You probably should pm him or see if you can email. He hasn't been on the 41 for a long time and restricts his posts to the 33

I do not believe that you can test frame flex losses with two power meters. And subtracting two big numbers is only half of the problem.
The bigger problem is that the front power meter does not accurately measure the power input into the system. Correct me if I am wrong, but it computes the power from measuring force and pedal position. However, the latter is measured relative to the bottom bracket. Therefore, whatever number your power meter gives you it is already your power minus the loss.

Jee, this is a blast from the past. What happened to Waterrockets?

OTOH, I tried measuring how much energy could possibly go into the frame.
I sat on my bike with the front break applied, shifted my weight from the saddle to the pedal in 3 o'clock position and measured the vertical pedal deflection, which came out to be 3-5mm with me weighing 69kg. Assuming that all my weight pushes the pedal it makes it roughly 1.4J of energy. This transfer happens twice per revolution and with 120rpm cadence we get 5.6Wt going through the frame. That might seem like a lot at first, but pushing with 700N at 120rpm is _A_LOT_ of power. Even if I inefficiently push the pedal straight down, I get 4*0.34*700 = 950Wt of power. And 5.6Wt/950Wt = 0.6%. Not that much.

PS. all measurements done on a steel frame from Columbus SL

well.. there is still no answer in the thread =)