Wow - looks like i started a profound technical argument!. I started looking at the same question the other night and almost went blind trying to reverse engineer the actual circut from the unit i have un-potted. I gave up due to time and fatigue. But I can tell you there ARE many more components in the circuit than are shown in that schematic. Several bias resistors, some bypass caps, and other protection diodes. If i didnt make a mistake, there are two (10K, 1k? not real sure if its orange or red) resistors from the two AC lines to the SCR gates that would keep them OFF unless Q3 overcomes the bias (through some other resistors and protection diodes not shown). Of course if i erred and these SCR bias resistors come from opposite sides of the AC, it would turn them ON in the absense of Q3 control!
I was about ready to argue the diode leakage would get the Q4/Q5 conduction "started" without a battery until i found the SCR "reverse bias" resistors.
Here's a Thomas Edison solution:
Anybody got a working tractor with this regulator? Go start it up, then turn ignition off (leave motor running), then disconnect battery, then disconnect/reconnect regulator (just to make SURE it turns off), then turn ign switch on, and turn lights on - if electricals/lights work, the circut is "self starting."