On the 14th I did an A-B coast-down test of a full Kammback and found it was worse than none. (A- Aero Hitch Box, B- Aero Hitch Box plus full Kammback.) I'm hesitant to report in detail since there were a bunch of problems with the test (10 MPH winds with 17 MPH gusts, and cardboard and duct tape Kammback was falling apart, I only coasted down from 45 MPH instead of 70). Although the data was crummy and I was still learning how to do a quality coast-down, I also did a tuft test and felt that the results combined with tufts flowing into the gap indicated that gap filler needed to be addressed first. No point in getting into detail on the Kammback until the gap is filled.
So this Thur the 20th I did a much more detail-oriented A-B-A test with a gap filer. A- Aero Hitch box; B- Aero Hitch Box plus gap filler on the sides and top. I will have to start a new thread sometime about coast-down testing to discuss the many details, and I'm not finished analyzing my data either (and I may need to pick an even better road before I claim definitive accuracy), but I think I've applied due skepticism to the following chart and I think it holds water, at least in relative A-B-A terms, (not absolute accuracy):
A- average: .360
B- average: .335
Improvement: 7%
(Reminder: I think the 7% improvement I'm reporting is supportable, but I know I have a few more items to address to get a good absolute Cd, so take .335 with a grain of salt for now.)
So the gap filler appears to be mission-critical, and I've started fab on a road-worthy version. I would suspect the gap filler becomes even more critical with increasing yaw angle, so the 7% might even be lower than real-world.
Although I thought the spandex idea in the Aero Trailer TDI thread would probably work well for me, I wanted to try coroplast first. With a little luck I'll have some pictures to post early this week.