1: \begin{abstract}

2: Due to the divergence-instability, low-order conforming finite element methods (FEMs) for nearly incompressible elasticity equations suffer from the so-called locking phenomenon as the Lam\'e 

3: parameter~$\lambda\to\infty$ and consequently the material becomes more and more incompressible. 

4: For the piecewise linear case, the error in the $L^2$-norm of the standard 

5: Galerkin conforming FEM is bounded by~$C_\lambda h^2$. However, $C_\lambda \to \infty$ 

6: as~$\lambda \to \infty$, resulting in poor accuracy for practical values of~$h$ if $\lambda$ is sufficiently 

7: large. In this short paper, we show that for 2D problems the locking phenomenon can be controlled by replacing 

8: $\lambda$ with $\lambda^\alpha$ in the stiffness matrix, for a certain choice of~$\alpha=\alpha_*(h,\lambda)$ 

9: in the range~$0<\alpha\le 1$.  We prove that for

10: this optimal choice of~$\alpha$, the error in the $L^2$-norm is bounded by~$Ch$ where $C$ does not

11: depend on~$\lambda$.  Numerical experiments confirm the expected convergence behaviour and show that, for 

12: practical meshes, our locking-free method is more accurate than the standard method if the material is 

13: nearly incompressible.  Our analysis also shows that the error in the $H^1$-norm is bounded by $Ch^{1/2}$,

14: but our numerical experiments suggest that this bound is not sharp.

15: \end{abstract}

16: