I believe there are some fundamental flaws in Russ video in the flow of the VIC. I am working on the premise that this coil arrangement is correct. However, here is how I believe that it is functioning.
First off, let's look at the initial pulse on the primary core. It will create a field on the core creating a potential difference across the L2 coil. Now the one thing that they are not seeing is that this also creates a voltage potential across the L1 and secondary coils (which are in series) also at the same time since they are in parallel with L1. This means that the voltage created on L2 will develop a field on the secondary core (which have L1 and the secondary connected in series) at the same time as L2 purely because of the voltage potential developing across L2.
Now, because L1 and the secondary are connected in such a way so that their current will oppose each other, the net effect will be that they cancel each other out, preventing current flow and leaving that side of the water cell with a nearly static electrical state.
However, L2 is not static. It continues to receive a pulse from the primary and induces current opposing the static state on the opposite side of the cell creating the pulse stretching effect on the building charge of the cell; essentially, the unipolar pulse. This is where the L2 must be variable to tune into the resonant properties occurring at the static side of the cell. Once they match, you will have a proper unipolar pulse across the cell.
When tuning the frequency, the resonant pulse will have a tuning range at resonance. I believe that it will need to be tuned in such a way while monitoring the wave form of the feedback coil so that peaks are reduced only until pulses and gate time reduce to form the side by side pulses indicated in Meyers "inductive coupling diagram". I have done this and created this coupling effect.
EDIT:
Also, as I indicate in my past KB post TSKB00006 at
http://www.ionizationx.com/index.php/topic,2409.msg22744.html#msg22744, L2, being variable, may need to be exactly tuned to the corresponding inductance value for its side of the water cell, whose capacitance is different from the opposite side due to the difference in plate area so that matching resonance occurs on both sides of the cell.