MAGNET WITH REVERSED TELESCOPE. 



49 



certain that this magnetic axis is parallel to the optical 

 axis of GDI The only thing that we can do is, so to 

 arrange observations in a different state of the magnet 

 that whatever error is produced by want of parallelism 

 in the first case shall be exactly reversed in the second 

 case ; and the mean of results will then be free -from 

 error. It is merely necessary to invert the magnet in 

 respect of the side on which it carries CD, as shewn in 

 Figure 19; if the optical axis of CD pointed too much 



Fig. 19. 



to the west of magnetic north in the first case, it now 

 points too much to the east of magnetic north in the 

 second case, or vice versd. 



