@ martinr33, e61_phil, Kleinstein, syau:
The null-meter requirement for 720a, 752a is to be uncertain to <= 500nV or <= 200nV respectively, (see manuals), which in practice includes resolution and noise.
For sure, nV-DMM fulfill both requirements, but they may be used only, if their bias current is < 1pA.
This is also not specified for those under consideration (34420, 182, 2182, etc), as low noise and low bias are always contradicting parameters.
Newer, low noise Chopper OpAmps also have high bias on the order of 20pA and more, and therefore are not suitable at all. 2057: 30pA typ, up to 300pA!
Only older ones like the LTC 1052 might do the job, but must be selected to below their typical bias, i.e. 1pA typ., 30pA max.
Here again, lower bias means higher noise.
Those old optical chopper null meters are the appropriate instruments for this job.
Frank
Edit: I just had a look into my recent tests on the 752A. From that I have to emphasize, that a proper shielding configuration, following the 752A manual addendum, is much more important for noise figures, and in turn for zeroing correctly, than having an ultra-low noise null-meter.