te completeness of Ptolemy's, but it is free from his enormous
errors, and it has all the advantage of science, however imperfect, over
brilliant guessing.

Of course, even in Ptolemy, this guess-work pure and simple only comes
in at intervals and does not so much affect the central and, for his
day, far more important tracts of the Old World, but we have yet to see
how, in the mediaeval period and under Arabic imagination, all geography
seemed likely to become an exercise of fancy.

The chief Greek descriptions of the world, we must clearly remember,
were before the mediaeval workers, Christian and Moslem, from the first;
these men took their choice, and the point is that they, and specially
the Arabs, chose with rare exceptions the last of these, the Ptolemaic
system, because it was the more ambitious, symmetrical, and pretty.

Let us trace for a moment the gradual development of this geographical
mythology.

Starting with the notion of the world as a disc, or a ball, the centre
of the universe, round which moved six celestial circles, of the
Meridian, the Equator, the Ecliptic, the two Tropics, and the Horizon,
the Arab philosophers on the side of the earth's surface worked out a
doctrine of a Cupola or Summit of the world, and on the side of the
heavens a pseudo-science of the Anoua or Settings of the Constellations,
connected with the twelve Pillars of the Zodiac and the twenty-eight
Mansions of the Moon.

With Arabic astrology we are not here concerned; it is only worth noting
in this connection as the possible source of early Christian knowledge
of the Southern Cross and other stars famous in the story of
exploration, such as Dante shows in the first canto of his _Purgatorio_.
But the geographical doctrines of Islam, compounded from the Hebrew
Pentateuch and the theoretical parts of Ptolemy, had a more immediate
and reactionary effect on knowledge. The symmetrical Greek divisions of
land into seven zones or climates; and of the world's surface,[10] into
three parts water and one part _terra firma_; the Indian fourfold
arrangement of "Romeland" and the East; the similar fourfold Chinese
partition of China, India, Persia, and Tartary: all these reappeared
confusedly in Arabic geography. From India and the Sanscrit "Lanka,"
they seem to have got their first start on the myth of Odjein, Aryn, or
Arim, "the World's Summit"; from Ptolemy the sacred number of 360
degrees of longitude was certainly derived, beautifu