112 
RECENT EXPERIMENTS ON THE CRYSTAL- 
LISATION OF MINERALS. 
LTHOUGH the crystallisation of alloys and of 
minerals must in its nature be essentially similar to 
that of the more ordinary solutions handled in the labor- 
atory, the ranges of temperature and pressure involved are 
so far different as to make any experimental study a 
matter of considerable difficulty. In the case of the 
metallic alloys, the difficulties incident on the production 
and measurement of high temperatures have in recent 
years been overcome by the use of platinum-resistance 
thermometers, as in the investigation of the copper-tin 
alloys by Heycock and Neville, or by the use of thermal- 
junctions of platinum with a platinum alloy, as used so 
effectively by Roberts-Austen and his colleagues in the 
work of the Alloys Research Committee. As a result of 
these investigations, the conditions under which the 
different constituents separate from a liquid alloy, and the 
changes which occur as the solid ingot cools, are as fully 
known as the conditions which determine the separation 
of ice or salt from an aqueous salt solution. 
The study of the crystallisation of an igneous mineral 
from a liquid magma has proved to be a task of very 
much greater difficulty. The temperatures of crystallisa- 
tion are much higher, and frequently lie above the melting 
temperature of platinum; the minerals to be examined are 
not easily obtained in a pure state; they are poor con- 
ductors of heat and—perhaps the most serious difficulty 
of all—many of the minerals are so viscous when first 
melted that several minutes elapse before even the corners 
of the crystals become rounded; conversely, the melted 
materials often cool to a glassy mass, and only reluctantly 
develop a crystalline structure. Difficulties such as these 
render almost inoperative the methods that have proved 
so effective in the study of metallic alloys, but new 
weapons have been provided by the perfecting of the 
radiation pyrometer as an exact method for the measure- 
ment of high temperatures, and by the commercial pro- 
duction of iridium melting at a temperature at least 600° 
above the melting point of platinum. 
A quantitative study of the crystallisation of the lime- 
silica series of minerals has recently been published by 
Messrs. A. L. Day and Shepherd, of the Geophysical 
Laboratory of the Carnegie Institution of Washington 
(Journ. Amer. Chem. Soc., xxviii., pp. 1089-1114, 
September, 1906). The results they have obtained are so 
far in advance of anything that has previously been accom- 
plished as to mark the opening of a new period in the 
development of experimental mineralogy. 
Dealing first with the two pure substances from which 
this series of minerals is derived, it may be noted that 
lime melts at so high a temperature that it is not yet 
possible to make a satisfactory determination of the melt- 
ing point; measurements can only be made with mixtures 
containing at least 20 per cent. of silica, and even these 
melt at temperatures ranging from 1400° to well over 
2000° C. The melting point of silica lies below that of 
platinum, but the melting is so slow that when a charge 
of quartz was heated in an iridium crucible in an iridium 
tube-furnace to the melting point of platinum (1709°) the 
grains did not coalesce, although they became tightly 
sintered together. Incipient melting could, however, be 
detected at a temperature nearly 100° lower, and the melt- 
ing point is fixed by the authors at 1600° C. 
Silica is a dimorphous compound, the two mineral 
varieties being known as quartz and tridymite. At 
temperatures above 1000° both quartz and amorphous 
silica change to tridymite. This is, therefore, the form 
which is stable at the melting point, and the melting 
temperature of silica is thus properly the melting tempera- 
ture of tridymite, and not of quartz, as is commonly de- 
scribed. Occasionally by rapid heating quartz can be 
partially melted without inverting to tridymite, but it would 
hardly be possible by any known method to determine a 
separate melting point for unchanged quartz. 
The converse change from tridymite to quartz is less 
easily observed. In presence of a catalyst, such as sodium 
tungstate, vanadic acid, or a mixture of potassium and 
lithium chlorides, amorphous silica was found to crystallise 
to quartz below 760°, and to tridymite above 800°; by 
NO. 1935, VOL. 75 | 
NAROT OE 
[ NOVEMBER 29, 1906 
heating for five or six days the direct change of quartz to 
tridymite was proved at 800°, and from tridymite to quartz 
at 750°. The change is therefore reversible, and there is 
a true inversion point at about 800° C, 
The melting-point curve for mixtures of lime and silica 
was explored by heating mixtures of definite composition, 
well mixed by grinding and repeated melting, on a 
platinum or (for higher temperatures) iridium strip, and 
noting the order of fusion. In this way two maxima and 
three minima were found, and these were subsequently 
investigated in such a way as to determine the exact com- 
position and temperature at which each occurs. The 
maxima at 48 per cent. and 65 per cent. CaO correspond 
with the composition of the metasilicate CaSiO, and the 
orthosilicate CaSiO,, but no indication could be obtained 
of the separation from the melt of the compounds 
2CaO.SiO, or 3CaO.SiO,, or of the silicate 4CaO.3SiO, 
analogous to the mineral akermanite. 
Both the metasilicate and the orthosilicate are poly- 
° o oz oF ov os 03 on oe oc oo: Sors 
O01 0s og OL og os ob oe oz 1 0 of5 
9,009 
Of5 + to'ssD 
Vs $19 "o's's> {+g do foisvog + ziuwno 
9,008 © 21uvnd Oo. 
BAIWAGIad 40 NOISUAANIS | OOF 
fgiseog + aiiwacina 
0008 
o°9n+ *o1s*ea-g 
OOTe 
ee owes fp 
40 NOISHSANI 
“ols @>0 % B4twAdion 
2oIt) to 1s*e3 do 
Gone 
(40 NoIsHaANi DoLiel id1493910a 
oo98 
oO 8D "016 785 40 
o0G* 
ooze 
ore 
Fic. tr. 
morphous. The metasilicate crystallises at 1512° in a 
pseudo-hexagonal form, and inverts at 1200° to a form 
identical with the mineral wollastonite. The orthosilicate 
crystallises at 2080° in a monoclinic a-form of density 
3:27, and inverts at 1410° to an orthorhombic B-form of 
density 3-28, and again at 675° to a monoclinic y-form of 
density 2-97.' | The latter change involves an expansion of 
1o per cent. in the volume of the substance, and is thus 
responsible for the disintegration or ‘‘ dusting’’ of the 
orthosilicate and all mixtures containing more than 51 per 
1 It is unfortunate that the authors have reversed the convention which 
obtains in the case of iron, whereby the y-form is that which is stable at th= 
highest temperatures, and the a-form that which is stable at atmospheric 
temperatures. 
