I'd try open the screws up to about 2 +1/2 turns then slowly turn one of them in at some point the revs will just start to drop, (the engine may speed up a bit before slowing down if 2+1/2 is too rich) you can just about hear the engine slowing down. I'd then open the screw about 1/8th of a turn. You then need to do the same with the other screw. (Get the engine up to temp first)
After that I'd kill the engine and screw each one fully home counting the number of turns. Write this number down as you will forget! (then re-set them).
The above method seems to get the A/F ratio to about 13.5:1 which is a CO reading of about 2.5%.
My 4.6 runs at 2 turns to get 13.5:1. I think Grahams 3.5 will need to be around 1+7/8 or 1+3/4 turns to get the same reading. His car is on 2 turns at the mo (unless he's changed it and it was running at 12.5:1) You can easily see the difference 1/8th of a turn makes using a lambda probe.
The "proper" method would be by CO readings but the above method is not bad.
I'm in the process of setting my 4.6 lump up using a lambda probe and the settngs that appear to be near the mark are way off the standard settings. (Secondaries are now at 104, up from 94. They might even have to go to 107!). I've ended up "thinning" the needles that I'm running by 1 thou on the cruise step).
Its a long process and has cost about 150 quid in fuel! (Power runs down my test road are burning it at a huge rate!
).
Regards,
Pete