2) As far as I know, nobody's ever found a foolproof method of getting a spring to fly away. Based on experimentation with belts and the Delay parameter, this is what I think happens: a spring, when landed on, is programmed to descend a bit (at least if given a delay higher than 0) and then return to its original position. Through some mishap -- the traditional way of triggering it involves having a powerup bounce on it, although presumably other gravity-affected crates should work just as well? -- the spring's internal memory of where it's supposed to return to gets reset to 0,0 and so it takes off towards the upper-left corner at the slow speed it uses to return from a Delay.