332 REPORT — 1851. 



or removed ; and also at noon and midnight, when the registering plates were 

 changed or reversed. It having been noticed on one or two occasions with 

 certainty, that a disturbance of the magnet occurred at the times of" chan- 

 ging the plates, it was conjectured that these anomalous motions might be due 

 to concussions generated in the apparatus by the necessary manipulations, 

 and that those at sunset and sunrise might arise from a similar cause. In 

 the end of April the instrument was fixed more securely to its supports. The 

 four brass columns which connected the lower slab to the upper were re- 

 moved, and the slab cemented to an additional massive corbel : the upper 

 slab was at the same time cemented to its supporting corbels. Care was 

 taken to preserve as nearly as possible the previous relative positions of the 

 different parts. The effect of this change was at once to remove the disturb- 

 ances which had occurred about noon and midnight; those at the placing 

 and removal of the lamp, however, still remained. Whilst making a more 

 careful trial, as to whether there was any magnetic action connected with the 

 lamp, the real cause of this particular class of disturbance occurred to me ; 

 and in a few minutes it was traced to the different positions of the iron bars 

 of the window-shutters, when these were closed or opened. It had, in fact, 

 been remarked previously, that the disturbances at sunrise and sunset oc- 

 curred always in opposite directions. This source of error having been dis- 

 covered, it was remedied by the immediate removal of all the shutter-bars to 

 the basement story. These anomalous dislocations in the curves were after 

 this time scarcely ever experienced. 



Other errors have exhibited themselves, the causes of which we have not 

 yet discovered ; one of these is, — a tendency of the magnet to change its 

 mean position from day to day, and always in one direction, showing an ap- 

 parent gradual diminution of the vertical force. This change has generally 

 been to an extent which quite precludes the idea of its being due, either to 

 a real magnetic change, or to a loss of magnetism in the needle. That the 

 latter is not the cause, we have only to refer to the two observations of the 

 time of oscillation in the horizontal plane given above : from these we see 

 that, considering the higher temperature at the time of the second observa- 

 tion, and the fact that an additional weight had in the interim been put on 

 the magnet, the loss of magnetism has been trifling. 



Another, and perhaps a more serious error is the inconstancy of the 

 height of the centre of gravity, as shown by variation of the time of vibra- 

 tion in a vertical plane. In nearly every case of adjustment during the six 

 months, it has been found that, after some days, the time of one vibration 

 has diminished to a very large extent. The following are a few of the cases 

 in which this has been shown : — 



April 1. At adjustment, the time of vibration was 20 sees. 



... 29. 28 days after, ... ... 11 ... 



May 5. At adjustment, ... ... 25 ... 



June 4. 30 days after, ... ... 13 ... 



... 5. At adjustment, ... ... 25 ... 



... 20. 15 days after, ... ... 17 ... 



... 26. At adjustment, ... ... 23^... 



July 11. 15 days after, ... ... 11 ... 



Again, the knife-edges having been in the interim re-ground : — 

 Aug. 14. At adjustment, the time of vibration was 23' 1 sees. 

 ... 25. 11 days after, ... ... 22-4 ... 



Sept. 9. 26 days after, ... ... 16-2 ... 



Very remarkable changes of this nature have been previously observed ia 



