This is an observation about the behaviour of the 34465A DMM which is different than the 34411A predecessor.
Both DMMs have the DIGITIZING option (standard with the 34411A) and 2M Memory options fitted.
When the 34465A is set to maximum sampling rate, (SAMPLE:TIM MIN) then read back (SAMPLE:TIM?) it shows 2E-05 (20 uSec).
Using a sig-gen using 50Hz (< 0.001% error) square wave, and viewing the sample log it shows that there are close to 1002 samples for one 50Hz period. So the sample rate is actually about 0.2% less (19.960 uSec) from what is displayed on the DMM
With the 34411A DMM setting the SAMPLE:TIM MIN and then reading back shows the correct sample rate, eg around 19.960 uSec).
Also with the 34411A DMM you can set the SAMPLE:TIM 1.9960E-05 and it will accept it, but on the 34465A it will give a SCPI range error.
This gets very confusing when you have a mix of both models in group of 10 of the DMMs in an ATE system in which we are relying on the timing accuracy to high degree.