fact that planets of M dwarfs are all situated so close to the host star. But if the average M dwarf has only 1 or 2 planets, the chance of it being within the HZ is still not very great.

  • Ronald January 7, 2013, 11:57

    Off-topic, but relevant with regard to planetary system stability:

    Wide Binary Stars Wreak Havoc in Planetary Systems, Astrophysicists Find: http://www.sciencedaily.com/releases/2013/01/130106145751.htm

    Planetary system disruption by Galactic perturbations to wide binary stars: http://www.nature.com/nature/journal/vaop/ncurrent/full/nature11780.html

    Very wide (> about 1000 AU) binary stars can seriously disrupt each other planetary systems over time, leading to very high eccentricity and even ejection of planets.
    However, this is mainly the case when the binary orbit is highly eccentric (i.e. very elliptical) and leading to a (very) close approach to the other’s system.
    Furthermore, it mainly disrupts outer planets in wide orbits, and in planetary systems that extend beyond about 10 AU.

    “The researchers note that this (…) could only be reproduced well when they assumed that the typical planetary system extends from its host star as much as 10 (AU). Otherwise, the planetary system is too compact to be affected by even a stellar companion on a very eccentric orbit.”

    So, the common compact systems and inner parts of more open systems should not be affected too much.

  • Rob Flores January 7, 2013, 14:55

    I can’t I magine any moons of jovians suviving a deep planetary migration,
    or for that matter even being created by being captured by said Jovians on
    their way down.

    One you get closer to the Primary, Don’t the orbital mechanics devolve into
    a three body (Primary, Jovian, moon) problem. Because these kinds of
    three body problems are inherently unstable and for the lower mass moon the tendency is to be ejected or less likely sent on a collision course to the larger bodies.

    It is certainly possible to have habitable enviroment on a moon,
    in a HZ. But such a moon would have to have vast oceans and a generous

    Obcourse what I would really like is for Kepler to find ONE earth like world.
    That way we could make an educated estimate of what is the average distance between such systems. So far from the cumulative planet hunting data from all souces it looks to me like 250-350 LY, (at the G.H. Zone at any rate), as a reasonable guess.

  • coolstar January 7, 2013, 16:31

    Ronald, that’s a good summary of the main points about the paper but I disagree with one of your interpretations. Given the sizes and location of most of the M dwarf planets found by Kepler, it would seem that most formed beyond the frost line for their systems and migrated inward, which is obviously not a good thing for terrestrial type planets in the HZ. You’re right that this has to be an almost complete sample. Given the short orbital periods (and larger transit depths per planetary radius) this is why I expected the first terrestrial type planet in a HZ to be found among this sample, and at least a couple of years ago! This is also not good news. Everyone should keep in mind that super-earths and ice giants (as most of these planets seem to be or that’s at least how they formed) in the HZ are not really all that interesting, from an ETI perspective. Even with the problems of tidal locking and chromospheric activity (neither of which are necessarily absolute show stoppers) it had been thought that K & M dwarfs might be very good abodes for life (given the very long MS lifetimes and high numbers of such systems). The data seems to show that this view is quite overly optimistic. I know that Geoff Marcy monitors this blog so it would be nice to get his take on the existing data about these systems.

  • Ronald January 7, 2013, 17:14

    coolstar: you probably refer to my sentence “The fact that most of these planets are in close orbits seems a good thing, because M dwarfs have narrow and close-in Habitable Zones.”
    Yes, I realized afterward that this was overly optimistic, because of the inward migration, as clearly argumented in the paper.
    Again, it would be nice to see an overview, table or so, of planets in this sample which are in the host stars’ HZ. From the mentioned graph 6 I get a rough impression that few earthsized planets are in the HZ, also given the general ‘planet mass/semimajor axis (AU)’ relationship, most of the smaller planets are in way too hot territoty, even for M dwarfs.

