“Oh, my sweet summer child," Old Nan said quietly, "what do you know of fear? Fear is for the winter, my little lord, when the snows fall a hundred feet deep and the ice wind comes howling out of the north. Fear is for the long night, when the sun hides its face for years at a time, and little children are born and live and die all in darkness while the direwolves grow gaunt and hungry, and the white walkers move through the woods”.

-George R.R. Martin, A Game of Thrones

Well, this is a very interesting quote, about a very interesting situation.

But, I'm going to have to ask; how?

Assume the following:

  • It causes snowfall a hundred feet deep within a region similar to Europe. This is not "the snow falls over a long period of time and eventually builds up to a hundred feet; Old Nan's wording implies that it falls a hundred feet deep in one go.
  • It causes a powerful wind to blow from the north, reducing temperatures enough to freeze bodies of water as per "ice wind". Presumably, that wind causes ice.
  • It blocks the sun for years at a time - long enough for "little children [to be] born and live and die all in darkness"

Now, a sufficiently large asteroid impact or a global thermonuclear war or a supervolcanic eruption might cause such an event in terms of the last point, but I kind of doubt that it would result in the first two. Snow a hundred feet deep? Really? That's the about height of a nine-story building.

So, what phenomenon could do such a thing?

It doesn't matter if you have to inflate the scale at which it occurs; just ensure that life is still possible in areas of the planet not effected by the winter.

EDIT FOR CLARITY: Any actual physical or meteorological phenomenon capable of doing this that isn't magic.

ANOTHER EDIT FOR CLARITY: It's not supposed to be a regular but exceptionally long winter that's a part of a recurring cycle of seasons; it's supposed to be a long and unnatural one.

  • 1
    $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$
    – L.Dutch
    Nov 2 '21 at 19:52
  • $\begingroup$ Does this answer your question? What natural phenomenon could lead to greatly varying season lengths? $\endgroup$
    – Mazura
    Nov 3 '21 at 0:44
  • 6
    $\begingroup$ "the snows fall a hundred feet deep" doesn't have to (or even typically) mean "a single snowstorm drops 100 feet of snow in one go" (if anything "snows" favors the multiple storm interpretation). It just means it's so cold the snow almost never melts and many storms drop that amount of snow cumulatively (with no modern plows and ice melters, a year with 4" a day is enough to build up well over 100' of snow). The physics of dropping 100' of snow at once would be essentially impossible (greater cold is incompatible with the amount of atmospheric moisture needed to produce that much snow). $\endgroup$ Nov 3 '21 at 3:52
  • 5
    $\begingroup$ Also worth noting, neither Old Nana, nor any living person in GoT has witnessed a long winter. $\endgroup$ Nov 3 '21 at 14:02
  • 1
    $\begingroup$ "Snow 100 feet deep" could also be the kind of hyperbole that comes from oral transmission of an event beyond living memory. Consider, for example, Great Flood myths (such as from Mesopotamia, the likely source of the Biblical Noah story), where "the whole world" is drowned. $\endgroup$ Nov 4 '21 at 15:05

17 Answers 17



The magnitude of variable stars vary over time - up to 4 years. This is a bit short for Game of Throne-style long winters, but in a fictional setting, it is not a far stretch to imagine periods of up to a century.


The Earth is subject to regular ice ages that last thousands of years, tens of thousands years apart. The last one lasted around 13,000 years and ended 13,300 years ago and covered parts of Europe and North America with ice sheets up to 4 km thick. The causes are not well known; it could be orbital changes, atmospheric changes, fluctuations in ocean currents, variations in the sun's output, and more. This gives leeway to imagine shorter, less severe ice ages maybe a century apart, lasting decades. It is not a long stretch to imagine clouds hiding the sun for much of this time - the snow has to come from somewhere.


From Wikipedia:

An object that orbits the Sun more closely than Earth would normally have a shorter orbital period than Earth, but that ignores the effect of Earth's own gravitational pull. If the object is directly between Earth and the Sun, then Earth's gravity counteracts some of the Sun's pull on the object, and therefore increases the orbital period of the object. The closer to Earth the object is, the greater this effect is. At the L1 point, the orbital period of the object becomes exactly equal to Earth's orbital period. L1 is about 1.5 million kilometers from Earth, or 0.01 au, 1/100th the distance to the Sun.

The L1 point is hence about four times as distant as the Moon, and an object with four times the Moon's diameter would be able to completely eclipse the sun. This would make it of nearly the same size as the Earth, which breaks the Lagrangian dynamics, which are based on the mass of the intervening body being negligible. Yet let us not despair: There will still be a point where the forces of the bodies cancel out, though it is not as simple to calculate.

So, imagine a planet orbiting between our "Earth" and its Sun, fairly close to the "Earth's" orbit. Periodically, this planet will be caught directly between the "Earth" and its Sun, causing an eclipse. The semi-Lagrangian dynamics cause it to be caught there for an extended period of time, creating a very long winter. The L1 point, however, is not fully stable, and eventually the planet moves away and ends the winter. It could be caught again at regular intervals or somewhat unpredicably, as you prefer.

We can imagine the planet to be rather lighter than the "Earth" even at the same diameter, being gaseous rather than rocky, or just carbonaceous rather than mainly iron-nickel like our Earth. If the latter, it could even be a very dark or dim planet that could have escaped the notice of all but astronomers/astrologers.

