While mechanics was developed under the idea of reciprocal action (interactions), electromagnetism, as we know it today, takes a form more akin to unilateral action. Interactions call for spatial relations, unilateral action calls for space, just one reference centre. In contrast, interactions are matters of relations that require at least two centres. The development of the relational electromagnetism encouraged by Gauss appears to stop around 1870 for reasons that are not completely clear but are certainly not solely scientific. By the same time, Maxwell recognised the equivalence in formulae of his electromagnetism and the one advocated by Gauss and called for an explanation of why such theories so differently conceived have such a large part in common. In this work we reconstruct and update the relational electromagnetism up to the contributions of Lorentz guided by the non-arbitrariness principle (NAP) that requests arbitrary choices to be accompanied by groups of symmetries. We show that a-priori there must be two more symmetries in electromagnetism, one related to the breaking (in the description) of the relation source/detector and one relating all the perceptions of the same source by detectors moving with different (constant) relative velocities. We show that the idea of electromagnetic waves put forward in concept by Lorenz (1861-1863) before Maxwell (1865) and in formulae (1867) just after Maxwell, together with the ``least action principle'' proposed by Lorentz are enough to derive Maxwell's equations, the continuity equation and the Lorentz' force, and that there is a dual formulation in terms of fields of the receiver (as opposed to fields of the source). While Galilean transformations are associated with removing the arbitrariness implied in the election of a reference space, they will not explicitly appear in a formulation based upon a relational space although we occasionally mention their usefulness. In contrast, Lorentz' transformations will emerge in this formulation involving the relations between the perceived fields of different receivers. Moreover, the role of the full Poincaré-Lorentz group as a group of transformations of the perceived actions is elucidated. In summary, we answer Maxwell's philosophical question showing how the same theory in formulae can be abduced using different inferred entities. Each form of abduction implies as well an interpretation and a facilitation of the theoretical construction. This work relies heavily on logical concepts as abduction put forward by C. Peirce, needed for the construction of theories.