VII 



OF THE EFFECTS OF SHRINKAGE 



563 



rayed star with rounded angles. The block has dried unequally; 

 its corners and its edges were naturally the first to dry*, and the 

 twelve dried and hardened edges began to play the part of the wire 

 frame in Plateau's soap-bubble experiment. The shrinking cube is 

 tending towards the identical configuration shewn in Fig. 212; it is 

 a minimal configuration, partially reahsed in a coarse material, but 

 reahsable to perfection in a film. 



A shrinking cylinder (as Plateau knew) she)vs various 

 phenomena, depending on its proportions. A low, squat cyhnder 

 begins to show a pulley-hke groove — a catenoid— around its periphery, 

 precisely like the soap-film between its two wire rings in Fig. 108 ; 

 and as the groove deepens, the plane surfaces of the cyhnder also begin 



Fig. 217. Shrinkage of cube and cylinder. 



to dimple in. They become spherical as they grow more concave, and 

 the deepening groove of the pulley passes from a catenoid to a nodoid 

 curve — so at least theory tells usf, for, beautiful as the experimental 

 configurations are, they hardly lend themselves to precise measure- 

 ments of curvature. But this shrunken cyhnder is now wonderfully 

 like the " amphicoelous " vertebra of a cartilaginous fish, the simplest 

 and most ''primitive" vertebra of all. A series of cracks, or sphts, 

 around the circular groove in the vertebra seem to be a final result 

 of irregular shrinkage, not shewn in the more homogeneous gelatine. 

 A long cylinder, or thread, of gelatine tends to become fluted, 

 with three or more ribs or folds, and it is in this way that threads 



* Just as, conversely, the prominent parts of a crystal tend to grow more 

 rapidly than the rest in a super-saturated solution, and to dissolve more rapidly 

 in one below saturation; cf. 0. Lehmann, Ueber das Wachstum der Krystalle, 

 Ztschr. f. Krystallogr. i, p. 453. 



t See p 369. 



