One thing to check for is any high spots on the end cap mating surfaces or on the mating surface that holds the bearing stack in the quill. In actuality the surface that contacts the bearing stack should place a load on the outer races to retain them in place with the surface that comes close to the quill body having a .005" gap between the two surfaces. This is pretty standard practice with precision spindles. Often with precision spindles there is a torque setting for the screws holding the end cap on that can not be exceeded because it may cause ditortion in the spindle housing. IIRC the preload on a BP spindle is all in the bearing set them selves with the spacers being equal length. One trick I used to do on BP heads when rebuilding them was to make a new set of spacers that were shortened to allow the use of a triple set of bearings in the spindle with the bottm two set in a tandem mount so that they are both back to back with the third bearing. This greatly stiffened the spindle leading to better finishes with the additional benefit of not getting beat to death when the idiots would run fly cutters. For anybody interested in doing this the new spacers have to be dead nuts flat. They can't just be faced in the lathe they need to be ground properly on a surface grinder while being held on a set of ground at set-up magnetic parellels. Grinding the parallels at set-up ensures the surface you are working off of is as flat as the grinder can produce
The optimist says the glass is half full, the pessimist says it's half empty. The paranoid in me says somebody put a hole in it.
Remember pessimists are at heart opptomists. They know things can and will get worse.