No. 6.| INTRODUCTION. CHRONOMETERS. XXII 
azimuth. For the determination of the apparent altitudes the refraction corres- 
ponding to the meteorological conditions was of course used; while an error 
affecting the true and the apparent altitude alike has only an insensible effect 
on the calculation of the true distance (being multiplied by the Sine of the 
difference between true and apparent altitude) an error in the refraction or 
parallax would affect the true distance by a quantity of the same order as 
the error itself. 
The results are not satisfactory. In some cases when the observations 
have been taken with intervals of a few days or weeks, the chronometers are 
unanimous in protesting against the deduced Greenwich times. As the tempe- 
rature during these observations was only once (1896 April 22) as high as 
—16° C., and on all the other occasions between —27° and —43°, it is prob- 
able that the sextant was affected with errors that would not have been of 
great importance for ordinary altitudes of the Sun, but which proved fatal 
to the delicate operation of determining the longitude by Lunar Distances. It 
is also to be remarked that the index error was not determined on each occa- 
sion but for some time considered as constant, because a determination in 
August 1893 in the Barents Sea and another off the mouth of Lena shortly 
before the enclosure in the ice had given identical results, 
It would of course have been better to use the altazimuth for determining 
the difference of azimuth between the Moon and a star or the Sun, and thence 
deduce the Moon’s right ascension. But as it happened that the planet Jupiter 
was circumpolar during all the 3 years of enclosure in the ice and so was 
always at hand when the Sun was absent, it was found to be a much more 
ready means of getting an approximate longitude to observe the eclipses of 
Jupiter’s Satellites and compare with the predicted times in the Nautical Al- 
manac. 
The results of the Lunar Distances are included in a table below (Tab. ce) 
containing the results of these Eclipses. 
Eclipses of Jupiter's Satellites. 
The observed moment of commencement or end of an eclipse of a Satel- 
lite is dependent on many circumstances, the aperture of the telescope being 
perhaps the most important. As the predicted times are sometimes seriously 
