It is finicky enough to set up two identical antennas that have virtually identical characteristics (aka twin). The bigger the difference in the antennas (or perhaps the less twin like is a better way of saying it?) the more difficult it will be, and the more skewed the output will end up being, but that is not to say that it can't be done.
I guess the question I have is, with as little of a difference a cophased setup will have over a single antenna (in most cases a difference so small you will never notice it) along with the additional problems of setting one up, why would you want to add in an even bigger layer of difficulty like this? Or were you planing on doing some other type of phasing?
The DB