Message written at 8 Jun 1999 08:21:36 +0100

I have just seen all this mail about use of double precision.  It
seems like a small experiment is in order with the main sources.  I
will look at it when the marking has finished, but I am basically out
of things for at least another week.
  I think a little minor massage would allow Csoudn to be built either
32 or 64 bit floats (assuming the machine supports that).  Perry Cook
did something like this for his physical modelling code.......ho
hum.... and I still want to do the binary score.srt file to gain
precision.