The occurrence of nine earthquakes with Mw greater than 8.5 between 2004-2012 raised interest in the long-term global rate of great earthquakes.   Studies exploring the significance of apparent clustering (e.g., Michael, 2011) focused on analysis of earthquakes since 1900.  Prior to this time the catalog is less complete and magnitude estimates are more uncertain.  Yet substantial information is available for earthquakes prior to 1900.  Reviewing the catalog of historical earthquakes back to 1700, including numbers of events and moment release rates, I suggest that approximately half of all Mw≥8.5 earthquakes are likely missing or underestimated in the 19th century.  The 8 February 1843 Lesser Antilles earthquake is an example of a known historical earthquake that could have been significantly larger than initially estimated.  I reconsider the felt effects of this earthquake, including a first thorough assessment of felt reports from the United States.  One recent study estimated a preferred magnitude of 8.5 based on near-field macroseismic effects (Feuillet et al., 2011), but the generally accepted value has been 7.5-8.  A consideration of the regional and far-field macroseismic effects reveals a felt distribution comparable to those of recent great (Mw ≥ 9.0) earthquakes.  A modest tsunami is described by two witnesses; two other accounts describe uplift of a stone wharf in Antigua, and along the northern coast of Guadaloupe.  For the 1843 earthquake, the far-field intensity distribution provides a stronger constraint on magnitude.  The observations support the inference of a high (Mw≥8.5) magnitude, and significant moment release towards or possibly around the northern corner of the Lesser Antilles Arc.   The results suggest that incorporation of best-available catalogs of historical earthquakes will likely lead to a significant underestimation of seismic hazard and/or the maximum possible magnitude in many regions, including parts of the Caribbean.