'Before the earthquake there were 440,548 buildings, and of these 316,087 have been destroyed or damaged. The population of Tokyo before thé disaster was 2,031,391; the number of refugees and homeless is officially given as 1,356,740.... There are still 135,000 persons without shelter of any kind in Tokyo.'

With some hesitation I approach the attempt to give an idea of what is recorded by these beautiful seismographs. Much of it is too technical to interest the general reader, and yet without some illustrative figures he can scarcely visualise the interesting problems we have been able to attack by utilising these records.

There are many ways in which a disturbance can be communicated from a distance, but for the present we will limit attention to three of them, usually denoted by the letters P, S, and L. The P waves are the first to arrive (unda Prima), and though the S waves are not really second, that was the designation intended, since they follow the P waves after an interval, which is of great importance and is denoted by S-P. In each case, by a fortunate coincidence, the designating letter is a reminder of the nature of the disturbance, as well as of its time of arrival; for P may stand for 'push' or 'pull,' and S for shake' or 'screw.' When we give a pull at one end of a rope we cause an effect at the other end much more quickly than when we give a shake, which travels in a snake-like manner. The longer the rope, the greater would be S-P, the interval by which the shake follows the pull; and though the solid earth does not much resemble a rope, yet earthquake shocks are communicated through the body of the earth along definite curved paths which may be compared with ropes. The letter L signifies long waves, and represents much slower or longer fluctuations than those of P and S. But it may also stand for last waves, so that in each case the letter has a double significance, reminding us of the order of arrival (prime, second, last) and of the nature (pull, shake, long). And again the L waves arrive last because they come the longest way round,' which in this particular instance does not turn out to be the shortest way home. They came in fact not through the earth but round it-along the surface, and are sometimes called Rayleigh waves, because the late Lord Rayleigh first explained their nature.

The following short table will give an idea of the times taken by these three kinds of waves to reach various distances from the starting-point or focus of the earthquake, supposing that point to be close to the earth's surface. The distances are measured round the surface of the earth and are given in three different ways: first of all in degrees, of which 180° take us round to the opposite point of the earth, and 90° take us round a quadrant; secondly, in kilometres, of which 20,000 take us to the opposite or antipodes, and 10,000 round a quadrant; thirdly, in miles at 5 miles to 8 kilometres, which is near enough for our purpose. The most convenient unit of the three is the first, since we can apply it to a globe of any size.

The actual tables used by seismologists are, of course, much more detailed and extensive than this; but enough is given to show their nature and the way in which they can be used. Suppose, for instance, that on developing the seismograph film we find the record of a distant earthquake. We notice from the time marks on the trace that the very first disturbance of the usually quiescent trace was at 5h. 3m. 228. This is the time of arrival of P. The trace then continues to be disturbed for some little time to a moderate extent, when suddenly S arrives (say, at 5h. 13m. 26s.) as is shown by a much more extensive agitation of the trace, which continues till about 5h. 30m. or 31m., when the nature of the agitation changes, the movement being slower and longer (L), and gradually developing into quite extensive

swings from side to side, which may carry the spot of light-or pen or whatever it is that does the registration-quite off the paper in the case of a big earthquake.

And now what can we infer from these readings? First, notice that S-P, the interval by which S follows P, is 10m. 4s. which happens-though it was not altogether an accident-to occur in the above table. It is the interval corresponding to 80° or 8889 kilometres, or 5556 miles; and we infer at once that the earthquake shock took place at this distance from the seismograph. We can also infer the time when it took place; for the table shows us that the P wave takes 12m. 19s. to travel this distance, and hence the shock must have occurred 12m. 19s. before 5h. 3m. 22s., or at 4h. 51m. 3s. We could get the same result by subtracting from the arrival of S, 5h. 13m. 26s., the time for S, 22m. 238.

