Page Content (OCR)
APPENDIX 243 account for the 923 landing reports in our 1969 compilation, the theory can no longer be supported today. Neither is the figure of 5000 a good estimate. Many indications converge to show that only one case in ten may actually get reported. Therefore the number of close encounters we need to explain is proba- bly of the order of 50,000. This does not take into account the fact that the overwhelming majority of our sources are located in Europe, the American continent, and Australia. It is logical to assume that the phenomenon is worldwide, and that we are missing the true magnitude of the problem at least by a factor of two. This leads to a figure of 100,000 events. If we remain faithful to a strict interpretation of the ETH, even this very large figure still underestimates the real number of actual landings. Shouldn't we assume that extraterrestrial explorers would land on our planet without regard for the presence of human witnesses? In fact Poher and I found (using independent data bases) that the geographic distribution of close encounters does indicate a pattern of avoidance of population centers, with a higher relative incidence of landings in deserts and in areas without dwellings (5). If we follow this line of reasoning, then it would be conservative to multiply our number by a factor of ten to account for the high ratio of sparsely populated over densely populated lands. This would place our estimate at one million landings to be explained. In other words, if human witnesses were equally distributed over the surface of the land, and if they reported every close encounter they observed, the data universe should contain one million records. This number still does not take into account another important pattern in the phenomenon, namely its nocturnal character. First pub- lished in 1963, this pattern shows no significant variation between older and more recent cases, and even yields the same distribution when a very homogenous sample of previously unreported cases from a single region is analyzed. Figure 1 shows the frequency of close encounters as a function of time of day for three different, nonoverlapping samples, namely (A) an