  • 3
    $\begingroup$ Wouldn't that need to be 16x the Moon's diameter, being 4x farther? So larger than the Earth? Lagrange points are only really a thing for items much smaller than the other two items. $\endgroup$ Nov 2 '21 at 20:16
  • 2
    $\begingroup$ @MichaelRichardson: No - 4 times the diameter means 16x the area. And as I wrote, I am talking about semi-Lagrangian dynamics. Whatever the size, there will be a point where the forces on the middle object cancel out. $\endgroup$ Nov 2 '21 at 20:49
  • 1
    $\begingroup$ If you want an object 4x farther than the moon to eclipse the sun, then you will require its absolute area to be 16x larger. If it is a natural object, then it is likely a planet 4x the size of Earth. If it is a construct designed to shade the planet, then it could just be a very large sheet of foil. $\endgroup$ Nov 3 '21 at 0:58
  • 1
    $\begingroup$ I don't think a large foil sheet in space would be very realistic in a fantasy setting. Measures would also be needed to make it always show a flat side towards the planet. Of course, it may be that planets in Martin's universe are flat. :-) $\endgroup$ Nov 3 '21 at 10:31
  • 1
    $\begingroup$ @KlausÆ.Mogensen dust cloud could work, esp. if you add a pinch of handwavium. $\endgroup$
    – Edheldil
    Nov 4 '21 at 10:02

For it to have "little children [to be] born and live and die all in darkness" then the planet would be in darkness for ~20 years; if this is regular event then most life would die.

If you are after severe planet wide winters, during which the planet cools dramatically, with the sun shrinking in the sky, howling winds, seas of ice growing from the north and the south, then a planet with a more eccentric orbit would would work (the images show orbits with different eccentricities, earth is ~$0.02$).

enter image description here

On the part of the orbit where it is further from the sun the planet, would receive less sun light an cool down, as such it would be planet wide phenomenon, also at the planet is further away from the sun, the sun would be smaller in the sky (not necessarily by much). Depending on how eccentric the orbit is, the effects could range from a global winter. All the way to when $e$ is close to $1$, when for most of the year the planet is far from the sun, and it is a frozen ball of ice, where even the atmosphere may start to condense (ie raining liquid nitrogen/oxygen), where the sun is a small dot in the sky.

hopefully that helps

  • 9
    $\begingroup$ The problem with eccentric orbits is that it would give very short summers and very long winters. The closer the planet is to the sun, the faster it moves. Kepler's Law states that an orbiting planet covers the same angular area in the same period of time, or, in other words, its orbital speed is inversely proportional to its distance from the sun. Far more time would be spend at the apogee than at the perigee. $\endgroup$ Nov 2 '21 at 9:42
  • $\begingroup$ @KlausÆ.Mogensen I was aware of that and felt that it added to the increased severity of the winters. But would probably be good to add to answer. $\endgroup$
    – Nyra
    Nov 2 '21 at 10:11
  • 11
    $\begingroup$ I interpreted "little children [to be] born and live and die all in darkness" to mean children probably didn't survive until adulthood, which would be more like a kid might die at 5 years old having been in winter the whole time $\endgroup$ Nov 2 '21 at 14:22
  • $\begingroup$ @KlausÆ.Mogensen It would spend more time near the apogee, could for the lifeforms on the planet that could be normalized, right? They don't know about orbital mechanics, the would probably have evolved to best fit the median day, and so since the planet spends most of it's time far from the sun, their perception of normal is pretty far from the sun. $\endgroup$
    – Zwuwdz
    Nov 2 '21 at 14:34
  • $\begingroup$ @Zwuwdz: Yes, in that case, far away and cold would be the norm, and the short, hot summer would be the exception. Rather the inverse of what we want. $\endgroup$ Nov 2 '21 at 20:51

Planetary shenanigans.

If you have a planet with a "wobble", or a significant axial precession, in addition to a (slightly) eccentric orbit you could have periods of time where a region/hemisphere of a planet is in a rotational winter + wobbly winter at the same time. If the wobble and the rotational period is not synced, but close you would experience several rotational winters gradually worsening and then diminishing and then correspondingly the same for summer.

The eccentricity and the wobble could be explained by the same planetary impact, hypothetically.

Earth has a wobble, but it takes 26000 years to complete and is for most things un-noticeable (though, every some hundred/thousand years people tend to pick different "polar stars" to represent true north)



The OP clarifies It's not supposed to be a regular but exceptionally long winter that's a part of a recurring cycle of seasons; it's supposed to be a long and unnatural one.

This rules out elliptical orbits because then there wouldn't be "ordinary" short years, and the big winters would be utterly predictable to the extent that all life would have evolved instinctive coping mechanisms.

So, it's a Supervolcano on the other side of the globe, that these people do not know about or don't understand the implications of. And for reasons to do with plate tectonics on a different world, it erupts every few centuries (not regularly), rather than every few hundred millennia like Earth's multiple supervolcanoes. When it explodes, dust darkens the sky globally, and an ice age starts, but it doesn't persist after the eruption is over and the dust settles out of the atmosphere allowing the Sun to warm things up again. (Note the "little ice age" here on Earth in C17, possibly caused by a lesser volcano)

I wonder what Mons Olympus on Mars was like, back when Mars was habitable? And might such a supervolcano be connected to the evolution of Dragons?

