I don't know the physics, but I think it has to do with what is hitting what as well. A steel ball bearing has low COR in that high school experiment, but guys have been using them to test the rebound of anvil faces for a while now.
http://www.anvilfire.com/article.php?bodyName=/FAQs/anvil-5.htm
Seems a ball bearing only rebounds about 35% when it's dropped on an old colonial style anvil, but the same bearing rebounds to 93% on a modern Peddinghaus anvil with an incredibly hard steel face.
My anvil is a near-mint Hay-Budden, and a ball bearing rebounds about 85% of it's dropped height according to the Anvilfire site. This quality makes it a joy to use because you don't have to use as much energy to raise the hammer each stroke. If I ever sell all my tools but one, my Hay-Budden will likely still be sitting on it's stump in the workshop.
Mass has a lot to do with it as well. That's why everybody wants to find a 500# anvil if they are using heavy hammers on thick stock.
With pool balls I thought the COR was up around 96% or higher for ball-to-ball collisions. Hopefully Dr. Dave will show up and straighten us all out.