Sounds like your best option is to decide for which subject distance you want to optimize it, and move the front group around, seeking the best image on ground glass. Perhaps you could use a cardboard tube or some PVC to cradle the elements. (I have some fancy lens mounts in the lab, we use them to set up things on an optical breadboard, but that's a pricey way to go about it; if you have a physics dept nearby, you can probably find somebody there with a breadboard system and they can help out)
Idea! If you use PVC or similar, note that you can get PVC which has mating screw threads, which might make a nice ghetto helicoidal adjustment system

You really need some sort of cradle mechanism because having the lenses slightly off axis or tilted will very quickly degrade performance. (OTOH it might degrade in an interesting way!)
The "correct" answer depends what focal length and focal distance you want to optimize for. You might say the only "correct" choice is what duplicates how the lens was produced.
Probably the best thing would be to get your hands on a duplicate lens and inspect that... you could at least measure the distance from the front and rear surfaces with calipers.
Are you planning to cement it back together?