  • Ronald January 7, 2013, 18:01

    And quite on-topic, the latest Kepler candidates;

    NASA’s Kepler Discovers 461 New Planet Candidates: http://www.jpl.nasa.gov/news/news.php?release=2013-005

    Total now stands at 2740.
    Present size distributions are:
    – Earthsized: 13%
    – Super-earth: 30%
    – Neptune-class: 47%
    – Gas giant: 10%

    “The most dramatic increases are seen in the number of Earth-size and super Earth-size candidates discovered, which grew by 43 and 21 percent respectively.”

    I checked how this relates to the above report by the Planetary Habitability Laboratory with its 15,847 planet-like detection events (Kepler Threshold Crossing Event, TCE).
    This gives as total results:
    – Earthsized (and smaller): 34%
    – Super-earth: 39%
    – Neptune-class: 19%
    – Gas giants: 9%

    The main discrepancies are that in the PHL/TCE report the % of Earthsized is much greater, the % of Super-earths somewhat greater and the % of Neptunes much smaller than in the Kepler release.

    The somewhat higher % of Super-earths for PHL can be explained by a different definition of Super-earth: they define it as 1.25 – 2.6 Re, whereas the Kepler team defines it as 1.25 – 2 Re.
    This, at the same time, also makes the Neptune class fraction smaller for PHL, because this category now starts at 2.6 Re (instead of 2 Re). However, I wonder whether this can explain the huge difference of only 19% versus 47% for Kepler.

    Truly amazing discrepancy is the Earthsized (and smaller) category: 34% for PHL versus only 13% for Kepler, despite a similar definition of this category.
    Is this because of different methodology by PHL and/or is this a harbinger of future Kepler releases, with more and more small planets in wider orbits being discovered?

  • Ronald January 7, 2013, 19:01

    And this re-analysis of the first 16 months of Kepler data;
    At Least One in Six Stars Has an Earth-sized Planet: http://www.cfa.harvard.edu/news/2013/pr201301.html

    It concerns mainly close orbits, but:
    “Extrapolating from Kepler’s currently ongoing observations and results from other detection techniques, it looks like practically all Sun-like stars have planets.”
    “The team (…) found that 17 percent of stars have a planet 0.8 – 1.25 times the size of Earth in an orbit of 85 days or less. About one-fourth of stars have a super-Earth (1.25 – 2 times the size of Earth) in an orbit of 150 days or less. (…) The same fraction of stars has a mini-Neptune (2 – 4 times Earth) in orbits up to 250 days long.
    Larger planets are much less common. Only about 3 percent of stars have a large Neptune (4 – 6 times Earth), and only 5 percent of stars have a gas giant (6 – 22 times Earth) in an orbit of 400 days or less.”

    And: “The researchers also (…) found that for every planet size except gas giants, the type of star doesn’t matter”.

