Initial investigation is very tentative. Aircraft is carbon fiber reinforced polymer (CFRP) covered by a coating inside and out and I'm not sure if it is isotropic or anisotropic but the latter makes sense.
Using an ohmmeter and measuring resistance between random points a couple of feet apart on the outside, on the inside, and between outside and inside indicated infinite resistance. Testing between several unconnected metal pass-through fittings showed low resistance. Obviously the coatings are non-conductive but the CFRP appears to be conductive and can be accessed by using a conductive fitting that is attached to it.
The airplane manufacturer rep said his experience was that ADS-B antennas could be installed with no ground plane. That supports the CFRP conductivity assumption.
Here are two papers by the Technical University of Wien that appear to be oriented at investigating the use of 1-10 GHz antennas in car bodies.
https://www.researchgate.net/public...as_antenna_ground_plane_material_up_to_10_GHz
https://www.hindawi.com/journals/ijap/2017/6152651/
They make the point that anisotropic CFRP is necessary to avoid distorted signal distribution.
A good friend who is a retired Collins Radio antenna engineer suggests that while CFRP may be OK as a ground plane in the 1-10 GHz range as supported by the papers, that may be different in the 3-30 MHz range of HF.
Another good friend, also retired from Collins, was a project manager on the Collins radios for the Beech Starship project a couple of decades ago. He says
"We used UHF and VHF antennas on the Starship with no big problems, aside
from the problem I mentioned about keeping a corrosion-free contact with
the carbon fiber material itself. I'm not primarily an antenna person, but we
always thought of it as a wavelength-dependent property. Where the
wavelength is short with respect to the airplane it seems to work; no
serious problems with antenna gain, range, pattern, etc. with VHF Com and
Nav or with UHF transponder, DME, GPS, GLS, Marker, etc. (GLS was a
symmetrical horizontal dipole, not using a ground plane.)"
He goes on to note directionality in the ventral fin mounted VOR antennas which had bearing and bank-dependent errors at ranges over about 75 miles ahead. He attributed it to the VHF signal not "bending" as it followed the skin of the Starship. (It could also be affected if the Starship layup is isotropic but that seems unlikely to this layman.) (Note he mentions problems with maintaining a good connection between the antenna feed and the CFRP, which could be an issue to those who install an ADS-B antenna with no ground plane, relying on good conduct with the CFRP via the through-fitting.) For my airplane, the wave-length is decidedly not short in regard to the ground plane.
The best producer of through-hull fittings for my needs seems to be Dayton-Granger. Their web site is not helpful but they sent me a thumb-drive of their catalog which is very useful.
A dipole mounted from each wing-tip to the top of the vertical stabilizer will match very nicely with the 20m band, about 14 MHz. If the CRFP hull would be an adequate ground plane I could probably match a 40m or 7.0-7-3 MHz antenna well.
The radio is likely to be a Yaesu FT-857D because I have one and there isn't much better at an inexpensive cost. I have an acceptable tuner. They will be mounted in the baggage area and a control cable run forward to the control head. The microphone will be picked up by a PS Engineering TAC 12100 which handles the impedance matching and fed to a PS Engineering 8000C so it acts just like any other aviation radio and can be selected as one of the radios, heard through the headset and uses the headset boom mike.
I'm going to model both a monopole using the CRFP as ground plane and a dipole. This is not going to happen for a couple of weeks or even months at the earliest. I'll post the results when I have them.
Thanks for those who made suggestions.