Most of the canals conform apparently to the type of the Djihoun rather than to that of the Ganges. Careful consideration of them fails to find any increase or decrease of distance, between the pairs of the same canal at different times, which cannot be referred to errors inevitable to observation of such minute detail. In short, the double is made by the addition of a second line in a particular position and not by a growth out to it of a line coincident to begin with with the first. I have said that the average width between the two lines of the doubles was about 3°. It must not be supposed that this average width denotes anything more than an average; or, in other words, that it denotes anything in the nature of a norm. The remark is important in view of a suggestion which I have heard made that we have here a system based on fundamental Martian units, in which, or in multiples of which, the dimensions of the canals are implicitly expressed. Such, however, does not seem to be the case. In some instances, indeed, we have certain evidence to the contrary and that the width of the double is conditioned solely by antecedent place. The Phison and Euphrates offer a case in point. These two important arteries in duplicate leave, as we saw, from two carets in the Mare Icarium, the Portus Sigaei, held in common tenancy by both. Each pair then proceeds down the disk inclined at its own particular angle to the meridian in order to reach by a great circle course a certain spot; the Pseboas Lucus in one case, the Luci Ismenii in the other. As one of these angles is thirty-five degrees while the other is only three, they must, from the circumstances of their setting out, have not only different widths, but widths determinately different in advance, since each is, roughly speaking, foreshortened by the degree of divergence from the meridian. The one, therefore, must be about four degrees to the other’s something less than three and a half. This is what they actually are as determined by measurement from observation. That the calculated value agrees with that found from observation helps certify to a community of starting-points, but it completely does away with comprehensive design in the question of their widths. For if the one were so settled, the other could not be.