APPENDIX I 
no other check been applied there would have been the probability 
of large errors in the longitudes. For the purpose of checldng the 
chronometers a number of observations of occultations were observed 
during the winter of 1915. An occupation is really the eclipse of a 
star by the moon. A number of such eclipses occur monthlj'^, and are 
tabulated in the " Nautical Almanac." From the data given there it 
is possible to compute the Greenwich time at wliich the phenomenon 
ought to occur for an observer situated at any place on the earth, 
provided his position is known within a few^ miles, which will 
always be the case. The time of disappearance of the star by the 
chronometer to be corrected is noted. The actual Greenwich time 
of the occurrence is calculated, and the error of the chronometer is 
thus determined. With ordinary care the chronometer error can 
be determined in this way to within a few seconds, which is accurate 
enough for purposes of navigation. The principal difficulties of this 
method lie in the fact that comparatively few occultations occur, and 
those whicli do occur are usually of stars of the fifth magnitude or 
lower. In the Antarctic, conditions for observing occultation are 
rather favourable during the winter, since fifth-magnitude stars can 
be seen with a small telescope at any time during the twenty-four 
hours if the sky is clear, and the moon is also often above the horizon 
for a large fraction of the time. In the summer, however, the method 
is quite impossible, since, for some months, stars are not to be seen. 
No chronometer check could be applied until June 1915. On June 24 
a series of four occultations were observed, and the results of the 
observations showed an error in longitude of a whole degree. In 
July, August, and September further occultations were observed, 
and a fairly reliable rate was worked out for the chronometers and 
watches. After the crushing of the ship on October 27, 1915, no 
further occultations were observed, but the calculated rates for the 
watches were employed, and the longitude deduced, using these rates 
on March 23, 1916, was only about 10^ of arcinerror, ju^dgingby the 
observations of Joinville Land made on that day. It is thus fairly 
certain that no large error can have been made in the determination 
of the position of the Endurance at any time during the drift, and her 
course can be taken as known with greater certainty than is usually 
the case in a voyage of such length. 
355 
