This paper investigates leaderless consensus control problem for a group of Euler-Lagrange systems with unknown control directions under an undirected connected graph in the presence of parametric uncertainties. A distributed adaptive controller is presented using the backstepping technique and a Nussbaum-type function. Moreover, the controller is distributed in the sense that the controller design for each system only requires relative information between itself and its neighbors. The projection algorithm is applied to guarantee that the estimated parameters remain in some known bounded sets. Lyapunov stability analysis shows that the consensus errors converge to zero asymptotically. Simulation results on multiple two-link planar elbow manipulators are provided to illustrate the performance of the proposed algorithm.