I think that in Special Relativity, the drift of the perihelion can be calculated by cumulating elementary Lorentz transformations along the (Newtonian, unperturbed) trajectory. I read that the result of this calculation is much smaller than the experimental value. It is also well known that general Relativity produces the exact result, within the error bars. My questions are: What happened and what can we learn from that ? How and why did GR perform better ? Furthermore, is that not somehow in contradiction with the Equivalence Principle ? Could we not imagine that the EP applied to this system would imply the same result for both SR and GR ? Thanks a lot for your feedback, Michel PS: I would also be interrested by a reference about these calculations, I am a bit lazy ! Specially about calculating the drift as a perturbation of the Newtonian case. Can that really be done by Lorentz tranformations?