One of the several reasons this storm was allowed to go undetected during critical times in its progression toward the coastline arose out of an unfortunate limit in the knowledge at the time about hurricanes themselves. Perhaps more appropriately, this hurricane also defied many conventions that guided the successful tracking of many hurricanes previous. In some ways, The Great Hurricane of 1938 provided-- only in retrospect, the necessary clues to decoding its path, and thus the potential track of another deadly storm.

It may be helpful, at this point, to examine the general ingredients for creating and maintaining a hurricane, some of which can be found here...

In the 1930s, forecasters had more tools to track a storm than they did to predict them. Barometers, thermometers, anemometers, and the like are only really useful for forecasting if a significant number of them are able to measure in hourly sync with each other. Meteorologists did not have satellites or radars at the time to constantly assess the movement of a storm. For all of tropical forecasting in times previous, there was no ability to see the steering currents in the atmosphere, which end up being critical to understanding the movement of the storms. But they did have experiential trends, copious logs of storms come by, to establish patterns and apply them to future storms. This becomes meaningless, however, the moment a storm with new intentions arrives.

In the late 1930s, forecasters started using weather balloons, to help them understand the forces at work thousands of feet overhead (Schwartz et al, 2007). The Weather Bureau, in October, 1938, officially adopted the term "radiosonde" to encapsulate this use of radioed weather information from instruments tied to a lofting balloon (Samuels, 1938). The weather observation deck atop Mount Washington, more than 6,000 feet above sea-level, was a source for constant updates about conditions aloft, if only for one location. Aircraft pilots could also make their own weather observations, but at the time, this was not too reliable, because no pilot was going to deliberately fly into a storm, and no measurement was going to be made at a regular frequency, location, or altitude.

  High Pressure Makes the Call

Extrapolated from the little data that was available, the first upper-air weather maps of the event (starting with September 16th) showed a trough of low pressure toward the Great Lakes in the west, and a deep area of high pressure in the southern Altantic (Pierce, 1939). By the 17th, this area of high pressure was lifting northward, and much moisture was becoming available for ingest into any willing tropical system. Beginning early on the 18th, the hurricane enters the western Atlantic, roaming under the base of this large area of high pressure as it itself continues to retreat northward. It is important to note here that with each movement of this area of high pressure away from the coastline, there are fewer data points from which to observe it, leaving a forecaster of the day at a huge disadvantage. On the morning of the 19th, the storm was heading a little more north of west, and moving into an area where it was assumed, adhering to the relatively short knowledge span of hurricane tracking, either would impact Florida, or recurve out to sea. This curvature continued, and forecasters effortlessly deemed this storm a threat primarily to shipping, and, in some ways, attempted to wash their hands of it, having worked 24-hour shifts for several days, (Scotti, 2004). The upper-air weather map presented by Pierce depicting conditions early on the 20th certainly looks ominous, however to a forecaster on the ground, it would be very difficult to infer the strengthening and entrenching area of high pressure to the east of the storm, having moved so far away from its initial location so as to have thousands of square miles of weather determined by a single data point (Pierce, 1939). Indeed, at Mount Washington, NH early on the 20th, previously tropical air had been replaced by the colder polar air advancing from the west. Signs were looking quite clear that this trough along the Appalachians would continue to push out to sea, and carry the hurricane with it like many storms before. But this progression did not continue. The combination of the strong blocking high pressure to the east, and the attractive invitation of the winds aloft from the upper air low to the west, instead enticed the system at a more rapid speed and kept it close, dynamics that were surely unbeknownst to the meteorologists at the time. As the hurricane halted it's recurvature, it began to push back a little at the advancing polar air from the west. Mount Washington, NH went back into warmer air by the evening on the 20th, and the Blue Hill observatory in Massachusetts noted a south-to-north movement in the higher clouds (Pierce, 1939). By the morning of the 21st, the hurricane was a train off the tracks, moving too quickly to be affected by anything further, and creating it's own meteorology as it raced toward New England.

  The New England Hurricane and it's Fronts

Upon the retrospect of months pouring over a hurricane that didn't follow the rules, forecasters discovered a remarkable trait to the New England Hurricane that was to later fashion the understanding of similar storms: It's creation and maintenance of surface fronts. At first it began as a peculiar oddity, all sorts of math and atmospheric physics attempting to explain the movement of temperature and moisture around the advancing hurricane ended in impossible, one even declaring that in order to verify the observed pressure reading at 10,000 feet from a radiosonde over Washington, DC on September 20th, the average temperature through that entire column of air had to be about 136 degrees warmer (Pierce, 1939). What followed would be a fantastic exploration into a new frontier of tropical meteorology.

