Skip Navigation Links
NOAA NOAA United States Department of Commerce

The Deadliest Atlantic Tropical Cyclones, 1492-1996


  1. Introduction
  2. Tropical Cyclone Terminology
  3. Casualty Information
  4. Storm Lists and Statistics
  5. Conclusions
  6. Acknowledgements and References
  7. Appendix 1: Cyclones with 25+ deaths
  8. Appendix 2: Cyclones that may have 25+ deaths
  9. Notes to the Appendices
  10. References to the Appendices

2. Tropical Cyclone Terminology

The United States National Weather Service technical definition of a tropical cyclone (National Weather Service Operations Manual C-41 1993) is: "A nonfrontal, warm-core, low pressure system of synoptic scale, developing over tropical or subtropical waters and having a definite organized circulation." In practice, that circulation refers to a closed, counterclockwise (in the northern hemisphere) airflow at the earth's surface.

Meteorologists generally recognize three classes of tropical cyclones stratified by their highest one-minute average surface wind speed. Tropical Depressions have maximum wind speed less than 39 mph (and, in practice, generally greater than 20-25 mph). Maximum wind speed from 39 to 73 mph characterizes Tropical Storms. Hurricanes have wind speeds of at least 74 mph. Of the defining criteria, the closed nature of the circulation in weak systems, the thermodynamic structure, and the precise intensity cannot always be determined objectively. For this compilation, the publication Tropical Cyclones of the North Atlantic Ocean (Neumann et al. 1993) and the associated NHC Best Track data set served as the final authorities for Atlantic tropical cyclone histories back to 1871.

These definitions are more quantitative than the terminologies of the past. Many early reports, especially from non-meteorological sources, referred to "hurricanes" without providing elaboration. Sometimes, hurricane meant any storm of apparently exceptional ferocity (such as a powerful high-latitude storm of non-tropical origin or a "severe" thunderstorm) that, perhaps, produced what we now consider hurricane force winds. Others used subjective terms like "a terrific gale" or winds "blowing a perfect hurricane" (e.g., Milner and Sowerby 1863). It is unclear in these instances whether the current requirements for a tropical cyclone were satisfied. Occasionally, however, an especially descriptive account added confidence to the interpretation, as in a summary printed in the 6 November 1761 issue of Lloyd's List 3 (note: in keeping with the style of the original text, "§" is used to approximate the archiac form of the letter "s"):

Capt. Young, arriv'd at Bri§tol from Guadalupe, came out the 17th of Sept. in Company with a Fleet of 26 Sail, mo§t of them for England, under Convoy of the Griffin Man of War, who was to §ee them as far as Lat. 28; but on the 27th ditto, in Lat. 22, they met with a heavy Gale of Wind, which began at the N. W. and veered all round the Compa§s to the S. E. in which the Fleet were §cattered, and §everal lo§t their Topma§ts. The next Morning he §aw only nine Ve§§els with the Man of War; and the Captain adds, That by the Smartne§s of the Gale, and the Wind's flying about round the Compa§s, he apprehends it was the Tail of an Hurricane.

Information about storm duration was helpful, too. The very long duration of the inclement weather described in the following passage is more consistent with a "cut-off" low than with a tropical cyclone:

Falmouth, 6th January. Arrived the Hyena, Captain Thompfon. Left St. Kitts on the 30th November, with about thirty fail of Veffels under her Convoy; but a Tempeft of Wind, on the 17th of December, in Lat. 32 feparated them; a Storm of an uncommon Sort, that lafted from that Period to this Day; the Damages of the Hyena are fo great, it was with difficulty fhe was brought into Port, and much is to be apprehended for the Fleet.
(Lloyd's List, 11 January 1782)

Accounts that included weather observations, such as ship reports based on the Beaufort scale (introduced in 1805) or barometric pressure measurements, helped to clarify the nature of some rough weather events. These data were most often found in meteorological studies, like Ludlum (1963) and Millas (1968), which provided many well-documented and corroborating descriptions.

