This consisted in the fact that with curved surfaces, while at large angles the centre of pressure moves forward as the angle decreases, when a certain limit of angle is reached it travels suddenly backwards and causes the machine to dive. The Wrights had known of this tendency from Lilienthal's researches, but had imagined that the phenomenon would disappear if they used a fairly lightly cambered--or curved--surface with a very abrupt curve at the front. Having discovered what appeared to be the cause they surmounted the difficulty by 'trussing down' the camber of the wings, with the result that they at once got back to the old conditions of the previous year and could control the machine readily with small movements of the elevator, even being able to follow undulations in the ground. They still found, however, that the lift was not as great as it should have been; while the drag remained, as in the previous glider, surprisingly small. This threw doubt on previous figures as to wind resistance and pressure on curved surfaces; but at the same time confirmed (and this was a most important result) Lilienthal's previously questioned theory that at small angles the pressure on a curved surface instead of being normal, or at right angles to, the chord is in fact inclined in front of the perpendicular. The result of this is that the pressure actually tends to draw the machine forward into the wind--hence the small amount of drag, which had puzzled Wilbur and Orville Wright. Another lesson which was learnt from these first two years of experiment, was that where, as in a biplane, two surfaces are superposed one above the other, each of them has somewhat less lift than it would have if used alone. The experimenters were also still in doubt as to the efficiency of the warping method of controlling the lateral balance as it gave rise to certain phenomena which puzzled them, the machine turning towards the wing having the greater angle, which seemed also to touch the ground first, contrary to their expectations. Accordingly, on returning to Dayton towards the end of 1901, they set themselves to solve the various problems which had appeared and started on a lengthy series of experiments to check the previous figures as to wind resistance and lift of curved surfaces, besides setting themselves to grapple with the difficulty of lateral control. They accordingly constructed for themselves at their home in Dayton a wind tunnel 16 inches square by 6 feet long in which they measured the lift and 'drag' of more than two hundred miniature wings. In the course of these tests they for the first time produced comparative results of the lift of oblong and square surfaces, with the result that they re-discovered the importance of 'aspect ratio'--the ratio of length to breadth of planes. As a result, in the next year's glider the aspect ration of the wings was increased from the three to one of the earliest model to about six