when you pass a numeric value greater than 255 to chr, it must return a wide character.

There is no "must" about it. It should be the case that unless I specifically ask for Unicrap, characters should be assumed to be 8-bits.

I'm afraid I don't quite understand the reason(s) for what happens when the "use bytes" pragma is added -- if I've done it right, the only difference is to eliminate the warning message about the "wrapped character in unpack"

You're right. It does just enough to lull you into a false sense of security; then sneaks around behind and kicks you in the nuts!

Rant:

Is it the case that you got the particular pattern of zeros and ones you expected, and were just complaining about the warning message?)

No. I wanted the shift to discard the high bit, as it does with integers:

$n <<= 1; print unpack 'B*', pack 'N', $n;;
10101011010101001010101101010100

$n <<= 1; print unpack 'B*', pack 'N', $n;;
01010110101010010101011010101000

$n <<= 1; print unpack 'B*', pack 'N', $n;;
10101101010100101010110101010000
[download]

Unfortunately, Unicrap (and Perl's implementation of Unicrap) conspire such that you can no longer rely upon simple byte semantics.

The idea that a string (a good old array of bytes) can suddenly contain a random Unicrap character in a program that doesn't (and doesn't want to) use any Unicrap, is a farce!