  • $\begingroup$ Beat me by 5 minutes! $\endgroup$ Nov 3 '21 at 10:35
  • $\begingroup$ This isn't a particularly satisfying answer, but I don't think GRRM has left much room for anything else. The length of winter is erratic. The length of summer is erratic. Hard to explain with merely celestial bodies. $\endgroup$
    – Michael W.
    Nov 4 '21 at 18:21


The record for yearly snowfall in the United States is was in 1998 when 95ft of snow fell around Mount Baker in Washington between July 1, 1998 and June 30, 1999. Keep in mind that this is a measure of precipitation not the height of snow banks. Strong winds have been known to blow 22 inches of snow into drifts of over 50 ft.

Also remember that when people talk about snowfall, especially before modern meteorology, they would talk about what accumulated on the ground and stuck around rather than strict inches of precipitation.

Ice Winds

When it gets cold things freeze. Not much more to say really. Pretty standard winter stuff.

Years of Darkness

This is harder to explain but anything that blocks out the sun for long enough would work. You already have some ideas suggested in your question. Pick any explanation that suits your fancy.

Or if you're feeling lazy just say your world is in the midst of a glacial period.

  • 1
    $\begingroup$ Sorry, but I downvoted because I feel like this doesn't really answer anything in the question: the snowfall is a cool trivia but it's a record amount, i.e. unusual, so all it does is prove that it's not infeasible, while OP calls for a regular occurrence; I'm pretty sure ice winds refer to very cold winds that freeze everything in an exceptionally short time (think WWs appearing and freezing everything instantly), and the last point is pretty much asking OP to figure stuff out themself... $\endgroup$
    – Neinstein
    Nov 2 '21 at 17:19
  • $\begingroup$ @Neinstein I didn't call for a regular occurrence; hence the things like "global thermonuclear war" or "asteroid impact" in my question. $\endgroup$
    Nov 2 '21 at 18:00
  • 1
    $\begingroup$ @KEY_ABRADE Fair enough - though the quotation itself implies a regular occurance. $\endgroup$
    – Neinstein
    Nov 2 '21 at 18:20
  • $\begingroup$ Extreme cold will NOT make extreme snow. Most of Antarctica is a desert, vs Washington state with it's abundant rain in it's winter (the altitude turning rain into "warmish" snow). $\endgroup$ Nov 4 '21 at 20:44

La Niña, but bigger

Climate sciences are hard. You can always have a huge ocean, handwave "big ocean, big currents", and have an unpredictible temperature oscillation that does this.

  • 2
    $\begingroup$ "moon-Jupiter orbit of 1.01 (or 0.99) year" -- a period of one year from Jupiter requires an orbit at more than forty light-seconds. You can have a moon there (Carpo has a 456d orbit around Jupiter), but Jupiter would cover way too little of the sky for that to make a difference. Maybe, an orbit around Jupiter's L2 -- but that wouldn't be stable. $\endgroup$
    – LSerni
    Nov 2 '21 at 17:55
  • $\begingroup$ A 1-year moon orbit around a planet whose orbital period is also 1 year would require a moon to be far outside of the Hill Sphere, regardless of planet mass, as long as the planet's mass is very small relative to the Star's mass. Carpo only fits inside Jupiter's hill sphere because Jupiter's much father from the Sun; it wouldn't be possible at 1AU. $\endgroup$
    – notovny
    Nov 3 '21 at 18:24
  • $\begingroup$ @LSerni, notovny, okay, it was just a bad idea. I don't see any way to make it work. My bad $\endgroup$
    – Vincent
    Nov 4 '21 at 8:17
  • $\begingroup$ Climate sciences are hard, +1. This is meteorology (for which on WB I've seen about two good answers in seven years), not orbital mechanics, and 100' of snow has nothing to do with GoT. $\endgroup$
    – Mazura
    Nov 14 '21 at 0:11

Binary Star System

Your planet exists in a binary star system, either orbiting one small star which itself orbits a larger star, Helliconia-style, or orbiting around the barycenter of both stars.

In the case of orbiting around one star, you have short, Earth-like seasons based on your orbit around that star, and possibly generations-long hyper-seasons as that star orbits around its parent star. In the Helliconia series, these seasons were marked with a mass die-off of most of the population, while the remaining survivors either packed on fat in order to survive the harsh super-winters, or lost it during the super-summers. (This was caused by a pair of diseases that either killed you or vastly altered your physiology in order to survive the approaching major climate change.

In the case of orbiting both stars, your hyper-seasons need not be quite so long, maybe years instead of generations. When the planet is roughly equidistant to both stars, you get a long summer due to being warmed by both stars, but when it's on the far side from one star, you get a long winter since only one star is close enough to provide much energy. You'd have 2 summers and 2 winters per single orbit. With a suitably eccentric elliptical orbit, you could alter the lengths of those such that 1 winter is longer than the other.

Also introducing an axial tilt and precession could add more variation to both models, similar to the primary cause of Earth's seasons.

