The notion of deformation involves that of rigidity. In relativity, starting from Born's early definition of rigidity, some other ones have been proposed, offering more or less interesting aspects but also accompanied of undesired or even pathological properties. In order to clarify the origin of these difficulties presented by the notion of rigidity in relativity, we analyze with some detail significant aspects of the unambiguous classical, Newtonian, notion. In particular, the relative character of its kinetic definition is pointed out, allowing to predict and to understand the limitations imposed by Herglotz-Noether theorem. Also, its equivalent dynamic definition is obtained and, in contrast, its absolute character is shown. But in spite of this absolute character, the dynamic definition is shown to be not extensible to relativity. The metric deformation of Minkowski space by the presence of a gravitational field is interpreted as a universal deformation, and it is shown that, under natural conditions, only a simple deformation law is possible, relating locally, but in an one-to-one way, gravitational fields and gauge classes of two-forms. We argue that fields of unit vectors associated to the internal gauge class of two-forms of every space-time (and, in particular, of Minkowski space-time) are the relativistic analogues of the classical accelerated observers, i.e. of the classical rigid motions. Some other consequences of the universal law of gravitational deformation are commented.