If you have an antenna with a matching network that can adjust reactance and impedance separately it isn't to hard. A quarter wave whip is known to have about 36 ohms of radiation resistance (the good loss). If at resonance it is presenting a 50 ohm impedance at the feedline the remaining 14 ohms or so of impedance will add up directly to ground losses. The "perfect" X=0 R=50 on the mfj analyzers are not always the best radiating point for a given antenna. Sure the connection between the feedline and antenna is happy, as is the radio, but the antenna is loosing some of that power that has been transferred to it as heat.
Another thing to remember, just because some of the RF is reflected back down the transmission line does not necessarily mean that power has been lost. If the transmission line has minimal losses most of that power will be re-reflected at the transmitter (or transmatch or amplifier) and come back to the antenna.
The DB
Thanks for clearing that up.