102 REPORTS ON THE STATE OF SCIENCE. 



T]ie second group of times dealt with is the P3, or commencement of 

 the principal portion of the disturbance. These records group themselves 

 most satisfactorily along a straight line, representing a rate of trans- 

 mission of about 29-1 seconds per degree of arc. Professor Imam ura does 

 not record the time of maximum, but this would be later by about live or 

 six seconds per degree. Probably 35 seconds per degree represents very 

 closely the true rate of propagation. 



It must be remembered that only four out of the twenty-four earth- 

 quakes dealt with originated on, or so near, land that the place and time 

 of origin could be determined with accuracy by direct observation. The 

 other twenty originated under the sea, and their time and place of origin 

 had to be inferred, with a consequent liability to error in each case ; but 

 these errors probably compensate each other to a great extent, and, used 

 with judgment, it is believed that the table will prove useful. 



X. Diurnal Changes in Level at the Royal Alfred Observatory, Mauritius, 



By T. P. Claxton. 



A Milne seismograph for recording unfelt earth-tremors has been in 

 use at the Royal Alfred Observatory, Mauritius, since September 1898. 

 It was first mounted in a small wooden hut, 12 feet square by 18 feet 

 high, on a brick pillar built up from a concrete floor 9 inches thick ; but it 

 was found that with this mounting a very large diurnal inequality of level 

 was recorded, a daily range of 8'' being of frequent occurrence, and 

 during heavy rains the boom occasionally drifted off the sheet. 



A more substantial foundation was constructed in November 1899. 

 The pillar was removed, and a hole dug 6 feet deep by 4 feet square ; 

 this was filled up with 4 feet of concrete, and a tapering column built up 

 from the latter without touching the earth on any side. 



In addition to the wanderings of the boom due to unstable mounting, 

 the registers are vibrated at night and early juorning by tremors, which 

 appear to be due to radiation from the concrete pillar, the tremors being 

 most active during rapid cooling. In order therefore to decrease the 

 daily range of temperature within the hut in the month of March 1 900, 

 the latter was completely enclosed by a straw thatching at a distance of 

 3 feet from the walls and roof ; but this did not entirely destroy the 

 tremors, and the nocturnal radiation was further checked by means of a 

 lamp which, lighted before sunset and extinguished soon after sunrise, 

 has remedied this defect, but has unfortunately introduced another. 



For the study of the diurnal and secular change of level a second 

 pendulum, registering N.-S. tilting, was added to the instrument in 

 February 1904,^ and a preliminary discussion of its recoi-ds revealed a 

 well-marked tilt to south, commencing at the time of lighting the lamp. 



It is not easy to distinguish the effect of the lamp on the E.-W. boom, 

 as the turning-points occur at about the times of lighting and extinguish- 

 ing the lamp. The effect, if any, wojald be to acccelerate both morning and 

 afternoon turning-points, as the boom is at its most easterly position 

 when the lamp is lighted, and most westerly when the lamp is extinguished. 

 In the E.-W. records, as affected by the lamp, the afternoon turning-point 



' This pendulum registers on the same cylinder as the E.-W. pendulum, with 

 which its booms run parallel. The discs at the end of the booms have been reduced 

 in width to 15 mm. Silk threads stretched across the registering slit at every second 

 millimetre produce a fine scale on the paper, by means of which the hourly ordinates 

 are readily and accurately measured. 



