When in doubt, always stick to the math. There were 2, maybe even 3 different decode equations that the BBC came up with for H during it's research and short run, then they went to different math for HJ when they punted on H. Also, there are most likely some small details about H and HJ that never made it to the tech papers, as by some accounts the BBC tech. folks were actively monitoring mix sessions when they were in progress. Thankfully they didn't leave us with too many recordings that need to be rescued.Good on you Duncan.
The articles in my Matrix H book mention in the clippings that the BBC engineers and Mr Mackenzie preferred a 60 degree phase shift in the decoding of H , which differed over the encode which conformed to a 47.5 degree shift.
This was the end result with the BBC Engineers and Angus MacKenzie's preferred aural reception of the quadraphonic broadcasts.