As Azz seems rather relaxed about answering posts here, today I've reported to Lance @ Zen that 186708 produced three failures one after the other, but a test on 161357 grinds on in its usual way, but does at least provide an answer.
The two tests are obviously being routed to different locations as the failing one doesn't report the IP Profile figure whilst testing, but the good one still does.
The bad one also starts the test immediately whereas the good one probably takes over a minute to start.
Both these lines are on the THCN exchange but the good one is on an Alcatel DSLAM and the bad one is on a TI DSLAM.
They are probably both linked to a Reading RAS but I haven't yet collected the log data to prove that.
Kind regards,
Walter