which is drying. Beyond this limit, however, there is far more open ground on the N.W., W., and S.W. sides of the station than there is in opposite directions, and the pronounced easterly motion may therefore be due to the unloading on the western side (fig. 9).

All round station N for a distance of at least fifty yards there are either high trees or shrubs. Beyond this limit in a westerly direction but at a lower level there are a pond and flat ground. In an opposite direction

on the same level there is at a distance of about one hundred yards a smaller area of open ground facing the Exhibition buildings. The movement is therefore as if the ground on the side of the largest evaporation area had arisen. But the movement is small (fig. 10),

Near to the instrument at S on the west side the ground is covered with green corn about one foot in height, while towards the S.E. there is a strip of bare ground perhaps fifty yards wide and one hundred yards in length. The movement, which, however, is slight, is towards the area most open to the sun.

Neglecting this strip, then, there is very much more open ground on the western than upon the eastern side of the station, and the movement may be explained on the assumption that this side, because it loses the most weight, rises relatively to the other (fig. 11).

Near to O the ground is somewhat more open on the north side, and it would appear that the motion was towards this side. Because the movements are slight, they might equally well be explained as a component of a south-eastern tilting due to greater relief of load upon the S.E. side, which is more exposed than that in the opposite direction.

Although a diagram may be modified by contraction following desiccation in the immediate vicinity of an instrument, this detailed examination of the observations in relation to the localities at which they were

made tends to the conclusion that diurnal waves are in part distortional effects of the earth's surface due to unequal relief of load from various areas by evaporation. When the movements have been absent or small, the instruments have been, at stations on the solid rock, well protected by trees or on an open plain. Many anomalies occur which still require an explanation, the most remarkable perhaps being the smallness of the motions at A, where the hypothesis requires that they should have been pronounced. The intermittent character in the movement at R may possibly be connected with the deep cutting on its western side, which breaks

the continuity of the ground upon that side, which it is assumed in order to produce the easterly deflection must rise.

I put forward these conclusions simply as being, for the present at least, the best that I am able to arrive at as explanatory of my own observations. The conclusions reached by Dr. E. von Rebeur-Paschwitz only partly confirm my results. In the British Association Report for 1893, on p. 316, he says that the range of motion is on an average very nearly proportional to either the quantity of sunshine or the maximum oscillation of temperature during the day.' This and the fact that the movements at Teneriffe, where the observatory appears to have been founded on and surrounded by soil and rock, were very much more pronounced than they were at Potsdam and Wilhelmshaven, where the soil was comparatively soft, apparently support the view that the diurnal wave may be a distortional effect due to evaporation. On p. 320 of the same report, however, he says that at Potsdam as well as Orotava the average range of daily motion agrees most remarkably with those meteorological elements which we may consider as a measure of the intensity of solar radiation. But I must not omit to remark that the single days do not show this coincidence equally well. For cloudy days occur with a large range of oscillation, and clear days with a small range.'

Although my observations in Japan have shown that when it was cloudy and wet the diurnal wave has been absent, it is not impossible that there may be cloudy days when, in consequence of wind, evaporation may occur, and in consequence the daylight distortion may be marked.

3. Effects due to Condensation (Night Effect).-It has been shown that at a favourably situated station the evaporation effect which has been marked during the morning may late in the afternoon be the means of starting a retrograde movement. It, however, remains to be explained why a motion possibly commenced in this manner continues slowly during the night until about 6 A.M. upon the following morning. Because this movement is comparatively small it may be produced by the addition or removal of a comparatively small load.

The precipitation of dew, which on a uniform area like evaporation follows in the wake of the sun, represents a feeble load, but the retrograde motion continues when dew is not visible. But although dew may not be visible, if we look beneath a board which has been lying on the ground all night it is usually found to be very wet. This observation suggested the idea that just as moisture is condensed beneath a board, so it may be condensed in the ground within one or two inches of the actual surface.

During a hot day moisture is evaporated from the soil, which is perceptibly heated to a depth of about one foot. Shortly after sunset the surface to a depth of one or two inches is chilled or in winter it is frozen. The result of this is, that moisture rising as vapour and by capillarity from water-bearing strata is condensed on the underside of the chilled surface. Like the dew we see on a uniformly covered surface the underground dew should be first precipitated on the eastern side of a station and subsequently upon the western side, and therefore during the night the surface on the former side gains weight at an earlier hour than the latter.

