Two low-level capacitors of 100 femto Farads, each (the left-most capacitor also possessing a precharged condition of one microvolt), provides for a switching signal which is simulataneously emitted by both capacitors producing

bounces back and forth among these three low-level capacitances while inductor, L1 (to the far-right in this circuit), accumulates a high-impedance of voltage. Together, this cap-gap-cap sandwich substitutes for this circuit's lack of a voltage source (and rightly so).
Both neon bulb spark gaps also provide for the generation of negative current (polarized in opposition to the polarity of sign for voltage) which noticeably occurs at low-impedance inductor, L5. This generation of current (especially at L5) makes possible the rise of amplitude (power measured in watts) since this compliments the accumulation of voltage at inductor, L1.
This circuit must store this escalation of wattage somehow. So, frequency rises in an attempt to store greater and greater quantities of wattage per smaller and smaller units of time.
This is how this particular variety of overunity circuit manages to override the misplaced assumption (popularized by a socially engineered ignorance within our "common" sense) that an inverse relationship exists between frequency and amplitude driven by the Conservation of Energy. This is a misplaced presumption carried over from the behavior of paired inductors on a transformer which does not apply to this situation since this circuit is overwhelmed by capacitive reactance. I know this due to an experiment in which I used a single sine wave input (of one microvolt) magnetically coupled to inductor, L1, which produced a triangular wave (not a sine wave) at inductor, L1.
[??]
Only by providing for two sine wave inputs of slightly differing frequencies was it possible to override this circuit's preponderance for capacitive reactance and create a sine wave at inductor, L1...
[??]
This proves that inductive reactance is superseded (in this circuit) by the capacitive reactance (of negative current) which discredits the equally presumptive (but popular) reasoning that "there is no such thing as a free lunch" with regards to the general topic of energy. This popularized dogma assumes that an inverse relationship exists between an increase in either frequency or amplitude, but not both at the same time, requires paying a cost for this increase by a reduction in either amplitude or frequency (and vice versa). This assumption is limited to mass-media descriptions of the positive (not negative) wattage of a transformer model which, alone, must satisfy this presumption. But this type of circuit avoids this limitation through the use of the inversion of polarity for current relative to the polarity of voltage in a circuit dominated by capacitive reactance. This type of wattage does not possess a direct relationship between frequency and amplitude, but an inverse relationship (instead).
This is a grave oversight whenever the topics of "free energy" and "free lunches" are entertained, and quickly dismissed, by our collective ignorance. All of us are liable (to some extent) for this ignorance. How often do we take the trouble to correct it? Do we, even, try? Or, are we too ignorant to bother...?
This is not an overly simplified, flashlight circuit catering to a throw-away culture of disposable batteries.