52 MESSRS. C. T, HEYCOCK AMD E. H. XEYILLE OX 
expected to find evidence of the ground having been transformed into S, but this has 
occurred, and here and tliere are copper-rich gaps in the massive S of the ground, in 
which the S fern-leaf is seen growing out from the edge of the more massive S. One 
of these gaps is visible in the figure. This ingot of Sn 21 would therefore appear to 
show the real nature of the ground as S. In addition to the quite black bands of 77 , 
there are, in places, paler bands. These may owe their paleness either to a different 
orientation or to a somewhat smaller content of tin. 
Sn 21 . 0/till at 588°, not specially slowly cooled (not reproduced). 
This shows, perhaps, a little less of the bars, hut it has some paler patches in 
addition to the black bars, also one or two small spots of S granulation or fern-leaf. 
The bars are well seen to have the orientation of the grain of solid solution from 
which they were formed. There are also places where white S bars crystallised on a 
darker ground. 
Sn 21 . Chill at 580° (fig. 57). 
This ingot, chilled on the D' line, was not a slow-cooled ingot before chilling. 
When etched with FeCl 3 , it has a larger fraction of the surface covered with bands 
than fig. 56, but these vary in tint from black to the palest grey. The bars can, in 
some cases, be made to alter a little in shade by tilting and rotation, but not enough 
to make it possible to explain their very varied appearance as cine to differences in 
orientation of a pure material such as SnCu 3 . Moreover, considering the percentage 
of tin in the alloy, the bars are far too abundant for sncli a hypothesis. It seems 
more probalile that when the temperature of the transformation curve was reached 
at G0(J°, the solid alloy began to separate into two, oi-, through imperfect eciuilibriiim, 
into several phases with different percentages of tin, all the more tin-rich phases 
re-crystallising in a form isomorphous with 77 , and therefore all to be classified as 77 . 
An analogy for this may, perhaps, be found in the breaking-up of a liquid into 
conjugvites during rapid cooling. In such a case a whole series of lic|uid phases must 
through imperfect transformation exist together, and if a process analogous to 
chilling could be effected they would be stereotyped as we see them here. In the 
whitest parts, which in the FeCl 3 etch are the most co]iper-rich, the S crystallisation 
has taken place. The bars, which are really sections of plates, have definite 
crystalline shape, and present many good examples of twinning. As Mr. Feaenstdes 
has pointed out to us, it resembles the twinning seen 111 soda felspars, but the fact 
that the bars darken at different rates when heat-oxidised, is some indication that 
they are not merely twins in a homogeneous substance. 
Sn 21 . Sloiv-cooled chill at 570° (not reproduced). 
This has a decidedly smaller area of the new crystals than the chill at 590 , and the 
