Wish I’d had a camera at the time but a cartoon will have to suffice, representing two actual incomplete rainbows that stop in midair where they meet, lit only by a setting sun.

This seemed so paradoxical, I honestly wondered if it was a dream. After all, on a normal rainbow with two arches, the arcs do not touch and the larger one is very much fainter with a reversed spectrum. The arches schematized here were indeed accompanied by typically-faint concentric larger arcs that also stopped abruptly where they, too, met, exactly above the main meeting point.

Being awake, as it turned out, I discovered the simple explanation for this, related to an often-subtle and less dramatic everyday phenomenon that is readily understandable in nonscientific terms. I had enough information at the time to solve this like a puzzle, though, and now you do too.

      What was that simple explanation for this odd pair of rainbows?

Notes. Only air was between the point of view and these rainbows. The less complete rainbow, to the right, is slightly brighter and both arches brighten the air directly below them. The picture has been revised to more clearly represent how the rainbows end at slightly different places, muddled by their overlap.   The larger arch ends a little to left (and slightly up) from where the smaller arch ends.   The rainbows were not as thick as depicted (none are) and had further characteristic features that scientific photography would reveal to extend a vertical pattern of similar meeting points.   Safe to guess that this effect was not observed by humans more than a century or two ago, although tiny animals may have experienced it over the course of eons.   Some details of the real-life story have been altered in an attempt to stymie internet searches.

Here is an added diagram meant to emphasize that the position of a rainbow depends on the direction of incoming sunlight and that the size and distance of any rainbow are intrinsically ambiguous because the mist that creates a rainbow is rarely at a single distance and the resulting image occupies the same portion of the visual field regardless of actual distances to individual mist droplets. More information can be found all over the internet, such as Wikipedia and (less laboriously) earthsky.org.


  • 11
    $\begingroup$ I'm voting to close this question as off-topic because this belongs on physics.stackexchange.com. $\endgroup$ – Ian MacDonald Oct 20 '16 at 23:31
  • 3
    $\begingroup$ Both that there are two and that they stop in midair can be bewildering $\endgroup$ – humn Oct 21 '16 at 5:17
  • 4
    $\begingroup$ Meta discussion about the on-topicness or off-topicness of this question. $\endgroup$ – Rand al'Thor Oct 23 '16 at 1:02
  • 3
    $\begingroup$ A real life puzzle supercedes a contrived one every time. I still can't figure this one out. :) $\endgroup$ – John Oct 25 '16 at 5:13
  • 2
    $\begingroup$ Since it isn't the answer I figured I would at least show some cool photos as I eliminated the Louvre and Chicago's cloud gate. i0.wp.com/www.sistavoyage.com/wp-content/uploads/2012/06/… and assets.thecreatorsproject.com/blog_article_images/images/000/… $\endgroup$ – John Oct 26 '16 at 0:44

10 Answers 10


Original solution by YowE3K   (who later turned it into this community wiki)

This isn't an answer to the exact question, but the following link is to an image that I thought was worth looking at anyway:


And @justhalf found another image which looks even more like the one in the question, except rotated 90 degrees:

Atmospheric Optics

I'm thinking that the actual answer is something to do with

light reflecting off the windows of a high-rise building

Added layout from puzzle’s poser

The rightmost diagram represents what can happen when the sun sets in a direction almost parallel to the wall of a high-rise building (whose corner is the gray rectangular area with writing) and reflects off its windows. A second rainbow is created by the sun’s reflection, as if there were a second sun, but neither rainbow’s right half is illuminated for different reasons.
• A portion of sunlight for the directly-lit rainbow is blocked by the building.
• Reflected sunlight that would complete the second rainbow (dimmer and left shifted) is not present due to the viewpoint’s position and the sun’s angle.
The building’s vertical wall plays the same reflective role as horizontal bodies of water in the photographs above, which is why the doubling is oriented 90° differently.

For comparison, shadows are often doubled in a related manner:

  • 1
    $\begingroup$ This explains why the larger bow is fainter. If we imagine that there's a mirror to the right, perpendicular to the screen, with the Sun behind us and to the left; then the reflection of the Sun would be to the right, and the (dimmer) reflection rainbow would appear to the left of the main bow. (And the rainbows would disappear where they were shadowed by the mirrored object on the right.) $\endgroup$ – 2012rcampion Oct 21 '16 at 0:47
  • 1
    $\begingroup$ This one fits the image better: atoptics.co.uk/rainbows/bowim47.htm $\endgroup$ – justhalf Oct 25 '16 at 7:52
  • $\begingroup$ Can't believe I didn't notice your comment until now, @2012rcampion, but it would make a valuable addition to the body of this post. $\endgroup$ – humn Oct 25 '16 at 8:25
  • 1
    $\begingroup$ @humn - Finding answers to this question is certainly causing me to find some extremely pretty examples of rainbows!! (I just found one where the rainbow is vertical.) (And now one where the rainbow stops mid-air cdn.zmescience.com/wp-content/uploads/2016/02/Lunar_Rainbow_3_-ORION_L-Victoria_Falls-_Calvin_Bradshaw_3-1024x768.jpg - oops - actually that is a moonbow, not a rainbow.) $\endgroup$ – YowE3K Oct 25 '16 at 19:16
  • 1
    $\begingroup$ @humn - :) I guess I am going to have to start going outside at sunset (or sunrise) more often - but first I am going to have to find a building at juuuust the right orientation! And, obviously, a different building for every day of the year. (P.S. I am fairly amazed that I actually understood the explanation - I was expecting it to be more complicated than that even though your question stated it wasn't.) $\endgroup$ – YowE3K Nov 3 '16 at 19:10

