The Earth's penumbral shadow is larger than would be expected from simple geometry, a phenomenon first observed by
Philippe de La Hire in 1707. The precise amount of enlargement varies over time for reasons which are not fully understood, but likely involve the amount of dust in certain layers of the Earth's atmosphere. Various eclipse almanacs have used different assumptions about the magnitude of this effect, resulting in disagreement about the predicted duration of lunar eclipses or, in the case of penumbral eclipses of very short duration, whether the eclipse will occur at all. In 1989,
NASA published a lunar eclipse almanac that predicted a short penumbral lunar eclipse to occur on 18 August 2016. However, the French almanac
Connaissance des Temps used more conservative assumptions about the size of the Earth's shadow and did not predict an eclipse to occur at all. The
Bureau des Longitudes in France continued to refine their lunar eclipse models; NASA's 2009 edition of its lunar eclipse almanac was based on their values, which effectively reclassified nine eclipses between 1801 and 2300 as non-events, including the one in August 2016. Some resources, including the
HM Nautical Almanac Office's online canon of eclipses, continued to list the 18 August 2016 event. Despite not appearing in NASA's printed lists of eclipses since the 2009 revision,
AccuWeather reported the upcoming eclipse and projected this was the final member of
Lunar Saros 109. == Visibility ==