To determine how far superficial soils gain in weight by an action of this description, independently of moisture precipitated from the atmosphere or condensed as it rises out of the ground, the following experiments were made. Two boxes each 1 ft. 6 in. square and 2 in. deep were

balanced on the extremities of beams carried upon knife edges. One box had a bottom made of tin and the other of fine wire netting, and each was filled with earth. Excepting when they were weighed, by placing weights at the other ends of the beams, they were allowed to rest on the soft earth of my garden. Sometimes it was found that during a night both boxes would lose weight, but at other times it was found that the weight of the box with the tin bottom had not changed, whilst the one with the wire netting had gained from 2 to 2.5 ounces, which apparently showed that there had been a condensation of moisture coming up from beneath of 10 ounces per square yard, or about one-eighth of that which might have been removed during a day by evaporation. As my notes upon these experiments were destroyed by fire, what is here said can only be taken as indicating the character of a phenomenon which hitherto has not received attention.

Whether the causes which have been described are sufficient to account for the diurnal movements of a horizontal pendulum remains for future investigators to decide.

The gradual taking away of weight, followed by a gradual addition of weight unequally on the two sides of a pendulum during each period of 24 hours, will account for the observed movements, and in the evaporation of moisture during the day and the precipitation of moisture on the surface, together with its condensation beneath the surface during the night, we have phenomena which relieve or load surfaces in the required

manner.

(h) Tremors.

In the third Report to the British Association, 1883, after observing tremors with the ordinary Italian form of tromometer, I attributed their origin either to the effects of high winds or to small but rapidly recurring variations in atmospheric pressure, such as may be observed during a typhoon.

After analysing a long series of records of these movements, which were obtained from an automatic tremor recorder, and comparing the results with observations made in Italy, the conclusion arrived at was that tremors were at a maximum when the barometrical gradient was steep, no matter whether at the place where the tremors were observed the barometer was high or whether it was low.

This relationship between tremors and the state of the barometric gradient, although it did not explain the origin of tremors, tended to destroy the distinction between tremors which occur with a low barometer, and are called baro-seismic motions, and those which appear during periods of high pressure, which are called volcano-seismic disturbances.

An examination of the photograms obtained from the horizontal pendulums, which permit of more accurate analysis than those previously obtained, although it does not show that tremors only occur at the times when the barometric gradient is steep, shows that at such times tremor storms are marked. These same diagrams, however, on account of the relationship they show between tremors, the changes in the position of a horizontal pendulum, barometrical pressure, and the diurnal wave, lead me to withdraw the suggestion that, because steep gradients are usually accompanied by wind, such winds, whether they are local or distant, may be the immediate cause of tremors. In their times of occurrence winds

and tremors undoubtedly show a close relationship, and therefore the former may, by its mechanical action upon buildings, trees, and the surface of a country, produce slight tremors, and influence the character of a record.

The points which are marked in connection with the recent observations are as follows:

1. Relationship of Tremors to Localities and Instruments.

Tremors have been pronounced at Station A, the instrument at which station, however, was the one most sensible to changes of level. At stations on the surface in Tokio they have been feeble, but have varied in their intensity. Underground upon the rock they have never been observed. This latter observation, which is based upon records obtained from five instruments, is in direct opposition to the observations made at Rocca di Papa in Italy, where, I understand, tremors are as pronounced underground as they are upon the surface.

At Station A, an instrument which showed tremors even in a more marked manner than the large horizontal pendulum, was a similar instrument made of a few millimetres of aluminium wire, a small mirror, and a needle point, weighing only a few grammes, a comparatively large form of which is described in the Report for 1892. On account of the manner in which the spots of light reflected from the mirrors of a pair of these instruments placed side by side would come to rest, and then start suddenly to move in the same direction, I was led to the conclusion that they were actuated by an intermittent tilting, and therefore that tremors, rather than being elastic vibrations, had the character of wave-like undulations. The fact that the instrument most sensitive to changes of level gave the most pronounced records appeared to strengthen this supposition, and I was led to call these movements earth pulsations.

The only effect produced by heavy gusts of wind striking the building, or the beating of a steam hammer at a distance of fifty yards, is to produce a temporary vibration in the pointers of an instrument; but there is no angular displacement, and consequent swing, which characterises the movements during a tremor storm.

An important observation made at Station A was that tremors were produced when two tons of water were taken out of a well distant about thirty yards. The operation caused the ground upon the well side to rise, and the horizontal pendulum was gradually displaced in an opposite direction. From this it may be inferred that either the pendulum took up its new position intermittently, or that the level of the ground changed intermittently. Whichever it may have been, it may be concluded that whenever a rapid change in the inclination of the ground takes place, horizontal and probably other pendulums may be caused to swing, and, as will be seen in the next section, at least a portion of the tremor records may be explained on the supposition that they are due to such causes.

2. Relationship of Tremors to the Diurnal Wave.

(1) Even when a horizontal pendulum is steadily following the diurnal wave and no tremors are visible, slight tremors nearly always appear about 6 or 9 A.M., just at the time when its easterly excursion has been completed and it turns to commence a relatively rapid motion towards the west (figs. 9 and 13).

The instruments were under the same cover. See Appendix.