I can't see images now, but from the optics laws ang geometry it follows that the center point of the rainbow's arc is opposite to the light source with respect to your eye. In other words, if the rainbow is directly in front of you, the Sun is directly behind you. If you see two rainbows intersecting, you must have two suns behind you, relatively close to one another.

If one rainbow arc was above the other, its respective light source must have been below the other one. If the arcs were side-by-side, the two suns were at the same height.

Probably one of them was a reflection of the original one in quite a big, almost flat surface. Did you have a lake, or a glass-covered sky-scraper behind you or in front of you when seeing the two rainbows?

  • $\begingroup$ Have you had access to the image yet? It might trigger a change in your current concluding question. $\endgroup$ – humn Oct 20 '16 at 11:06
  • $\begingroup$ What do you mean "had access to the image"? What image? $\endgroup$ – Gareth McCaughan Oct 20 '16 at 11:28
  • $\begingroup$ @Gareth McCaughan: The image in the puzzle, which CiaPan hadn't seen when writing the answer (based on just the words of the puzzle, I guess, wish I had described it more thoroughly). $\endgroup$ – humn Oct 20 '16 at 11:51
  • $\begingroup$ Oh, I see. I don't actually see anything in the answer that would be invalidated by looking at your image -- except, I suppose, that it doesn't address the "stopping in mid-air" bit which is a particularly striking feature of your rainbows. (But which is also clearly mentioned in the question text.) $\endgroup$ – Gareth McCaughan Oct 20 '16 at 12:26
  • $\begingroup$ Sorry to report a mistake with the rainbows' meeting point in the picture. The puzzle statement has been revised and a new picture is in the works. $\endgroup$ – humn Oct 26 '16 at 7:21

If it is only lit by the setting sun

I would guess that the second one came from the sun's reflection from the moon.
Since the sun was setting, odds are the moon was starting to show up already. And since it's only the reflection of the sun, it would explain the different intensity of the colors.
Since the light's point of origin is different, it would explain the different angle of the rainbow. For the 2 rainbows which came from 2 completely opposite source of lights to be generated in a cross arc like this must be very rare indeed.

  • $\begingroup$ Sorry to report a mistake with the rainbows' meeting point in the picture. The puzzle statement has been revised and a new picture is in the works. $\endgroup$ – humn Oct 26 '16 at 7:22

Probably this answer might be stupid but thought to give a try. Is it

Prism based rainbow?


on a normal rainbow with two arches, the arcs do not touch and the larger one is very much fainter with a reversed spectrum. These lines some how tells that it's not a normal/natural rainbow which you tried producing using two prisms. One created a normal looking rainbow while the other created the brighter one. Reason being, there is a little pink color in the rainbow which can be produced due to red and violet combination. Phenomenon may be called "Dispersion of light" Also with small animals and insects the phenomenon called "Visible spectrum" comes naturally.

  • $\begingroup$ ^vote with a note: Interesting take, with a bonus lesson about nature. The pink you mention might be my last-minute touchup attempt to fade to uncolored sky but the edges of the rainbow still have too much contrast to be realistic. (Had to learn a new graphics system just to get the overlapped parts to come out as anything but utterly bizarre.) $\endgroup$ – humn Oct 25 '16 at 18:45
  • $\begingroup$ Sorry to report a mistake with the rainbows' meeting point in the picture. The puzzle statement has been revised and a new picture is in the works. $\endgroup$ – humn Oct 26 '16 at 7:21

It may be a

solar eclipse.

Not sure about the science of it, but here's a shot.

The moon blocks out let's say half of the sun, so only half of the rainbow from the sun is seen. Then there is a secondary but dimmer rainbow coming from the reflected light from the moon, which due to the moon being closer has a wider arc. The only source of light is still the sun, but some is being reflected by the moon. And as to the history of them, maybe people only figured out how to view them without being blinded within the past couple centuries.

