60 Professor Forbes's Experiments on 



the needle itself, which the improvements in question do not 

 affect. Until by a regular and long-continued series of ob- 

 servations, such as those likely to be undertaken at Green- 

 wich, magnetism shall be reduced to more of a science than it 

 is at present, we must beware of pretending to illusory accu- 

 racy in a traveller's detached experiments. Those about to 

 be detailed in this paper, will sufficiently indicate the degree 

 of comparability of observations made with Hansteen's instru- 

 ment, such as it is, and which is by far the best test of their 

 real value. It has certainly rather exceeded than fallen short 

 of my expectations. 



§ 1. Adjustments and Method of observing. 



6. Hansteen's instrument contains no provision for securing 

 the horizontality of the needle, which is of considerable im- 

 portance. The needles have, indeed, sliding collars of sus- 

 pension, which may be altered with change of dip, but the 

 box has no adjusting levels. I have always* used a small spirit 

 level for adjusting the bottom of the box, and then, as care- 

 fully as I could, made the needle hang parallel to it, but the 

 adjustment was troublesome and unsatisfactory. 



7. The needle being levelled and allowed to come to rest, 

 it was drawn out of its position of rest f, but always in a 

 horizontal plane, by the approach of a piece of iron or steel 

 (usually a penknife), and the process repeated until the semi- 

 arc of vibration exceeded 20°, if 300 vibrations were to be 

 observed, or 10° if 100 only were observed, as was more 

 usually the case. This last deviation from Professor Han- 

 steen's practice was not adopted without due consideration. 

 The large commencing arc necessary in order that the vibra- 

 tions might be distinguishable at the close of 300, increased 

 greatly the errors pointed out by Mr. Harris. Moreover, 

 there seemed less chance of error in combining several series 

 of 100 vibrations taken in succession, than in using a single 

 series of 300, which, from the time it occupies, is more liable 

 to be interrupted and rendered useless by a gust of wind, or 

 a momentary relaxation of the painful attention required to 

 be exerted by an unassisted observer. Besides, the mere error 

 of the observed time, depending on the eye and ear of the ob- 

 server, will not exceed even in 100 vibrations the uncertainty 

 arising from causes impossible to eliminate, — indeed falls 



* I cannot answer however, for two or three of the first observations 

 hereafter to be quoted. 



f As the torsion of the silk fibre must have some influence, it is not un- 

 important to remark, that the same thread which was adapted to the in- 

 strument in August 1832, has been used ever since. 



