The Q3 collector voltages reported by mdc do look like really typical MkIII territory. Now, it looks like for mir9, there were a variety of zeners that could get you in that neighborhood depending on the transistor. That tells me leakage is a factor, so anyone building this may need to audition a couple of different zeners to get in that 2.5v territory. On the other hand, historic MkIIIs could be anywhere from 1.5v to 3v--I've tweaked ones I've built within that range, and they pretty much always sound good. So maybe we just say 4.7v zener and leave it at that?
Thanks for everyone's work on this, by the way! I had more or less forgotten about this variant, so am excited to see it come together.