Proceeding from the equality constrained nonlinear programming problem, the authors consider merit or line search functions for determining a steplength in a sequential quadratic programming algorithm so that desirable convergence properties are retained. They belong to the class of differentiable exact penalty functions and possess the descent property in the neighbourhood of a solution. If the iteration sequence converges Q-superlinearly, then the steplength one will be accepted when approaching an optimal solution. The results of some numerical experiments are included showing the sensitivity of the resulting algorithm with respect to a merit function parameter and that a steplength of one was chosen close to a solution.