All the changes so far observed on the planet's disk are, I believe, capable of explanation either by errors of observation or by seasonal change. For, as is the case with the Earth, not only must vegetation produce different appearances according to the time of year, but its aspects would vary somewhat as between year and year. This seasonal variation would affect not only the visibility of any one canal at any particular time, but might easily produce apparent alterations of place; visibility of one canal, combined with visibility or invisibility in its neighbors, being competent to simulate any shift.
The Araxes is a case in point. On Schiaparelli's chart there is but one original Araxes and one great and only Phasis. But it turns out that these do not possess the land all to themselves. No less than five canals traversing the region, including the Phasis itself, were visible this year at Flagstaff, and I have no doubt there are plenty of others waiting to be discovered. These cross one another, at all sorts of angles. Unconscious combination of them is quite competent to give a turn to the Araxes one way or the other, and make it curved or straight at pleasure.
Unchangeable, apparently, in position, the canals are otherwise among the most changeable features of the Martian disk. From being