APL ⎕RVAL real number generation wrong

[Lin Shile]

APL ⎕RVAL real number generates numbers (.5, .75).
possible cause: IEEE 754 float is defined as "se...ef1...fn", which
represent s 1.f1...fn 2^(a function of e...e), instead of
s f1.f2...fn 2^(a function of e...e).