280 Influence of Bending Needles on Magnetic Dip. 



advantage in retaining the longer ones. The bending might 

 be diminished to an almost inappreciable amount by having 

 the needles about three times as wide, but it would then 

 become more difficult to magnetize them strongly. If the 

 magnetic moments of the needles were measured before every 

 observation, the error due to bending could be calculated to a 

 sufficient degree of accuracy, as Young's modulus will not 

 differ much in different specimens of steel used for magnets. 

 This course would, however, entail a good deal of trouble, 

 and it almost seems to me as if an empirical correction would 

 answer the purpose almost equally well. We might, for 

 instance, determine the correction to be applied in any par- 

 ticular year with sufficient accuracy by finding the mean 

 difference of dip shown by the needles in a certain number, 

 say five, of preceding years. As the influence of bending 

 for the three-inch needles is negligible, and as the magnetic 

 properties of the needles, which are always magnetized in the 

 same way, will probably not alter during a course of ^Ye 

 years, a fairly accurate correction may be deduced. Thus, for 

 instance, during the five years 1882-86 the mean dip of the 

 needles in the order of their length was 67° 30' 30" ; 

 67° 30' 00" ; 67° 29' 35". We deduce a correction of 55" 

 for the longest needles, and one of 30" for the intermediate 

 ones. The observed dip in 1887 was 67° 27' 13" ; 67° 26' 20"; 

 67° 25' 45"; and, applying the correction, we find for the 

 three needles the now closely-agreeing values of 67° 27' 13", 

 67° 26' 50", 67° 26' 40". The mean of these three values, 

 viz, 67° 26" 54", will probably be as near the truth as the 

 errors of observation will allow. 



I have not brought this matter forward in any spirit of 

 criticism towards the Greenwich observations. The fact that 

 the small effect of bending shows itself persistently from year 

 to year to a nearly equal degree, is in itself a proof of the 

 excellency of the observations. The manner in which the 

 magnetic records are reduced, and the form in which they are 

 published by the Greenwich authorities, may well serve as a 

 model to other Observatories. But as there seems no doubt 

 that the accuracy which is aimed at in the dip-observation is 

 appreciably affected by the cause of error which Joule has 

 pointed out, it seems to me advisable to eliminate that error. 

 The exact way in which the correction is carried out may well 

 be left to the proper authorities. 