This study adhered to several guidelines that minimized subjectivity and simplified the analysis. Every entry in the Appendices had a documented association with bad weather that was, or could reasonably be, related to a tropical cyclone. This requirement eliminated many cases from further consideration, even those where the remaining evidence (in the example below, the date and location of a loss of multiple ships) tempted us to attribute the disaster to a tropical cyclone:

The Duke of Cumberland, (Captain) Ball, a Letter of Marque of Briftol, laft from the Canaries for Virginia, was loft in September laft nine Leagues to the Southward of Cape Henry; the Captain, Surgeon and twenty three Men were drowned, and 21 faved. about the fame time were alfo loft a Snow and a Brig, Names unknown, and all the Crew of the former perifh'd.
(Lloyd's List, 11 November 1757)

Wherever helpful, the data and descriptions provided by the sources are reprinted verbatim. (Unfortunately, by doing so, we also pass along some information that either originally [or over the years] was [re]recorded incorrectly. Conflicting accounts were noted in, and by, several sources and the associated uncertainties are reflected in Appendix 1. We hope, however, that by providing all relevant reference information, the reader will gain as thorough a documentation of the event as possible.)

Footnotes are included to point out special conditions. For example, the footnote "c" indicates that the tropical nature of a storm was in doubt for at least part of the event. Often, it applies to storms moving poleward from about 40-45N, where weather systems generally encounter relatively cold ocean waters (°C) and tropical cyclones transform to extratropical cyclone status.

The track data of Neumann et al. (1993) show that Atlantic tropical cyclones are almost exclusively a warm-season event, as implied by the mariner's poem (Inwards 1898):

June too soon.
July stand by.
August look out you must.
September remember.
October all over.

The last line may be more ambiguous than helpful. In some Octobers, "all over" seems to describe the spatial distribution rather than a certain cessation of activity. The NHC officially defines the hurricane season to run from June through November. Tropical cyclones outside that period are relatively rare and mostly limited to low latitudes. In this study, when lacking evidence to the contrary, storms between December and May were eliminated from further consideration.

Only in obvious circumstances was a report purportedly about a tropical cyclone rejected outright. The following account refers to a "Hurricane", but the storm's date and location are inconsistent with our expectation of a tropical cyclone:

Plymouth (England), Jan. 5. Laft Night it blew a Hurricane; almost every Ship in the Harbour drove.
(Lloyd's List, 7 January 1791)

The concept of storm track and the difference between storm motion and circulation remained obscure until Benjamin Franklin's conclusions of the mid-18th century (see, Ludlum 1963, p. 22) were extended and formalized by Redfield (e.g., 1836), Reid (1841) and others. In addition, with communications generally limited for centuries to the line of sight, storms almost always moved faster than did the information about them. The first words about "The Great Hurricane" of 10-16 October 1780 did not appear in Lloyd's List (published twice a week at that time) until the 19 December issue, and new reports appeared through 13 April 1781.

These limitations certainly contributed to the peril of people in the path of an oncoming storm. One impact on this study was to introduce uncertainty in some instances about whether contemporary storm accounts from a region referred to a single tropical cyclone or possibly to multiple systems. (The Lloyd's List issues from December 1780 through April 1781 describe losses in the Caribbean Sea and adjacent islands. We now know that in addition to the Great Hurricane, two more of this hemisphere's most notorious storms occurred in that region during October 1780; see, Millas 1968). Another example occurred in 1785 when a storm devastated the area from St. Croix to Cuba during the last week of August (The Daily Universal Register). On the 2nd of September, a "savage" storm struck the Delaware coast (Seibold and Adams 1989). Two disturbances could be responsible for these events. Alternately, the tracks of more recent storms suggest that a single tropical cyclone could have been the culprit. Cases where uncertainties persist about the number of storms involved were entered into the catalog and assigned the footnote "z".

(Please see the NHC Glossary page for other terminology in this report.)

3 This account, like several that follow in the text and in Appendix 2, is shown in an older style of English, presented by the source, where "f" sometimes represents "s".

Next: Casualty Information