There are line breaks at a fixed interval so that it wraps at 80 chars. I imagine that makes up the other ~300KB.
Edit: Line-breaks do indeed make up the bulk of the overhead, but it's also an HTML page so there's some boilerplate markup at the beginning and end of the file.
Edit: Line-breaks do indeed make up the bulk of the overhead, but it's also an HTML page so there's some boilerplate markup at the beginning and end of the file.