16 
r^IESSRS. C. T. IIEYCOCK AND F. H. NEVILLE ON 
from an original centre of crystallisation a uniformly oriented scaffolding may stretch 
out in three dimensions until it meets similar structures starting from other centres. 
The figures we give of an unchilled ingot of Sn 1, of the chill of Sn 6 at 966°, and of 
Sn 12 at 805°, all illustrate tliis process. If the centres of crystallisation were 
uniformly scattered through the liquid, and if the velocity of crystallisation were 
everywhere the same, we could predict the shape of the outer hoiuidary of a crystal 
skeleton, Imt, as a rule, on account of unequal velocities of crystallisation, due to 
local temperature differences and other causes, the skeletons vary very much in size 
and shape. If the sul)stance is a pure or nearly pure metal, or an alloy that 
solidifies to a uniform solid solution, the skeletons become solid bv the thickenino- of 
the Ijars and the growth of new orders of twigs, and in the wholly solid ingot when 
cut and polished each skeleton is found to have given rise to a grain. The sections 
of the grains are irregular polygons, and, although the original skeleton can often no 
longer be detected in the polygon, yet a uniform orientation persists throughout a 
polygon and often causes it, wdieu etched, to reflect light at a particular angle. Thus 
Avhen parallel light falls on the surface, some polygons will appear dark and others 
bright, but each polygon will have a uniform degree of brightness throughout its 
area. The chill of Sn 2 at 957° (Plate 1, fig. 3 a) shows skeletons wdiich have grown 
until they almost fill the whole alloy. The same surface (fig. 3) when illuminated 
l)y obli(|ue light, shows that there are several grains differently oriented. 
If the substance contains an impurity, which is not isomorphous with the material 
first solidifying, this impurity will be principally found as an envelope round each 
grain, and in a section of an ingot it wull appear as a slender network round the 
polygons. Some of this mother-substance must, however, from the peculiar skeletal 
manner of growth of the grains, exist in minute particles enclosed in the grains, 
though it may not always be possible to detect it. If there is very little impurity it 
may not form a complete network round the polygons, but only be found in isolated 
patches where three polygons meet. The chill of Sn 28 (fig. 73), and tlie unchilled 
Sn 6 (fig. 12), are good examples of this. If there is a good deal of impurity, so that 
the suljstance first crystallising ceases to form before the metal is solid, and if the 
remainder of the li(juid produces an entirely different solid, then the original skeletons 
never fill up, and can be detected in the polished and etched sections. They then 
appear as fern-leaf luarkings, as gridirons, or as combs, distinguished by colour or by 
texture from the mother-substance that solidified at a later stage. The chill of Sn 9 
at 777° (Plate 2, fig. 15) is a good examjfie of the phenomenon, tlie pale skeletons 
l)eing composed of a copper-rich substance Avhich. ceased to form some time before the 
whole mass was solid. It is surrounded by, and sharply divided from, the tin-rich 
matter, dark in the figure, Avhich solidified later. The existence of such combs in an 
alloy often enables us to form a correct inference as to the material that solidified 
first, but we shall see in the course of this paper that it is not always safe to assume 
that the material of the fern-leaf or comb pattern was the first to solidify. 
