In recording a spatial distribution some information is inevitably lost. In designing research it is important to ensure that techniques of data collection do not obscure those portions of the variation in spatial distributions which are of interest. This paper explores the relationships between data collection procedures and information loss. Formulae are presented which allow requirements concerning quadrat size or accuracy of measurement in point data to be calculated when the relevant components of spatial pattern can be specified. (See also <b>AATA</b> <i>21-8</i>) -- AATA