In addition to the oversights mentioned by other posters (panels in partial shade, putting panels on the north side where they will never see any sun), it seems that he also neglected to attach a load to the outputs of the cells. This greatly skews the results in favor of the tree arrangement of the cells. If a solar cell does not have a resistive load attached, applying a small amount of light to it quickly raises the voltage across it to a maximum value, and increasing the intensity of the light does not increase the voltage any further. This is visible in the graph of the voltage on his standard design: on most days, the voltage quickly jumps from 0 to nearly maximum voltage and stays constant over the day. If the voltage measured actually described the incoming power, it would show a clear maximum at noon and gradually decrease at earlier and later times as the sunlight becomes less direct. What he has effectively measured is the number of cells being exposed to any sunlight at all. For his "standard" design, only half of the cells are ever in the sun, where for the tree design, it would seem that more than half are in the sun on average, since none of the panels seem to face downward as they would if the orientation were random. I think he would find that, if he would actually measure the power produced, the south-facing panels would produce more, since they face the sun more directly, on average, than somewhat randomly oriented panels.

If the VHF comm was transmitting around 121.19 MHz, the 13th harmonic of this transmission would fall right on the L1 frequency of GPS at 1575.42 MHz ( http://en.wikipedia.org/wiki/GPS_signals ). Given how weak GPS signals are at the surface of the earth, it isn't surprising that even a high harmonic of a multiple Watt signal could interfere with it. But it does seem that the interference only happened because of this overtone. Looking at the GSM frequencies given here ( http://en.wikipedia.org/wiki/GSM_frequency_bands ), it seems there's no integer fraction of either the L1 or L2 frequency of GPS that falls into any of the GSM uplink bands, which would mean that at least GSM phones could not interfere in this manner.

And a counterfeit warrant wouldn't just have the number of an accomplice on it?