When extracting files from your website, can you check what types of content are in the two files? I would expect a different behavior if they showed a different type (perhaps the web server interprets the data and sees different triggers in each file).
I checked the headers of http://web-sniffer.net/ and the second file no longer exists, so I can not compare.
Another wrinkle that can affect things is if you don't configure the user agent to match your Android devices, you can get a different result in your desktop browser or web sniffer.
Not sure if this is causing your problem, but worth checking out.
source share