The Gregorian calendar is used for all dates from 1582 Oct 15 onwards. Before that date, the Julian calendar is used. For more information on this topic, see Calendar Dates. The Julian calendar does not include the year 0. Thus the year 1 BCE is followed by the year 1 CE (See: BCE/CE Dating Conventions ). This is awkward for arithmetic calculations. Years in this catalog are numbered astronomically and include the year 0. Historians should note there is a difference of one year between astronomical dates and BCE dates. Thus, the astronomical year 0 corresponds to 1 BCE, and astronomical year -1 corresponds to 2 BCE, etc..


The coordinates of the Sun used in these predictions are based on the VSOP87 theory [Bretagnon and Francou, 1988]. The Moon's coordinates are based on the ELP-2000/82 theory [Chapront-Touze and Chapront, 1983]. For more information, see: Solar and Lunar Ephemerides. The revised value used for the Moon's secular acceleration is n-dot = -25.858 arc-sec/cy*cy, as deduced from the Apollo lunar laser ranging experiment (Chapront, Chapront-Touze, and Francou, 2002).

The largest uncertainty in the eclipse predictions is caused by fluctuations in Earth's rotation due primarily to tidal friction of the Moon. The resultant drift in apparent clock time is expressed as ΔT and is determined as follows:

  1. pre-1950's: ΔT calculated from empirical fits to historical records derived by Morrison and Stephenson (2004)
  2. 1955-present: ΔT obtained from published observations
  3. future: ΔT is extrapolated from current values weighted by the long term trend from tidal effects

A series of polynomial expressions have been derived to simplify the evaluation of ΔT for any time from -1999 to +3000. The uncertainty in ΔT over this period can be estimated from scatter in the measurements.