NA rURE 



[January 2, 1908 



existence of tension in tlie substratum and the manner of 



final rupture were noted. 



Experiments were made on five models : — 



(a) Water gelatin model, Vyrnwy type, tested to rupture, 



;ind illustrating preliminary difficulties. 



Fig. ;.— Kuptur< 



ry of a Lirge jelly-dam of the Vyrnwy type. 



(6) Control dam, Vyrnwy type, very stiff jelly. 



(c) Dam of Vyrnwy type, moderately stiff jeliy. 



(d) Dam of Assuan type, moderately stiff jolly. 



(e) Dam of Assuan type, moderately stiff jelly, tested 

 by optical methods. 



The pressure was applied 

 to the face of the models 

 in three ways : — (i) By a 

 board pressing against the 

 face and loaded normally 

 at the centre of pressure 

 by means of a stirrup and 

 shot bucket. A rubber 

 tube was placed between 

 the board and model to 

 •distribute the load so as 

 to imitate water pressure 



(2) By a board as in (i) 

 without tubing, the jelly 

 itself distributing the load. 



(3) By an elastic water 

 bag resting on the front 

 of substratum and against 

 the face of the dam, the 

 sides of the bag being ■|o 

 rigid and independently j^P 

 supported. The last 

 method gives a true dis- 

 tribution on the face and s 

 substratum, and obviates ' 

 the necessity of . piling 

 vi^eights on the latter to 

 maintain equilibrium ; but \ 

 the pressure cannot be 

 ■varied relatively to the 

 density of the dam as in ^ 

 methods (i) and (2), nor c 

 can the dam be tested to 

 rupture. The strains 

 were, in fact, only , 

 measurable by delicate i 

 optical means. 



The Vyrnwy model , 

 tested to rupture, behaved 

 as shown in Fig. 2. The 4 7 o u ". 



initial cracking gradually Curves of slide on the horizontal sections 

 •extended across the entire D, E, F. 



front, being followed by 

 the tearing of the jelly 



■from the cement or of the cement itself at the base con- 

 nection. Then followed the third and last stages, the 

 dam finally toppling over the tail, which separated at the 

 ■vertical cracks. This experiment indicates weakness in the 



NO. 1992, VOL. 77] 



substratum, and tension across vertical sections at the 

 tail. It may be remarked that the failure occurred at 

 the section of discontinuity of contour, where the stresses 

 in any actual dam become distributed through the 

 surrounding earth in an indeterminate manner, and further 

 that the extent to which the connection of such a model 

 to a rigid base (though good experimentally) imitates the 

 bedding of the substratum of an actual dam is also un- 

 known. The existence of tension on vertical sections is 

 emphasised by the authors. 



Distribution of Shear. 



This was determined by measuring the angular displace- 

 ment of lines ruled on the ends of the models. It is 

 admitted that no great exactitude is possible, and the 

 following sources of possible error are indicated : — 



(o) Inaccuracy of ruling on a jelly face ; (b) ruling model 

 while resting on its side, giving error especially near the 

 contour ; (c) distortion due to elastic effect of weight of 

 vertical jelly ; (d) set due to weight of jelly, or to shrinkage 

 in course of time. The authors think that (a) and (b), 

 while sensible, are not very large, but that (li) is more 

 important, and not easy to determine. If the model is 

 photographed in a vertical position the change of the angles 

 from 90° may not represent the slide due to weight only, 

 since errors (a), (6), and {d) affect the result. Similarly, 

 the combined effect of weight and water pressure cannot 

 be found exactly ; but if the model is photographed on the 

 same occasion, with and without water pressure^, the errors 

 are largely eliminated, and the change of angle on the 

 two photographs gives the slide due to water pressure 

 alone. The mid-third rule is applied to the cases of 

 reservoir empty and reservoir full. Hence it must apply 

 to the water pressure alone, and the change of angle as 

 measured should give a parabolic curve if the linearity of 

 normal stress holds. 



Typical results for the Assuan dam model are shown 

 in Fig. 3 for the sections D, E, and F between verticals 

 b and 14. The dotted lines correspond to doubtful 

 measurements near the contour, which, it may be noted, 

 appears to be rather large at section F. There are indica- 

 tions of negative shear near the contours, but the liability 

 to error and uncertainty of measurement must be re- 

 membered in any atternpt to interpret 'the curves in this 

 region. There is no resemblance to a parabolic distribu- 

 tion. The curves of shear due to weight, and to weight 

 and pressure combined, shown in the memoir, are 



19 /a ly /e IS It IS /2 /I 10 9 a 7 6 c f i z 1 



Key figure to the network on the A'suan model. 



Fig. 3. 



extremely wavy in outline, and nothing short of a quartic 

 would in any way represent them between verticals 6 

 and 14 ; but bearing in mind the special liability to error 

 in these cases, the curves are not of the same interest as 



