Oscillations of the Horizontal Needle. 7 
brations in a rare medium, and confining them within very small 
ares,—as small as possible, so that the are be consistent with accu- 
racy of observation, (as shewn in Table III, observations on the 
Shade, p. 22) ; with a magnet about 5 inches in length, and about 
1th of an inch square, suspended from one of its edges, I have 
succeeded in obtaining between 200 and 300 vibrations, before 
the arc became reduced from 5° to 3°; the time of 100 vibrations 
deduced from 200 vibrations, and estimated from the Oth to the 
100th vibration, from the 10th to the 110th, and so on, up to 
the 200th vibration, according to the method pursued by Profes- 
sor Hansteen, being upon each 100 vibrations exactly the same. 
Indeed, when such irregularities as those already stated (a) (b) 
are considered, and which almost always happen to a greater or 
less extent, when the magnetic oscillations are observed in air, 
and taken in large ares, it is not unreasonable to infer that ano- 
malies and slight differences of time may appear on a great num- 
ber of vibrations, notwithstanding the theoretical corrections pro- 
posed. to neutralize them. It may not, therefore, be safe to attri- 
bute the cause of minute differences in time to variations in the 
terrestrial intensity, before the instruments of research are made 
so perfect as to be altogether without the limits of the above 
mentioned causes of error. 
12. The mode of suspension and other mechanical conditions 
connected with an oscillating bar, are not less worthy of conside- 
ration than the disturbing influence of the air. 
In all experiments with the horizontal needle, it is evident 
that we do not measure actually the magnetic intensity of a given 
place, but only one of its resolved portions, except the place 
should, be directly in the magnetic equator. It is therefore of 
consequence to the experiment, to preserve the bar in the same 
relative position. This we suppose to be horizontal, butsince 
every freely suspended magnet inclines more or less in various 
places, it becomes necessary either to measure accurately the angle 
of inclination, or otherwise correct it in some way. The latter 
