The 1938 hurricane in the open sea was like many hurricanes of old, a hidden creature of the Atlantic, popping up on ship reports as a dire storm at sea, if the ship itself was able to survive the onslaught. Over the years, combing backwards over statistics and anecdotes, the puzzle assembles a picture that on the surface puts the storm in another of the pattern of "Cape Verde Hurricanes" established long before (Redfield, 1854). Though documentation exists to further explore the birth of the storm into the South Central Sahara (Hubert, 1939), ship reports indicate that the hurricane gales were first experienced by the Brazilian Ship S.S. Alegrete on the evening of September 16th (Tannehill, 1938), still well west of the Caribbean Islands, but also unusually north of them as well for a hurricane that would later strike the US mainland. The Netherlands S.S. Socrates next spotted the same storm on the 17th 500 miles northeast of the Leeward Islands, moving slightly north of west, but with increasing winds and a barometer reading of 29.29 inches (992mb).

--It's important to note here, that in the open water prior to airplane reconnaissance in 1944, ship reports were perhaps the sole indicator for the placement of storms, and due to lack of available forecasting (but fortunately for storm trackers), there were plenty of involuntary opportunities for such encounters. These ships would report by wireless telegraph whatever they experienced, with various degrees of accuracy (their own position and barometric pressure readings being of most help, wind reports a little harder to calibrate), and if they survived the event, further detail may be retrievable in the ship's log. It is clear that these reports are best used for identifying the track of a hurricane, rather than their exact wind speeds (Jarvinen et al, 1984). It is easily conceivable that storms that never tracked through shipping lanes were never discovered, and a particular storm would be 'lost at sea' until the next ship has their unfortunate run-in perhaps days later, with no information in between.

Continuing on its journey to the west-northwest, the storm next roughs up the British S.S. Corales, who reports a pressure reading of 27.90 inches (945mb) on the 18th (Tannehill, 1938). By then, these reports (and many others from ships in the western Atlantic) were getting noticed, and the US mainland, particularly Florida, was becoming concerned. On the heels of the Great Labor Day Hurricane of 1935, 25-radio stations simulcast the first state-wide storm warning with the news that the storm, "traveling 20 miles-per-hour, in a westerly direction...should reach the Miami-Dade area sometime Tuesday morning [the 20th]...every precaution should be taken" (Scotti, 2004). And that they did, buying stores out of all manner of supplies.

Meanwhile, the hurricane continued to strengthen attaining a Category-5 power (the Saffir-Simpson hurricane wind scale was not published until 1973) at about 2am Monday the 19th. Wind estimated from ship reports indicate a 160 mph wind during the day (Sheets et al, 2001). Forecasters on shore listen with nervous anticipation for the next radioed report, almost like the call of the next pitch in a pivotal playoff game. On the afternoon of the 19th, forecasters started noticing a trend in the reports they were getting; pressures weren't falling as quickly on the islands in-between Florida and the path of the storm, winds were beginning to shift to directions that would occur only if the hurricane was veering toward the north, Cuba and the Bahamas by all practical means were being missed. A few more confirming ship reports later, and the Jacksonville weather office radioed out 'all-clear' for South Florida at 10pm on the 19th (Scotti, 2004).

It appeared at this point to most experienced forecasters that this storm would now take on the usual path by most 'Cape Verde' hurricanes: The Recurve. Though not much had been definitively known about hurricanes at the time (and what had been gleaned wasn't obtained all that long before), at least one hurricane tracking pattern had been put to paper-- and that was an oft observed bend in a hurricane's path beside the Bahamas (Redfield, 1854). If it starts recurving out to sea even an ounce, it will eventually finish the job. Even the strongest, most legendary tempest would then have an innocuous end of little consequence.

Tuesday the 20th saw the storm continue it's bend, now moving to the north-northeast. A slow weakening has taken place as well, with winds down to a suddenly un-alarming 138mph. Forecasters had in mind to simply throw out a few stern warnings to ships to remain in harbor, and follow-up with more of the same along the east coast, until the shipping lanes were clear (Tannehill, 1938). There's an unfortunate problem with all vessels properly heeding these warnings in 1938: No ships, no reports. For almost a full day it was silence in the open waters, a silence that should have been deafening were it not for the exhausted forecasters breathing sighs of relief on the presumed brilliance of their predictions. Certainly, they were later to discover that this hurricane wasn't cooperating.

