Unusual and potentially paranormal experiences were reported in a residential building in New York City. Occupants believed that a "ghost" was inhabiting the premises. Several methods were used in an attempt to learn about the reported phenomena. The primary technique, the basis for which originated with Schmeidler in 1966, involved having sensitives and skeptics tour the building and mark on floor plans the locations that they sensed or inferred might be related to the "ghost." The participants also responded to a check­list, which contained both true and false descriptions of what witnesses had reported. Statistical comparisons were made between the locations reported by building occupants and the floor-plan re­sponses of participants. Similar comparisons were made for the de­scription-checklist measures.

Several secondary methods were used to further explore the phenomena. A random number generator (RNG) test device was brought to the building to determine whether "ghostly" sites (per witnesses' reports) would yield different RNG output than sites where no "ghostly" activity had been reported. In addition, explor­atory tests were conducted with infrared photography, Polaroid photography, portrait art, motion-picture film, video, and psychometry. Psychological tests were given to witnesses.

Four of eight records collected from sensitives showed a signifi­cant or suggestive correspondence to what witnesses had reported (p = .013, p = .028, p = .052, p = .098; uncorrected for multiple analysis) while equivalent records from skeptics showed only one marginally suggestive correspondence (p = .098). However, pooled scores of sensitives were not significantly higher than pooled scores of skeptics for either floor-plan or checklist measures. Pooled scores for both sensitives and skeptics were, in fact, significant on the floor­plan test (p = .017 and p = .040, respectively). RNG tests conducted blind at locations designated as sites of ghostly activity gave a suggestively higher variance than tests conducted at control sites [F(12195) = 1.668, p = .086]. Other exploratory measures failed to provide meaningful evidence for or against a "ghost" hypothesis.