> Presumably the test generator is producing bogus numbers from time to
time.

Thank you for the report: it's actually a problem with the library code!
(The `write-hexadecimal-float` procedure accidentally used `write`
instead of display for special cases, so the test was correctly
complaining that "#e#x\"0p0\"" cannot be read as a number.)

I have incorporated the other suggestions (along with other changes).

Thank you for the comments,

-- Peter McGoron