the result being that it is seldom possible to be confident of their time records. As Professor Milne very justly remarks': "It does not seem likely that, until we are in possession of a series of records taken at long distances apart on the surface of the globe by means of instruments which are similar, which have sufficient sensibility to record preliminary tremors, and which record upon surfaces moving sufficiently quickly to allow of accurate time determinations, that our present knowledge will be greatly increased."

The question has often been raised whether the speed of propagation is constant. Theoretically it should be constant only when propagated through media in which the elasticity-density ratio is constant. That this ratio is a variable one is always taken for granted. This is equivalent to saying that constancy cannot be expected. It has, how

ever, been inferred by many that the speed increases with the distance, and by a few that it diminishes with the distance. Both views, especially the first, can find support in the computed speeds of different quakes. Instances can be cited which appear to indicate in the first few hundred miles a decrease of speed, followed at greater distances by an increase of speed. But these cases really prove nothing. When we remember that the instruments vary much in delicacy and pick up a wave or wave-group early or late according to their sensitiveness, we may expect to find evidence supporting all kinds of views on this subject.

1 1 Rep. Brit. Assoc. for Advancement of Science, 1895, p. 169.

CHAPTER XII

LONG-DISTANCE WAVE SPEEDS

Origin of Present Pendulum Observations-Von Rebeur-Paschwitz's Investigations of Minute Deflections of the Plumb-Line-Milne's ResearchesIdentification of Such Disturbances with Distant Earthquakes-Dr. Charles Davison's Observations-Von Rebeur-Paschwitz's Suggestion of Co-operative International Inquiry upon this Matter-Researches of the Italian Seismologists—Accumulation of Long-Distance Seismograms -Their Study by Prof. C. G. Knott-Results of the Study-Different Groups of Waves-Preliminary Tremors-The Following Phases-Dr. Oldham's Researches-Speed of Surface Waves-Speeds of Preliminary Tremors-Dr. Knott's Remarkable Results-Method of Determining the Origin of a Distant Earthquake

A

S far back as 1883, Professor Milne, animated, perhaps, by the brilliant success he had even then attained in devising instruments for seismic investigation, ventured to predict (in substance) that the time would come when a great earthquake would be recognisable at any point on the terrestrial surface, even as far away from its origin as the antipodes. His meaning appears to have been that a forcible quake generates vibrations which permeate the whole earth within and without, and that its recognition at any point is only a question of suitable instruments. It would seem as if he had never lost sight of his prediction, which has been fulfilled sooner, perhaps, than he might have been justified in expecting.

This recognition has been gradually reached by the cooperative results of two lines of experimental inquiry, both involving a common problem, viz., the constancy or inconstancy of levels. Astronomers had long been aware of certain anomalies in the records of their instruments which were explainable only by minute changes in the levels of the piers on which they were installed. These suspected changes of level were usually temporary and were soon recovered, being, in fact, periodic. They could not be attributed, therefore, to the settling of their foundations, nor to any permanent changes of the ground beneath them. Some of them have been found to correspond with the daily oscillations of the barometer; others, of annual period, to the seasonal changes. The amounts of change seldom exceeded one or two seconds of arc, but the increasing standard of accuracy in astronomical work made this a very important matter. Systematic inquiry into these changes was, therefore, begun in England, France, and Germany as far back as 1865 by means of instruments of extreme sensitiveness. The work of the seismologists in Japan and Italy was well known, and the investigators in the interest of astronomical science were naturally on the watch to ascertain whether seismic disturbances might not be revealed in the analysis of the records obtained from their instruments for detecting minute fluctuations of level. Such disturbances were definitively recognised in 1889 by Dr. E. von Rebeur-Paschwitz at Potsdam and Wilhelmshaven.

Meantime investigations were in progress by Professor Milne in Japan, having for their purpose the analysis of the various classes of tremors which were represented in the

traces given by horizontal pendulums. He was led to the conclusion that the column on which the instruments rested was subject to periodic tilting by movements of the ground. resembling the long, flat swells of the sea, originating in a distant storm. These tilts repeated themselves at intervals of two or three seconds, and the deflections of the pendulums were equivalent to tilts varying from one to four seconds of arc. There were also indicated other movements of two or three minutes in period, and these periods were remarkably regular. These traces were sent to Von RebeurPaschwitz, and the comparison left very little doubt that both were dealing with the same kind of phenomenon. It was suspected that these records were caused by far-distant earthquakes; but considerable time must elapse before the suspicion could be verified or dispelled by the accumulation of more evidence. And the evidence was at length forthcoming.

As the observations of Von Rebeur-Paschwitz were primarily for astronomical purposes he secured the co-operation of Professor Kortazzi in the Russian Naval Observatory at Nicolaieff, to whom he furnished one of his very delicate horizontal pendulums described in Chapter VI. Similar ones were set up at Potsdam, Wilhelmshaven, and Strassburg. It was soon shown that a very large proportion of the suspected disturbances were common to all four places. Indeed, as Von Rebeur-Paschwitz says:

"It is a comparatively rare occurrence that when an earthquake figure, however small, appears on one of the photographs, it is not equally visible on the other. It often happens that the curves are not sufficiently distinct, owing to variations in the

intensity and figure of the light-point, and faults in the paper, or where a general microseismic movement is more pronounced at one of the stations than at the other. In such cases small disturbances may at first escape detection, but are often found when notice is given from the other station."

It was also found that these disturbances followed a few minutes (less than half an hour) after some earthquake of very considerable force in a distant part of the world. Generally the time interval between a great quake and the appearance of the indications of it upon the photographic records of the pendulums bore definite proportions to the distance traversed by the impulses. But these definite proportions were found to hold only in one and the same earthquake recorded at different distances from the origin. As between different earthquakes it appeared that the more powerful ones were propagated a little more rapidly than the weaker. The general agreement in the character of the indications given by the various pendulums was a strongly corroborative feature.

In the columns of Nature, December 27, 1894, Dr. Charles Davison called attention to a publication of Dr. von Rebeur-Paschwitz, making public for the first time, so far as known, the diagrams obtained from the latter's pendulums. The accompanying Fig. 60 is reproduced from that article, and shows two photographic traces, one obtained at Strassburg, the other at Nicolaieff. The numerals on the lower line indicate hours of Greenwich mean time. It will be noted that both traces are interrupted by gaps which are due to several causes, the chief one being the wide oscillations of the light-point which often carried it off the band