  • $\begingroup$ There's also the research paper (though check the submission date in the pdf) ui.adsabs.harvard.edu/abs/2018arXiv180311390F/abstract that describes a suitable elliptical orbit. $\endgroup$
    – Rob
    Nov 3 '21 at 11:15
  • $\begingroup$ @Rob Though if that paper is talking about Westeros, I don't think either the books or the show ever made any suggestion of there being a second sun. So unless it's so far away or dim as to be indistinguishable from other stars, I'm not sure that explanation holds. (Maybe if it were a neutron star? But not sure if that would output enough energy to account for the extreme climate change.) It would explain Tatooine though... $\endgroup$ Nov 3 '21 at 13:26

GOT is a GOAT among fantasy worlds for me. I've imagined how the long winters could have come about many times and never heard or read any single explanation cover it fully. There are constraints to work within though... -GOT is vague as far astronomy but does mention a sun and moon like ours. -Years are measured by path around the sun and constellations. Ancient tales mention a 2nd moon that broke up(?) -Maps appear to only cover the north hemisphere and very little of the south.

Imagine a world with a variable star as its sun. A planet with a moon in a newly established (from a geologic/astronomical prospective) tidally locked semi-polar orbit. The inhabitants of northern hemisphere would never see it. Only if they traveled to the polar north or very far south.

The southern sea could contain all kinds of weird currents or flora /fauna. When South America broke away from Antarctica, and the circumpolar current started flowing, it threw a MAJOR monkey wrench into earth climate.

Imagine a volcanic region that lifts and subsides an area of land allowing a circumpolar flow to be cutoff or established 'randomly'.

Also lookup "The Azola Event". Basically duckweed took over the South Atlantic ocean, sequestered billions of tones of CO2 out of the atmosphere in a relative instant and... yeah Monkey wrench. This of course took more than a few years but in an unstable or delicately balanced environment who knows, a chain reaction of smaller synergistic events could add up to a quickly changing system

Of course all of this could seem random to the short lived human eye and memory. and could be just a small part of a longer repeating cycle that the inhabitants are too close to see.

  • $\begingroup$ I once asked Martin if his world had a regular summer/winter cycle in addition to the greater one. He said that it wasn't something he had thought about. For all Martin's quality as a writer, I think that is rather poor world-building. I don't recall if the age of any of the cast is mentioned in years, but if so, a regular cycle would sort of be required - or something else that could mark the passage of years. $\endgroup$ Nov 5 '21 at 11:19
  • $\begingroup$ This is true, however the OP did ask for a reasonable explanation. I did want to clarify that in GOT, while there isnt seasons, time does flow prity much as we are used to. Seconds hours minutes. However no names for moths or days. $\endgroup$
    – Gillgamesh
    Nov 5 '21 at 11:41
  • $\begingroup$ I don't think you can have a moon over only the southern hemisphere (assuming north/south is defined by your axis of rotation). $\endgroup$ Nov 5 '21 at 22:06
  • $\begingroup$ I might be wrong. But like I said if it were in a tidally locked near polar orbit a portion of the planet wouldn't ever see it. ie: if N America were Westeros, and the orbit carried the moon over the north pole and down around Antarctica and over the southern tip of S America. The moon wouldn't ever be visible in North America. The only time it were in those latitudes it would be on the opposite side of the plaint. $\endgroup$
    – Gillgamesh
    Nov 8 '21 at 12:55

Not strictly an answer, but even on Earth the climate in the past was somewhat different. For example, at some period the Baltic sea was freezing for long enough to build an inn in the middle of it, to serve passing travellers. You can imagine that it had to be combined with much higher snowfall than what we typically experience now (and old nan is most probably referring to accumulated snow, not a single snowfall).

The freezing wind is definitely to be expected then (not that the wind itself freezes the bodies of water, it's simply perceived as super cold by humans and that's absolutely enough).

I would say the only challenge that remains open is the Sun disappearing for years. The two things that come to my mind is:

  • changing tilt of the axis, leading to a change what region experiences polar winters. Not exactly causing the described effect (still the polar night will last no more than half of the year) but in the stories of old nannies it can get greatly overestimated
  • a strong volcanic eruption causing millions of tons of ash to be thrown into the atmosphere, blocking the Sun or at least changing its perception. The Krakatoa eruption actually decreased average temperature of Earth and caused changes to how the Sun was visible

Keep in mind GoT involves magic so you may want to use it to amplify effects even you do not want it to be the main cause.


The best solution is likely to combine several mechanisms, both to tune the desired effect and to achieve the desired unpredictability.

The simplest way to make winters longer is to simply increase the orbital period.

Jupiter has an orbital period of almost 12 years, so at the distance of Jupiter you get cold seasons that last 6 years. To get a more habitable climate, simply increase the strength of the star to compensate for the distance. You might object that a winter still spans half a "year", even though it's longer, but in that case you can just have your inhabitants measure time in some other way, such that their units translate well to our years.

To make seasons more extreme, simply increase the tilt of the planets axis. It doesn't have to be by much. This gives a larger proportion of the planet closer to the arctic circle, and winter nights would thus be longer.

To make winters more severe relative to the summers, simply increase the eccentricity of the orbit. A small increase to perhaps 0.1 or 0.2 would have a very noticeable effect on the seasons.

To make seasons more irregular, add some other very large planets to the solar system*. Possibly inside the orbit of our planet of interest. This will cause small perturbations to the orbit

Now for the fun part: Combine these effects! There are many possible configurations.

If the direction of the axial tilt aligns with the major axis of the elliptical orbit there will be one pole with mild winters and one with severe winters. If they are perpendicular one side will have dark seasons that are colder towards the end, while the other has dark seasons that grow progressively warmer.

