CollinsGuitar: Hi Michael, good to see you here. I attach the top first and measure specific mobilty of the top, the deflection and the peak monopole frequency T(1,1)2. I know from experience that I can expect something like 20Hz drop of T(1,1)2 after full assembly. With just the top on, I can play with the bracing and/or top thickness easily. This is especially helpful with the Selmer guitars you and I deal and those little sound holes. The top is more important than the back, so I favor putting the top on first to be sure I get this where I want it (for better or worse
You raise a good point, however, the back is important too and I've been spending a lot of time thinking about how to voice it properly before attaching, assuming the top goes on first. Because we can't just reach in a shave braces, we need a different work around. Temporary attachments have their appeal. Taping the back to the top is reported to work, I tried it once and I recall the results were closed to the glued results. The tape has to solid and complete Thinned hide glue and paper between the joint is a time honored method for making an easily reversed joint especially if you get it apart before it sets up rock hard. I've not tried this, but I plan to this weekend. Attach the back, get it where I want, remove it, put the top on, voice it, then put the back on and hope for the best.
I'm all for letting the math do the work. I need to go back to G&G and see if the math on the back can be applied to the a full size laminated back.
Getting accurate data using a jig to temporarily hold the plates is difficult because the mass of the jig significantly affects the voice of the plate. This is quite noticeable and Trevor talks about the effect of the mass of the sides on the top and back extensively in the book. Unless the mass of the jig and hold downs is very close to that of the sides, it probably will give you a good result.