Note: these instructions are my memory of a conversation with a former naval engineer who is into this stuff as a hobby. I made one of these that worked OK, but I've never read up on the underlying science. It works and the explanation sounds plausible, but take it with a pinch of salt.

1) Make your electrodes: Get two sticks of graphite (probably from an art supply shop). To each one, attach a piece of insulated wire longer than your jar is deep. Make sure that the join is protected from water: the easiest way is to expose some wire, wrap it tightly around the graphite, then cover the connection and all of the exposed wire in hot glue. You're going to dip these in water, so it's important that all of the metal is protected from water and that most of the graphite is exposed.

2) Get a big (clear?) jar, something like 2-5 litre capacity. Fill it about 1/3 of the way up with mud. I'm told that mud from the bottom of a river or pond with ideal, but there's a lot of flexibility: I made a working cell using builder's sand with a torn-up cheese and lettuce sandwich stirred through it. You won't get much power out: think in the same sort of range as sticking electrodes into a stack of lemons. But it's enough to measure, and maybe power a small LCD clock.

3) Bury one of the electrodes in the mud, nice and deep. The wire from that electrode should come up through the mud and out of the jar.

4) On top of the mud, fill the rest of the jar with water. I used bottled water, as you want to avoid too much chlorine. Pond or river water would probably be great too. Dangle your second electrode mid-water, again with the wire coming up and out of the jar.

5) Now connect an ammeter or multimeter between the two wires and wait for a day or two. You're looking for a current measured in milliamps.

Deep down in the mud, there isn't (or soon won't be) any oxygen for the bacteria to use, and new oxygen can't easily travel through the inches of mud to the bottom of the jar. Instead, some of the species already present will switch over to a form of anaerobic respiration which involves dumping excess electrons into their environment. So your buried electrode will have a source of free electrons. And as well as blocking oxygen, your layer of mud is rubbish at allowing electrons to flow far, making a charge accumulate wherever these bugs are growing.

Up in the water, other life forms (to my shame, I don't know which ones) would be all to happy to accept those electrons, if only they could get them. So electrons dumped onto or near your chunk of graphite in the mud will tend to flow into it, along the wire, through the ammeter, then into the electrode that's hanging mid-water and out of that to whatever's there to accept them. You now have a current flowing through the wire, powered by energy produced by bacteria munching stuff in the mud and water.

If you're lucky, algae will grow in the water and start to thrive. These will use sunlight to manufacture sugar (and, indirectly, other tasty stuff), some of which will gradually rain down into the mud and keep your electron-producing bacteria well fed. Although I didn't try it, the guy who told me about this claims that you can make these airtight and, given enough sunlight, they'll stay alive and producing electricity pretty much indefinitely. Occasionally you'll get one where the ecosystem crashes and everything dies, but he says that most of his go through a couple of algal blooms/die-offs then stabilise.

His dream is to see this sort of thing established along canals, rivers and reservoirs, effectively turning all decent-sized bodies of water into solar power sources. I've no idea whether this is possible or what the side-effects might be, but it's an intriguing thought.

Give it a go! It's a bit of a slow-burner - an hour to set up, then a few days or weeks of watching the ecosystem develop and mature, and monitoring the current. It'd probably be handy for teaching kids all sorts of things about electricity and ecology, if you're into that sort of thing.