I did I pretty comprehensive test once upon a time with .38 HBWCs loaded backwards. I used both .38 and .357 brass, and varied the velocity from 650 fps to 850 fps. I used a 2.25 inch SP-101, a 4 inch GP-100, and a 6 inch GP-100, and shot 6 round groups from a rest at 7 yards. In every case with every load tested, at least 3 of the 6 rounds of each group keyholed (Hit the paper sideways). Another example is when the wadcutters were loaded correctly with 2.7 grains of Bullseye, the 6 rounds went into one hole at 7 yards. When loaded backwards with the same load, it shot 4 inch groups at seven yards with 4 of the six rounds keyholing.
When loaded base down or using factory ammo, the .38 148 HBWC is a great load for defensive use in small frame snubby revolvers, teaching new shooters, or for competition. I have not been convinced that the HBWC loaded backwards lives up to the hype.
Just my .02,
LeonCarr
I tend to agree with your accuracy assessment, but I didn't really experience much keyholing at 7 yds. although I do agree that base down was far more accurate. My buddy and I were loading with Red Dot and I can't remember the load, but I also loaded with 3.2 grains of Hodgdon Trap 100 (Win 452AA). With base down loads I could put 5 shots in a 2.5" diameter circle at 25 yds from a rest. That was with my Dan Wesson M15, the most accurate revolver I've ever owned (still kicking my self for selling it). My Ruger Security-Six wasn't quite as accurate, but close.