Nice to play with a meter but like you said, it is not a scientific test. Though, some things are easy to exclude by doing the right measurements. For instance use a DC source and a nulldetector (indeed like the standard cells) Some spectroscope meters also use a DC bias.
The pressure can be EMF, warmth of your hand, increasing the leakage current by sweat ect. But the test is simple. Just put it on the bench en press with a piece of teflon on it instead with your fingers.
An other very interesting thing you can do with that trendplot. Put it in 10G mode. Put it in the shadow and in such a way the sun will shine through a small hole and travels in time over the battery. For instance when the suns goes down. We do not want to fry it. You will the the voltage climb and go down. Eveb indoors inspring and a late evening sun that little bit of light was enough. Very funny. But you can use that oven you made too.
After 5 minutes, Dave talks about the 1 kHz impedance. But that is not even close to be correct. The impedance is measured by a 1kHz sinewave and not using a pulse during a very small time. Also R does not allways stay constant. It changes with current and charge. Enigizer has a pdf over impedance of their batteries.
Impedance is made out or a pure resistive part ( materials like metal, and other stuff used) and an ionic part. The chemical proces ( electrolyte conduction, ion mobility, electrode surface area)
There are several tests:
DC resistance, first load the battery with a very small current and let it stabilize (the chemical proces), Then load it with a larger current and measure the voltdrop. The best way is to do the loadpart with a pulse during 100 ms and a longer recovery so deeper layers have some time to distribute charge to the surface layers again
There are two subvariants. One is repeating pulses (100 ms) that load it and pull a "normal" current. The other is shorting the battery 1 time with something like 0.1 Ohm and measure the Voltdrop but this is very hard to measure. This is also called a flash amp test.
AC resistance, for those small batteries this is most times done at 1 kHz. It measures |Z| and that will be lower as the DC resistance. Sometimes much lower. DC resistance varies with load (and current) and charge left . That also explains the behaviour in the video. With such a small load, resistance will be much higher so the voltdrop higher as expected. ( Think of it as the the ESR of a ceramic capacitor, very low at 100 kHz upto 100KOhms's at 50 Hz ) For instance an AA cel I tested (varta) had 6 Ohms DC resistance when (pulse) loaded with 100 Ohm but it droped to 0.9 Ohm when loaded with 15 Ohm (and inbetween with a very light stabilizing load) So loaded with 10M it will be higher than 6 Ohm. (at 10GOhm ther leakage from battery plus to minus pole over the dirty fingerprints, dust etc on the outside can be higher as the meter loading it)
Spectroscopy, here they measure all parts of the impedance and split them up in Z = ( R+jX ) and it tells you something about the charge or health too. I allready measured this before I knew that name. It was strange I could not find anything about the way I measured and the results that give you an indication of the health of the cell. But this week I found out they do it but it is called spectroscopy in the battery world.
I used an impedance bridge and a VNA. However both methodes give the same parameters. It measures magnitude and phase.
Randles model of a battery gives two resistors in series and a capacitance parallel over one resistor. A charged battery or new battery has a certain Rs and allmost no capacitance. It is a even a tiny bit inductive if it is a larger cell. If charge drops the Rs increases and the imaginair part becomes more capacitive. A bad rechargable battery will show more capacitance as a good discharged one. But capacitance of a disharged one is always higher as when charged. I read somewhere a rule of tumb would be 1F per 100 Ah for Randles model.