  • madscientist January 7, 2013, 21:18

    Been reading the discussion here and it’s quite interesting to see that many of you have either glossed over something very important to this matter. Everything being discussed and calculated upon is being done on what is basically a statistically insignificant sample of star. Look at it this way, Kepler is studying around 160000 stars in an area covering a very tiny fraction of the overall area of the galaxy. Miniscule, in fact. For argument’s sake, let’s just say there are 112000 M class stars in the sample Kepler is looking at (70% of all the stars present in the study). It doesn’t matter how many planets orbit those stars, or, what types of planets they are, or, at what distance they orbit their parent stars, or, how they formed. Even if you take the very low value for the number of stars in this galaxy (1oo billion), that 112ooo M class stars is 0.00000112% of the entire sample. Or 0.0000016% of the entire population of M class stars in the Galaxy. Given that there’s more likely 200-400 billion stars in the Galaxy, that number is even smaller. Saying that you can confidently calculate the number of planets orbiting M class stars and all the other characteristics you may append to them on the basis of such a small sample is stretching the case just a wee bit too far. It’s like saying you can confidently calculate the eye colour of everyone on the planet just from observing 2 or 3 people in a random sample. Meaningless. We are in no position whatsoever to be making any sort of statements, grand or otherwise, about the numbers of planets present in this Galaxy, or any other for that matter. The only thing we can say is, all things being equal, our pitifully tiny sample tells us that planets MAYBE as common as stars, and MIGHT be more so. It’s the same with the types of planets and their orbits. Now, we would be somewhat more confident if we restricted our arguments to saying that WITHIN THE SAMPLE we have X, Y and Z, but that would be about the only situation where we could be reasonably confident to place our bets, so to speak, about our knowledge of what was there. I am rather disconcerted with the trend that has been occurring in science all too often of other scientists making assumptions about observations being made on small samples sizes and then projecting those assumption onto much larger systems without taking into account the real, factual limitations of taking such tiny samples and then basing all their theories on what is occurring on a much larger scale. Even using their own statistics, it makes no sense. I would suggest that before we go making definitive statements about anything here, we should first make the effort to observe and study are far larger and more representative sample of stars over a much wider region of the Galaxy. That way, once we’ve got a few billion stars under our belts, we maybe in a position to have not only a statistically significant representation of the numbers of stars having been studied, but also a far better idea of the numbers and types of planets in orbits about the various stellar classes and how many of them are within the HZ’s of their respective stars.

  • coolstar January 7, 2013, 21:40

    Re the latest Kepler data release. I think my old friend Steve Howell might be a wee bit optimistic with “It is no longer a question of will we find a true Earth analogue, but a question of when.”
    I certainly thought this was true at the start of the mission, but now my personal, somewhat conservative odds would be right around 50/50. It might take an extended-extended mission to make those odds better. Does anyone know off hand how long before Kepler’s orbit takes it too far away and thus the data transmission rate too low to be really useful?
    Today’s press releases from the AAS meeting don’t really seem to add a lot, unfortunately (and I have to get in a word arguing strongly AGAINST the AAS (and anyone else, especially Nature) embargoing papers.)

  • coolstar January 8, 2013, 0:15

    This is off-topic for the discussion of the Kepler-32 system but I thought this paper from Monday’s session of the Longbeach AAS meeting could be interesting (as are LOTS of other papers…..):
    149.29. Detectability of Tidally Heated Exomoons Using Direct Imaging Techniques
    Mary Anne Peters; Edwin L. Turner
    The abstract only is available at aas.org as it was a poster paper.

  • Col January 8, 2013, 7:15

    Ronald, your above about the number of planets per M dwarf (1,0 +/- 0.1), average planets per M dwarf system (2) and majority of M dwarf systems (3/4) being a single planet raises one question for me:

    Do we have data on whether M dwarf planetary systems tend to be sufficiently coplanar for the above trends to be inferred from a transit derived study? i.e. what is to say that many M dwarf systems don’t have additional non-transiting planets, particularly beyond the “snow line”?

    My laymans view is to, as always, view results such as these with considerable caution.

  • Ronald January 8, 2013, 11:42

    Col, good question, From what I get from the paper the answer is yes, the M dwarf systems are sufficiently coplanar to derive those statistics reliably. The paper mentions the coplanarity of Kepler-32 and its being highly representative for the M dwarfs sample. A further quote from the discussion: “inclinations (…) consistent with values for the entire Kepler planet candidate ensemble (1.0◦ – 2.3◦)”.

  • Ronald January 8, 2013, 11:51

    madscientist, I do not agree with you stating that this Kepler sample would be statistically meaningless, on the contrary.
    The Kepler team selected this sample and region of the galaxy, because it would be highly representative of (solar type stars of) the galactic disc. So, we are indeed not talking about the halo or galactic core here.
    Telling also is the fact that Kepler findings are rather well in line, and increasingly so, with HARPS data, which covers a very different region (i.e. relatively nearby stars).
    It is indeed possible to learn statistically significant things about the entire US population from research of a few thousand. Not perfectly so, but significant all the same.

