In the actual implementation the transformer would transform the 50ohms load to 12.5ohms as seen by the collector. This would result in the theoretical output of 5 watts QRP from a 12volt power supply.
Previous tests in terminating the 3dB attenuator pad in a 50ohm load showed that the attenuator is accurate and indeed has an insertion loss of 3dB. As can be seen in these results the power measured at P2 should therefore have been 360/2 = 180mW. In fact the measured result was 193mW. Indicating (I think) that in fact the input impedance to the transformer is something other than 50ohms. So the actual input impedance would be R=V^2/P = 3.11^2/180mW = 53.7 Ohms. ie 7% error. The load resistance was 200ohms as measured so the error must be in the transformer.
Looking further at the measured output power of 176mW it can be seen that there is a loss of 8.8% power in the transformer. Again the 200ohm load is accurate at least at DC level!
This teaches me that even the simplest measurements at RF will yield some very instructive results!
Using a return loss bridge (which I MUST build now) would yield the system performance with less maths!