“…A natural space for local minimizers is the class of functions u from W 1 p (Ω; R nN ) such that Ω f (∇u) dx < ∞ for each subregion Ω Ω, and so one is interested in the regularity properties of local minimizers u which means that one asks for higher integrability of ∇u, Hölder-continuity of u or even Hölder-continuity of ∇u provided that f satisfies additional smoothness and convexity assumptions. In general, the hope for positive results increases in the scalar case but counterexamples of [Gi2] and (later) of [Ho] show that even for N = 1 unbounded minimizers exist, when q is too big with respect to p. On the contrary, there is a long list of authors investigating the different aspects of the regularity theory, we mention (without being complete) the works of Acerbi and Fucso ( [AF]), Fusco and Sbordone ([FS]), Marcellini ([Ma2]), Choe ([Ch]) and the papers [ELM1], [ELM2] of Esposito, Leonetti and Mingione, where the interested reader can also find further references. Typically, in the above mentioned works either a bound of the form q < c(n)p (1.2) with c(n) > 1, but c(n) → 1 as n → ∞ is required, or a dimensionless restriction like q < p + 2 (1.3) occurs together with the assumption that u is a locally bounded function.…”