ON SEISMOLOGICAL INVESTIGATIONS. 51 



satisfactorily. The diagram (Plate II.) shows the character of the photo- 

 graphs that are now being taken. The instrument is not well adapted for 

 the record of earthquake waves, but two small tremors are shown ; the 

 second one is not to be found on the ordinary earthquake film. 



Some difficulties have been introduced by the greater sensitiveness, 

 and some were made more apparent, but these will probably disappear 

 with greater experience. One difficulty was to determine the linear dis- 

 placement of the boom due to an angular tilt of the instrument, for the 

 smallest angular motion which it was possible to make with accuracy 

 moved the multiplying style off the scale. It seemed necessary to reduce 

 the sensitiveness by a known factor — that is, by increasing the distance 

 between the supports of the bifilar portion. There are objections to this 

 plan, and up to the present the results have been left in the form of the 

 actual measured displacement. Another difficulty arose from a long 

 slow movement of a very minute order in one direction, probably masked 

 in the less sensitive instrument, but now distinctly noticeable in a con- 

 tinued series of observations. To explain this creeping it may be men- 

 tioned that the whole seismometer is mounted on a slate slab on the top 

 of a drain-pipe, two feet in diameter. This form of stand was preferred 

 by Professor Milne, because it avoided the drying of mortar or cement, 

 which, in a brick-built pier, would take a very considerable time. The 

 observed creeping may be due to some motion of the stand or of the hill 

 on which the Observatory is built, akin to the annual variation in the 

 azimuthal error of the transit instrument. While the instrument has 

 been in use the temperature has been increasing. Observations in the 

 second half of the year may clear up this point. 



It must not, however, be overlooked that one possible cause for 

 this creeping may be found in the seasonal shift in the direction of 

 the north-south barometrical gradient, accompanied by a seasonal 

 change in the mean sea-level. In summer time the region of high 

 barometrical pressure lies to the north of Great Britain, whilst in 

 winter it lies considerably to the south. 



The amplitudes on the diagrams seem sufficiently large to warrant 

 an attempt to determine the tidal constants by means of harmonic 

 analysis in the same way that the records of a tidal gauge are used. 

 It may be said here that it was hoped originally to determine from the 

 residuals between the computed and observed curves the direct effect 

 of the moon's tide-generating force. At the present moment such an 

 inquiry is no doubt rendered difficult owing to the slow creeping of 

 the pendulum towards the north. The problem resembles that of trying 

 to find the height of the tides from readings on a scale that is continually 

 sinking into the ground, and at a rate which cannot be determined and 

 which may not be uniform. There are also other practical difficulties 

 connected with the winding of the clock, attending to the illumina- 

 tion, &c. It is by no means certain that after a disturbance the boom 

 returns to the position originally occupied with no greater error than 

 the small quantity sought. The discussion of the results, so far as they 

 have gone, is useful as emphasising these difficulties, and with that view 

 they are printed here. The observations from April 14 to April 28 

 seemed as free from objection as any that have been made, and. as a 



e2 



