This post from 2007 was about a highly entertaining YouTube clip of some people pulling arcs from a long string of nine-volt batteries. With those neat little clip connectors, 9V batteries are just begging to be clipped together into very long, very dangerous daisy-chains. And, in that particular case, they had 125 batteries in series, by my count. That adds up to a nominal 1125 volts DC.
But now, unfortunately, that video's been removed.
So I went hunting for more experiments of this type.
Here's a string of 19 (for 171 volts DC, nominal) running a compact fluorescent lamp:
The experimenter boldly holds the thin-insulation alligator-clip leads in his bare hands, but that's as exciting as this video gets. Interesting to see that these lamps run from DC as happily as from AC, though.
Here's some fun with 52 batteries:
That'd give 468V if the batteries were all at their nominal 9V, and could easily make it to 500V with fresh batteries. But apparently these were discarded "8.4... ish" cells of unknown provenance (my money would be on a company replacing the batteries in all of its smoke detectors). 52 times 8.4 gives a mere 437 volts, open-circuit.
(All of these voltages will plummet when you close the circuit, to start striking arcs, because the more current you ask for the further the terminal voltage will sag, and alkaline nine-volters aren't meant to deliver more than a very little current. Energizer, for instance, don't provide a maximum-current rating on the datasheet [PDF] for their standard alkaline nine-volters, but the maximum current on the load-versus-capacity graph is half an amp, at which discharge rate the capacity drops from a 25mA-load maximum of more than 600 milliamp-hours, to a little more than 300mAh. If you buy a Big Bag of Innocent Unsuspecting 9V Batteries the cheap way, by getting carbon-zinc "super heavy duty" batteries instead of alkalines, the rated current [PDF datasheet] is now only about 5mA, and the highest current on the performance graph is only 25 milliamps. You're not going to be able to pull a multi-amp arc out of a string of those poor little things for long. Ex-smoke-detector alkaline batteries will probably work a lot better for this sort of Unwise Experiment than will brand new carbon-zincs.)
Here we have 48 batteries in series - so, maybe more than 460 volts open circuit - molesting a coin:
(With, again, not as much attention paid to safety as might have been.)
Here, though, is what we've been looking for!
Four hundred and ninety 9V batteries, baby!
490 times nine volts gives 4410V; fresh batteries would add up to more than 4700V. These are more ex-smoke-detector batteries, though; the video description says they only added up to "almost 4000V".
Even four thousand volts can't strike a very long arc by itself. The dielectric breakdown strength of dry air is about 33 kilovolts per centimetre (around 84 kilovots per inch). So four kilovolts, even with humid air helping it (and hindering electrostatic experiments...), can only strike an arc a few millimetres in length, at the very most.
Once you've struck a spark with the terminals close together, though, you can draw it out into a much longer arc, because the ionised air between the terminals - which may include vaporised matter from the terminals - is much more conductive than un-ionised air. That's how arc welding works (and Jacob's Ladders too, for that matter), and that's what's happening in the video clip.