We consider exploiting space-time correlations of passive RFID tags for localization and tracking. Consider tags with storage memory distributed densely in space. Users move through the field, scanning the tags. Subsets of these tags are space-time correlated with respect to the user's trajectory. For example, if two adjacent tags are scanned close together in time, they are highly correlated.We can represent this correlation value as a weighted virtual arrow pointing from the earlier scanned tag to the later scanned tag. A user continuously calculates these arrows, and stores them locally in the tags, forming a digital path, for localization and tracking of herself by other users. We specify several of these correlation functions and evaluate them experimentally. We provide statistical measures of performance, as well as visualizations of the resulting arrow fields created by users.