of ships. Thus Captain Murdoch, of the Denbighshire, when in the neighbourhood of St. Paul Island, recorded two severe shocks. I quote from his log, which is one of a number collected and discussed by Rudolph.

"The first shock was like a jarring of everything in the ship. On deck it appeared as if the chain cables were running out and the topmost yards were coming down by the run, and it seemed as if every step we took on deck we must fall down. This shock lasted 30 or 40 seconds. All hands had rushed on deck, thinking the ship was on shore, and while sounding the pump the second shock occurred. It was sharp and instantaneous, as if a large cannon had been fired immediately below the ship. . . It was a volcanic eruption or explosion. The noise that accompanied the first shock was like the low groaning of distant thunder, but yet it appeared near and about us.'

The land surface of the globe is only a small part of the whole surface of the earth, and there must evidently be a very large number of earthquakes originating below the sea for which no observations by observers on land are available. It is, therefore, of special importance to have records of seismic phenomena at sea. The ships at sea, however, are comparatively few in number, and indeed are few and far between compared with the great stretches of ocean over which they navigate, and the records of earthquakes occurring under the ocean beds are necessarily more incomplete than the records of other earthquakes.

The

In spite of all these difficulties there is a certain amount of information available as to the frequency with which earthquakes have occurred during the last forty years in different districts of the earth's surface. The seismic maps of the world are of interest in this connection.* A well-known earthquake region is in Italy and the Alps, which has, according to de Ballore, as large a number of earthquakes for its size as Japan. strongly marked regions appear to be situated on the borders of continents and in areas where geological changes are known to be in progress. Thus, for example, there are areas of strongly marked seismic activity all round the Pacific Ocean, from the East Indies, the Philippines, Japan and Alaska, and along the west coast of America. Thus statistics collected of recent years

* See Knott, The Physics of Earthquake Phenomena, p. 97.

E

show that for every earthquake felt in Great Britain there were (roughly speaking) 50 in Japan and 158 in Greece, the areas of these countries being taken into account.

We have already mentioned the discovery made by Milne in Japan which formed the starting-point of Seismology as an exact science dependent on accurate observations and capable of development in accordance with the general principles of science. In Milne's seismograph we have, as its fundamental characteristic, a horizontal pendulum fitted at its inner end with an agate cup which presses against a steel pivot-point screwed into a vertical iron pillar cast in one piece. The pendulum is supported at its outer end by means of a fine steel wire which passes to a pin at the top of the pillar. Now, when an earthquake is in progress, it is found that there is, in general, a definite movement of the pendulum. In fact, even when the earthquake is at a great distance away (and it is by no means unusual for a seismograph of this type to record seismic movements occurring at a distance of 10,000 kilometres), there is apparently a definite movement of the earth which can be detected by a pendulum, provided that the adjustment of the pendulum is sufficiently delicate. There are many very important mechanical devices for measuring accurately and conveniently the motion of the pendulum, and, in fact, in the Milne seismograph the pendulum carries at its outer end a small transverse plate of aluminium with a narrow slit parallel to the pendulum, which, by means of an ingenious arrangement of a slit in the case which covers the instrument, and an illumination from above, enables a small dot which corresponds to the intersection of the slits to be cast on the surface of some bromide paper, which is wound on the surface of a cylinder made to revolve uniformly, the speed of revolution of the paper being nearly 4 mm. per minute. In this way an accurate and convenient record is obtained of the motion of the tip of the pendulum. It is interesting to see the type of record made on a Milne seismograph when a great earthquake is in progress. There are, first of all, certain oscillations which are called the primary phase, and after a time, which varies for earthquakes at different distances away, the seismogram changes its type, and there is then usually a large movement denoted by S, which initiates the second phase. Its incidence is less sharply marked than P, and it is sometimes very indistinct. This second phase also lasts for a time, depending on the distance away of the earthquake, and then the whole appearance of the

seismogram changes and assumes a strongly periodic character. This phase, which is called the long-wave phase, is usually marked by a few waves of period about 20 seconds, gradually increasing in amplitude. After reaching a maximum amplitude, the waves subside and pass through a succession of maxima before merging into the "tail" or "coda" of the earthquake.

