I got interested in whether the chassis as a ground is as good as a negative side power cable. Since copper is such a better conductor than steel, it seems a reasonable question if one should run a ground cable or not. If your battery is next to the engine this is not worth contemplation, but my battery is in the front.

I ran some rather low current tests with cables I had laying around and concluded maybe a ground cable would be a good idea. My experiment did not include cranking the engine, so it was not a definitive test. But the scraps weren't pretty/sexy so I bought a length of #2 at the local parts store. My negative battery lead is run to close by to the right side steering rack mount frame/bolt. Cleaned, shiny and tight. The engine ground strap is at the chassis left engine mount support.

Engine is an H6-3.0 liter, connected to a 98 Forester manual 5 speed. Starting was done with cold engine, temp was about 50F, with the clutch engaged as I had not installed the clutch slave cylinder yet. Transmission was in neutral. Test equipment is Fluke 87 meter at starter 12V terminal to starter frame. A high end Wavetek meter at the battery and a Beckman DC current clamp set to 400AMP range.

Battery is a used one, and not exactly the most healthy, but a big Group model used in our Dodge Charger V8's, and I think delivered ok for this test:

No ground cable, chassis ground only -
crank = 160 Amperes, 11.3V at battery terminals, 10.53V at starter

Added ground cable, #2 guage copper -
crank = 160 Amperes, 11.3V at battery terminals 10.60V at starter

While you may get different results, or in current lingo YMMV, I am thinking I don't need the ground cable. On the other hand, it will get some red shrink at each end and be my new, bigger battery positive cable front to back.