The effects described hereafter were accidentally discovered in the course of analyzing results obtained from numerical simulations of fast fracture and crack arrest in DCB specimens. The simulations were performed as part of an investigation concerned with the evaluation of the dynamic material toughness KID. To this end the finite difference code presented in [i] was employed while measurements of the crack extension time history, as given in [2], served as input data for prescribing the crack growth in the numerical scheme. Numerical calculations of the energy balance during crack propagation were expected to yield the sought-for information on KID. One of the possible ways of checking the validity of the numerical solution is by comparing between the calculated and the measured dynamic stress intensity factors, KI~yn. So far, best fitting Kidyn history curves, drawn through numerically obtained points, were found to be in good agreement with the corresponding experimental curves. Yet, compared with experimental observations, the fluctuations of the numerically calculated KI ~yn values about the just-mentioned curves are too high for the simulation used to be regarded as satisfactory. The disturbing fluctuations may be attributed to the intermittent one-mesh-size crack extension inherent in the present code. It is anticipated that these fluctuations will decrease considerably by using a modified code which is now being developed and which would allow for a smoother numerical crack growth. Referring to results obtained from the as yet unmodified scheme, there was detected a sudden and remarkable decline in the Kidyn values in each of the run-arrest simulations investigated. It occurred only once in each case, at a time which appeared to be strongly dependent on the crack growth-time history of thq specific case treated. The intensity of the sudden change in the KI ayn level, being far beyond the current Kidyn fluctuations, suggests that the phenomenon described here is a diffraction of stress waves which takes place when the steep front of a specific pulse, that has been developed in and propagates through the specimen, reaches the propagating crack tip. Considering the special DCB test conditions (i.e., quiescent start, high initial stress intensity factors, fixed grips, compact geometry) one is obliged to adopt the following scenario as a reasonable explanation for the above-described events: Part of the stress waves generated at the crack tip when fracture initiates, are reflected back from the specimen boundaries finally to reach the relatively slow running crack end while they are still sufficiently strong to effect the Kidyn level drastically. In view of the fact that the waves' source is located at a bounding surface - the initial crack surface - it could have been anticipated that of the two kinds of waves generated, Rayleigh surface waves and P or S body waves,