Strange behavior and motion of bubbles in a glass of beer Very unintuitive observation:
I pour myself a Guinness and the bubbles in my glass seem to move down toward the bottom of the glass instead of rising directly to the top of the glass as foam.
How can this be explained? Why is it that I observe this behavior drinking Guinness and not other carbonated drinks? What system properties (ie, temperature, nature of the solute and solvent) would affect this behavior?
 A: Yes, this is a real effect. In my hometown of Pilsen which gave the beer its name, every kid knows the physics because they're taught to drink and observe Pilsner Urquell, Gambrinus, and other top brands already when they enter the kindergarten. See e.g.

http://www.stanford.edu/group/Zarelab/guinness/StanfordReport.htm
http://www.stanford.edu/group/Zarelab/guinness/

The explanation is that the bubbles in the middle of the glass go up immediately - because it's easier for them to move if they're in the middle. But by going up, they're pushing the liquid in the upward direction, too. Obviously, the liquid has to return to the bottom in some way - to the place that was initially occupied by the rising bubble.
What is the path through which the liquid is returning over there? Well, the path goes near the boundaries of the glass - near the glassy material itself - where the bubbles are easier to be seen. So the rising bubbles at the center create some circulation patterns that go in the opposite direction (down) near the glassy boundary and that's where the bubbles are very easy to be observed (they're not hiding behind other bubbles and opaque liquid). With some exaggeration, the circulation patterns may look like this:

The effect doesn't last long because the bubbles ultimately achieve a higher speed than the circulation speed (note that the bubbles moving up also accelerate, at least for a while). After some time, the liquid is returning to the bottom in between the bubbles, more or less uniformly in the whole horizontal area of the glass.
I suspect that dark beers are more likely to be biased in the direction that the "nearby bubbles, near the glass, are easier to be seen" which is why the visual impression that "the motion down prevails" should be stronger for darker beers such as Guinness.
A: It's a beautiful thing to look at, isn't it?
I always assumed that it was an observation artifact, basically a visually observable "hole vs. electron motion in semiconductors".
I have seen the same effect in lighter coloured beers too.  It seems to me that a high population density of small bubbles is needed.
(@Lubos I've never been to Pilsen and have aways wanted to try the beer there.  And later today I will be less than 100km away, hmmm...)
