Spiraling tentacles of galactic doom!

By Phil Plait | October 30, 2007 7:00 am

So I’m sitting here trying to finish my book, writing about the inevitable collision of the Andromeda Galaxy with the Milky Way, when literally I take a few seconds to check my email. And what do I see?

Beauty. Sheer, incredible, perfectly coincidentally-timed beauty. Behold:

Holy Haleakala! Click on it for access to a much, much more beautiful version. This small one does it no justice at all.

This newly released Hubble image shows the intertwined pair called Arp 87, two big galaxies undergoing a collision. Halton Arp is an astronomer who cataloged peculiar galaxies in the 1970s, and many of them have been found to be colliding, or at least interacting in some way. This pair consists of NGC 3808A, the big spiral on the right, and NGC 3808B, the odd cigar-shaped dude on the left (actually, it’s an edge-on spiral). They are about 300 million light years away in the constellation of Leo, for those keeping track at home. They are separated by about 100,000 light years. The other edge-on spiral is a background galaxy, apparently.

Evidently, sometime in the past they swept past each other. Gravity from 3808B drew out a long tendril of stars, gas, and dust from 3808A. That stuff is now orbiting 3808B, forming a ring perpendicular to the plane of the galaxy’s disk. In the hi-res image, you can trace the material around the galaxy at least twice! It’s amazing; the dust from the tendril torn off 3808A can be seen blocking light from 3808B when it’s in front, and the tendril itself disappears when it’s behind 3808B. I would guess that the collision must have happened about 200 – 300 million years ago, to give the tendril time to wrap around the galaxy. Given their separation and typical speeds (200 kps), they may have passed each other as little as 160 million years ago. So those numbers are consistent. I’d love to know for sure, but oddly, I couldn’t find much in the scientific literature about this pair.

What’s obvious from the image (to the experienced eye) is that there are scads of stars being formed in 3808A. All those bright blue blobs are areas where newly born massive stars are lighting up the gas around them. This is expected: when galaxies interact, gas clouds can collide (they’re big, and smash into each other) or the conditions inside the galaxies allows clouds to collapse and form stars. These are called "starburst galaxies".

Poetic, but fitting.

The other galaxy, which is edge-on, looks reddish to me. That means there is a lot of dust in it, which is also an indicator of lots of star formation.

Eventually, though, the activity will die down. Given how far apart they are, it looks like they may actually drift apart, and not fall back in to each other (though this is impossible to know for sure by eye). In a few billion years, we won’t be so lucky. The Andromeda Galaxy is headed our way like a hundred billion solar mass freight train. It’ll smash into us (well, we’ll smash into each other), and the two galaxies will draw out long filaments from each other just like in Arp 87 here. But unlike these two galaxies, we won’t have enough velocity to escape each others’ pull. We’ll draw apart some, but then recollide, maybe twice more, and eventually merge into one giant elliptical galaxy. Weirdly, stars almost never directly physically collide in galaxy mergers; stars are small and very far apart. It’s the clouds that collide, forming new stars.

Take a good, close look at these two galaxies that make up Arp 87. In two billion years or so, that’ll be us.


Comments (36)

  1. AndreH

    A really nice picture! I aleasy loved galaxy pictures. Thanks for the pic and link.

  2. Now that’s cool.

  3. Quiet Desperation

    “perfectly coincidentally-timed”

    Clearly it was an act of IE – Intelligent Emailing

  4. WOW! That’s an amazing, stunning image! When I see those pics I always feel like buying myself an airbrush and spend the rest of my life reproducing it on a big wall or something…

    But one thing struck me as weird, maybe a very silly question as I am a complete not-even-amateur-astronomer:
    The big one (A) turns counterclockwise here. It seems to have hit B on the right. But, my very unexperienced eye here tells me the remains of A around B also seem to turn counterclockwise. Now, I would expect the remains of A to have been kinda “split up” in the direction of A’s travel, being caught by the gravity of B, now turning CLOCKwise around B. (Otherwise they must have been stopped and changed direction).
    Appearently I am thinking/seeing/interpreting wrong here, but where is my fault? ūüėé

  5. Probably they didn’t stop and change direction so much as their primary attracting gravitational influence changed in the ‘collision.’ I know I have a galaxy collision simulator program somewhere, but I don’t know if it’s tweakable to the extent of simulating observed events… anybody have something they can recommend?

  6. Rasputin


    You’re just sucking up to the minions of PZed now aren’t you?

    Cool pic though

  7. Gary Ansorge

    WOW! What a pic! I note that I am now using the iMac my Son sent me to view these pics and the monitor definition is awesome. I saved the pic to my iMac. Maybe I’ll use it for a screen saver(I’m still learning the ins and outs of this system).

    Yeah, Phil you should really upgrade. My Son used to work for MicroSoft, but after only a couple of years in Seattle, he quit and joined up with Apple in Cupertino. When I asked him why, he just said,” I got tired of working on crap that doesn’t work,,,”.

