This paper proposes a method for optimizing rectilinear tasks on a 3 degrees-of-freedom (DoF) modular serial metamorphic manipulator. The overall experimental process was designed in order for theoretical assumptions and previous experimental results, regarding the characteristics of a class of reconfigurable manipulators to be verified. The optimization procedure undergoes two stages. In each stage, the tasks are initially simulated and the optimal solutions obtained are afterward evaluated in the manipulator. Optimal task placement in the configuration space of the reference anatomy is concerned in the first optimization stage. Two different kinematic manipulability measures are utilized to form the objective function of the genetic algorithm (GA) used. Determination of the optimal anatomy for each task execution is concerned in the second stage. All feasible anatomies are exhaustively evaluated, and the anatomy with minimum execution time achieved in simulation is considered as optimal. The simulated tasks are executed for the reference and the optimal anatomy extracted. Overall task execution time reduction is measured. For tasks executed, Tool Center Point (TCP) position and velocity are obtained from navigation equations using measurements from an inertial measurement unit (IMU) sensor. In order to obtain more accurate solutions from position and velocity equations, a Kalman filter (KF) algorithm is implemented. Finally, conclusions are made based on the results of each task execution. Overall the metamorphic manipulator can achieve higher kinematic performance and minimize task execution time for the optimal anatomy calculated. Optimal task placement for the reference anatomy also reduces the task execution time.