III. 



EXPERIMENTAL INVESTIGATION. 

 A. APPARATUS AND METHODS. 



The first series of experiments was made with a model of the Fram on 

 the scale of 1 : 200 in a glass tank 120 cm. long, 25 cm. deep, and 15 cm. 

 broad. The results, however, seemed to show that on such a small scale the 

 influence of viscosity was rather too great, and especially that the length of 

 the tank was too small to allow the velocity of the model to become uniform. 

 In the summer of 1900 a new tank, 350 cm. long, was therefore made, and 

 in it, it was possible to experiment with a model of the Fram on the scale 

 of 1 : 100. From this time all experiments were made in this large tank. As 

 its construction required considerable forethought and care, and it has since 

 proved very suitable, I think it is well to begin the account of the experi- 

 mental arrangements with a description of this tank. 



The tank (Figs. 1—3, PI. I) is built up of five panes of plateglass (Fig. 1, g), 

 forming its bottom and four sides, which are mounted in a wooden frame 

 (Figs. 1 and 2). Its inside dimensions are 350 cm. in length and 40 cm. in 

 breadth and height. Owing to the considerable length, care was required to 

 prevent the long panes from breaking. To prevent warping, the beams of the 

 frame were therefore made up of well-seasoned spruce-fir, sawed into boards 

 and glued together in such a way that the structure of the whole was the 

 most symmetrical possible. To reduce the give of the tank under the weight 

 of the water, as much as possible, it was on Prof. V. Bjerknes' advice sup- 

 ported at eight points, the ends of four wooden springs, instead of at four 

 points only. Two of these springs ss', ss are seen in Fig. 2; the position of 