    ‘Nuff said,,,

    Gary 7

  8. Those galaxies are paying their respects to mighty Cthulhu with their tentacly appendages.

  9. Moose

    Oh wow. Now that’s what it’s all about.

  10. Dave Salovesh

    Stunning image – who woulda thunk it could be out there for millions of years and yet arrive in your inbox at such an appropriate time. It has to be more than simple coincidence, right? RIGHT?

    Anyway, I think I have a similar curiosity as Anse above:

    The tendril from A is wrapped at least twice around the GC of B – maybe even three times, if what I’m seeing as an inner wrap isn’t formed by a different process. But when the objects stripped from A fall under greater influence from B, getting wrapped as they are, are they not also picking up B’s rotation? The tendril seems to be staying in the plane of A, instead of twisting around while wrapping as I naively suppose it should.

    Also, in A there appears to be ghost spiral limb of young stars that wasn’t displaced by the close encounter – I can see how the limb was deformed into the tendril by the influence of B, but the young stars are about where I would expect to see the limb if it wasn’t deformed.

  11. Jason Nyberg

    Here’s a little secret. You know that “galaxy” hack in xcreensaver? It has math problems…

    Apply these changes to fix it, and watch galaxies collide, merge, and get torn apart in your screensaver the way they were meant to! You’ll often see images surprisingly similar to the one heading up this thread.

    (Basically, the wrong masses are used in the force calculations. I’ve known about the bugs for years, but am too lazy to push them upstream…)

    The “greater-than” fixes are the right ones; these are based on ver. 5.03, the sources of which can be found here: http://www.jwz.org/xscreensaver/download.html

    [jnyberg@jnyberg-231l hacks]$ diff galaxy.c ~
    mass / (d * sqrt(d)) * DELTAT * DELTAT * QCONS;

    > d = gtk->mass / (d * sqrt(d)) * DELTAT * DELTAT * QCONS;
    mass * eps;

    > d = gtk->mass * eps;
    mass * gt->mass / (d * sqrt(d)) * DELTAT * QCONS;

    > d = 1 / (d * sqrt(d)) * DELTAT * QCONS;
    mass * gt->mass / (EPSILON * sqrt_EPSILON) * DELTAT * QCONS;

    > d = 1 / (EPSILON * sqrt_EPSILON) * DELTAT * QCONS;
    vel[0] += d0 / gt->mass;
    vel[1] += d1 / gt->mass;
    vel[2] += d2 / gt->mass;
    vel[0] -= d0 / gtk->mass;
    vel[1] -= d1 / gtk->mass;
    vel[2] -= d2 / gtk->mass;

    > gt->vel[0] += d0 * gtk->mass;
    > gt->vel[1] += d1 * gtk->mass;
    > gt->vel[2] += d2 * gtk->mass;
    > gtk->vel[0] -= d0 * gt->mass;
    > gtk->vel[1] -= d1 * gt->mass;
    > gtk->vel[2] -= d2 * gt->mass;

    Jason Nyberg

  12. Richard B. Drumm

    Holy schistosomiasis!
    It’s like the Flying Spagetti Monster is caressing the little guy with his noodly appendage!
    That’s a tidal tail to beat all tidal tails!

  13. I just read about the image at the ESA webpage, and just wondered: “Hmm… Will Phil Plate write something about this? That would be great…”

    And here it is, hehe. Great explanation, as always. Thanks for that.

    Just a little question: you said that the Milky Way and Andromeda will collide in 2 billion years or so. I’ve been trying to find that information, but I got numbers from 3 to even 15 billion years for that to happen… all very unclear.
    So is 2 billion years the valid term now?

  14. DarkSapiens: there is a fantastic scientific paper on the collision between the MW and the AG online at http://arxiv.org/pdf/0705.1170. It’s a bit technical, but it says the collision starts in two billion years, and will continue for several more.

  15. Pierre

    Jason, I’m interested in your patch, but the less-than signs of the diff output were interpreted as HTML code, above; if you tell me the lines identified as changed (with greater-than signs) are all OK, I can figure out how to fix the code, otherwise can you post it again? (To everyone else, I know posting code is not really the point of this blog, but it’s just for this one time). Thanks for the fix, galaxy.c has always been one of my favorite. (Orbital simulation code is a hobby of mine, eh).

  16. One of the things I always think about when I see pictures like that…

    Some multi-tentacled life forms on a planet orbiting a mediocre star 2/3 the way out from galactic centre of the galaxy on the right in that picture, are looking at an image of our Milky Way and the Andromeda galaxy on their brain-link screens with a little narrative that reads something like:

    “Astronomers observing these galaxies believe that in 1-2 billion quatlons, these galaxies will interact much like our own, and eventually coalesce into an elliptical galaxy. This would be a spectacular sight for any life in those galaxies.”

  17. Duane

    Hint for the link posted above: take out the “.” at the end of the link.

