Seasonal changes seemed the only thing to account for the phenomena. And in a general sense this was undoubtedly the explanation. To learn more about the matter, to verify it if it existed, and to particularize it if possible, I determined to undertake an investigation permitting of quantitative precision in the case. A method of doing this occurred to me which would yield results deserving of consideration from the amount of data upon which each was based and capable of being compared with one another upon an equal footing from which relative information could be derived. It seemed wise to determine from the drawings the degree of visibility of a given canal at different seasons of the Martian year, and then to do this for every important canal during the same period of time. The great number of the drawings suggested this use to which they might be put. For from a great accumulation of data a set of statistics on the subject could be secured in which accident or bias would be largely eliminated and the telling effect of averages make itself felt. To render this possible it was necessary that the drawings should be alike numerous, consecutive, and extended in time. These conditions were fulfilled by the drawings made by me at the opposition of 1903. Three hundred and seventy-two drawings had then been secured, and they covered between them a period of six months and a half. They were also as consecutive as it was possible to secure. During a part of the period the planet was seen and drawn at every twenty-four hours, from April 5, namely, to May 26, or for forty-six consecutive days. Though the rest of the time did not equal this perfection, no great gap occurred, and one hundred and forty-three nights were utilized in all. Furthermore, as these drawings were all made by one man, the personal equation of the observer—a very important source of deviation where drawings are to be compared—was eliminated. But even this does not give an idea of the mass of the data. For by the method employed about 100 drawings were used in the case of each canal, and as 109 canals were examined this gave 10,900 separate determinations upon which the ultimate result depended. That each of these determinations was independent of the others will appear from a description of the method itself on which the investigation was conducted. To understand that method one must begin a little way back.