By the use of pendulum seismographs it has been shown that during an earthquake the ground may move in one, two, or several directions (see p. 21); and it is, generally speaking, only in those cases where we experience a decided shock in the disturbance that we can determine with any confidence the direction in which the motion has been propagated. Such directions are usually indicated by the major axis of certain more or less elliptical figures which have been drawn, which in themselves appear to indicate the combination of two rectilinear movements. Results similar to those indicated by the records of pendulum seismographs have also been obtained upon moving plates with a double bracket seismograph. Thus, in the earthquake which shook Tokio at 6 A.M. on July 5, 1881, there were indications of the following motions:— Near the commencement of the shock the motion was N. 112° E. One and a half second after this, the direction of motion appears to have been N. 50° E. In three-fourths of a second more it gradually changed to a direction N. 145° E., and after a similar interval to N. 62° E. Half a second after this it was N. 132° E., and four seconds later the motion was again in the original direction—namely, N. 112° E. These particular directions of motion have been selected because they were so definitely indicated. The commonest type of earthquake which is experienced in Japan, and probably also in other earthquake-shaken districts, is the compound or diastrophic form. That earthquakes often have motions compounded of two sets of vibrations, has also been proved by the analysis of the records obtained from two component seismographs. From an analysis of a record of this description, Professor Ewing has shown that in the earthquake felt in Tokio on March 11, 1881, there were approximate circular (somewhat spiral) movements. This leads us to the consideration of the twisting and wriggling motions which are said to be experienced by some observers. Motions like these, which by the Italians and Mexicans are called _vorticosi_, are usually supposed to be the cause of objects like chimneys and gravestones being rotated. These phenomena, it will be seen from what is said in the chapter upon the effects produced in buildings, can be more easily explained upon the supposition of a simple rectilinear movement. That at the time of an earthquake there may be motion in more than one direction has been recognised since the time of Aristotle; and it is possible that two sets of rectilinear motion, as, for instance, the normal and transverse movements, may have led observers to imagine that there has been a twisting motion taking place, and this especially when the two sets of movements have quickly succeeded each other.