  • $\begingroup$ ^vote for taking into account so many parts of the puzzle. Turns out, though, that a partial sun produces just as much of a rainbow as a full sun, only dimmer. $\endgroup$ – humn Oct 26 '16 at 0:18
  • $\begingroup$ What if >! It was also very windy, such that the rain drops are flattened. The shape of the rainbow corresponds to the shape of the rain drops, so if the wind makes the drops essentially hemispheres, then the bow would be half of an arc. $\endgroup$ – Spuds Oct 26 '16 at 0:27
  • $\begingroup$ Woops the spoiler didn't work. New to this stuff. Any way to delete/edit that last comment? $\endgroup$ – Spuds Oct 26 '16 at 0:29
  • 1
    $\begingroup$ @Spuds, glad this was at least asked. But I thought wouldn't create the crisp demarcation necessary. $\endgroup$ – John Oct 26 '16 at 0:41
  • $\begingroup$ Super interesting idea about irregular drops, $@$Spuds and @John. Can imagine how that would kinda work. $\endgroup$ – humn Oct 26 '16 at 1:24

Here's another try.

You woke up on a boat. This explains why you thought it was a dream. The image is sideways of what it would be from the perspective of someone standing upright. The sea was calm so it reflected the second, dimmer part. As to the recent (last couple hundred years) aspect, this phenomenon may occur at extreme latitudes. All the pictures I've come across were from places very north. It was mentioned in comments that there is a human made component, which I think is climate change. There is now more (calm) sea up north so it is plausible to wake up to a reflected rainbow.

"tiny animals may have experienced it over the course of eons and written poetry about it"

The earth's last warm period before some old ice age? This is still confusing. Were creatures smaller back then? Did they write poetry?

Sorry for answering twice but it's a completely separate thought.


They don't line up because the actual reflection is happening past the horizon.

  • $\begingroup$ (Removing the schlocky wisecrack about poetry from the puzzle statement.) $\endgroup$ – humn Oct 26 '16 at 2:59
  • $\begingroup$ Your "Edit:" idea could provide a very mysterious premise for a future puzzle. $\endgroup$ – humn Oct 26 '16 at 3:06
  • $\begingroup$ Sorry to report a mistake with the rainbows' meeting point in the picture. The puzzle statement has been revised and a new picture is in the works. $\endgroup$ – humn Oct 26 '16 at 7:20

We know from the other answers, that ...

... two light sources are needed to create two non-concentric rainbows. We also know from the other answers, that one of the light sources must be left or right from the other.

The first thing that came to my mind were the headlamps of a car, but the question states, that the sun was the only light source. However, the OP confirmed that he was sitting in a car when he saw the rainbows for the first time. One possible explanation would be, that the low standing sun was reflected by two things on the car. This could be either the headlamps (not sure if this would work) or the side mirrors. In the latter case the OP could see the rainbows in the inner mirror while sitting in the car. The rainbows would stop in mid air, because the light is reflected in the form of cones.

  • $\begingroup$ OP also said that "Only air was between the point of view and these rainbows", so it wouldn't be seen through a mirror. $\endgroup$ – Spuds Oct 26 '16 at 15:08
  • $\begingroup$ I took this to mean that the sunlight itself might have been reflected by the car's mirror, which I can imagine producing an interesting effect on a smaller-than-human scale $\endgroup$ – humn Oct 26 '16 at 17:00
  • $\begingroup$ @Spuds He also said "that's exactly where I was, and had to go outside to make sure that wasn't causing it". $\endgroup$ – Sleafar Oct 26 '16 at 18:50

You could be

At Niagara Falls (or at a place with multiple waterfalls)


Waterfalls can create rainbows:

enter image description here

But where there are multiple waterfalls, two rainbows could be created, one behind the other, so if you positioned yourself just right then they would appear to touch. One rainbow is slightly brighter because it is closer.


They are cut off as they go into the waterfall

  • $\begingroup$ ^vote with a note: This is where physics/geometry pulls a fast one, as the rainbows would always line up, just as if the one in front were smaller as well. The apparent size and position of a rainbow depends only on our relationship to the sun rather than on the exact position of the reflecting mist, as long as there is some mist at some distance in the right direction(s). Thanks for the gorgeous picture too! $\endgroup$ – humn Oct 25 '16 at 21:39
  • $\begingroup$ Sorry to report a mistake with the rainbows' meeting point in the picture. The puzzle statement has been revised and a new picture is in the works. $\endgroup$ – humn Oct 26 '16 at 7:20

Could this be something to do with wearing spectacles/bifocals and seeing one through plain air and other through glasses unless spectacles violate the "only air" assumption

  • $\begingroup$ ^vote with a note: Yes, eyeware is meant to be excluded by the "only air..." phrase, but now I want to see a rainbow through 3D-glasses! $\endgroup$ – humn Oct 26 '16 at 21:00

Community gallery of almost-answers, begun from comments— feel free to add.

Moonbow that stops mid-air


Carrousel du Louvre, Paris


Cloud Gate, Chicago


Who cares about double rainbows?


Run-of-the-mill everyday double rainbow


Low antisolar point


  • $\begingroup$ (Some of these might not be 100% unaltered) $\endgroup$ – humn Oct 26 '16 at 6:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.