expended in throwing up a jet of water 330 feet in height into the air. A point well worthy of notice is the fact that the gunpowder waves had a more gradual increase than those observed in shocks produced by dynamite; in other words, the former had a closer relationship to what is so often observed in the records of actual earthquakes than the latter had.

3. Experiments of MM. FOUQUÉ and LÉVY.

In the experiments of MM. Fouqué and Lévy the velocity of vibrations. on the surface and underground was determined by recording the intervals between the shock which was usually produced by the explosion of from 4 to 8 kilos. of dynamite, and the displacement of an image produced by a ray of light on a photographic plate moving with uniform motion. The ray of light was reflected from a surface of mercury at the receiving station. The highest velocity was obtained between a point underground and the surface, along a line of 383 metres in length, which gave a velocity of 2,526 metres. In this case the shock was due to an explosion of 8 kilos. of dynamite.

The general results obtained were as follows :

The velocity evidently increased with an increase in the amount of explosive employed, and it was greatest in the more elastic rocks.

The discrepancy which exists between the above and Mallet's determination for granite (507 m.) only disappears if we compare it with the second maximum in the photographic record (325 m. to 543 m.).

The second set of experiments, considering the nature of the material in which they were obtained and the smallness of the charges employed, give remarkably high results, the velocity for the first maximum exceeding that obtained by firing a larger charge of dynamite in granite on the surface. In a single experiment to determine the velocity between a lower and a higher level underground, the direction of the wave path is unfortunately not very different from that of the stratification, and therefore is not comparable with those velocities along paths from the upper level nearly transverse to the stratification between it and the surface. If we are allowed to accept the results of Mallet's experiments, which show that the velocities in these two directions are in round numbers as 1.8 1.0, then we may conclude that the velocity between the lower level and an upper level was markedly greater than it was from the latter upwards to the surface.

These experiments show that the velocity between two points on the surface is less than it is between the surface and a point underground. They also indicate that the velocity with which vibrations are transmitted may vary with the depth of the wave path.

1895.

M

4. Observations of Prof. J. MILNE and Prof. T. GRAY.

In the author's experiments, which were commenced in conjunction with Professor Thomas Gray in 1881, and continued at various times during the next four years, the object was not simply to determine the rate of transmission of earth waves, but also to determine their general character. Usually the movements resulting from the fall of a heavy weight, or the explosion of dynamite or gunpowder, were recorded by seismographs. The weights employed varied from 1,710 lb. to 2,000 lb., while the charges of dynamite, which were exploded in holes 8 or 10 feet in depth, seldom exceeded 2 lb. Although the ground in all cases excepting one was soft, the resultant vibrations up to distances of about 600 feet were sufficiently large to be recorded as clear diagrams by bracket and other seismographs.

At various stations usually in a straight line joining them with the focus of the explosion, seismographs were installed, which wrote their movements on the smoked surface of a long plate of glass, the motion of which was controlled by clockwork. One seismograph was placed so that it wrote the movements parallel to the line of installation. These are called normal vibrations. A second seismograph was arranged to record the movements at right angles to such a direction. These are called transverse vibrations. A vertical lever seismograph was occasionally employed to give the vertical motion. A fourth pointer actuated by an electromagnet in connection with a short pendulum swinging across mercury gave a broken line marking small but equal intervals of time.

By the depression of a contact key, the receiving plates at all the stations were set in motion, the pointers of the seismographs drew fine straight lines on the smoked surfaces, while the pendulum indicated intervals of time. A few seconds later a second contact was made and the charge exploded, and the seismographs gave open diagrams of the resulting vibrations. When the earth motion had ceased, all the plates were stopped and were ready to receive a second diagram without any readjustments. One observer controlled all the stations, and the only errors due to human interference may have arisen from slight differences in the sensibilities given to the recording instruments. This, however, disappears when velocities were determined, not from the commencement of a disturbance, but from the sharp commencement of the violent vibrations or from the intervals of time between the appearance of particular waves at the different stations.

Observations were also made with seismographs having single indices by observing the disturbance created in similar dishes of mercury, and with other arrangements.

The results of observations made respecting velocity of propagation were as follows:

1. The velocity of transit of vertical vibrations near to an origin decreases as a disturbance radiates. Normal vibrations, although they have shown a decrease in velocity between the second and third stations, have also shown a decided increase. This latter observation has been marked with the transverse motions.

2 Near to an origin the velocity of transit varies with the intensity of the initial disturbance.

3. In different kinds of grounds, with different intensities of initial

disturbance, and with different systems of observation, I determined. velocities lying between 630 (192 m.) and about 200 (61 m.) feet per second.

4. In my experiments the vertical free surface wave had the quickest rate of transit, the normal being next, and the transverse motion being the slowest.

5. The rate at which the normal motion outraces the transverse motion is not constant.

6. As the amplitude and period of the normal motion approach in value to those of the transverse motion, so do the velocities of transit of these motions approach each other.

(b) Observations on Earthquakes.

