Here's little more information than I had the other night. Not much mind you.
Look down near the very bottom of the page above for the two-color subtractive information. This is the page that got me thinking about using a plate beam splitter and a front surface mirror. It is boatloads easier than finding or fabricating a Phillips prism.
I did toy with the idea that you could just use several beam splitters to get your three images, but that's a lot of loss in the system. Basically you'd loose at minimum 2 stops if they system is perfect. And the beam splitters farther down the chain have to get pretty big or you have to stick with small negatives. When I said earlier that a plate beam splitter was affordable I didn't mean 8x10! The Phillips prism avoids those losses.
Another link for two color:
Sorry there's no hard core data here. I guess we have to rediscover it and make it up for ourselves.