AVGresponding, Kleinstein:
I finally took the time to do the measurement as promised. I tested the 34401A input in >10G DCV mode with oscilloscope and 10V source. Fall time seems consistent enough and average to 16.5ms, so input capacitance should be 750pF (fall time defined as 90% to 10% and tau~=fall time/2.2). It's a reasonable value, so perhaps the experiment is correct. I tried 10M DCV mode and fall time got to 8.2ms as expected, I tried ACV mode and (this was expected but not to this extent) 34401A is awful here with a fall time of 5.14s, that translates to an effective 240nF, not according to 34401A specs that say 1Mohm in parallel with 100pf. In the schematic (page 158 of the document 34401-90013) I see C301, a 220nF capacitor in series with a 1Mohm resistor going to a OPAMP virtual GND, thus we are reading the 220nF discharging to an effective R of 11Mohm, measuring it through a resistance divider 11M/10M. Taking everything in account, effective measured C is 230nF. Good, it's C301.
I also tested the Fluke 177. DCV: wonderful 590us resulting to a 53pF (effective Zin 5Mohm). Here the oscilloscope probe should be taken in a account with its 15pF, so we have 38pF, something less because also the cables count in this case. ACV: the circuit looks similar to the 34401A, but with a 10Mohm in series to a C that looks to be 12nF.