Now the appearances of seismograms to the trained eye present a curious uniformity in spite of the minor variations which occur. And it is found that these characteristic phases of the primary waves and the secondary waves and the long waves are capable of a very important, and at the same time simple, explanation, if we assume that the earth is an elastic solid. For it is known that an elastic solid is capable of transmitting various kinds of waves. It can have a longitudinal wave which moves with a certain velocity V1, and it can have a transversal type of wave which moves with a different velocity V2, and when the elastic solid has a shock at a certain point waves of both these kinds are sent out from the centre of the disturbance. In the first kind of waves the various particles move backwards and forwards in the line of wave propagation. In the second kind of waves the particles oscillate backwards and forwards at right-angles to the direction in which the wave is moving. Now, if the earth is an elastic solid, we may expect two trains of waves travelling with two different velocities to be sent out if there is a shock at any point. The starting-points of the two phases P and S can now be interpreted as the arrival of the two types of waves-first the longitudinal waves, which move faster, and then the slower transverse waves. And with this hypothesis we find that the velocities of the two waves are about 51⁄2 km. per second and 3 km. per second in the case of fairly near earthquakes. It is a great achievement to have obtained this amount of agreement between the actual seismograms of various earthquakes at many different stations and the predictions of the theory of elasticity which asserts the existence of two waves of these types. This chapter of seismology, indeed, shows the tremendous difference between a set of observations which have been welded into a science and observations which are discrete and disconnected and which have no underlying theory behind them. Science really begins when some generalization is made which is capable of covering data already obtained and which predicts other data. Now in Seismology, the moment it became reasonably probable that these well-marked P and S phases represented the arrival

of the two kinds of wave, it at once became possible to predict within limits the time each type of wave would take to go a certain distance. And evidently, also, the quicker wave gains a definite amount for every kilometre traversed. Thus a certain definite interval separating the two types of wave in any record betokens a definite distance away of the disturbing cause. Thus, a record at Edinburgh, say, might show a time difference of a certain number of seconds in the start of the P and S phase; from this we could deduce the distance away of the disturbance. If, then, records at other observatories are also available, we may be in a position to assert that the earthquake was at x km. from one station, at y from another, and at z from a third, and, therefore, that it must lie in the neighbourhood of Tokyo. In this way the many earthquakes already treated have been located. It is to be noticed that the larger the number of records available the larger is the degree of accuracy to be expected in the location of the earthquake itself. And it is also plain that a fair distribution of seismographs all over the world is to be desired, not merely the excessive equipping of stations which lie for the most part in one or two continents only. It is for this reason that Milne was so anxious to establish seismographs in countries not hitherto making any records of seismic phenomena, and, owing to his great zeal, fifty of his seismographs have bcen distributed all over the world, so that no continent, and few large countries, have remained unrepresented in the international seismological service of the world. Seismology is essentially a science which needs the co-operation of many countries and peoples, and it provides a strong link between the people of the East and those in the West whose scientific pursuits have led them into these absorbingly interesting fields of study.

In spite of the large number of observations of earthquakes which are now available, there is a real need for more material. It is by no means an easy task to deduce the velocities of the P wave and the S wave from the mass of observations, for in the case of each earthquake we do not in general know the exact location of the earthquake. And even if we know the point of the earth's surface under which the earthquake occurs, we do not know how far down the actual disturbance took place. The place at which the disturbance takes place is generally called the "focus," and the point of the earth's surface directly above it is generally called the "epicentre." Thus, in the case of any earthquake, we have to find not only the epicentre—which, in the

case of a large earthquake, may be only too obvious if much damage has been caused, but which is not known to any degree of accuracy in the case of small earthquakes but also the focus. Now, it is clear that if reliable information were available as to the rates at which the P and S wave travel it would give us some help in our task of making an estimate of the depth of the earthquake focus, for the deeper the focus the longer the path traversed by the waves in getting from the disturbance to an observing station at a certain specific distance from the epicentre of the earthquake. It is, therefore, of great importance to obtain information as reliable as the circumstances permit of the velocity of the P and S waves. It was with a view to obtaining information of this kind that Dr. Jeffreys and I undertook an enquiry into the waves caused by a great explosion at the works of the Badische Anilin und Sodafabrik, at Oppau, in the Bavarian Palatinate, on September 21st, 1921. Oppau is about 5 km. north-west of Mannheim and stands in the Rhine valley. The shock of this tremendous explosion was so great that waves of the P and S type were started in the earth's crust, and these waves, which spread out in all directions, were recorded at Strasbourg, which was 110 km. away, at Nördlingen, 175 km. away, at Zürich, 240 km. away, and at München, which is 282 km. away, by the seismographs which were at work in the various observatories. Now, as we already knew that the disturbance took place at a certain definite place, and took place, in fact, on the surface of the earth, there was no ambiguity at all about the focus or epicentre of the disturbance, and it was, therefore, a simple matter to deduce the velocities with which the primary and secondary waves travelled through the earth's crust. We found the velocities to be 5.4 km. per second for the primary waves and 3.15 km. per second for the secondary waves. With this information, it is now possible to obtain more reliable information as to the precise location of earthquakes which are not more than 200 or 300 km. from the recording station. If it were possible to make a similar investigation in the case of even greater explosions which are sufficiently strong to enable a record to be obtained at much greater distance, further information would become available which would materially increase the probability of making a more correct estimate of the distance away of seismic disturbances, even when these disturbances are --as in fact they generally are in practice at a far greater distance away than 300 km. But there appear to be grave