Its C3 that does the timing, both for the delay (charging) and for the pulse length (discharging via the other path). The lack of higher value tanalum beads at Maplin is a tad disappointing (they probably noticed that they were useful, and therefore followed their long standing policy of ditching them), but I was probably just being picky. Regular radial cans are a lot better than they used to be, and the actual timing is not hyper critical.
Upping the value will increase the range of timings available, but note that the increase will likely not be linear - the input sense of the 555 does take some current which does affect the timing somewhat. When designing circuits with a long timing, it was common to use the CMOS version, or use a short repeating pulse supply and count the pulses. For what is wanted, this would be overkill. Anybody into programming PICs would probably laugh his socks off, then spend the next 40 seconds or so writing a routine to do that exact job.