I haven't seen the rest before, but the Cleveland and McGill paper is excellent, and well worth the read. They make some very interesting conclusions about the effectiveness of commonly used graph types, and concrete recommendations for alternatives. The criticisms of 'patch maps' is very strong, and makes an interesting point.

If you can find copies, its worth reading all the Tukey papers cited by the Cleveland paper, especially "The Future of Data Analysis".

Some nice papers there, and I clearly am not up on my classic Viz papers.

NB If you want a dense, difficult, crazy even but (perhaps) ultimately rewarding paper you could try:

http://scholar.lib.vt.edu/ejournals/SPT/v7n2/araya.html

The author goes right back to the idea of what it means to measure and display and view geometry within a two dimensional screen.

What, no Edward Tufte?? ;-D

Not directly, but check the references in each of the papers!