The general rules for two elements are:
A voltage fed element requires equal voltages with proper voltage phase lag or lead.
A current feed element requires equal currents in each element feedpoint, with proper current phase diffeence.
Phasing antennas at heights other than those that have current maxima or voltage maxima at the feedpoint can be very complicated, because neither voltage, current, or phase are what you might think. Because of that, I think you are biting off more than you should to try end-fire phased 5/8th wave antennas. I would stay away from it, and get my feet wet with 1/4 wave verticals first.
With 1/4 wave spacing, the optimal phase difference is around 110-120 degrees for skywave. This will place a null cone over the rear, and increase gain and improve apparent F/B ratio over 90 degree phasing.
Phase can change over a reasonable amount, perhaps from 95 to 125 degrees, without noticed changes. Improper current maximums, where they are not equal in current, will kill F/B ratio much more than small phase errors.
Read this carefully:
http://www.eznec.com/Amateur/Articles/Current_Dist.pdfand this
http://www.eznec.com/Amateur/Articles/Simpfeed.pdfbut keep in mind OPTIMUM F/B on skywave does NOT occur with 180-S degee phasing. In other words, optmum phase with 90 degree spacing is NOT 90 degrees, nor is it 120 degrees with 60 degree spacing, or any other 180-S combination.
Unless you actually want less gain (wider pattern) or less rear rejection of signals, you really want the null offset slightly to the sides at zero elevation, so the null forms a cone around the rear.