Thanks for the replies, everyone.
Here's an isometric sort of cross-section that may make it easier to discern what's what.
The alternative to threaded rod is to thread the ends of steel rod -- and that's why I was looking at what lengths and variety of studs were available from McMaster-Carr. This would allow for round shaft to ride in the bearings, while threaded shaft was exposed from either side.
I understand also that there's some chance of the coupling nuts tightening against the bearing blocks. That will occur if it's easier to tighten the nut on the input side than it is to spin the shaft -- a situation most likely to occur at start and at stall. I don't have a good idea about whether it'll be easier to compress the coupling nut against the nylon spacer and bearing block than to overcome static friction, say, or what effect something like Loc-tite will have in binding the nut to the threads. Conceivably, I could pin the nut through the shaft to prevent it from moving at all, but the goal while designing this was that it'd require the absolute minimum amount of maching work possible. The left-handed bolts in each end of the shaft are to prevent the coupling nuts from backing off when the drive spins counter-clockwise.