76 INSTRUMENTS AND RECORDS. 



7. On December 31, 1902, an important change was made in the Vertical-Force magnetograph. 

 That instrument is fitted with three auxiliary magnets. Two are thin bars, almost wires, one on either 

 side of the Vertical-Force magnet, with their centres near its level. The third is a short, much thicker 

 bar, in a brass piece which screws on to the base of the Vertical-Force box, at some distance under the 

 Vertical -Force magnet. These auxiliary magnets are intended apparently to assist in reducing the 

 temperature coefficient. The use of auxiliary magnets is one of which the wisdom is open to much doubt. 

 The results are of a somewhat complicated character, and any unrecognised change of moment or position 

 in an auxiliary magnet may cause error and confusion. The auxiliary magnets in the present case may be 

 described as of " soft " iron, but still they possess very appreciable " permanent " moments. In December, 

 1902, the temperature changes in the Antarctic were large, and Mr. BERNACCHI, noticing a large temperature 

 effect on the Vertical-Force trace, made a serious attempt to reduce it. After trying several less heroic 

 remedies, he finally removed the short magnet from its place below the magnet box and placed it above, 

 where it remained until the instruments were finally dismounted. 



When the tabulations were commenced at Kew, it was feared that nothing could be made of the Vertical- 

 Force records. There were occasions when the instrument was obviously out of action, while on many 

 occasions the Vertical-Force trace appeared so extraordinarily quiet compared to the others as to raise 

 suspicions. Eventually I decided to have two months' trace measured, selecting December, 1902, and 

 February, 1903, as months when the Vertical-Force changes seemed specially large and the information 

 as to the scale values most complete. Even a superficial inspection of the sheets at times of large 

 temperature change showed that the influence of temperature on the Vertical-Force trace must be 

 considerable. But having had the 2-hour readings of Antarctic temperature through my hands, I knew 

 that the regular diurnal variation of atmospheric temperature had a range of at most 3 or 4 F., and I 

 thence inferred wrongly, as it proved that inside the Magnetic Hut the regular diurnal inequality of 

 temperature would be so small even at Midsummer that its effect on the Vertical-Force diurnal inequality 

 might be neglected without risk of serious error. I realised, of course, that this neglect might prejudice 

 seriously individual curve readings, but for my immediate object that did not matter. Accordingly, 

 diurnal inequalities were got out for December, 1902, and February, 1903, from the Vertical-Force curves 

 uncorrected for temperature. The ranges appeared surprisingly large, and the inequality in the one 

 month was inverted as compared to the other. This pointed to something being wrong. On inspecting 

 the curves it was apparent that rise of temperature was associated with movement of the Vertical-Force 

 trace down the sheet in 1902, but movement up the sheet in 1903. The explanation that may appear 

 most natural -viz., that there had occurred an actual change in sign in the temperature coefficient 

 postulated conduct so contrary to my previous experience of magnetographs, that after reflection I 

 inclined to the view that the Vertical-Force magnet must somehow or other have got turned end for end 

 between December, 1902, and February, 1903. Minute inspection of the curves fixed the date of this 

 supposed occurrence as December 31, 1902. Investigation of the written records then showed that on this 

 afternoon Mr. BERNACCHI had made the alteration mentioned above in the position of the short magnet, 

 and as the alteration entailed a rebalancing of the magnet whose ends are closely alike, he agreed with 

 me in regarding the occurrence as at least a possible one. Experiment showed that the magnet worked 

 equally well whether its N. end were east or west. The natural way of settling the question was to refer 

 to the information on the sheets of scale-value determinations as to the position of the deflecting magnet, 

 whether N. pole up or S. pole up. Information on this point was given, however, only on some of the 

 sheets, and the different sheets for 1903 contradicted one another. When making the deflection 

 experiment, the observer had to hold the magnet close up to the ceiling in the dark, and had to rely on 

 his memory as to how he held it. Naturally it never occurred to him that circumstances might introduce 

 an uncertainty as to the sign even of the Vertical-Force change. 



The difficulty, of course, would not have arisen if temperature coefficients had been determined in the 

 Antarctic before and after changes of the instrument, but no determinations had been found practicable. 



8. Though I did not at first think it possible that differences so large as those presented by the diurnal 

 inequalities found for December, 1902, and February, 1903, could arise from temperature alone, inspection 

 of the curves led me to suspect that the regular diurnal inequality of temperature in the Magnetic Hut 



