I'm not sure, if this is a bug, or intentional
behaviour, because on one hand RFC states that field
names for HTTP headers are case-insensitive, on the
other hand, it's not what server really returns to client.

Example:

print response.getheaders()

Yields this (output formatted a bit for clarity):

[('content-length', '5998'), 
('accept-ranges', 'bytes'), 
('last-modified', 'Sun, 30 Jan 2005 14:36:09 GMT'), 
('connection', 'close'), 
('etag', '"26d79-176e-41fcf0d9"'), 
('date', 'Thu, 24 Mar 2005 18:14:07 GMT'), 
('content-type', 'text/html')]