In any case, the relationship between the direction of the axial tilt and the major axis of the orbit will likely drift due to interactions wit your heavy planet (or better: planets!). This leaves open the possibility of legends of back when things where way worse than they are nowadays! Check out the Wikipedia page on orbital resonance for hints about periodical interactions!


Short Answer:

I think that my rough calculations show that it might be possible to design a system where there are two co-orbital planets with very similar orbits, and where every few decades the outer planet suffers an eclipse from the inner planet lasting for at least one who year of hte outer planet.

Long Answer:

I once read that George R.R. Martin did write that there was an explanation for the suuper long winters and it would be given sometime in th future. I also hear it is taking him a long time to write the latest books in the series, and I suspect there is no guarantee that the explanation will be given before the last paragraph in the last book. And maybe not even then.

And I wonder whether George R.R. Martin aleady knew the explanation when he wrote that there was one, or if he is just stalling for time until he can think of one which will be good enough for him and for his fans!

I don't know whether Martin will use a scientific and astronomical explanation for the long and bad winters or some sort of magical explanation.

But I have an idea for a scientific and astronomical explanation.

The planet in ASOIAF, and any similar planet that KEY_ABRADE might be thinking of writing about in a series of their own, apparently has a fairly normal cycle of seasons.

At least the people there know of years, since the winter is said to last for years. I believe that in ASOIAF people's ages are given in years, and ages inASOIAF years seem to corerespond fairly well with the number of Earth years those characters seem to have lived.

Possibly there is a study somewhere about the possible range of ASOIAF year lengths compared to Earth year lengths. If so, that would be very helpful. For now I will assume that ASOIAF years are approximately the same length as Earth years.

I haven't read the books or seen the series, so I don't know if the world in ASOIAF does have a normal cycle of spring, summer, autumn, winter, and spring again, with occasional super winters which last for several years.

Not being familiar with the story I can't say for certain whether the planet does have normal seasons, including normal winters, betweent he terrible long winters that last for years. A world with little or no axial tilt - https://en.wikipedia.org/wiki/Axial_tilt - would have very mild seasons which people might not notice much.

In temperate zones the cycle of the seasons is very important, and people notice it, and that is the main evidence for the length of a a year.

But on Earth, people lived in tropical regions which the seasons were often much less important. Some tropical places have important and noticeable wet and dry seasons, but maybe some do not. So in some tropical regions people might not have noticed any vary important cycle of the years to show them how long a year is. Do those tropical cultures use years in their native calendars?

People can notice the length of a year without having a noticeable cycle of seasons.

A midnight the stars on the line between due north and due south through the zenith will be in the opposite direction from the direction to the Sun.

Over the course of a year, the celestrial sphere will seem to rotate, though actually the line between noon on Earth and midnight on Earth will be doing the rotating. After six months the stars seen on the midnight line will be 180 degrees, halfway across the sky, from the stars that were seen there six months earlier. And after another six months the original set of stars will be seen on the midnight line in the sky.

So that is one way which people could use to figure out the approximate length of a year.

Ancient Babylonian astronomers and astrologers were able to plot the postion of the Sun against the background of the stars despite not mbeing able to see the stars in the daytime sky. They knew the Sun at noon would be almost exactly opposite to the stars on the midnight line on the previous and following midnights. So ithey wound it i aeasy to plot the suns's position against eh unseen stars whose positions "on" the celstial sphere had long been mapped.

So it is possible that people know the length of a year, regardless of how noticeable the regular and ordinary seasons are on the ASOIAF planet.

My theory is that possibly the ASOIAF planet could sometimes be eclipsed for years by a planet in an interior orbit around their star.

The gravitational interactions between and star and its planets create forbidden zones around each planet. If an astronomical body enters the forbidden zone of a planet its orbit will be gradually perturbed until it takes a new orbit outside the forbidden zone, or crash into the star or antoher planet, or is ejected from the star system. Such things are believed to have happened more than once during the first few hundred million years of the solar system, ecjecting or colliding many asteroids, comets, planetesimals, moons, and even planets.

So the forbidden zones between neighboring planets make it hard to squeeze planetary orbits close enough for one planet to clipse all of another planet, or even any part of the other planet.

But there are theoreticall and real examples of astronomical bodies which share the same orbit, called a co-orbital configuration.

For example, there could be two planets in trojan orbit around their star, one planetbeig about 60 degrees ahead or behind the other planet in their mutual orbit around the star.

But a trojan planet wouldn't cast a shadow on its co-orbital planet.

Saturn has examples of moons in Trojan orbits with other moons, which are no good for this question, and it has anotherand more interesting set of co-orbital moons.

Epimetheus has dimensions of about 129.8 kilometers by 114 kilometers by 106.2 kilometers and its orbit around Saturn has a semi-major axis of about 151,410 kilometers.

Janus has dimensions of about 203 kilometers by 185 kilometers by 152.6 kilometers and its orbit around Saturn has a semi-major axis of about 151,460 kilometers.

Note that the difference between their orbits is less than their sizes. That seems like a recipe for a collision, even though they have existed in a co-orbital configuration for thousands, millions, or billions of years.

If perfectly circular, the orbit of Epimetheus would have a circumference of about 951,336.28 kilometeers, and the orbit of Janus would have a circumf951,650.44 kilometers.

