14 AKSEL S. STEEN. TERRESTRIAL MAGNETISM. [NORW. POL. EXP. 



It will be seen from this table, which contains 122 different determinations, 

 that the values found vary from 44'4' to 11*6', without its being possible 

 to find any decided change with time, if the values are plotted graphically. 

 We may therefore certainly assume that the error of collimation has in re- 

 ality remained constant all the time, and its most probable value will thus 

 be the mean of the 122 determinations, i. e. 30'!', which accords fairly 

 well with the value found in Hamburg in 1893. If we omit from the table 

 all the values whose difference from the mean is greater than KX, and 

 which are indicated by brackets [ ], this has no effect upon the result, as the 

 mean remains the same, 30'!'; and if the values obtained by the deter- 

 minations in Hamburg in 1893, and in Wilhelmshaven in 1897, be added, 

 + 30'9' and 37'7' respectively, the mean still remains unchanged. I have 

 therefore deemed it advisable to take this mean, 301', as the constant value 

 of the error of collimation. The sometimes great deviation shown by a few 

 of the figures in the table, may be easily explained by the mobility of the 

 needle owing to the low force of direction, and the magnetic disturbance which 

 constantly prevails in the polar regions. Captain SCOTT-HANSEN has informed 

 me that the needle was always oscillating more or less widely and quickly; 

 and if we moreover consider that the needle was not suspended by a thread, 

 but rested upon a pivot, the results here given may well be deemed as satis- 

 factory as it was possible, under the circumstances, to have them. 



THE SMALL NEEDLE. 



In the passage from point 7 in Dr. NEUMAYER'S manuscript quoted in 

 the introduction, a case is mentioned containing two reserve needles for de- 

 clination observations. Only one such needle accompanied the apparatus, and 

 it was moreover intended more especially to act as the deflected magnet in 

 deflection observations for the determination of the horizontal intensity. This 

 needle is a little shorter than the declination needle proper, having a length 

 of 70 mm. ; and it weighs 21*2 grammes. As it cannot be reversed, its error 

 of collimation was found during the determination of constants in Hamburg 

 in 1893, by several series of comparative observations with this and the true 

 declination needle (double needle), the result being that a declination deter- 

 mination with the small needle gave a declination that was 8.6' more easterly 

 than the true one. A correction of 8'6' must therefore be added to all 



