>   I thought that float representation varried rather a lot between
> processors. The number of bits of mantissa, etc.

A lot of modern systems use the same format - it might even be
an IEEE standard.  This is probably because the initial version
or the processor used 'off the shelf' hardware (or software)
and everything has stayed compatible.

So if you have 2 known systems you might be able to move
fp values between them.

Remember a while ago there were systems that didn't use 2s
compliment integers.  I wrote code for one system that
did arithemtic on decimal numbers (up to 10 digits) with
a sign 'overpunch'.

	David

-- 
David Laight: david@l8s.co.uk