  18. Capt. Action

    I’ve been wondering how the Milky Way and Andromeda collision will coincide or how it will be effected by the local group’s pull by the “Great Attractor”
    Since we’re all heading for the same region of space could we expect further collisions with other galaxies after the Milky Way/Andromeda collision in 2 billion years?
    I know, too difficult to look that far into the future. However, I would think around that time frame a lot of other Galaxies/ Dust clouds would be in our vicinity.
    Just wondering how it will all play out for small patch of sky?

  19. A great picture, really puts you in awe of our magnificent universe!

    To bad it has to share the webpage with personality test ads featuring Dianetic by L Ron Hubbard…

    Keep up the great work!

  20. zeb

    Hey, I can see my house!

  21. Jason Nyberg


    Here’s the patch in a different format; to apply just run “patch galaxy.c”, paste in this text, and hit ctrl-d exactly twice (for some reason.) Remember, this is for ver. 5.03, but I don’t think the lines in question have changed, well, forever.

    To build, just “.configure;make” and then you can run “./galaxy” directly. I suggest using the flags “-count 2 –

    I’m kind of surprised nobody has ever found the error…

    (BTW, while looking at it again I found 1 more bug, fixed here, at line 373. :)

    [jnyberg@jnyberg-231l ~]$ diff -e galaxy.c.orig galaxy.c
    gt->vel[0] += d0 * gtk->mass;
    gt->vel[1] += d1 * gtk->mass;
    gt->vel[2] += d2 * gtk->mass;
    gtk->vel[0] -= d0 * gt->mass;
    gtk->vel[1] -= d1 * gt->mass;
    gtk->vel[2] -= d2 * gt->mass;
    d = 1 / (EPSILON * sqrt_EPSILON) * DELTAT * QCONS;
    d = 1 / (d * sqrt(d)) * DELTAT * QCONS;
    d = gtk->mass / (eps * sqrt(eps));
    d = gtk->mass / (d * sqrt(d)) * DELTAT * DELTAT * QCONS;

    Jason Nyberg

  22. occam's comic

    Haldon Arp’s work on “connected” galaxies is interesting not just visually but also because several of “connected” galaxies (he found many) had rather different red shifts, something that should be impossible if red-shift = distance. Is this pair one of the “impossible” ones?

  23. DarthTed

    I mis-read the title as Testicles of galactic doom, and then looked at the picture and thought the title made sense. Only then did I realize it was tentacles “nt, big difference”

  24. Chris

    I’m speechless. It’s amazing that we can see things like this.

  25. Capt. Action: yes, the Milky Way/Andromeda elliptical galaxy (or merger) will collide with some other galaxies when heading towards the Virgo supercluster. That would be really nice to see, too…

    And Phil, many thanks for the paper. I’ll try to read it as soon as I can.

  26. Walabio

    Oh Gary # 7, The Bad Astronomer has a MacBookPro.

  27. tussock

    So, all I have to do is wait up for a billion years or so and the sky will be getting much more interesting.

    Cool, unless there’s some sort of delay for the southern hemisphere, though NZ may have drifted north by then anyway. Might have to look it up sometime before then.

  28. ipgrunt

    I would imagine the probability of a collision between two stars is quite small “when galaxies collide” (so much space, so little mass), but has anyone actually attempted to calculate these odds?

    I have no idea what factors should be in such a formula. Can any of the astronomers here venture a guess?

  29. Nigel Depledge

    Once again, BA, thanks very much for bringing a superb astrophotograph to our attention.

    BTW, I have to ask the obvious question: how many times have you been pondering galactic collisions and NOT had an image of galactic collisions appear in your inbox, hmm?

  30. forrest noble

    Good question,

    Observed galactic collisions occur at a much less frequent rate than encrouching speeds might suggest. Why? Because current theory is wrong! Why? ask me! I think some of Halton Arp’s theories will prove to be right– after his death, similar to my own theories! se la vi.

    forrest noble, forrest underscore forrest at netzero dot net.

  31. Pierre

    Thank you Jason, I successfully applied the patch. I can see the behavior of the colliding galaxies is slightly different, indeed. Still, it’s clear the simulation is far from ideal, because once you look at a few runs you see that if it was correct the intergalactic space would be filled with stars flung in all directions whenever two galaxies get close, which is obviously not the case. I guess it’s just a numerical problem: not enough precision in the floats that are used, and bad handling of real close encounters. Not much we can do, it’s what happens when we try to model a continuous phenomena using digital approximations… Anyway. Thanks.

    Have you seen the program Celestia? I love it, it’s great. Once you get used to how to manipulate your camera properly, it’s wonderful.

  32. Jason Nyberg

    Pierre, glad to hear it.

    Re: the fidelity of the simulation, the galaxy hack is a poor approximation of how a real galaxy would behave, as it doesn’t model the gravitational effects of the individual stars at all, only a single central point mass per galaxy. The stars are more like field “tracers”… My own version puts all the stars in a single plane, which I think results in a more coherent view of what’s happening in the gravitational field. I “fix” the start conditions in other ways to produce “better” near-misses that look cooler…

    I have played around with Celestia a little bit.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


See More

Collapse bottom bar