  • Ron S January 8, 2013, 12:31


    Do not start making the very same errors of which you accuse others. The quantitative questions are: the quality of the sampling, the calculated standard deviation, and the explicit enumeration of the assumptions that power the calculations.

    For example, ET samples a dozen humans out of 7 billion and notes that the average number of limbs is 4. They carefully evaluate their sampling technique and how it might be influenced by population distributions, evolution migration and environmental factors. They also make, and state, some assumptions about the experiment. Within those bounds they can state that 4 is the expected value of limbs for humans, with a standard deviation based on the sampling statistics. They do not need to sample a billion humans before making this claim; they just need to make their clear their techniques and assumptions. They could still be wrong. Of course they could still be wrong after sampling a billion humans, they are just less likely to be wrong (lower sd).

    Are you claiming that the researchers hid their analysis and assumptions, or that they haven’t but you think their calculated statistics are wrong?

    Keep in mind that the sample is indeed invalid (e.g. globular clusters and the inner galaxy could easily show different planetary statistics), but then I do understand that they were somewhat careful in explaining, and calculating, the implications (i.e. statistics and assumptions).

    What’s so wrong with that?

  • Ronald January 8, 2013, 12:39

    With regard to the above-mentioned Harvard-Smithsonian Center for Astrophysics (CfA) study (At Least One in Six Stars Has an Earth-sized Planet) and its conclusion that “17 percent of stars have a planet 0.8 – 1.25 times the size of Earth in an orbit of 85 days or less”, it would be very interesting if they (or others) had done extrapolations for these earth-sized planets to wider orbits, in particular HZ orbits.

    I am thinking something along the lines of Catanzarite and Shao (The Occurrence Rate of Earth Analog Planets Orbiting Sunlike Stars) and Traub (Terrestrial, Habitable-Zone Exoplanet Frequency from Kepler).

    Catanzarite and Shao derive a rather low Eta Earth (1-3%), because they notice a significant decrease of planet occurrence with increasing orbit (beyond a certain low orbital period), even when corrected for observational bias. Traub comes to a much higher Eta Earth.

    It would be so fascinating to update these estimates and extrapolations with more recent Kepler data.

  • coolstar January 9, 2013, 21:40

    John Johnson, the second author on this paper and the leader of the lab that produced the work, is this year’s winner of the Newton Lacy Pierce Prize, which is an award granted by the AAS to an astronomer under the age of 36 who has completed outstanding achievements in observational science.

  • Eniac January 12, 2013, 12:37


    Everything being discussed and calculated upon is being done on what is basically a statistically insignificant sample of star. Look at it this way, Kepler is studying around 160000 stars in an area covering a very tiny fraction of the overall area of the galaxy.

    This quote betrays a gross misunderstanding of sampling statistics. What matters for the accuracy of sampling is the absolute number N. It is completely irrelevant what the “potential” sample size is (e.g. all stars in the galaxy, in the universe, etc.). Statistical inferences are made from a sample of size N to an assumed infinite population. The two things that can make such inferences inaccurate are 1) small sample size N, and 2) a biased sample. At a sample size of N=160000, the former is definitely not a problem. The latter, however, has to be carefully evaluated. Given that stars are mixed up pretty well in the galaxy, there should be very little bias based on where we look. Much more serious is bias in our detection methods.

    I still do not understand how anyone can claim that there are no more than 1 Earth-sized planet around a given type of star, when we are still almost completely blind to Earth-sized planets in orbits as wide as Earth or wider. Who is to say there isn’t an average of 100 Earth-sized planets between 5 and 10 AU? Our system could be special because we have giant planets that ate all of these. Maybe smaller stars do not form giant planets, and their mass is instead distributed among hundreds if not thousands of smaller planets. Maybe Earth-sized ones. Admittedly this is an arbitrary proposal without any evidence in favor, but can it be excluded with the data we have?