As shown in the attached screen capture, capacitance measurements can be rather unstable. This is an intermittent issue.
Keysight told me: "R&D has seen this issue (intermittently). At the time they also changed power outlets to ensure the DMM was not sharing the outlet with other instruments, motors, or 3-phase devices. There was some improvement, but not a "resolution" as fluctuations continued to be intermittent.
What did end up improving performance when measuring a "physical" capacitor was to use a shielded coaxial cable with the shield connected to the grounding screw on the rear panel of the DMM."
Has anyone else experienced this?