{
    surfaceVectorField n = mesh.Sf()/mesh.magSf();

//     traction = (n&fvc::interpolate(sigma));

    // surfaceTensorField sGradU = 
    //     ((I - n*n)&fvc::interpolate(gradU));

    // traction =
    //     (2*mu + lambda)*snGradU
    //   - (mu + lambda)*(snGradU&(I - n*n))
    //   + mu*(sGradU&n)
    //   + lambda*tr(sGradU&(I - n*n))*n;

//     traction =
//       (2*mu + lambda)*fvc::snGrad(U)
//     - (mu + lambda)*(n&sGradU)
//     + mu*(sGradU&n)
//     + lambda*tr(sGradU)*n;

    // philipc
    // I am having trouble with back-calculation of interface tractions from solid interface
    // procedure (in multiMaterial.C), the tractions have quite large differences from each
    // side. Interpolating sigma is OK for now
    // traction = rheology.law().interfaceTraction(n, U, gradU, rheology.mu(), rheology.lambda());
#   include "calculateDEpsilonDSigma.H"
    traction = (n&fvc::interpolate(sigma+DSigma));

    // forAll(traction.boundaryField(), patchi)
    //   {
    // 	if (mesh.boundary()[patchi].type() == "cohesive")
    // 	  {
    // 	    forAll(traction.boundaryField()[patchi], facei)
    // 	      {
    // 		Pout << "face " << facei << " with traction magnitude "
    // 		     << mag(traction.boundaryField()[patchi][facei])/1e6 << " MPa and traction "
    // 		     << traction.boundaryField()[patchi][facei]/1e6 << " MPa" << endl;
    // 	      }
    // 	  }
    //   }
}