Platform zero did not appear on the station map, but commuters began finding it whenever the last northbound train was delayed. The station has platforms one through six, numbered consecutively, and a service corridor at the western end that is not open to passengers.
Twenty-three people have reported reaching a seventh platform through that corridor. All twenty-three reports describe the same conditions: a delayed final northbound service, after eleven at night, and a door that is normally locked standing open.
The station operator has confirmed that the door is fitted with a magnetic lock that releases on power failure. It has also confirmed that the western corridor terminates at a disused platform, constructed in 1903 and taken out of use in 1962, which is not numbered on current signage.
The platform is real
The existence of the platform is therefore not in question. What is in question is everything else.
A woman in a gray coat offered each visitor one precise warning. The descriptions of her are consistent on the coat, the approximate age, and the manner, which multiple witnesses characterized as businesslike rather than portentous.
She offered each visitor one precise warning.
By morning, every warning had become local news. This is the claim that has attracted attention, and it is the claim that survives scrutiny least well. Of the twenty-three reported warnings, eleven were recorded in writing before the events they describe.
Of those eleven, four correspond to subsequent events with a specificity that is difficult to attribute to chance. Three are sufficiently general that they could be matched to any number of occurrences. The remaining four did not correspond to anything.
The warnings that survived scrutiny
The four specific cases have been examined individually by a statistician at the regional university, who has published a short paper on them. Her conclusion is that the cases are interesting but that the sample is compromised by the method of collection.
The warnings were reported to a community forum thread, and she notes that the thread's existence created an incentive to report matches and no corresponding incentive to report failures. She estimates the true number of warnings issued may be several times the number recorded.
One of the four specific cases involves a warning about a burst water main on a named street, recorded in a forum post at 11:48pm and reported by the water authority as occurring at 6:10am the following morning. The statistician's paper addresses this case at length and does not resolve it.
The witness in that case, a hospital administrator named Iris Kelleher, has declined all interview requests except one, in which she said that she had not been frightened at the time and had become so afterwards, and that she would prefer the platform to remain closed.
The 1903 platform remains in place behind the new lock. The operator has no plans to bring it back into use, citing the cost of bringing a structure of that age up to current accessibility standards, and has described the recent interest as unhelpful.
