> > when in doubt, blame Microsoft
> Are you saying that MS *should* have dowloaded 4125 bytes, but did it right because it misinterprets the content type ? ? ?

No, MS should have downloaded 4125 bytes, but did it wrong because it misinterprets the content type...

The webserver is telling it that it's a plain text file. When downloading a text file, a browser should adapt the line endings in the file to the local custom. UNIX uses single LF characters (0x0A), DOS uses CRLF pairs (0x0D0A), so Netscape correctly adds a CR to any single LF it finds, which is why the file is bigger, and also why it doesn't work anymore...

The problem here is that the webserver is not configured correctly. It should be reporting a binary content type such as application/x-octetstream. It's the webserver's resposibility to determine the type of a file, not the web browser's...