In my experience, it's not just atmospheric water, and most drift is associated with tramp water and on-going reagent degradation, especially if titration systems are left empty and only used occasionally.
If drift was mainly atmospheric, then leaving the system sealed should result in a continual downward drift to near zero, but I find the equilibrium time doesn't change much whether I add a sample or not, and if I put a dry Nitrogen purge in place, the drift only deceases by about 30%. I suspect most is tramp moisture, along with some inefficiencies of the desiccant.
My volumetric titration Mettler DL31 sits on a drift of about 4-5 ug/min, as it's used fairly infrequently. If I warm glassware and use a small nitrogen positive pressure, rather than just molecular sieves, it can drop to about 3 ug/min, and the Hydranal booklet say that 0.01 ml of reagent ( my reagent is 5 mg/ml )/ is the maximum drift expected from a "perfectly dry" titration cell. That would be 5 ug/min for my instrument, which is similar to what I had for a Mettler DL18 system, using the same composite 5 reagent...
The Hydranal book also says a coulometric system should have a maximum drift of 10 ug/min, with 4 ug/min as ideal.
The advantage of most modern reagents and systems is the speed of titration, which then minimise the effect of drift.
I highly recommend getting the free Hydranal CD from Sigma-Aldrich, it's full of good stuff about the reaction and the reagents, as well as plenty of hints on how to minimise issues, such as drift, stirring, and titration speed..
Please keep having fun,
Bruce Hamilton