Epimetheus has an orbital period, during which it travels 360 degrees around Saturn, of about 0.694333517 Earth days, while Janus has an orbital period of about 0.694660342 Earth days.

Epimethmeus travels about 518.4828201 degreees along its orbit every Earth day, while Janus travels about 518.2388834 degrees along its orbit every Earth day. So Epimetheus gets about 0.2439376 degrees ahead of Janus every Earth day. So after 1,475.78251 Earth days, or about 4.040485285 Earth years Epimetheus gets a full circle ahead of Janus and so catches up with Janus from behind.

So Epimetheus should have smashed into Janus after no more than four years of sharing an orbit.

Epimetheus's orbit is co-orbital with that of Janus. Janus's mean orbital radius from Saturn is, as of 2006 (as shown by green color in the adjacent picture), only 50 km less than that of Epimetheus, a distance smaller than either moon's mean radius. In accordance with Kepler's laws of planetary motion, the closer orbit is completed more quickly. Because of the small difference it is completed in only about 30 seconds less. Each day, the inner moon is an additional 0.25° farther around Saturn than the outer moon. As the inner moon catches up to the outer moon, their mutual gravitational attraction increases the inner moon's momentum and decreases that of the outer moon. This added momentum means that the inner moon's distance from Saturn and orbital period are increased, and the outer moon's are decreased. The timing and magnitude of the momentum exchange is such that the moons effectively swap orbits, never approaching closer than about 10,000 km. At each encounter Janus's orbital radius changes by ~20 km and Epimetheus's by ~80 km: Janus's orbit is less affected because it is four times more massive than Epimetheus. The exchange takes place close to every four years; the last close approaches occurred in January 2006,[14] 2010, 2014 and 2018. This is the only such orbital configuration of moons known in the Solar System[15] (although, 3753 Cruithne is an asteroid which is co-orbital with Earth)


So could an Earth like planet and another planet be co-orbital like Epimethmeus and Janus?

If the inner planet was moving 0.2439379 of a degree farther than the other planet every Earth day, it would travel about 89.098317 degreees past the outer planet in one Julian calendar year of 365.25 days.

If the outer planet has an orbital period of about one Earth year, or about 365.25 Earth days, and orbits its star at at a distance of 1 AU like the Earth does, its orbit shoudl have a circumferance of about 939,950,349.2 kilometers, and each degree of arc along its orbit would be about 2,610,973.192 kilometers, and a planet 89.098317 degrees wid would be about 232,633,317.2 kilometers in diameter.

That would be too big to fit inside the orbit of the Earth. It would also be about 1,626.988 times the the 142,984 kilometer equatorial radius of Jupiter, and planets and brown dwarfs can not get much larger in diameter than jupiter. Adding more mass just makes them denser, not wider.

So let me design an extreme example of co-orbital planets that almost touch.

Earth has a radius of about 6,400 kilometers and thus a diameter of about 12,800 kilometers. Suppose there is a planet of that size in Earth's orbit, and another planet in an almost identical orbit which is about 20,000 kilometers smaller than Earth's orbit.

Earth has a semi-major axis of about 149,598,023 kilometers, so if its orbit was perfectly circular the orbit would have a circumference of 939,951,306.2 kilometers. Earth has an average orbital speed of about 29.78 kilometers per second. So it would take Earth about 31,563,173.48 seconds to complete one orbit.

The sidereal year of Earth, the time it takes to complete one orbit as seen from the distant stars, is 31,558,149.504 seconds, the difference being explained by Earth's elliptical orbit and varying orbital speeds.

So the orbit of the inner planet in my example would have a semi-major axis of about 149,578,023 kilometers and thus if it was perfectl circular it would have a circumference of 939,825,642.6 kilometers. It would have a slightly faster orbital speed than Earth, but I will make it about 29.78 kilometers per second, just like Earth's.

So the inner planet would take about 31,558,953.75 seconds for each orbit.

One degree of Earth's orit would be 2,610,975.851 kilometers, and one degree of the inner planet's orbit would be 2,610,626.785 kilometers.

One arc minute of Earth's orbit would be 43,516.26418 kilometers, and one arc minute of the inner planet's orbit would be 42,510.44642 kilometers.

One arc second of Earth's orbit would be 725.2710696 kilometers, and one arc second of the inner planet's orbit would be 725.1741069 kilometers.

So Earth would travel 2,572,992 kilometers in 86,400 seconds or one day. That would be 59.12713438 arc minutes per day. And the inner planet would also travel 2,572,992 kilometers in 86,400 seconds or one day. That would be 60.52611103 arc minutes or 1.008768517 degrees per day.

So the inner planet would go 1.39897665 arc minutes farther than the outer planet each Earth day. Since there are 21,600 arc minutes in a full circle it would take 15,439.85741 Earth days or about 42.2720 years for the inner planet to gain a full circle on the outer planet.

The inner planet would have an angular diameter of about 27.74 degrees or 1,664.50416 arc minutes as seen from the outer planet. If the star has an angular diameter of about 0.5 degrees or about 30 arc minutes, the planet will travel a total of about 55.98 degrees or about 3,359.00832 arc minutes from first contact to last contact during its eclipse of the star.

