Stacking distance
Two antennas stacked one above the other narrow the beam in the vertical plane and add up to about 3 dB of gain. Too close and their capture areas overlap, wasting gain. Too far and big side lobes appear. The optimum distance is D = λ ÷ (2 × sin(θ ÷ 2)), where θ is the antenna's half-power beamwidth in the stacking plane, the method popularised by DL6WU. Longer, higher-gain Yagis have narrower beams and need wider spacing. If you don't know the beamwidth, it's estimated from the gain.
The phasing harness
The simplest harness uses two equal lengths of 75 Ω coax, each an odd number of electrical quarter waves long. Each line transforms its antenna's 50 Ω to about 112 Ω, and the two joined in parallel give 56 Ω, close enough to feed with 50 Ω coax. Both lines must be exactly the same length so the antennas are fed in phase, and long enough to reach the centre of the frame. Cut them with an analyser to be sure.
Side by side or one above the other?
Stacking vertically narrows the beam vertically and lowers the angle of radiation, which suits tropo and weak-signal work. Side-by-side stacking narrows the beam horizontally, which can help reduce interference from other directions. For the antennas themselves, use the Yagi calculator, and for a single wider-band beam, the log periodic calculator.
Questions
How far apart should I stack two 2m Yagis?
It depends on their beamwidth. For a 12 dBd Yagi, around 3 m on 2m is typical.
How much gain does stacking add?
Up to about 3 dB in theory, typically 2.5 to 2.9 dB after harness losses.
Can I use 50 Ω coax for the harness?
Yes, with a different method: equal lengths of 50 Ω to each antenna, then a quarter-wave 35 Ω matching section, made from two 75 Ω cables in parallel, to 50 Ω.
Do both harness lines need to be the same length?
Yes, exactly. Any difference changes the phase and spoils the pattern.