Thunda - thanks for that, it's much more than I expected anyone to come up with.
This was obviously an official mod produced by Rolls Royce for Merlin 21s and 23s, but I can find no reference to it in otherwise authoritative titles such as Putnams' 'DeHavilland Aircraft since 1909' nor in 'British Piston Aero Engines' by Lumsden. As I said above, I believe the wartime use of N20 on a Merlin installation was unique to the Mosquito - a British equivalent of the German GM-1 equipment.
As I'm sure you know, the British generally preferred mechanical means for increasing charge density and cooling (via superchargers and intercoolers), rather than chemical injection methods, so I'm surprised to find N20 in use here.
I can only assume that the Merlin 21 series, being single stage, two speed engines, simply ran out of puff above 20,000ft and some expedient means of quickly increasing their rated altitude was required. From my reading it seems that the two-stage Merlin 61 series could not be fitted to the Mosquito because the latter's radiators were higher than the engines (not beneath them, as on Beaufighters, Lancs etc.), which required that the engine be redesigned to allow for 'reverse coolant flow' (the new engine design being labelled Merlin 70). I suspect there was a shortage of two-stage, Merlin 70s (most being fitted to photo recon Mosquito variants), so N20 equipment was retro-fitted to home-defence Mosquitos as a stop gap.
According to the Lumsden book, the 1390 h.p Merlin 21 was cleared for 14lbs boost in MS gear and 16 lb boost in FS gear, so the 12lb applied in the Cunningham/Bennett test was not the limit, but I suppose they didn't want to overspeed the engines in a dive and they reached compressibility anyway.
[I don't understand the reference to 'resins' in those reports...(!?)]