10 EEPORT ON THE COMPLETION OF 



the threads thus constructed must have performed very indifferently ; for his obser- 

 vations, which were made at the same time in the Paris Observatory, and in the 

 caves below, shew differences and variations in the monthly means which are not 

 explicable by anything within our later, and, as we may suppose, more accurate 

 experience. Since Cassini's time, the improvement of the suspension-thread 

 seems to have made very little progress. M. Kupffer, apparently despairing 

 of satisfactory results from a silk suspension, substituted silver wires in the 

 Russian declinometers ; a similar suspension has also been employed by M. 

 QuETELET at Brussels. This seems to me a step towards the cap and pivot sus- 

 pension. Indeed, M. Nervander of Helsingfors has found that such suspension 

 cannot be trusted, since the wires are so affected by temperature that, when an 

 unmagnetic bar is suspended, it has a considerable diurnal motion : a fact which 

 I had suspected, and had pointed out as a probable source of error in determining 

 the temperature coefiicient for the bifilar magnet. 



The suspension-thread acts in the following manner : — As the thread is com- 

 posed of a series of fibres more or less twisted, the plane of detorsion, that is, the 

 vertical plane in which an unmagnetic bar will rest when suspended, is deter- 

 mined by the composition of series of opposing forces : if the torsion of the indi- 

 vidual fibres be at all considerable, very small motions of the magnet will cause 

 them to occupy slightly different positions inter se, or moderate changes of humi- 

 dity acting to a greater extent upon the external than the internal fibres, and 

 upon some of the external fibres than upon others, will change the plane of 

 equilibrium, and in this way force the magnet from its true position. Changes of 

 the plane of detorsion caused in this way, and by the occasional breaking of 

 fibres, will explain the great discrepancies occurring in large series of observations, 

 and the consequent lessening of their value. The importance also of obtaining 

 properl}"- constructed suspension-threads will be at once evident, when it is remem- 

 bered, that unless such are obtained, the labour of years may be rendered of little 

 or no value. The conditions sought to be obtained in the construction of the thread 

 for the Makerstoun declinometer, were the following : — First, The absence of all 

 torsion from the fibre. The so-called u?itmisted compound fibre from the cocoon, 

 usually emploj^ed in observatories, receives a considerable twist in the operations 

 of drawing from the cocoon and reeling, as may easily be perceived by passing it 

 between the thumb-nail and index, the method which I employed to remove the 

 twist. Second, That each fibre when combined into the thread, shall bear an equal 

 portion of the weight. For this end the fibre was not cut into pieces, but a sufficient 

 length being obtained, free of flaws, it was wound round two smooth pins, placed 

 at the required distance, so that no twist should be introduced in the act of winding ; 

 when a sufficient thickness was obtained, and the ends were tied, a hook attached 

 to a weight was inserted in place of the lower pin ; the thread being formed of one 

 continuous fibre was thus free to move round the upper pin and the weight-hook 



