In a short series of papers in this Magazine on How to work in the Archœan Rocks, I briefly set forth the principles of correlation which seemed to me available, and pointed out the precautions which should attend the application of each test. I discussed the evidence from (1) Organic remains; (2) Order of superposition; (3) Included fragments; and (4) Mineral composition. The lastnamed criterion has the widest application, and therefore it was necessary to define its scope with great care. I showed that its value largely depended upon accessory considerations. I insisted that rock-groups should be compared as a whole, that the degree of metamorphism was an important factor in the evidence, that a valuable auxiliary test was to be found in similarity of succession, that the proof of a Pre-Cambrian age enormously reduced the chances of error, and that, in some cases, the origin of the deposits was a useful aid in correlation. I recall attention to these points, because, at a recent debate in the Geological Society, some of my critics seemed to believe that I had been attempting to determine theage of formations by the comparison of a few hand-specimens. They omitted to take fully into account the subsidiary evidence which has been accumulated during the last decade. But even if we entirely ignore this mass of proof, it is surely of interest to point outlithological resemblances between rock-systems, even if we are not prepared to correlatethem. Twilight is better than absolute darkness. We may say that the Pebidian of Hicks is like the Lower Taconic of Emmons, without asserting the contemporaneity of the two groups; and then leave the lithological hint at the mercy of the evolutionary laws which will either destroy it or develop it into a theory.