An examination of the above table leads to several important results. In twenty-four cases the motion underground has been greater than that

on the surface, while in six cases it has been less than that on the surface. These instances are taken from the three or four day periods. If the analysis was made for daily periods, the difference between the amount of motion recorded underground and that recorded on the surface would be yet more marked. Whenever the movements of the surface instrument A have been great, exceeding 1" in seven instances, its direction of motion has corresponded with the direction of movement of the instrument which is placed in a parallel direction F. In two cases the directions of movement between A and F have been opposite to each other. When, however, the movements of A have been small or less than 1" the cases of agreement and of disagreement in direction of motion are practically equal, there being 6 of one and 11 of the other. For January, February, and March rainfall seems to have been followed by considerable disturbances underground, the movements during dry periods being comparatively small. The instrument on the surface has, however, shown several marked exceptions to the latter rule, its pointer having moved from 8 to 12 mm. (2" to 3") at least five times when the displacement could not be attributed to the saturation of the soil.

During April and May, although a considerable amount of rain fell, the movements of the underground instruments were small, but it must be remembered that during these months percolation was in all probability very small as compared with that of January, February, and March. Instrument A, on the contrary, showed on several occasions very large movements between April 13 and 14, moving as much as 30 mm. or 6′′, and from what has gone before it is not necessary to assume that these disturbances were directly connected with rainfall.

Up to the date of writing this report I have not been able to obtain from the Meteorological Department factors which enable me to make any accurate estimate of the ratio of percolation to evaporation, but it may be taken, as a general rule, that percolation and the fluctuations in height of subterranean water are greater during the winter months than they are during the summer, and if the instruments partly owe their movements to movements of underground water, these movements ought to be most pronounced in winter, and this seems to have been the case. Since the commencement of May, up to June 6, E and F have wandered but little, the diagrams being fine straight lines like fig. 1, Plate II., and without tremors. It must also be observed that it has been the instruments in the underground chambers within 12 feet of water level which have moved the most. To throw additional light upon the part that subterranean water may have played in influencing the motion of the pendulums the following experiments were made :

1. The movements of water in an unused well were recorded.

2. A rough measurement of the rate at which moisture was evaporated from ground near to one of the instruments was made.

3. A well near to one of the instruments was twice emptied of its

water.

(k) Movements of Water in a Well.

From April 18 until June 8, I established a tide gauge in an unused well 80 yards to the east of the underground chamber. It consisted of a large wooden float carrying a baniboo mast 30 feet in length, the top of which projected through a hole in a lid which covered the top of the well. As the mast rose and fell a pencil in contact with a sheet of paper on a drum

recorded the motion. The diagrams obtained indicate the following facts. Very shortly after heavy rain the well commences to rise, and the rising continues for three or four days after the rain has ceased. The upward motion, which at times has been as much as 7 or 8 inches in 24 hours, eventually becomes slower for about two days, after which the water falls slowly. A more important observation, however, is that during any 24 hours there are fluctuations in the rate of rising or falling which when the well is nearly steady are distinct, but when the coming up or going down of the water is rapid they are barely visible. A number of these daily fluctuations are shown on Plate IV. About midnight and for some hours afterwards the water is at its highest, and it is again high during the middle of the day. It is lowest in the evening and the early morning, which is the time when the greatest quantity of water is being drawn from wells throughout the city. The nearest well from which water is drawn to the one in which the gauge is established is on the east side, about 60 yards distant.

In the following table the dates refer to the interval between noon of one day to noon of the succeeding day. The figures in the columns indicate the hours at which the well commenced to sink and then to rise in the afternoon and evening, and when it again commenced to sink and to rise in the morning. The time midway between the rising in the evening and the sinking in the morning may be taken as the crest of the night wave, the crest of the midday wave being halfway between the A.M. rising and the P.M. sinking of the next day. The omission of dates or hours indicates that the inflections on the diagram were indistinct or absent. The letters R, F, or S in the sixth column indicate whether the well was rising, falling, or steady.

The time at which the well commences to rise in the evening is fairly constant, about 8 P.M. This precludes the idea that the diurnal motion may be dependent upon the tides in the neighbouring bay, which is some two miles distant. The most irregular figures are those indicating the time at which the well commences to sink in the morning, which as summer approaches, when the city rises at an earlier hour, also tends to become earlier, and therefore assist in confirming the suggestion, that the rising and falling of the water are due to the facts that larger quantities of water are being used in the morning and evening than are being used during the middle of the day and the middle of the night.

In

The amount of these fluctuations has seldom exceeded 5 mm., and the day and night waves have about the same amplitudes. Professor Franklin H. King found that a heavily loaded train moving slowly past a well at a distance of 140 feet caused the water in the well to slightly rise, from which it might be inferred that the rising and falling of water in a well might be accompanied by a rising and sinking of the surrounding surface. If this were the case then during a day and night the horizontal pendulums in Tokio ought to show a double curve. some few instances there is a tendency to show such a double motion, as, for example, in fig. 5, Plate II. But because one of the curves is faint and because it is of rare occurrence, the 12-hour movement in the well is by no means sufficient to explain the daily wave indicated by the pendulums. It must, however, be remarked that the period of well observations coincides with a period when daily curves were not well marked, and what happened in the well when they were distinctly marked I have at present no means of ascertaining.