If the inner planet gains 1.39897665 arc minutes on the outer planet each Earth day, the inner planet will take 2,401.046737 Earth days or 6.5737 Earth years for the eclipse of the star. And according to my rough calculations about 3,299.00832 of those arc minutes will be during a total eclipse of the star, with all of the star covered, which should take about 2,358.158244 Earth days or about 6.4562 Earth years.

So that imaginary outer planet should spend about one seventh of its time in eclipse.

Except that during what would otherwise be the middle of the eclipse, the two palnets will be at their closest and will switch orbits, thus putting the former outer planet into the inner orbit and into the bright sunlight.

But then, 42 years later, when the former outer planet is eclipsing the former inner planet, they will switch places again and the former outer planet will be the outer planet again and will be eclipsed again.

So each planet should experience about 3 years of eclipse every 42 years, one fourteenth of the time.

I note that:

In addition to swapping semi-major axes like Saturn's moons Epimetheus and Janus, another possibility is to share the same axis, but swap eccentricities instead.[11]


I am not certain how that would work or whether that would allow one planet to always be in the outer orbit.

I note that this is a very extreme example of two co-orbital planets being very close together. If the fictional Earth like planet is supposed to have a moon in an orbital period about the same as Earth's, that orbit would be many times as wide as the distance between the planet's orbits.

For a planet to have a moon at the distance of Earth's Moon with a stable orbit, the inner planet would probably have to be at least 1,000,000 kilometers closer to the star. If the inner planet's orbit gets too much smaller than that of the outer palnet, they may cease to be co-orbital. And then they would have separate forbidden zones, and they would have to get much farther apart to be out of each other's forbidden zones.

Another possibility would be to make the inner planet a bit closer to the star, and make it a giant planet with a ring system. All four giant planets in the solar system have ring systems, though only Saturn has spectacular and easily seen rings.

Jupiter has an equatorial diameter of 142,984 kilometers. And no giant planet can have a diameter more than about 20 percent larger than Jupiter's, so no more than about 171,580 kilometers. Saturn is much smaller than Jupiter, but has a large ring system. The A ring, the outermost bright ring, has a radius of 136,775 kilometers and thus a diameter of 273.550 kilometers, and there are fainter rings much farther out than that.

Suppose that the inner planet has a ring system with a diameter of about 2,700,000 kilometers, about 10 times that of Saturn. Suppose that it orbits around the star 5,400,000 kilometers closer to the star than the outer planet, and thus at a semi-major axis of 144,198,023 kilometers, and an orbital circumference, if the orbit was perfectly circular, of about 906,022,134.2 kilometers.

The inner planet with the rings would have an orbital period of about 0.946185 Earth year or about 345.59407 Earth days according to this orbital calculator.


So the inner planet would travel about 1.0416845 degrees along its orbit during one Earth day, while the outer planet in the position of Earth would travel about 0.9856262 degree during one Earth day. So the inner planet would gain 0.0560583 degree on the outer planet during one Earth day, and it would take it 6,421.8857 Earth days or about 17.582164 Earth years to gain 360 degrees on the outer planet.

When the inner planet with the ring system passes closest the ring system will have an angular diameter of about 22.918332 degrees as seen from the outer planet. With the star having an angular diameter of about 0.5 degree, the ringed planet will travel a total angle of 46.336664 degrees during the eclipse. That will take it about 826.57989 Earth days or about 2.263 Earth years.

So I think that thes rough calculations show that it might be possible to design a system where there are two co-orbital planets with very similar orbits, and where every few decades the outer planet suffers and eclipse from the inner planet lasting for at leas tone who year of hte outer planet.

Of course it would take more expert calculations than mine to work out the details fo such a system and make certain it was possible.


Trinary Star System

The long-term evolution of a system of three roughly-equal masses is exceedingly hard to predict because the system exhibits chaos for most initial conditions. The same is true of a smaller mass caught between the three larger masses (i.e. a planet among three stars.) We have found (mathematically) a few known stable orbits in trinary systems, but for most initial conditions, we simply don't know what will happen in the long term. They could result in ejection, they could be periodic, but the period might be millions of years.

This means a fictional work has broad license to make up weird orbits, which could exist, for all we know. (Especially for all a human reader, who is not a supercomputer, knows.) For example, a planet could be in what appears to be a stable-ish orbit in the habitable zone for centuries, then take an excursion a little further away for a few decades, which could explain these sorts of long, unpredictable mega-winters that are survivable with effort. (You don't need a celestial object to eclipse the suns, heavy clouds induced by the sudden climate change suffice.)

This features prominently in Cixin Liu's The Three-Body Problem, which at one point describes a planet which is sometimes thrown so far away from its three suns that the atmosphere freezes, only to come back to a habitable orbit later.

  • $\begingroup$ Alright, so long as nobody comes up behind you and suggests "Quaternary Star System!" $\endgroup$ Nov 2 '21 at 20:06
  • $\begingroup$ @DarrelHoffman bit of a "1, 2, or many" situation really, 1 is predictable and simple, 2 is predictable and complex, many is both unpredictable and complex $\endgroup$
    – John
    Nov 2 '21 at 20:51
  • 1
    $\begingroup$ We don't know what will happen, but we can work out general trends. "Ejected at high speed" tends to feature prominently, while chaotic wandering tends to require careful setup. $\endgroup$
    – Mark
    Nov 2 '21 at 22:50

I don't have a complete solution but how about a frameshift:

The planet is orbiting a star that's too cool, Winter is the normal state for the planet.

