531 ‘ 
way displaced from it. This was tested by fastening a small plane mirror 
to the side of the needle with white lead, and counterpoising it with 
a little lump of bees’ wax, so that when the needle was placed on its 
agate plates, the reflected image of one of Gauss’s scales could be ob- 
served by a telescope. The scale was so placed that a millimetre on it 
corresponded to a minute of arc, and thus 1 Oths of a minute could be easily 
read by estimation. These dispositions were made in the small obser- 
vatory of Trinity College for absolute determinations, in which tremors 
were wholly avoided, and it was possible to keep the instrument undis- 
turbed for several days. From the observations made there, it appeared 
that, with a needle rolling on plates of agate, we can follow the daily range 
with great precision, and observe occasional disturbances, of which many 
were seen from one to six or eight tenths of a minute; and that when the 
needle is artificially swung by the attraction of a key through ranges 
varying from three minutes to four degrees, it returns to its position 
certainly within the tenth of a minute, and probably within a very small 
fraction of the tenth of a minute. 
Hence it is plain, that if we can bring the apparatus for laying down 
and removing the needle to a state of mechanical perfection corresponding 
with that of the axis, we shall be able to use dip needles rolling on agate 
plates for many purposes: for which they have hitherto been supposed 
not capable of affording determinations of the required degree of accuracy, 
but for which they are otherwise eminently suited. 
It appears essential that the endlong adjustment should be given to 
the axis by a part of the apparatus distinct from the Ys. Perhaps it 
could be successfully made by a little edge of agate mounted so as to 
rise into a carefully turned V-shaped groove in the thick part of the axis. 
It also appears to me that the Ys should be brought to bear on the care- 
fully turned cylindrical ends of the axis, and mounted so that they could 
be easily adjusted to let down the two ends of the axis strictly at the 
same moment; andthat each Y should be formed of two plates of agate, 
worked to a rounded edge, and placed at an angle as obtuse as is consis- 
tent with insuring that the axis drop always to the bottom of the Y. 
This would very much assist the Ys to disengage themselves com- 
pletely and abruptly from the axis as soon as it comes in contact with the 
edges on which it is to roll. I should also prefer that the frame carry- 
ing the Ys should rise vertically, and not turn on a pivot, which com- 
municates mechanically, and always in the same direction, a rotation to 
the magnet in placing it on the agate plates. A swing, to be communi- 
cated afterwards by magnetic influence from without, would seem to be 
much better. Of course, attention should be given to the parts by which 
this vertical motion is to be given, to make it smooth and without 
shake. 
With attention to these details, some of which I have tested with 
very promising results, it seems likely that a dip circle might be made 
which should allow the needle to be removed from the instrument, and 
replaced, without risk of any error exceeding a few seconds. 
* 
