1. Outline: The flow of the siphon spillway is very intricate and affected by vacuous condition. It is an intresting problem to make vacuum rapidly in_siphon, i.e, to provoke the perfect siphonaoge with the lowest water-level of overflow. The hydraulic experiments of siphon spillway have practiced at the Masan Lake, Haenam, the province of Junnam and gained rational and economical results which cannot be odtained by calculation. 2. Experiment: The model scale was a half of that of the prototype. 3. Results and discussion: In the experiments of 14 runs, washing, overflow water level, outlet, water level of cushion, phenomenon of water flying and water pressure of every part, etc. were studied. a) The relation of overflow water-level at siphon inlet and deflector When the elevation of the deflector is 10. 65m and the form of deflector O.4m long is the hypotenuse of an $45^{circ}$ isosceles triangle, the over flow water-level is the minimum, i. e. the siphonage was excellent. There is no effect by the rising of overflow water-Ievel between 11.95m to 1O.65m of deflector elevation (in the first plan, it is 11.05m). But the overflow water-level rises remarkably in the outside region of the above limits. b) The relation of overflow water-level, the length of cushion and standard height of the base. The reduction of the length of cushion brings the rising of overflow waterlevel, and the rising of the standard height of the base brings the rising of overflow water level. For the long cushion length and low standard height of the base, it cannot be expeted to have the falling of overflow water-level. The most satisfactory data were obtained at 5.20m of the base standard height and 6.1m of the length of cushion. The first vah,le planned was 5.70 m and 4.30m.