THE STRUCTURE OF METALS 89 



equally in all directions, even when the temperature is maintained 

 as uniform as possible. The growth of a crystal in a mass of 

 fused metal does not, as a rule, resemble the growth of a crystal 

 of chrome alum in an undisturbed solution. Instead of a 

 perfect or almost perfect octahedron being formed by the 

 gradual addition of layer to layer, so that the shape is pre- 

 served as the crystal increases in size, the accretion of solid 

 metal takes place principally along certain axes, a skeleton 

 or crystallite being formed. This behaviour is characteristic of 

 metals. 



In the case of certain salts, according to O. Lehmann, 1 

 the first visible mass surrounding the nucleus may be an 

 octahedron ; during the subsequent growth the added matter is 

 not deposited uniformly over the surfaces of the octahedron but 

 becomes attached chiefly at the solid angles, so that the particle 

 becomes star-shaped. Further growth at the now sharpened 

 angles accentuates the difference from an octahedron, the form 

 of which soon disappears, its place being taken by needle-like 

 prolongations of the axes. The effect has been satisfactorily 

 explained by Lehmann, in the cases examined by him, as being 

 due to the rate of growth exceeding that at which the super- 

 saturation or undercooling in the immediate neighbourhood 

 can be equalised by diffusion or convection ; it is not clear that 

 the same explanation will serve for slowly crystallising molten 

 metals. Whatever the cause may be, the skeletal mode of 

 growth is more frequent in metals than the normal mode of 

 accretion by successive layers. 



Before the prolongations of the axes attain to any great 

 length, secondary axes make their appearance, in the form 

 of transverse growths parallel with the other axes of the original 

 crystalline particle ; these are followed in turn by tertiary 

 axes and others of a higher order. The skeleton therefore 

 becomes more complex and more closely packed, approaching 

 more and more nearly to a compact mass. Given a sufficient 

 supply of liquid metal, the process of " filling up " continues 

 until the numerous axial growths are in perfect contact and 

 the mode of formation of the crystallite has ceased to be 

 apparent. If, however, the supply of liquid be restricted or 

 if the closing up of the outer parts of the crystallite be complete 

 before the inner part is solid, cavities may be left which afford 



1 Molekularfihysik, i. 326 (Leipzig, 1884). 



