III. On the uneliminated instrumental error in the observations of magnetic dip
A single reading of one end of a dipping-needle placed in a dip-circle provided with microscopes for observing is liable to a variety of instrumental errors, which are eliminated by taking the mean of the sixteen readings of the two ends in the eight different positions included in a complete observation. Nevertheless it is found that with the best modern instruments a mean value results from these sixteen observations different for each different needle, and that the difference between the results obtained with two different needles is not the same at all times. The irregularities in the values of the dip observed at Bombay with two needles of excellent character made by Barrow of London, led the author to investigate the effect of a hypothetical irregularity in the shape of the axle of the needle, such that a section of the axle by a plane perpendicular to its axis would be elliptical instead of circular in form. Another source of error, which was brought to the notice of the Royal Society many years ago in a paper published in the Proceedings, is the displacement of the centre of gravity of the needle from the centre of the axle, combined with inequality in the magnetization of the needle when the poles are direct and reversed. Experience has led the author to the conclusion that the usual method of magnetization, by a definite number of passes of the same pair of bar-magnets, communicates magnetism to the needle very unequally when the one end of the needle is made north and when the other end is made north. Consequently it is advisable to investigate the effects of ellipticity of the axle and of displacement of the centre of gravity at the same time, which the author proceeds to do.