Certain file types, can't draw them by byte ranges?

Can't get anyone to answer this in the developer groups, so....

I have built a downloading engine that draws files in 64K chunks, rather
than pull a file all at once. It does this by specifying which byte it wants
to start and end at. This is a memory management scheme designed to conserve
resources on the server.

It generally works really well - for image file types, PDF file types, and
raw data formats, I'm ecstatic about how well it works. However there's one
peculiar behaviour, and perhaps somebody has some insight.

For the Microsoft Office formats, the files do not arrive properly - they
are corrupt and unopenable. I've windiffed them against the originals and it
looks like there are huge gaping holes in the files - as though the data
offered by the web server has been truncated. I'm assuming this has
something to do with how IIS handles the mime types for PowerPoint and Excel
(the two formats where I noticed the troubles.)

has anyone got any insight on what might cause this, and how I might somehow
force these files to behave like normal binary files, offering up the right
bytes in the right ranges?

Thanks,

R.