Thanks for your prompt reply (not to mention the great work on the product). 
As an example, I have a p3 600 machine with a 10GB hard drive.  It downloaded
a site over the weekend that when I last checked was at 7 GB in size on the
hard drive (approx 1.5 gb left).  This morning it reported the mirror
completed (log = fatal error, giving up) presumably because it ran out of hard
drive space but the mirrored site is only 800 mb (and thus there is now almost
8 gb in free space).

Now, besides the undeniable fact that I should join the 21rst century with my
hardware, the .html.tmp files that I presume were downloaded, not parsed, then
deleted could be useful when I transfer the project to a bigger machine and
restart.