Measured replay pulses, recorded on metal evaporated tape, show a strong asymmetry and dependence on the direction of tape movement. It is found that pulse height and pulse width are unequal for both directions of tape movement. This is attributed to the nonlongitudinal easy-axis direction which causes differences in head-tape geometry for both directions, resulting in a different transition width. A simulation model is proposed. The differences in pulse asymmetry and pulse width between both directions of tape movement are very well reproduced by this model. From X-ray diffraction measurements and the results of the simulations, it is concluded that not the columns but the crystallites act as magnetic entities for this kind of tape.