The first recordings of the Novaya Zemlya (NZ) effect were made during Willem Barents' third Arctic expedition. Ray-tracing analyses of the three key observations, on 24-27 January 1597, show that all the reported details can be explained by adopting one common and realistic type of temperature inversion. In particular, the Moon-Jupiter conjunction could have been visible over the central mountain ridge of the island. We show that the NZ effect distorts the relative positions of Jupiter and the Moon in such a way that the looked-for fingerprint of the conjunction occurred almost 2 h after the true conjunction. The quoted direction for the apparent Moon-Jupiter conjunction is then found to be accurate to within 1 degree. This delay of the apparent conjunction largely explains the error of 29 degrees in their longitude determination. The truthfulness of these observations, debated for four centuries, now appears to be beyond doubt.