I went ahead and edited the article to use a 0.9 correlation instead. Again, 1 is the maximum correlation--and the ONLY case where you get a flat line. Every other case produces a negative slope.