444 FIRST REPORT UPON MAGNETIC WORK

carefully taken out every night at the hotel. The next morning be-
fore making the observations, if there was no wind stirring, the copper
weight was suspended, the are of vibration bisected and the torsion
removed by means of the torsion head; the amount of torsion developed
in the single fibre over night and while preparing for the observations
was such a trifle that it could be easily removed in this way when it
was deemed worth while. If the weather was windy, no further
attempt was made to remove the torsion beyond what had been done
the night before at the hotel. And so likewise when proceeding from
one station in the morning to another station in the afternoon, I gen-
erally made no attempt to remove the torsion developed since the
morning observation. From a series of experiments made at the base
station, extending over a week in all kinds of weather, sunshine and
heavy rains, with the magnet suspended by the single fibre all day
within the observing tent, I have convinced myself that the method
pursued with regard to the torsion did not introduce a larger error
than a fraction of a minute.

Sometimes it must be confessed, however, that the single fibre
proved disadvantageous in that it would occasionally break—on the
average about once for every six stations occupied. ¹ In such a case
the torsion of the new fibre, previously soaked in glycerine, had to be
removed in the field. In this event the other observations, e. g.,
magnetic inclination, would be proceeded with while the torsion was
being taking out. To provide for such emergencies, one part of the
work was usually kept in reserve to be carried out in case of accident
to the suspension fibre. In times of strong wind, however, it was
next to impossible to wholly remove all the torsion from a new fibre.
The observations were then proceeded with and on returning to the
hotel the amount of torsion in the fibre determined. Such accidents
might also occur, to be sure, not so frequently, with two fibres, and
in this case the uneliminated torsion would introduce a greater error

¹ It would undoubtedly prove of great advantage to have two glass tubes
with a fibre in each. It would be a simple matter to arrange the method of
fastening of the tube to the suspension box so that, in case of breakage of
a fibre, the tube could be readily replaced by the second tube with a fibre
whose torsion had already been largely removed.