In order to demonstrate that attention is distributed disjointly, we created a search task with a high penalty for attention to intervening areas. The area to which observers must attend is defined by a square-wave grating. A single target is located in one of the even strips (along with distractors), and ten false targets (identical to the real target) are located in the odd strips (also with distractors). To successfully report the location of the real target (versus the many false ones), the observer must both attend even strips and strongly ignore odd strips. With only two strips, one above, one below fixation (or one left, one right of fixation), it is easy to attend to one and ignore the other. As the number of strips increases, performance declines. Yet even with 12 strips, including six disjoint attended areas, analysis shows that all six strips are - to some degree - attended, and the intervening strips are - to a greater extent - unattended. A color-coded square wave grating (instruction grating) is used to indicate the areas to be attended and ignored. In one version of the task, the instruction grating is turned off before stimulus presentation. In another version, the instruction grating remains on, superimposed on the search stimulus. The data from both procedures indicate the attenuation of attentional modulation (between attended and unattended areas) with increases in spatial frequency. The results from these tasks are a critical ingredient for models of the distribution of spatial attention across the visual field as a combination of attentional modulation and visual acuity.