gerdes1723:
it makes sense that the capacitor didn't have time to charge or drain after the multiplexer switched.
That's why you use a 100n capacitor from the tap of the voltage divider to ground.
When the A/D muxer comes past to measure the divider, it sees a solid voltage (the 100ncap) to sample from.
Now you can even use 10Megohm resistors.
http://jeelabs.org/2013/05/16/measuring-the-battery-without-draining-it/
Edit
I seems the 100Megohm input impedance is just a number grabbed out of a hat.
I have measured >>1000Megohm, at room temp, with nothing else connected to adjacent pins.
Charge a low leakage large value film cap to e.g. 3.3volt, and measure it every second.
With noting else connected than the cap, the A/D value will stay there for days.
Leo..