A slight change had also been made in the method employed of attaching the Pénaud tail to the machine. In 1896 the tail was attached to the machine by means of a flat piece of wood (hickory) which had been steamed and bent to the proper extent to cause the rudder to have a negative angle of about 5 degrees, but no accurate note was made of its angle or stiffness, so that in 1899 no data were available as to exactly what the angle had been or how stiff the spring was. Owing to the fact that wood not only warps and twists, but also that any piece which has been steamed and bent gradually loses a certain amount of its curvature, it was decided in 1898 to change this method of attaching the tail, the wooden spring being replaced by a coiled steel spring attached to an upper guy-post and connected to the tail by a bridle wire fastened to the center of figure of the tail. After the experiment of July 7, 1899, a lower spring, consisting of small rubber bands, was connected by a wire to the lower part of the rudder and fastened to the guy-post, thereby more nearly reproducing the conditions obtained when using a wooden spring, which, of course, tends to return the rudder to its normal position when it is displaced in either direction. After attaching this lower spring to the rudder, the experiment of July 14 was made, and it was found that the aerodrome still flew with its midrod pointed upward at a very steep angle. It was, therefore, felt certain that the upper spring on the rudder was too stiff, and that it should not require so much as 1240 grammes to bring it to the horizontal. This spring was, therefore, replaced by a weaker one, and the angle of the rudder was also decreased until it had a negative angle of only 5 degrees and required only 200 grammes placed at its center of figure to bring it to the horizontal. From the description of the flight of July 19, it will be seen that these changes immediately corrected the tendency of the aerodrome to point its nose upward at such a sharp angle, and it will be later seen that after a further slight adjustment the flight of July 29 was made, in which the proper balancing was obtained and the aerodrome made a good horizontal flight. After these preliminary tests with the “overhead” launching apparatus, it was dismounted and the “underneath” one substituted and the experiments of July 28, 29 and August 3 were made. Everything connected with this “underneath” launching apparatus worked perfectly from the start and four flights of the aerodromes were made using it. [Illustration: PL. 37 EXPERIMENTAL FORMS OF SUPERPOSED SURFACES, 1898, 1899 (SEE ALSO PLATES 64 AND 65)] [p153]