Because I intentionally MADE it part of the circuit. Lower impedance. Basic ohm's law. The winch is STILL connected directly to the battery with both cables as recommended, plus - emphasis PLUS - the frame ground. This means the negative lead is now in parallel with the frame - lower impedance. The direct negative lead isn't done that way because its the lowest impedance, its done that way for reliability - and its still in place if one of the frame connections fails (again: high current grounds in this type of environment can be problematic). There is no cable that could be run in practicality that is as low of an impedance as the frame route, assuming one has good ground connections that haven't corroded. In fact, properly installed, more than half the current is going to go through the frame rather than the cable for that very reason.
Looking up the resistance of #2 wire shows it to be 184 micro-ohms per foot at 149 Degrees F. E=IxR - at 400 amps that's 73.6 millivolts per foot, assuming a 6 foot length that becomes 441.6 millivolts AKA .4 volts drop. And that's just 1 leg - both legs will be in excess of .8 volts. Adding the frame to the equation minimizes half of that. Yes, you still have - doing the math here - just shy of 37 mv drop at the 6 inch jumper around the motor mount, plus - say - two feet from the block to the battery giving 147.2 mv for a total of 184mv. Which is less than half of the wire alone. But wait - you actually do have that wire, so doing all the parallel resistance calcs - 47 micro-ohms in the frame ground circuit - assuming 0 for the frame itself because it basically is - in parallel with 1.1 milli-ohms, is 324 micro-ohms which gives a .1297v - or 129.7 milli volt drop at 400 amps which is better than either one alone.
I glossed over the math a bit, converting the voltage drops back to resistance (at 400 amps) in order to make the parallel calcs - anyone please feel free to check my numbers I could have very easily fat fingered something on the calculator, or slipped a decimal. 130 millivolts beats 441 - that makes the total round circuit drop including the + lead goes from .8832 volts to .571volts which is .3122 volts improvement, or about 2.6% - almost a third of a volt. It could be further improved by using larger wires for the leads, and/or for the motor mount jumper/block-to-battery wiring. A third of a volt isn't a lot - but when you're running that kind of amperage, that motor needs every microvolt you can give it! And its very cheap and easy to do - so why not?
Ground jumper - you can't see the star washers I used:
View attachment 533452