In most character sets, including iso-8859-1 and Unicode, the pound symbol is 163 decimal, or 0xA3 hexadecimal, so you should use "chr(163)". If you use the correct encoding on the output, that will be translated to the correct character.

E.g. this

perl -e 'binmode STDOUT, ":encoding(cp437)"; print chr(0xA3), "\n"' \
   | od -c -t x1
0000000 234  \r  \n
        9c 0d 0a
[download]

(Code page 437 is a common code page for a DOS box on Windows, but it can be changed with CHCP.)

and this

perl -e 'binmode STDOUT,":encoding(latin-1)"; print chr(0xA3),"\n"' \ 
   | od -c -t x1
[download]

gives:

0000000 243  \r  \n
        a3 0d 0a
[download]

But, note that 156 (0x9c) does not work with the Encode translations:

perl -e 'binmode STDOUT, ":encoding(cp437)"; print chr(0x9C), "\n"' \ 
   | od -c -t x1
[download]

Gives:

"\x{009c}" does not map to cp437.
0000000   \   x   {   0   0   9   c   }  \r  \n
        5c 78 7b 30 30 39 63 7d 0d 0a
[download]

If you are outputting to HTML, you also have to make sure the Content-type charset gets set correctly or output the characters as entities, e.g. £ or £