The geography of escape from the Upper Keys could be described in a single sentence: one track, running northeast, across open water, for forty miles.
The track ran the length of the Keys. There was no other way out. Every elaboration of it — the bridges spanning the channels, the causeways crossing the shallows, the stations where the train stopped, the passing sidings where northbound and southbound traffic could negotiate the single-track limitation — was a detail within the fundamental constraint. There was one way off the islands. It ran in one direction. It crossed water that a storm surge would render impassable long before the storm’s center arrived.
The Florida East Coast Railway’s Keys extension had been designed for commerce, not for evacuation. The track layout reflected the operational requirements of a railroad carrying freight and passengers between Miami and Key West on a regular schedule: single track with passing sidings at intervals sufficient to allow opposing trains to clear each other, stations at the principal Keys communities, maintenance-of-way facilities at strategic points along the route. The grade ran at approximately eight feet above mean high water on the embankment fills and at varying heights on the bridges and viaducts — some at fifteen or twenty feet above the waterline, others lower, the heights reflecting the navigation requirements of the channels they crossed rather than any assessment of the water levels a storm might produce.
The northbound journey from the Matecumbe Keys to the mainland required the train to cross approximately fifteen miles of open-water bridges and causeways and another twenty-five miles of island fills and short bridge crossings. Under normal conditions — clear weather, dry track, no obstructions — the journey took approximately two hours at the speeds the locomotive and the track geometry permitted. Under deteriorating conditions — rain reducing visibility, wind increasing lateral load on the cars, water rising on the causeway approaches, debris accumulating on the track from the mangroves and construction materials that the wind was beginning to strip from the islands — the journey time extended by increments that could not be predicted in advance because they depended on conditions that were changing as the train moved through them.
* * *
The track’s vulnerability was not in any single section. It was in the sequence.
A bridge that was passable at the beginning of the journey might be impassable by the time the train returned on the northbound leg. A causeway that was dry on the southbound approach might be under a foot of water two hours later when the train attempted the same crossing in reverse. The storm’s conditions were not static. They deteriorated continuously as the system approached, and the rate of deterioration accelerated as the center drew closer. A train that departed the Upper Keys with two hours of margin might arrive at a bridge crossing thirty minutes into the return journey and discover that the margin had evaporated — that the wind load on the cars exceeded the point of stability, that the water on the approach had risen above the rail, that the bridge structure itself was being tested by wave action from below while the wind pushed the train laterally from above.
The single-track limitation compounded the vulnerability. On a double-track railroad, a train could proceed in one direction while another train moved in the opposite direction on the adjacent track. On the Keys extension, the northbound return had to use the same track the southbound approach had used. If conditions deteriorated during the loading time at the Matecumbe station — if the water rose on a crossing that the southbound train had cleared an hour earlier — the train was trapped. It could not proceed north through the obstruction and it could not retreat south because there was nothing to retreat to. The islands behind it were lower and more exposed than the islands ahead of it.
The Archivist had seen bottlenecks before. He had not seen one so narrow, so low, and so dependent on a single rail line.
The causeways at Miami Beach in the 1926 storm had demonstrated the principle in a simpler form: two bridges connecting a barrier island to the mainland, rendered impassable by rising water before the storm’s center arrived, confining the population on the beach for the duration of the storm’s passage. The Keys configuration was the same principle at a larger scale and with a more consequential limitation. At Miami Beach, the confined population could shelter in the concrete hotels that Fisher had built. On the Matecumbe Keys, the confined population’s shelter options were wooden barracks at five feet above sea level.
* * *
The Overseas Highway, under construction alongside the railroad, was not yet a viable alternative.
The highway project — the same project the veterans’ camps had been established to support — had been building south from the mainland since the early 1930s, using the railroad’s right-of-way and in some sections its bridge structures to carry a two-lane road across the same island chain the railroad crossed. By September 1935 the highway was passable in some sections and incomplete in others. Gaps existed where bridge construction had not been finished, where the road surface had not been laid, where the connection between one island and the next depended on the railroad bridge because the highway bridge beside it was still under construction.
A vehicle attempting to drive from the Matecumbe Keys to the mainland on the afternoon of September 2 would have found the road passable for some distance and then impassable — a bridge deck that ended in air, a section of roadbed that had not been graded, a gap that could be crossed only by the railroad. The highway was not an escape route. It was a future escape route, still months from completion, that would eventually replace the railroad’s monopoly on the Keys’ connection to the world beyond the reef. In September 1935, the highway’s unfinished state meant that the railroad remained the single viable means of moving a large number of people quickly from the Upper Keys to the mainland.
One track. One train. One chance.
* * *
The logistics of the evacuation train were governed by the geography and by the railroad’s operational requirements.
The train would be assembled in Miami or Homestead. A locomotive and passenger cars — the railroad’s available rolling stock, augmented if necessary by freight cars fitted with temporary seating — would be gathered at the yards and dispatched south on Ghent’s authorization. The locomotive would lead the southbound train through Homestead and onto the Keys extension, running cab-forward with the engineer looking down the track ahead through whatever visibility the weather permitted.
At Homestead, the train would stop and the locomotive would be repositioned. The single-track extension required the locomotive to be on the north end of the train for the return journey — pushing the cars south on the outbound leg and pulling them north on the return. This repositioning required the locomotive to uncouple, move to a wye or a turntable, reverse its orientation, and recouple to the north end of the consist. The process consumed time — twenty minutes, perhaps thirty, depending on the facility and the crew’s efficiency. It was a routine operation for the railroad. It was not an operation designed for urgency.
From Homestead, the train would proceed south across Jewfish Creek to Key Largo and then down the island chain to the Upper Matecumbe station at Islamorada. The distance was approximately eighty miles. The travel time, under deteriorating conditions, was uncertain.
At Islamorada, the train would stop and the veterans would board. How quickly seven hundred men could be loaded onto a train in hurricane conditions — in wind and rain, from camps scattered across two islands, with some men a mile or more from the station — was a variable the evacuation plan did not address with precision. The plan assumed the men would be gathered at the station when the train arrived. This assumption depended on the camps receiving advance notice of the train’s departure — notice that would travel the same administrative chain that had authorized the train in the first place.
The return journey would begin when the loading was complete. The train would proceed north, locomotive first, pulling the loaded cars across the same bridges and causeways it had crossed on the southbound approach, under conditions that had deteriorated further during the loading interval.
Every step consumed time. The storm consumed it faster.