If the motor manufacturer has done the job correctly, there is a permanent match-mark (typically like a chisel or scribe line) on both the supporting ring that stabilizes the brush arms and some "fixed" point on the machine, usually at or near one of the clamping positions where the support is held steady. Line these up and you _should_ be at the best commutation setting. NOTE - This does NOT mean you are necessarily at the neutral axis; the rigging may be intentionally offset to accommodate other machine performance characteristics such as a wide field-weakened speed range.
If you have to do it yourself, do the following:
1) Set the mechanical position of the brush contact to be as close as possible to the main pole centerline (by eye). Apply a 120 V ac source to the main field, and using an auto-ranging voltmeter, measure the voltage at a specific point relative to the center of the brush contact. (A lot of folks who do this regularly use a non-conductive "scale" with a series of holes punched in it and test the voltage using a pencil-style contact. They center the scale on the center of the brush / commutator contact patch.)
2.a) If you're using the "scale" method, you'll get a curve of by plotting the different points in relation to the centerline of the contact. The idea is to get the lowest value to coincide with the midpoint of the contact.
2.b) If you don't have a scale, you'll have to manually move the brush rigging a very small amount (typically 5 or 6 positions in each direction from the starting point, and about 0.25 inch (6 mm) or so circumferential distance between points, measured out at the outside diameter of the support). Once you find the lowest point, set the rigging so that the middle of the brush contact is over that particular point.
The 120 V value works well for shunt wound motors (i.e. field resistance measures more than an ohm or so). For windings with lower resistance, lower voltages (and greater care) need to be taken to get the best setup.
And if there ARE series windings - note that the "running" neutral may not be the same as the "no load" neutral point due to the influence of the current in the series windings.