Spent the last two days "Prunin' and Tunin'" on this spin of the antenna, and I think it's about as good as it's going to get.
After looking at the data I collected, and seeing what was going on, I decided to go back to the full 88' length. This gives away some of the nice match it had on 15 Meters (21MHz), but improved it on 20 Meters (14MHz). The match on 40 Meters (7MHz) and 75 Meters (3.5MHz) is still far enough from 50 Ohms that I don't think I'll be able to transmit into it with this radio, but I can live with that for now. The final spin of this antenna will have an autocoupler at the feedpoint, providing a nice 50 Ohm match on all frequencies I intend to use. But for now, I'm getting my feet wet (again) with wire antennas of the non-resonant type. The only other time I've done this, has been with an autocoupler at the feedpoint, and since the radio always saw 50 Ohms, and I didn't own a graphical antenna analyzer, I couldn't measure what the wire by itself looked like, so I didn't care about it. I just put up as much wire as high as I could, and connected it to the SGC-230 autocoupler. So this has been an interesting learning experience, and a good source of exercise the last few days.
Final configuration for the "Mark-I" is:
88' wire, #14 gauge
9:1 balun at feedpoint
4 counterpoise/radial wires. Two are 100' long and six feet above the ground, and two are 50' long, about 2'~3' above the ground. The top ones are mounted to the fence pickets by threading them around every 5th picket or so, and the lower ones in the back run along the mid-height fence stringer, and the North ones are stapled to the fence. Going from two, 50' radials to four 50' radials, and then lengthening the top ones to 100' didn't change the shape of the curve very much, but dropped the curve down, resulting in a better match over the entire frequency range.
Here's the graph of the 'final' Mark-I version.