[/quote]

Every test that's performed on arrow performance (or any test for that matter)musttry to have only 1 variable and keep everything else as a contant when comparing items. Introduce too many variables and the test is not valid or the outcome is skewed.
[/quote]

You said it, there's the problem