Paul Vigay wrote: My guess is that it's actually happening at the HTTP transport layer, because of content type negotiation. Servers can store things like GZip'd HTML and the HTTP layer is supposed to transparently pass the unzipped result back to the browser without the browser having to know that this occurred. Transparent compression and decompression over the wire doesn't make as much sense as it once did, given the greater available bandwidth these days, but it's still a neat idea even if not widely implemented. I don't know enough of the relevant bit of the HTTP 1.1 specification to be sure, but it could be that the *server* isn't quite behaving properly when it sees a tar.gz file. Maybe it specifies things in its response to NetSurf that makes the latter believe it should transparently decompress the file with upper layers of the application unaware that this has taken place. It would be interesting to see what content type the HTTP layer passes back in such cases - assuming that this is indeed what's going on. Browse supported this feature of HTTP 1.1 many moons ago - you would know it was in action when the reported data transfer rate was several times higher than that supported by your connection. I too saw this happen with .tar.gz files on the odd occasion from certain servers. -- TTFN, Andrew Hodgkinson Find some electronic music at: All sorts of other bits and pieces at: http://www.ampcast.com/pond http://pond.org.uk/