The program is here though. It comes in two versions: the recommended one with everything in double precision; the other with a mixture of single and double precision, which should be faster on 32-bit machines.
You can also get the (fairly incomplete) notes that I used to send out with it.
October 1997: thanks to Markus Wobisch who found a small but important bug in the reconstruction of subjets within an inclusive jet. It is now fixed in the version above.
November 2000: thanks to Charles Currat who pointed out a potential machine-dependence. This is now fixed in the version above but is also the reason for now recommending the double precision version by default.
Mike Seymour