The RP25.
This is an example with the Goodmans loudspeaker. There are 2 versions of this fitted to the RP25 - early ones have 3 bolts on the rear of the magnet, later ones are without bolts. This is the latter.
Frankly, the Goodmans is a great AM speaker, but not so good for FM sets. Hacker recognised this, and replaced it with a Celestion model shortly before they changed to the RP25A. I don't have an example of this that is easily accessible, so I'll have to come back to that...
Subjectively, the Goodmans lacks treble. The bass is a bit tight too - earlier "bolted" models are more generous in that area. So what do the measurements tell us?
As mentioned earlier, I skipped the "raw", direct-to-loudspeaker measurements, and took two - tone controls set to flat (which is minimum on many Hacker sets, though not the RP72) - black trace - and after the tone controls had been set for best results on music (blue trace). And from here, I decided to use 1kHz as a reference point for both curves.
As expected, we see a great big lump in the middle (400Hz to 5kHz), with the bass and treble rapidly rolling away either side. But this one does go up reasonably far - better than I expected, and better than the RP72 unit, and it might be possible to use some dedicated equalisation to bring it up. It would be easier to add a small tweeter, perhaps.
At the bass end, we see the sharp rolloff below resonance (~100Hz).
Having adjusted the tone controls, things improve at the low end, and the HF is extended to just over 10kHz, but the 9dB spike at 3-5kHz is there, and it's not pretty.
So, slightly better than expected, but not a great result. Hacker can do better