Ive been testing this very thing for the last 4 months, just to observe SE behavior, and effect on
PR. I have 4 sets of domains, about 30 per set, no more than 3 domians per C class. In each set, all domians were purchased the same day, but I put one or two old domains I owned from previous SEO tests with some
PR, just to send spiders to each domain. I also use a cloaker to generate relevent keyword pages . Im doing this in a very competive two word combo.
In the 4 months, all of the the
PR from the new domains have had a gradual increase, from nil to 3, 4, even 5. Maybe there are errors in the way I break it down, but best I can tell, its the linking and the C class IP factor.