By inspecting either the wavy line or the resultant curve, it will be seen that since the ninth century down to the present time there has been a decided decrease in seismic energy. From the ninth century down to the fifteenth century this decrease is represented by a regular curve. At this point, however, the decrease becomes slightly more rapid, and is represented by a second curve. If, instead of calculating ordinates for my curve, in which intensity has been considered, simply the number of earthquakes are counted, a similar result is obtained. From this it appears that the rate at which seismic energy decreased during the last 500 years was about the same as that at which it decreased during the 500 years previous to this period. If the lists for the Italian and Turco-Hellenic districts could be similarly analysed, and the earthquakes of any particular district picked out from the others, it is very probable that a similar decrease or alteration in seismic energy might be observed. Provided that we have at our disposal records of the various earthquakes which have occurred in any given district during a sufficiently long period of time, one conclusion that we may expect to arrive at is that we shall be able to trace some variation in the seismic activity of that district. For the Kioto area, it has been shown that there is a diminution in seismic activity, In other districts, however, there may possibly be an increase.[94] _Relative frequency of earthquakes._—A question which is of great interest to those who dwell in shaken districts is as to how often disturbances may be expected to occur. From a general examination of this question, considering the earthquakes of the whole world. Mallet arrived at the following conclusions:— 1. While the smallest or minimum paroxysmal intervals may be a year or two, the average interval is from five to ten years of comparative repose. 2. The shorter intervals are in connection with periods of fewer earthquakes—not always with those of least intensity, but usually so. 3. The alternations of paroxysm and of repose appear to follow no absolute law deducible from these curves. 4. Two marked periods of extreme paroxysm are observable in each century, one greater than the other—that of greatest number and intensity occurring about the middle of each century, the other towards the end of each. The form of the curves which Mallet has drawn seem to indicate that seismic energy came in sudden bursts, and then subsided, gradually gathering strength for another exhibition. This is continually seen in the shocks experienced in various seismic areas—a large shock, or the maximum of the activity dying out by repeated small shocks on succeeding days. Mr. I. Hattori, writing on the large earthquakes of Japan, remarks that on the average there has been one large earthquake every ten years. They, however, occur in groups, as shown in the following table.