In M. Walter Maxwell's books he addresses this. The first of his Reflections books this is section 19-4, in the second and third of his Reflections books this is section 24.5. This is taken from Reflections 3.
Sec 24.5 The Effects of Line Attenuation said:
In real transmission lines having attenuation, all power entering the line is absorbed in the load, except for that dissipated as heat and lost due to line attenuation. When the matching is performed at the line input, the loss from line attenuation happens because in addition to the attenuation of the forward power, the reflected is also attenuated in the same proportion on its return to the source.
Should this concern us as amateurs? Probably not, if we are using a typical antenna system at HF. When the matched-line attenuation low, as it is in typical amateur installations, the additional loss because of Z0 mismatch is small, as explained in Chapters 1 and 6. The additional losses are so low that even with moderate to high SWR values, the difference in power radiated compared to that with a 1:1 SWR is too small to be discerned by the receiving station. The additional power lost versus SWR may be determined from Figs 1-1 or 6-1, or may be calculated using equation 6-1.
This is where it ends in Reflections, however in Reflections 2 and 3 there is a reference to Appendix 8 and Appendix 8A which is another way of calculating the same thing as the figures he mentioned above. However, in Reflections 3 there is another section that discusses this further.
Sec 24.7 The Conjugate Match-Is it Real said:
Along with the myth that no conjugate match can exist when the source of power is an RF power amplifier, another myth concerning the conjugate match has re-emerged. This myth had it that the conjugate match exists only in theory, because the assertion is that it can occur only in a lossless environment that doesn't exist. The claim is that any resistive loss in a circuit containing real reactive elements introduces additional reactances that destroy the true conjugate match. This is true to a point, but it is only the beginning of the story.
He then continues on for nine and a half pages which I am not going to type up here for several reasons, although it is a good read. I will, however, summarize.
When loss is present in the components this loss will change said component's reactance. This is where the claim comes from that a conjugate match can no longer exist. However, according to Maxwell that is where said claims stop.
Maxwell then goes on and talks about adjusting the reactive elements individually to compensate for their loss resistance, in effect "tuning to resonance". For the record, the method described mirrors how ham radio operators use their antenna tuners. I'll quote another part of it here.
Sec 24.7 The Conjugate Match-Is it Real said:
However, we must be careful to understand that during the conditions described above, although the conjugate match exists everywhere in the real network, it exists in only one direction -- forward.
Essentially, the conjugate match becomes directional. As long as the system is tuned to be a conjugate match with the source on one side and the load on the other, the conjugate match still exists. If something changes, say, you stop transmitting and start receiving, effectively reversing the source and the load, the conjugate match doesn't exist, in the case of this example, as long as you are receiving.
I hope this helps.
The DB