Participants acquired spatial knowledge of a fictitious island by studying either (a) a complete physical map, (b) a sequence of part maps each showing the outline of the island and a subset of the landmarks, or (c) a sequence of sentences each describing a part map. During test, they verified the direction between 2 landmarks. Spatial knowledge in the fragment condition was as high as in the complete map condition, and both were better than the text group. Response times showed a distance effect. Additionally, spatial relations really presented were judged faster than inferred relations. The author replicated these results in a condition in which only relative spatial information could be used. These results suggest that simultaneity is not critical for obtaining a map advantage in spatial learning and that the mental representation of the map is structured.