However, the star is part of a multi-star system. The other "star" is a black hole, the deadly polar jets are pointed far from the planet so it doesn't fry. This object has it's own companions, red dwarfs or brown dwarfs. They shed too little energy to be meaningful for warming the planet. However, they are all in elliptic orbits about the black hole and pass close enough that substantial amounts of gas get ripped off at periapsis. This gas feeds an accretion disk that sheds enough energy to keep the planet reasonable. Each time a companion passes the disk glows more (summer), then as it's digested it fades (winter). However, there's a time when there's no object coming by for a while, now you have Winter.


Out of universe, I strongly suspect that GRRM simply slightly misunderstood the relationship between axial precession and Earth's seasons, and so came up with a fantasy high concept of a planet where the planet could have an irregular or long-period "wobble" that occasionally put the northern hemisphere into winter for an extended period of time.

It is very common for people - even educated people - to incorrectly believe that winter is the result of a change in axial tilt, and not the effect of revolution around a star for a planet with a constant tilt.

I think GRRM's explicit statement that traditional physics has nothing to do with the phenomenon is a post-hoc rationalization of a simple error that would never have been an issue if the series didn't explode in popularity.

  • 1
    $\begingroup$ I think their winters are just supernatural. It makes sense thematically with ice fae It also makes sense as people still track things in years. This makes it seem like this was not always the way of things $\endgroup$
    – Andrey
    Nov 3 '21 at 13:41
  • $\begingroup$ @Andrey I also thought that the reader was supposed to infer that this is supposed to be related to whatever disaster destroyed the second moon. $\endgroup$
    – tbrookside
    Nov 3 '21 at 15:06

It's possible that Nan's statement is part hyperbole and part myth. Not all of it is to be taken literally. It could be that the "long night" isn't decades, but so bad that many children are born and die fairly quickly because of harsh conditions. There could be negative bias involved. This is where you tend to remember the bad things more than good things. This is probably because negative information seems to require more processing than positive.

It could be a volcano on an unexplored part of the world that erupts every so often(every 200 years, maybe) and if you could combine the hyperbole with this, it might seem to take longer than it actually does.

It also could just be "magical" in nature. Maybe there is some, unknown to the world, magical artifact that absorbs energy for a number of years, decades, or centuries and once it's "full"(or empty) releases a "pulse" of hand-wavium that puts the planet into a "freezing" state and also releases(multiplies, whatever) the evil that "comes howling from the North."


It is quite probable that Old Nan is telling a made-up story there, or greatly exaggerating actual events. 100ft/30m of snow "in one go" would kill about everyone, like an avalanche (chance of survival: less than 10% after 2 hours, assuming that you don't asphyxiate right away). Note that the Red Keep is plausibly tall enough to provide a supply of air to its inhabitants even with 100ft of snow around.

That said, an avalanche or winds (see sphennings's answer) can cause localized snow pileups. But in reality...

It needs to be pretty hot (!)

Snowfall is a form of precipitation, so, obviously, the air needs to be quite humid. If the whole planet is below freezing temperature, there will be no supply of water vapor (unless there are significant numbers of improbably strong geysers). Once the air is dry, no further snow will fall. So some very elliptic planetary orbit or variable star activity or world-wide shadowing won't plausibly give us 100ft/30m of snow.

So we need hot regions where water evaporates to supply cold regions with snow. Note that polar regions, despite being pretty cold, do not receive as much snow as other regions (IIRC, the Artic receives 0.2m per year, while Alaska and Greenland can go up to 2.0m/yr).

Now, an obvious "solution" to that issue would be a tidally locked planet (tidally locked to the sun, obviously). One half is frigging hot, the other half is darn cold. There will be an intermediate region where life would be possible. However, a planet doesn't get tidally locked and "tidally unlocked" at the convenience of some story which needs to be told. A tidal lock is there to stay. Of course, there's libration, which means that the planet might "wiggle a bit around" so the intermediate regions vary a tiny bit. But the biggest issue is that a tidal lock prevents day and night cycles. Viewed from the planet's surface, the sun would always be in the same spot.

So my assumption is that Old Nan is making things more interesting by a bit of exaggeration. 100ft/30m of snow have been observed for a complete winter season on Earth, just not "in one go", and we know of Dark Years. After all, Bran explicitly requested a horror story from her.


Just as a regular winter occurs. The continent that Game of Thrones takes place on might be close to one of the poles, and the planet might be tilted like ours (or more). If the planet orbited its star more slowly than the earth does, this would cause normal winters that were much longer than ours. Neptune's years are 165 earth years long, allowing for seasons that take place over generations, and midwinter nights years long close to the poles.

  • 1
    $\begingroup$ But this would be a regular pattern that happened every year; GOT’s hyperwinter happens randomly (as far as the characters know), and in between you have normal winters. $\endgroup$
    – StephenS
    Nov 2 '21 at 23:11
  • $\begingroup$ @StephenS where did you get the idea of normal winters? From my reading of GOT they always get years long summers and winters, it's just that sometimes they are not that long, or extra long. "Sweet summer child" refers to a child that has never seen a winter. $\endgroup$
    – Andrey
    Nov 3 '21 at 13:50
  • 2
    $\begingroup$ What is a “year” if they don’t have regular seasons? $\endgroup$
    – StephenS
    Nov 3 '21 at 13:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .