L54 SMITHSONIAN CONTEIBtrTIONS TO KNOWLEDGE VOL. 21 



years of development of these models had taughl the very valuable lesson that 

 in work of this kind where we have no margin on anything, but everything has 

 to be calculated on the " knife-edge " basis, it is an exceedingly unwise thing to 

 introduce any modification from what has been proved to be satisfactory, unless 

 such modification is absolutely necessary. 



The principal object in building the one-eighth size model of the large aero- 

 drome, as mentioned in the first part of this chapter, was to determine by act- 

 ual experiment whether the new form of " underneath " launching apparatus, 

 which had just been designed, was likely to prove as satisfactory as the original 

 " overhead " type, which had been used in the successful flights of the models in 

 1890. Yel after it was completed this aerodrome was found so very strong and 

 stiff, even though roughly constructed by merely tying the joints of the tubing 

 together with wires and soldering over the joints, that it was decided to equip 

 it with power, if a suitable form of power could be found which could be easily 

 applied. Just at this time liquid air as a motive power was attracting consid- 

 erable attention all over the country, and attempts were made to procure a 

 small power plant for operation by liquid air. After devoting considerable time 

 to the matter it was found impossible to do anything with it just at that time, as 

 the liquified air could not be obtained in Washington, and one of the chief ex- 

 perimenters in New York, who had been given a commission to make certain ex- 

 periments at his plant, so continuously delayed beginning them that it was found 

 necessary to give up the idea. 



However, after the completion of the tests of the launching apparatus some 

 experiments were made in flying the model as a kite. For this purpose a mast 

 twenty fed high was constructed and so arranged that it could be mounted at 

 the center of a small power launch. The model aerodrome was flown by a cord 

 connected to it by a bridle, the cord passing over a swivel pulley on top of the 

 mast and down into the boat, whence it could be played out or hauled in as 

 occasion required. By heading the launch into the wind it was possible to se- 

 cure sufficient relative velocity to cause the model to support itself and a num- 

 ber of tests were made in this way. It was found that when the bridle was at- 

 tached at the point at which the propellers would deliver their thrust, had they 

 been in use and driven by power, the model flew exceedingly well, maintaining 

 its equilibrium even during very strong ^usts. Owing to the rolling produced 

 by waves from the large boats which were continually passing in the part of 

 I he river where these tests were made, the power launch was often in danger of 

 being upset by its tall mast ; and finally, when the tests were just reaching the 

 point where accurate information was being obtained on the balancing of the 

 model, a sudden rolling of the boat caused the mast to snap off while the model 

 was in the air. Before it could be picked up from the water a passing boat had 

 swamped it and it was lost iii the river. 



