secondly, the fact that its previous invisibility was not due to any sort of obscuration. The persistent clear-cut character of the neighboring coast-line during the whole transformation showed that nothing of the nature of mist or cloud had at any time hidden the peninsula from view. A something was actually there in August which had not been there in June.
As yet nothing could be seen of Atlantis. It was not until the 30th of October that I caught sight of it. About the same time, the straits between the islands, Xanthus, Scamander, Psychrus, and Simois, came out saliently dark, a darkness due to contrast. The line of south temperate islands, with their separate identity, was then for the first time apparent.
Meanwhile the history of Hesperia continued to be instructive. From having been absent in June and conspicuous in August, it returned in October to a mid-position of visibility. Vacillating as these fluctuations in appearance may seem at first sight, they were really quite consistent; for they were probably due to progressive change in the one direction, a change that was manifested first in Hesperia itself, and then in the regions round about it. From June to August, Hesperia changed from a previous blue-green, indistinguishable from its surroundings, to yellow, the parts adjacent remaining much as before. As a consequence, the pe-
