I think I have "seen the light", concerning this. The chart I want really cannot be valid without lots of replicated patterns. I can build my pattern plate and shoot all the patterns I want, and arrive at my own conclusions.

Thanks to several of you who pointed that out. It really sunk in for me when I used the corn yield example. Replicated tests are absolutely necessary to have any useful data there, too. Once a farmer, always a farmer, I guess. smirk

Thanks all, SRH

May God bless America and those who defend her.