As far as I can see, you can either:
a) eyeball it, and judge that a given grey is middle, that another is between white and middle, that another is between black and middle, and so on. The problem here will be that of one's eyes accuracy.
b) slavishly follow a sequence of exposures and use the resulting tones. For example, expose a sheet of paper to a step wedge corresponding to 0.3 increases in density. The problem here is that tones are not linearly spaced on a paper characteristic curve, so that a given pair of successive tones are separated by, let's say 0.6 density, while another pair of successive tones are separated by 0.9 density.
c) use a reference grey scale (a Munsell chart or a Ross scale) and work to replicate the tones on the photographic medium. The problem here is that the medium used to make the reference scale may not have the same reflective property as a given photographic paper, so it's approximative at best.
My ideal method would be:
1) Find maximal black and white on a sheet of photographic paper
2) Do a first approximation of the scale by using either of the three methods above
3) Measure the intervals between successive tones
4) Normalize the intervals so that tones are equally distinct from each others
You end up with a series of greys G1...G10 that is matched to a series of exposure time t1...t10 (or densities d1...d10, depending on how you work), so that t1->G1 and so on.
It's really step 3) and 4) in the above method that I'm trying to figure out. Step 3) is definitely a technical problem (use a light meter? use a reflection densitometer? use a method that involves both the measurable and the perceptual aspects?).
But I think step 4) really is an aesthetic problem. Again, if I look at music, there are more than one way to space your notes, and not all temperaments equally space notes.
The real goal here is tonal control in printing: how can you establish a reference palette when you are printing, so that you can make the minute adjustments in dodging/burning/etc to help you with achieving a particular tone.