Message written at 24 Sep 1999 09:01:10 +0100
--- Copy of mail to f00f@freemail.hu ---
In-reply-to: <19990923143153.15605.qmail@freemail.c3.hu> (message from
	=?ISO-8859-2?Q?Varga_Istv=E1n?= on 23 Sep 1999 14:31:53 -0000)
References:  <19990923143153.15605.qmail@freemail.c3.hu>

I *think* the k-rate linseg not going to zero is a fence=post effect;
it has run out of segments and so it repeats the last value which is
the value at teh start of the previosu k-rate.  However testing this
hypothesis has given a problem as my test version gives totally
incorrect answers to the orchestra (while the production one has the
fence-post problem.  

Oh dear!