Day and Shepherd — Lime-Silica Series of Minerals. 267 



regions (represented by dotted lines in the diagram fig. 3), 

 but they have not the same accuracy as those portions of the 

 curve which are represented by full lines. All the mixtures 

 used were prepared with the greatest care and were chemically 

 pure within one or two tenths of one per cent. 



Lime-Silica Minerals. — Of the lime-silica series only one 

 well-defined mineral is known — wollastonite — which when 

 melted usually crystallizes in a pseudo-hexagonal form of the 

 same composition. This mineral has already been carefully 

 studied by Allen, White, and Wright,* and the relation between 

 the natural mineral and its second crystal form thoroughly 

 cleared up. Pursuing the conventional metho'ds of reasoning, 

 we might also expect to have an intermediate mineral for the 

 trisilicic acid, 2CaO,3Si0 2 an akermanite analogue, 4CaO,3Si0 2 ; 

 an orthosilicate, 2CaO,Si0 2 , and a tricalcic silicate, 3CaO,Si0 2 . 

 All of these minerals are deducible from the various hypothet- 

 ical silicic acids. We have undertaken to prepare and study 

 the entire series of mixtures of lime and silica. 



BoudouarcVs Measurements. — So far as known, there has 

 been only one serious attempt to determine the constitution of 

 this series of minerals. f According to Eoudouard, the freezing 

 point curve for the lime-silica minerals consists of four eutec- 

 tics and three maxima (compounds), the maxima correspond- 

 ing to the metasilicate, the orthosilicate and the tricalcic sili- 

 cate. Unfortunately, the method used by Boudouard was a 

 very inaccurate one. He prepared small cones of the various 

 mixtures and placed these in a furnace alongside of Seger cones. 

 The furnace was then heated and allowed to cool down again, 

 after which the crucible was opened and the cone observed to 

 see whether or not the mineral had melted. It is a common 

 method in industrial practice, but has rarely been thought ac- 

 curate enough for original determinations. 



This method is peculiarly unsuited for such investigations for 

 several reasons: Suppose the mixture to contain an eutectic with 

 a greater or less excess of one of the components. The cone 

 would begin to weaken as soon as the eutectic began to melt, 

 and its further progress would be governed entirely by the rel- 

 ative quantity of eutectic present and its viscosity after melting \ 

 No information whatever regarding inversions in the solid or of 

 the component in excess could be obtained, and errors of more 

 than 500° would certainly occur (in fact did occur in Boudou- 

 ard's case) in the interpretation of the softening temperature 

 in some parts of the lime-silica curve. Let us illustrate by a per- 

 fectly obvious hypothetical case (fig. 1) : Assume first that the 

 melting-points change rapidly following a steep liquidus AB 



* E. T. Allen, W. P. White and Fred Eugene Wright, ; 'On Wollastonite 

 and Pseudo-Wollastonite, Polymorphic Forms of Calcium Metasilicate, " this 

 Journal, xxi, 89, 1906. 



\ Boudouard, Journ. Iron and Steel Institute, 1905, p. 339. 