Finally, as a check on this result we deduce the time when the L waves should be expected. Their arrival is a much more indefinite phenomenon than that of P or S, which is indicated in the table by giving only the whole minutes, without any seconds; and it may be noted as a convenience that these minutes are just half the degrees in the first column. The L waves in fact travel approximately two degrees per minute round the earth's circumference, and thus without needing to refer to the table at all-if we remember this rule-we can say that the L waves are due at a distance of 80 degrees about 40 minutes after the shock (4h. 51m. 3s.) or at about 5h. 31m. 3s., which is approximately what was observed. If it be asked why this check is necessary the reply is that it is only too easy to make a mistake in identifying P and S, or in counting the minutes, or in some other way. The first waves P are often so small as to be lost in the constant movement to which most seismographs are subject-small oscillations called microseisms, whose cause is not yet known. They have been thought to be due to the beating of the waves on the coast, which sends tremors through the land; and the fact that they seem to travel from Western Europe across Russia into Siberia, dying out as they go, seems to accord with this origin from the Atlantic rollers on the western coast of Europe. But the late Prince Galitzin, after a most careful and extensive study of microseisms, summed up

against this explanation, for the good reason that big Atlantic rollers do not always concur with considerable microseisms. Sometimes there are big rollers with no microseisms; and sometimes microseisms with no big rollers as ostensible cause. So that the matter is still at least in doubt. We may, however, for the moment leave on one side this puzzle of the origin of the movements, for we are immediately concerned simply with their existence, as a good reason why the P waves of a small or even a moderate earthquake may be read wrongly, or cannot be read at all; and then perhaps S, being the first visible movement, is read as P, which gives false information. With large earthquakes these difficulties are, of course, not so liable to occur.

We have yet other sources of information in the trace. Suppose that we have at a station not merely one seismograph but two; and suppose them arranged at right angles, so that one is sensitive in a north-south direction, the other in an east-west. Then, to take an extreme case, an earthquake due east (or west) of the station would affect only the second instrument, leaving the first undisturbed, while an earthquake due north (or south) would disturb the first and not the second. An earthquake in any of the quadrants (N.E., N.W., S.E., S.W.) would disturb both equally. In fact, we see that, subject to certain ambiguities, we can make inferences as to the direction in which the earthquake centre lies, and, if the ambiguities can be removed, by combining these with our knowledge of its distance away, we can specify the actual spot at which it occurred. To remove the ambiguities we must have a third instrument which is sensitive in the vertical direction-a more troublesome matter than the horizontal components but not of prohibitive difficulty. To sum up, at an observing station fully equipped with three instruments we can infer both the time when, and place where, an earthquake occurred; but if, as is often the case at present, there is but one instrument, though we can still infer the time, we only know the distance of the occurrence, but not its direction. It is perhaps unnecessary to explain the reasons for this present shortage in equipment; shortage of funds will readily occur to even the least imaginative, and then there is the indirect effect of improvement in instruments, which

suggests, when there is money to spend, getting a new pattern rather than a second component to match the old.

It will also be seen that when there are a number of observing stations, even if each can specify only the distance of the earthquake centre without its direction, still by combining the information we can identify the actual spot. For let us take a terrestrial globe and describe round each station a circle with radius equal to the known distance of the earthquake centre. If our information is exact, the circles will all intersect at the requisite point on the globe.

It is desirable that the stations should be well separated; for otherwise the circles will not intersect so as to give a precise specification. Circles of nearly the same size and nearly the same centre are never very far apart throughout their whole circumferences, and moreover, a small error in the size of one of them will produce a great change in the point of intersection with any other. Hence it has occurred sometimes that when information was only available from one or two stations in the British Isles-or even in Europe, which is not a very large area on the globe-there was considerable uncertainty as to the earthquake centre. An extraordinary instance of this occurred on Dec. 16, 1920, when the largest earthquake of recent years was clearly registered at several stations in England. Unfortunately, these did not include our one completely equipped station, Eskdalemuir, where the seismographs were temporarily out of action; but three or four stations (Dyce, Bidston, West Bromwich, Oxford) possessing single seismographs could all give the distance of the disaster with considerable accuracy. The circles when drawn on a globe were nearly alike, but showed two regions of intersection, one in the far East, the other in the far West. (It will be remembered that any two circles intersect in two points.) What was wanted to complete the specification was information from a quite different locality, giving a circle which would cut across one of these regions and not the other. The only other information immediately forthcoming was an ordinary press telegram from the United States, stating that the centre must be only some 3000 miles from Chicago and Toronto. But