ON A QUAUTX r THl;KAl> CliAVITV HAI..V 



aiid the balance was brought back to the laboratory in the morning. The result of 

 the observations is shown in Plot 6. Without correcting for daily rate, the Sydney- 

 Hornsby differences of reading, as determined from these observations, are 13'5, 15*0, 

 12'0, 16'9, 16-4, 12'0 sextant minutes, and confirm the existence of a serious change 

 produced by travelling. We now know that the difference is 18'2, so that from one 

 of these observations the difference might be in error by 6'2 sextant minute* 



The only points of the instrument which we could think of as being affected by the 

 travelling, were : the end of the rosette spring to which one end of the thread is 

 attached, and the lever itself, the after-end only of which is clamped, leaving the 

 forward-end not incapable of vibration. Although the end of the rosette spring was 

 stayed by wires to four upright pieces of watch spring, it was still capable of some 

 vibration in a direction at right angles to the length of the thread. 



On October 29th, 1898, the inner framework was taken out. The arrester springs 

 were strengthened, so that the lever was held more firmly than before. The junctions 

 of the little wire stays of the rosette spring were paraffined at both ends to prevent 

 the slightest play in the links, and three fine glass hairs were attached with paraffin, 

 one end in the centre of the rosette spring and the other on one of the bars of the 

 framework, the three glass hairs lying in a plane at right angles to the line of the 

 thread. The point of the rosette spring may now be considered to have no freedom 

 of movement in a direction at right angles to the length of the thread. 



The instrument was re-mounted, and observations commenced on November 7th. 

 It was found that the temperature coefficient had increased, an increase being what 

 one would expect if the resilience of the spring system had increased, as it must have 

 done owing to the additional constraint imposed by the three glass hairs. 



Plot 7. 



fmonth? a 3 to n 12 a HO 16 



November 1090. 

 Headings reduced to 23" C. 



It was very soon seen that the changes due to travelling had been greatly 

 diminished, the change in the Sydney reading, after an observation at Hornsby on 

 November 9th, being only 2 '6 sextant minutes. It was found, however, that by daily 

 observation at Sydney, from November 10th to December 7th (those up to 

 November 15th being shown in Plot 7), that discrepancies amounting to 4 sextant 

 minutes on each side of the mean line now appeared in a most erratic way. This we 



vol.. < X..IIJ. A. '1 K 



