The mode in which the copies were taken was by running over and through their outlines a blue aniline pencil, and then pressing a wetted sheet of ordinary printing paper upon them, so that the impression was actually taken by the process of printing. During the two field seasons mentioned, with the aid of Mr. Creed, three hundred and fifty different engravings and groups of engravings were thus printed. Some of these prints were of large dimensions, and included from ten to fifty separate characters and designs. On the parts exposed in 1887 there were dates from 1800 to the current year, the number for the last year being much the greatest, which was explained by the fact that the wonderfully beautiful lake had been selected for a Sunday-school excursion. Over the greater part of the surface visible in 1887 there were few levels specially favorable for marking, and when these were found the double or treble use was in some instances noticed. After the writer had inspected the rocks and discovered their characteristics, and learned how to distinguish and copy their markings, it seemed that, with the exception of a few designs recently dug or chipped out by lumbermen or visitors, almost always initials, the only interesting or ancient portions were scratchings which could be made on the soft slate by any sharp instrument. The faces of the rocks were immense soft and polished drawing-slates, presenting to any person who had ever drawn or written before an irresistible temptation to draw or write. The writer, happening to have with him an Indian stone arrow which had been picked up in the neighborhood, used its point upon the surface, and it would make as good scratches as any found upon the rocks except the very latest, which were obviously cut by the whites with metal knives. As is above suggested, the peculiar multiplication of the characters upon the most attractive of the slates affords evidence as to their relative antiquity superior to that generally found in petroglyphs. The existence of two or three different sets of markings, all visible and of different degrees of obliteration or distinctness, is in itself important; but, in addition to that, it is frequently the case that the second and third in the order of time have associated with them dates, from which the relative antiquity of the faintest, the dateless, can be to some extent estimated. Dates of the third and most recent class are attached to English names and are associated with the forms of English letters; those of the second class accompany French names, and in some cases have French designs. Figs. 1 and 2, about one-fourth original size, are presented to give an idea of these peculiar palimpsests. [Illustration: FIG. 1.--Palimpsest on Fairy rocks, Nova Scotia.]