Thanks for your response.

For me, your string returned:

93 64 6f 75 62 6c 65 94
d1 20 d2
201c 64 6f 75 62 6c 65 201d
2018 73 69 6e 67 6c 65 2019
[download]

xd1 and xd2 are outside the range being checked (x80-x9F), are legal unicode (the cp1252/unicode chart gives the same codes) and encode_entities returned (as you found):

Ñ Ò
[download]

Note that, for example, ’ returned x2019 so wouldn't be mucked about by the cp1252 replacement.

The comment in the script:

# "replaces HTML entities...
#  with the corresponding Unicode character"
[download]

is from the H::E doc and had more significance than I first realised.

__But__ I too am surprised it appears to work. I wrote quite a bit of code to process utf8 and was almost a bit miffed that it seemed unnecessary!

This my first outing in these waters so will be pleased to be corrected if I've got any of this tangled up.

Again, thanks for your comments,
John

unicode.org cp1252 chart

update:

Extract from the chart:

cp1252  unicode
0xD1    0x00D1    #LATIN CAPITAL LETTER N WITH TILDE
0xD2    0x00D2    #LATIN CAPITAL LETTER O WITH GRAVE
[download]