# Understanding biplot for compositional data

1. Say in the biplot of some compositional data, I discovered that the links $\vec{AB}$ and $\vec{CD}$ intersect at roughly 90 degrees. So I can say that $\log {A\over B}$ and $\log{C\over D}$ are roughly uncorrelated. The questions is... so what? What's the use in discovering this?

2. Now, say in the biplot I discovered vertices E, F, G are collinear. Then my 'textbook' says they form a 1-d biplot. So they have unidimensional variability. The question is... what does this even mean?