Here's the deal. If you put the wires close together you get a lot of
interaction. The manifestation of the interaction is that the
higher-frequency dipoles end up considerably shorter than normal, and
they'll have a narrower bandwidth than an isolated dipole. The longest
one will also be affected by the others, but not nearly so much. You'll
also find that small differences in spacing can have quite an effect on
the dipole resonant frequencies, which is why a cookbook approach
usually doesn't work unless the writer is very careful to document the
antenna accurately and you're extremely careful to exactly duplicate it.
But you just about always end up having to tune it.

Tuning a close-spaced multiple dipole like this is time consuming. You
begin by adjusting the length of the longest one to resonance. Then you
adjust the next shorter one, and so forth. It might be necessary to
repeat the process after the first time through. And, as I mentioned,
you'll end up with some pretty narrowbanded antennas, and the lengths
won't be what common formulas predict.

The interaction decreases rapidly as you spread the dipoles apart. If
you can get them around 30 degrees apart, the interaction is minimal and
you can just about treat them like separate dipoles. A lot of
installations fall between these extremes, so the dipoles have some
interaction but it's not as severe as it is when they're very closely
spaced.

Roy Lewallen, W7EL