Pentcho,
"In the direction perpendicular to the motion, the time to reach the mirror and come back is calculated by solving (cT)^2=L^2+(vT)^2, so T=(L^2/(c^2-v^2))^(1/2)."
Wouldn't this calculation also apply for Feist's measurement? The values he presented in his Figure 7 do not confirm this calculation. That's why my essay called his experiment stunning. I looked for a plausible possibility to explain the undeniable and easily reproducible result of Feist's experiment. My reasoning is quite simple and hopefully understandable from my Fig. 5: The signal was emitted into the air with high directivity. It propagates within it with c. If the air was at rest, it would reach the position R_0 at the reflector after a timespan d/c. While the motion of air does not change this duration, the signal has meanwhile been shifted sidewards to the position R_2 by dv/c. The reflected from there highly directive component of the signal cannot return to the emitter E because it gets further shifted to the right and will arrive after once again d/c outside E. The emitter E can only see a diffusely reflected from R_2 part of the signal that compensates for the rightward shift during return. Seen from E it seems to come from a fictitious position R_2. Hence the length of return path R_2 to E amounts d sqrt(1+4v^2/c^2), and the total time T_2 of travel from E to R_2 and back to E (T_2= 2 + 4v^2/c^2) agrees well with the apparently dilated time alias contracted length in the direction of motion [T_1=d/(c-v)+d/(c+v)].
In my essay I wrote v/c=r with roughly r=0.1 in Feist's measurement. Motion of earth with about 30km/s re ether corresponds to about 30/300 000 = 0.0001.
With the expectation by Potier/Lorentz/Michelson, Feist should have measured cT_2/d = 1.005. He actually measured 1.010 plus-minus 0.0005.
Eckard