es not comprise an exact number of days, but
includes, as we have seen, one-third of a day in addition.

It will be granted, of course, that if the number of days was exact, the
_same_ parts of the earth would always be brought round by the axial
rotation _to front the sun_ at the moment of the recurrence of the
eclipse. But as there is still one-third of a day to complete the saros
period, the earth has yet to make one-third of a rotation upon its axis
before the eclipse takes place. Thus at every recurrence the track of
totality finds itself placed one-third of the earth's circumference to
the _westward_. Three of the recurrences will, of course, complete the
circuit of the globe; and so the fourth recurrence will duplicate the
one which preceded it, three saros returns, or 54 years and 1 month
before. This duplication, as we have already seen, will, however, be
situated in a latitude to the south or north of its predecessor,
according as the eclipse series is progressing in a southerly or
northerly direction.

Lastly, every eclipse series, after working its way across the earth,
will return again to go through the same process after some 12,000
years; so that, at the end of that great lapse of time, the entire
"life" of every eclipse should repeat itself, provided that the
conditions of the solar system have not altered appreciably during the
interval.

We are now in a position to consider this gradual southerly or
northerly progress of eclipse recurrences in its application to the case
of eclipses of the moon. It should be evident that, just as in solar
eclipses the lunar shadow is lowered or raised (as the case may be) each
time it strikes the terrestrial surface, so in lunar eclipses will the
body of the moon shift its place at each recurrence relatively to the
position of the earth's shadow. Every lunar eclipse, therefore, will
commence on our satellite's disc as a partial eclipse at the northern or
southern extremity, as the case may be. Let us take, as an example, an
imaginary series of eclipses of the moon progressing from north to
south. At each recurrence the partial phase will grow greater, its
boundary encroaching more and more to the southward, until eventually
the whole disc is enveloped by the shadow, and the eclipse becomes
total. It will then repeat itself as total during a number of
recurrences, until the entire breadth of the shadow has been passed
through, and the northern edge of the moon at le