There are many innate errors involved in the process of deconvolution. Of interest here is an analysis of one of these. In particular, the errors resulting from the pseudo randomness of the seismic records are calculated. These errors are present even assuming that the many assumptions involved in the deconvolution process are correct. Such assumptions are in themselves dangerous. The errors calculated thus represent the most optimistic conditions.A technique to calculate the useful amount of deconvolution is derived using the error information. The technique is applied, example results given, and the practical difficulties discussed. The technique is found to be of primary interest as a learning device and research tool. Its complexities at present limit its use in routine processing. It is found, however, to yield information which cannot otherwise be objectively obtained.