In 1834 the existence of an annual period of earthquake frequency was pointed out by Mérian, and his discovery has been fully confirmed by the work of later seismologists. No result of equal importance was, however, added to our knowledge until fifty years afterwards, when a valuable memoir was presented by Dr. C. G. Knott to the Seismological Society of Japan. As the present paper follows somewhat closely in the lines laid down by Dr. Knott, I will first give a short account of the results obtained by him, and in the next section will describe his method of investigation. One of the objects of Dr. Knott’s work was to ascertain whether any trace could be detected of a six-monthly, as well as of an annual, period. He examined for this purpose lists of earthquakes in Japan from 1872 to 1880 (Milne), Europe and the adjacent parts of Asia and Africa from 306 to 1843 (Perrey), the Grecian Archipelago from 1859 to 1873 ( S c h m id t), the East Indian Archipelago from 1873 to 1881 (Bergsma), New Zealand from 1869 to 1879, and Chili from 1873 to 1881 (Vergara). The chief results obtained.from this discussion are the following: (1) In five of the six districts selected there is a well-marked annual period, the maximum occurring about December or January in Europe, the Grecian Archipelago, and Japan, and between June and September in New Zealand and Chili; during the winter season in every case: while in the East Indies, lying close to and on either side of the equator, no such period could be discerned. (2) With the exception again of the East Indies there seems to be in the earthquakes of each district a semi-annual period, the maxima occurring about January and June or July in Europe, about March and September in the Grecian Archipelago, about June and November in Japan, about February and August in New Zealand, and about April (for the first maximum) in Chili, the second not being clearly defined. In the case of Europe, the semi-annual period is not very pronounced, and the smoothness of the curve representing it as compared with the others “almost suggests,” says Dr. Knott, “ that this semi-annual characteristic is almost accidental, and that with a sufficient number of observations it might vanish altogether ” ; “and yet,” he adds, “it seems too general to be accidental.”