One ship at sea, the Carinthia, a cruise liner that granted weekly trips for 1,650 passengers to the Bahamas from New York in the late 1930s, later torpedoed by the Nazis in 1940 (Braynard et al, 1991), did encounter this hurricane head-on and unexpectedly overnight on the 20th. When they managed to survive the events, they radioed in their experience the next morning, and a report of a reinvigorated hurricane with a pressure reading dropping to 27.85 (944mb) in less than an hour (Scotti, 2004). Despite this ominous and dire report, forecasters still expected this storm to continue its curvature out to sea.

And the storm certainly would have, were it not for an expanding area of high pressure blocking its way further eastward, and for an inviting jet stream collecting loitering showers along the eastern seaboard. Being squeezed between these elements, the end result would be a veritable sling-shot effect toward Long Island. During the morning hours of the 21st, the hurricane was spotted, barely 75 miles east of Cape Hatteras, bringing 50 mph winds and a pressure reading locally of 29.30 inches (992mb). At this time, forecasters put out warnings extending from there through to Eastport Maine (Tannehill, 1938), but these warnings used language like "gales", "whole gales", and "a tropical storm"...nothing that would match even the same call to action as what aired in Florida a few days prior. In fact, the word 'hurricane' never came up, just generic terms that the local fisherman heard many times before (Burns, 2005).

The storm accelerated as it moved straight north, at 40, then 50, then 60 miles per hour. Various land-based barometers up the coastline reported a rapidly changing picture-- 28.99 (982mb) inches at Atlantic City, NJ, then 28.71 inches (972mb) at Sandy Hook (Tannehill, 1938). None of the data newly concerned forecasters received could come in quick enough, nor come from a location close enough, to tell what's really going on. Being used to making forecasts and updates every 6 hours or so, it was clear this event was out of control, with the final storm warning issued at 2:00pm, still not using the word 'hurricane' (Burns, 2005), advising that a tropical storm was "likely to pass over Long Island...in the late afternoon" (Tannehill, 1938). The core of the tempest was already lashing the beaches that very hour, with hurricane force winds now being experienced as far away as Block Island, RI.

The center of the storm moved onshore in Long Island, as reported by a number of sources, before 3pm on the 21st (Pierce, 1939). By 4pm the eye of the storm, now a little filled in, was stomping through southern Connecticut, still moving at a furious clip. Though the storm has spent more than an hour over land, the pressure reading in Hartford still shows a remarkable 28.04 inches (950mb) at 4:30pm (Tannehill, 1938). By 5pm, the storm was already moving through western Massachusetts, perhaps faster than anyone can get there by car at that hour. Already in southern Vermont by 6pm, the system finally begins to slow down, and starts to lose its tropical characteristics. It approaches Burlington around 8pm, still a vigorous storm registering a pressure reading of 28.68 inches (971mb), and there are reports of sea salt coating the window panes in Montpelier (Scotti, 2004). The storm finally fizzled out over Canada on the 22nd, yet still keeping its residual gales and rain to the last.

The Great New England Hurricane is still unrivaled in its scope, its ferocity, and its speed. Striking all of New England in mere hours, for all intents catching all unaware...and setting still-standing records of deep pressure and wind gusts. The storm put down gusts of 150mph on the south shorelines (Weil, 2004), astonishing wind of 173, and later 186 miles-per-hour on top of the Blue Hill observatory in Massachusetts. The peak of Mount Washington in New Hampshire hit 136mph (Tannehill, 1938), with many gusts approaching 100mph in the lowlands (Schwartz et al, 2007). The surprise by which everyone was caught regarding the re-emergence of the storm and it's swift, dead aim on the New England coastline was immediately transformed into a research effort, which aided dramatic steps toward both understanding and predicting future hurricane movements. Which is a good thing, because if the same storm comes by again, the explosion of population and property values in the 70 years since would bring losses to near incalculable levels.