> I get the same problem on perfectly normal sites,
> reading perfectly normal HTML. Since HTTrack thinks 
it
> downloaded them right I can't even update them to 
full
> size. Perhaps some kind of 'after mirror verify' 
would
> be a good upgrade to HTTrack? In my case it might 
have
> something to do with the fact that I have a modem
> connection and the modem drivers don't seem to be 
fully
> stable.(ie. they might be dropping connections, and

Humm.. normally HTTrack always checks the size of html 
files (except if you selected 'Tolerant requests' in 
options settings), except when the server do not give 
any 'Content-length' in the response.

When the remote filesize is known, HTTrack look at the 
filesize at the one of the mirror: (quote from the 
sources)

              if (back[i].r.totalsize>0) {    // test 
totalsize
                if (back[i].r.totalsize!=back
[i].r.size) {  // not the same!
                  if (!opt->tolerant) {
                    if (back[i].r.adr) freet(back
[i].r.adr); back[i].r.adr=NULL;
                    if (back[i].r.size<back
[i].r.totalsize)
                      back[i].r.statuscode=-
4;        // recatch
                    sprintf(back[i].r.msg,"Incorrect 
length ("LLintP" Bytes, "LLintP" expected)",back
[i].r.size,back[i].r.totalsize);

The problem, when the engine don't know the file size 
sent by the remote server, is that the only solution 
is.. to wait and get data sent, until no data is sent 
anymore, THAT IS, until the connection was dropped. 
Therefore, the engine can't make the difference 
between completed pages, and connection broken due to 
remote or local error.

If you get such problems, erase all new.* and old.* 
files in the hts-cache folder, but you will loos all 
previous cached data.