The observations quoted in this section commence with those where the wave paths have not been more than a few hundred feet from station to station. These are followed by the results obtained from instruments separated from each other by distances of from three to six miles, a few hundred miles and so on up to velocities determined over paths equal to a quarter of the earth's circumference.

1. Observations in Japan.

For several years the author took diagrams of earthquakes at seven stations, each about 900 feet apart. These stations were in electrical connection, so that one pendulum marked time intervals upon each of the moving surfaces upon which diagrams were being drawn. From fifty sets. of diagrams, representing fifty different earthquakes, it was only in five instances that the same wave could be identified at the different stations. The result of these identifications led to calculation of velocities of 1,787, 1,302, 1,825, 869, and 501 metres per second.

Even these determinations cannot be accepted without reserve, because it is found that waves may spread out as they pass from station to station, a given wave splitting up into two waves, &c. Hence a velocity calculated from a wave (a) may be different from a velocity of a wave (b), and yet both are part of the same disturbance. In the diagram from one station a large wave may have a slight notch upon its crest, at another station this notch is seen to have increased in size, while at a third station it is so large that the single wave appears as two waves. As in the artificially produced disturbances, although an earthquake becomes feebler as it radiates, it apparently increases in its duration.

The same system of observation has recently been elaborated in Japan, but the distances between stations have been increased to several miles. Because the commencement of a disturbance at a given station varies with the sensibility given to the seismograph, the determinations of velocity depend upon the identification of particular waves upon the diagrams obtained from at least three stations. Up to the present this has only been possible on one or two occasions. On November 30, 1894, at 8.30 p.m., a velocity of 5 km. per second was obtained, other disturbances giving from 2-4 to 3.6 km. per second.

The following are examples of velocity determinations made in Japan between stations which have access to the telegraphic system of the country, and which are provided with seismographs and clocks which

automatically record the time at which a particular vibration was drawn. At each of the observatories it is therefore possible to calculate the instant at which a given instrument commenced to write its record.

In 1891, on December 9 and 11, strong shocks originated in the province of Noto on the west coast, which were observed in Gifu, Nagoya, and Tokio. The mean velocity determined from these records was 2.31 km. per second.

The destructive disturbance of October 28, 1891, which was recorded in Europe, was followed by many after shocks, the times of arrival of seventeen of which were accurately noted at Osaka, Nagoya, Gifu, and Tokio. The origin of the main shock was about five miles to the west of Gifu. To reach Tokio, a distance of 151 miles, took 120 seconds. The average time taken for all eighteen shocks was 118 seconds, and the average velocity was 2.40 km. per second, the rate of transmission to Osaka being the same as it was over the much longer path to Tokio.

The primary disturbance seems to have reached Shanghai at a rate of about 161 km. per second, and Berlin at about 2.98 km. per second. For the Shōnai shock on October 22, 1894, as a mean obtained by the method of least squares from observations at ten stations from 60 to 300 miles distant from the origin, a velocity of about 1.95 km. per second was obtained.

Giving these last determinations, all of which were computed by Mr. F. Omori, weights proportional to the number of observations each represents, the average rate at which disturbances are propagated over long distances in Japan is 7,560 feet, or 2-3 km. per second, a rate which fairly well agrees with that at which the large waves of similar disturbances travel from Japan to Europe.

2. Observations along Wave Paths of Great Length.

Next we will turn to earthquakes which have been noted at distances from their origins greater than those at which it has been possible to observe in Japan-a notable example of which are the observations made at the time of the Charleston Earthquake on August 31, 1886. Over 400 observations were made. A number of these were obtained from clocks which had been stopped, and as many of these were regulators which had daily been compared with a time signal, there is no reason to doubt their accuracy. All these observations, which were made on wave paths between 300 and 924 miles in length, were subjected to a rigorous analysis by Professor Simon Newcomb and Captain Charles Dutton, with the result that an average velocity of 5,184 m. was determined, and there was no indication of any sensible variation in speed. Considering the phase of motion which in the majority of instances was in all probability observed, this result is remarkably high.

A valuable study of the rate at which vibrations may be propagated through the earth's crust is one made by Dr. G. Agamennone of a series of shocks which in 1893 had their origin near to the island of Zante. These were recorded at the various stations mentioned along the foot of the diagram Fig. 15, the one farthest from the origin being Potsdam. The lengths of the various wave paths are indicated in kilometres. The time intervals measured vertically are on a scale of 20 seconds per millimetre. For five shocks straight lines connect a series of points indicating the differences in time between the occurrence of the shock near to its origin

and the time at which maximum motion was experienced at the various stations.

The dotted lines connect similar points for the commencement of the disturbance. The time of these commencements undoubtedly varied with the sensibilities of the recording instruments, and therefore it will be noticed that they have only been taken into account for Rome, Nicolaiew,

and Strassburg, where records were obtained from the great pendulum of the Collegio Romano, or from the horizontal pendulums of Dr. E. von Rebeur-Paschwitz.

The principal object in reproducing this series of observations is to see if there is any reason for believing that there is any variation in the velocity with which a given disturbance is propagated.