The Steiner Problem in its classical formulation reads as follow: given a finite collection S = {p 1 . . . , p m } of points in the plane find a connected set with minimal length that contains S. It is well known that minimizers are finite union of segments that meet in triple junctions forming angles of 120 degrees. Finding explicitly a solution, it is however much more challenging (even numerically). In 1995 Brakke [3] and more recently Amato, Bellettini and Paolini [1] introduced an alternative approach to the Steiner problem rephrasing it in a covering space setting: minimizing the perimeter among constrained sets in a suitable covering space of R 2 \ S is equivalent to minimize the length among all connected planar networks that contain the given m points. The covering, here denoted by Y , can be constructed by a cut and paste procedure that we sketchy describe here (see [1] for details). Consider a network in the plane that contains S and another Lipschitz curve Σ that connects the points and does not intersect the network. The union of these two (the network and the curve) creates a partition in m regions of the plane. Consider m copies of R 2 \ S and lift each of these regions to one of the copies. Moreover introduce an equivalence relation that identifies the points along the curve of the different copies in such a way that the m regions can be seen as one set E in the covering space Y (given by the union of the copies together with the equivalence relation). Then the perimeter of the set in Y is twice the length of the network. The set E has some special features: it is a set of finite perimeter in Y such that for almost every x in the base space there exists exactly one point y of E such that p(y) = x, where p is the projection onto the base space R 2 \ S. We denote by P constr (Y ) the space of all sets in Y satisfying the previous properties. Then we look forWe underline that this quantity is independent on the choice of the curve Σ in the construction of Y .Our first goal in [5] was to introduce a theory of calibrations for Problem (1).Definition Given E ∈ P constr (Y ) a calibration for E is a (sufficiently regular) vector field Φ : Y → R 2 such thatwhere with Φ i we denote the restriction of Φ on the i-th sheet of the covering Y . The main purpose of searching for a calibration is to show easily the minimality of a certain candidate. Indeed we have the following:Theorem If Φ : Y → R 2 is a calibration for E ∈ P constr (Y ), then E is a minimizer among all sets in P constr (Y ).Finding a calibration is not easy. In [5] we exhibit a calibration only in the case S is composed of 3 and 4 points, but for general configuration of points it seems to be an hard task. In particular it is a long standing open problem to find a calibration when S is composed of points lying at the vertices of a regular polygon [4]. Indeed, even if different notions of calibrations are present in the literature (see for instance [7,8]) and despite the effort of more than one author, this problem has never been addressed. This...