This paper studies the consensus problem of nonidentical Euler-Lagrange systems with parametric uncertainties, under switching directed graphs. We present an adaptive distributed control algorithm that solves this problem in the leaderless scenario, under weak assumptions on the switching interconnection between agents. It is shown that the proposed consensus protocol achieves consensus without using neighbor's velocity information and accounts for the discontinuity of the input torque that results from the switching interaction between agents. Simulation results are given to illustrate the effectiveness of the proposed control schemes.