S77 
iron arising from its rotation. 
made observations at every lo'* degrees of latitude on each 
side of the south pole. Before, however, giving these obser- 
vations, it is necessary that I should state the kind of reliance 
I place on them as forming a complete set. In order to make 
the observations near the pole, it was necessary to adjust the 
instrument as in fig. 3. and after having made the complete 
set, I suspected that, in the change from the one adjustment 
to the other, the centre of the plate had been nearer to that 
of the needle in making the observations near the equator, 
than those near the pole ; and that consequently, the deviations 
due to rotation in the former case, were proportionally too 
great. I was confirmed in this suspicion on comparing these 
observations with those which I had, in the first instance, 
made in lat o® and in lat. 90°; and still further on comparing 
them with others, which I subsequently made at the several 
distances 15, 17, 19, 20 inches; in the corresponding situ- 
ations. For example, in my first observations, the deviations 
due to rotation in lat. 0°, long. 0°, and in lat. 0° long. 180° 
were 3° 10', and 3° 14*', giving a mean 3° 12' in lat o ; and in 
lat. 90 S, 1° 31^; when the centres of the plate and needle 
had been carefully adjusted to the same distance 13.2 inches, 
in the two cases ; whereas the corresponding deviations in 
the table are 3° 43 ' and 1° 29^' ; and, by subsequent obser- 
vations, I found the sum of the deviations at the distances 
15, 17, 19 and 20 inches to be in these two cases, 7° 20' and 
3° 32', to which 3° 12' and 1° 31' are very nearly propor- 
tional. As however these differences do not in the least 
affect the conclusions which I at the time drew from this set 
of observation, and they were all made immediately follow- 
each other,' I prefer giving them as a complete set for the 
