Defining and controlling the support of a field measurement (the volume or area over which the measurement is averaged) are critical components of statistical and/or geostatistical analysis of a data base. For the present study, the seemingly simple task of collecting a water sample from a single fracture is used to illustrate potential difficulties in defining and controlling the support of that sample. Two definitions of sample support are considered for this problem. First, one could argue that the support is equal to the volume of water collected. Second, one could argue for a support defined in terms of the area of the fracture surface contacted by the water collected. Collection of water samples with constant volume is relatively straight forward in a field situation. Collection of a water sample which involves contact with a specified area of the fracture, however, is not as straight forward. In the case of negligible regional flow, a constant sample area can be approximated by requiring that the sample volume collected be linearly proportional to the locally averaged aperture (thus requiring an independent measure of the local aperture). In the case of regional flow with continuous sampling, a constant area within each sample can be approximated by matching the up gradient width of the capture zone of the sampler (thus requiring knowledge of the local aperture and hydraulic gradient). These examples illustrate how difficult it may be to define and control the support for even these simple measurements. Further, they illustrate that the choice of support may be closely related to the purpose for which the measurement is collected. Additional attention to the problem of defining/controlling sample supports has the potential to pay substantial dividends in the areas of scaling and field design/analysis. ¿ American Geophysical Union 1995