318 
Proceedings of the Royal Society of Edinburgh. [Sess. 
a greater regularity of arrangement. The increase in the size of the crystal 
grains, or aggregates of uniform orientation, is more clearly shown when the 
pieces are etched more vigorously, though such treatment gives specimens 
only suitable for low-power examination ; some of the grains appear dark 
and others bright, and the limits of most of them are well defined. Figs. 
6 to 9 were obtained by more deeply etching the specimens represented by 
figs. 2 to 5 respectively ; the average size of crystal grain, as determined 
from the bright areas of fig. 6, shows a continuous increase through figs. 7, 
8, and 9, the final linear dimensions, after three million revolutions, being 
5 or 6 times as great as in the original alloy. As regards the piece which 
experienced three million rotations, the scale of the structure varies a good 
deal over the prepared surface, and fig. 9 represents about the coarsest 
part of the area; the other three pieces figured were sensibly uniform. 
There is no doubt as to the increase in the average size of crystal grain 
throughout the series. 
Since the photographs illustrate four separate pieces of metal, it may be 
well asked whether the change of structure is real or only apparent — in 
other words, whether or not the structure of all four pieces was sensibly 
identical at the beginning. Unfortunately, the rattled pieces were not 
examined microscopically before rotating, but other pieces cut from the 
same bar showed approximately the same scale of structure as figs. 2 and 6. 
If the larger grains of the rotated pieces had been due to slower cooling of 
the metal in the corresponding portions of the original casting, one would 
have expected their skeleton structure to be coarser also. Now, an inspec- 
tion of figs. 2 to 5 shows that as the size of the skeletons increases, their 
branches become distinctly finer — a result which is not so astonishing if the 
growth has taken place within the solid metal. 
If the change of structure is real, it may be due to causes other than 
vibration — namely, distortion and rise of temperature caused by the blows. 
The rattled pieces of the copper-tin alloy gave but little evidence of dis- 
tortion except in the immediate neighbourhood of their external surfaces. 
On the contrary, the copper-zinc, in w r hich exfoliation was very marked, 
showed dark striations upon the primary skeletons even in the centre of 
the piece. The occurrence of distortion is known to be favourable to sub- 
sequent recrystallisation, and it may be that the growth noticed would 
have been less if deformation of the pieces had been prevented. But, on * 
the other hand, the crystals did not exhibit any noticeable growth when 
the pieces were not rotated. The period of vibration lasted for five months 
after the date of casting the specimens. A comparison of rattled and 
unrattled pieces was made at the end of that time, and again after three 
