Only one problem remains. Just exactly what produced the images? Of the many possible explanations, we first considered the simplest, an astronomical source. The UFOs appeared low in the western sky at 8:12 P.M. E.S.T., about forty-five minutes after sunset. The night was dark, for the moon had just entered its last quarter and did not rise until much later. Apparently the only planet that could have been involved was Mercury. Setting a little more than an hour after the sun, it should have been visible above the western horizon at the time of the sighting, but since it was not particularly brilliant, having a magnitude of a little more than +0.6, we put aside the astronomical theory, for the moment, as improbable. We next explored the possibility of multiple reflections in the glass windows of the cockpit, produced by a light source inside the plane (such as a cigarette), or in the air outside (such as the bright-red exhaust trail of one of the jets in the area). Like the astronomical theory, this idea was set aside as improbable. Learning to distinguish between a reflection and a real light seen through a cockpit window is part of every pilot’s training. When he sees a strange light, he automatically makes the proper checks. Furthermore, Nash and Fortenberry had observed the disks through three separate windows having different orientations. Accepting the overwhelming probability that the source of the UFOs was outside and below the aircraft, we concluded that it was almost certainly on the ground. The densely populated coastal region near Newport News and Norfolk, with several airfields and military installations, included countless possible sources such as a searchlight, an illuminated advertising sign, an air beacon. Stratified clouds or inversion layers of temperature and/or humidity could have multiplied such a light into a series of glowing disks (see Figure 19). [Illustration: _Figure 19._ Searchlight shining on clouds. A, Through slightly foggy or dusty atmosphere, light cone plainly visible; B, through multiple thin cloud layers and foggy or dusty atmosphere; C, on cloud layer through clear atmosphere, no light cone visible; D, on multiple thin cloud layers, no light cone visible.] The soundness of this theory depended on the prevailing weather conditions. According to the reports, on the night of July 14 roughly a third of the sky at 20,000 feet was covered with thin cirrus clouds, practically invisible; at lower altitudes the night was cloudless and sharply clear, there was no apparent haze, visibility was unlimited, and no temperature inversion existed. Under such conditions the suggested mechanism would obviously not operate.