I got this idea while pondering over this news :
http://www.overunity.com/index.php/topic,3388.0.html
Capillary action (http://en.wikipedia.org/wiki/Capillary_action) requires no external energy. (Correct me if I'm wrong). Water rises 14 cms in a tube of 0.1 mm radius all by itself.
If one takes an inverted Y shaped tube, with one leg dipped in the water and other leg(shorter, so that it stays above the water surface) with two gates at the junction and opens the gate which is not immersed in water periodically, one can make a tiny waterwheel turn using this falling water perpetually.
Water won't fall out of the capillary, so one needs a material which can also expand/contract, so that by changing the diameter, water flows out.
Now the question is - whether the energy gained will be greater that that required to open/close the gates and expand/contract the tube?
What do you guys think about it? I guess there is a catch, otherwise someone would have done it already ;)