It turns out that the movement of moisture and temperature within a hurricane as it advances beyond about 30-degrees latitude can be best described by a system of fronts. In a normal mid-latitude cyclone (a term describing a typical non-tropical storm system), a warm front would be carrying moist air northward, overrunning denser cold air, and wringing out precipitation, while a cold front would be on the opposite side advancing colder, drier air, cutting into loitering warm air, and causing precipitation that way. With a hurricane in the north Atlantic, an extreme version of this dynamic becomes apparent, with the theoretical warm front wrapping all the way around the storm, primarily oriented along the north axis of its movement. This has the added effect of enhancing the overrunning qualities of a typical warm front, and it is often observed that the lion-share of rainfall associated with a storm of this type occurs to the west of the storm propagation. The cold front itself is all the way in the eastern-half of the storm, whipping up winds over the relatively frictionless open waters. With the 1938 hurricane, the heavy forward speed participates in a number of unique ways to the storm, in this context by effectively adding to the wind speed on the eastern side of the storm. If a ball is slowly rolled forward within a traveling car, its speed perceived within the car is negligible, but to the vantage-point of a stationary  observer, it's the fastest moving object around. Relating this to the 1938 hurricane, and its assessment at times of a nearly 70mph forward speed near landfall (Pierce, 1939), a forecaster could effectively add the internal wind speeds of the hurricane to the movement speed of the system itself on the eastern side of the storm-- and perhaps subtract on the western side. It is clear by all manner of damage and wind reports, that this dichotomy of wind field was indeed observed with the 1938 storm (Tannehill, 1938).

   Extra-tropical Transition

When comparing this storm to what was known about hurricanes of the day, it was expected that this storm should maintain an exquisite symmetry about it, whereby rainfall distribution and wind radii are concentric about the eye of the system. Any deviation from that utopian characterization becomes suspect. With the 1938 hurricane, the landmark discernment of fronts within the storm, each moving distinctly different air masses, combined with the asymmetry with its wind-field, would easily lead the forecaster to pronounce a storm as 'extra-tropical' (Pierce, 1939). However, most modern tools to ascertain this extra-tropical quality are rather subjective, among them an understanding of the sea-surface temperatures the storm has been traveling through, together with an opinion of the storms appearance on satellite imagery (Hart et al, 2001). Neither of those are very helpful with a storm over land occurring before Sputnik.

Nevertheless, the characterization of frontal formation, solidifying movement of distinctly separate maritime tropical air and continental polar air about the system is a good one from forecasters of the day, though perhaps its a more common thing to hurricanes that find their way above 30°N on their way to New England (Hart et al, 2001). The reality arrived at ends up being a rather murky purgatory of 'Extra-tropical transition'...whereby one could, with some credibility, simultaneously argue a storm's persistent hurricane-ness and extra-tropical finality with equal vigor. An 'extra-tropical' system need not always dissipate rapidly, on the contrary, the reduction in energy available by ingestation of closely available warm ocean-water (and from disruptive terrain-friction) is sometimes balanced-- or even exceeded by the newly arrived energy of air-mass distribution (and a favorable tail-wind of upper air cyclonic circulation) (Pierce, 1939). The 1938 storm made ample use of this once transitioning upon land, helping it to maintain some considerable force for some distance. There is a point, however, in this transitional section of a storm's lifetime-- and the 1938 hurricane is no different, where a finality of extra-tropical character becomes most evident, which would be the moment of occlusion. In an occlusion, the colder polar air will have advanced sufficiently around a cyclone so as to have effectively elevated all incoming warm tropical off the surface level. It could be posited that, at some level, other internal dynamics notwithstanding, that the surface fronts of the transitioning 1938 hurricane was in the process of occluding as it made its way up through western New England, and at the first point of occlusion, there would no longer be a residual tropical core of air recognizable at the surface, and the storm itself would then be feeding solely on energy common to all mid-latitude cyclones of a non-tropical character.

It could be said that this core maintained itself, while the rest of the system, working its way inward toward the central-dense-overcast region of the storm, gradually participated in the more extra-tropical advection of air-masses. Pierce recognized that the tremendous forward speed of this storm was an inhibiting factor toward quick occlusion (Pierce, 1939), though once the process begins at all, the dynamics of those nascent conditions (dropping sea-surface temperatures, appearance of an internal front system, asymmetric wind fields, and a marriage to polar-based steering currents) forcibly see it through to completion with New England storms, as was also the case with the 1938 hurricane.