I have up loaded the paper Dec 1975 by Peter Walker in Wireless World.
Here is the Quad 405-2 circuit in its last iteration.
Now at first glance your reaction is going to be the same as mine was, and most others looking at these circuits the first time, will be "But it's not stable!" But my friend it is.
However stability is a big issue in this approach. Peter and I were in continuous correspondence in the early years, as I was the importer for Western Canada.
In order to make the original 405 stable, slow power transistors had to be sourced. This had the result of adversely affecting the slew rate. Further iterations, and there were at 14, to my recollection improved the situation.
Now to keep it stable bandwidth does need to be limited pretty much to the audio range.
In all iterations of the 405 and 405-2, the slew rate was a limiting factor. So at the higher frequencies full power was not available. However, as Peter pointed out music never calls for full output in the upper frequencies! So they sounded fine at least with the 405-2. The sound of the 405s was slightly veiled.
However as devices improved, and with tinkering by Peter and especially his successors, this problem was solved. From the 606 and up there is no high frequency power bandwidth limitation.