NECESSITY OF PURE MINERAL TYPES 153 



the quantitative studies of mineral solutions so far undertaken in the 

 laboratory have brought to light no serious exception to these qualitative 

 generalizations.^^ Quite the contrar}^ The enormous increase in the 

 range of temperature required for mineral investigation compared with 

 low-melting salts has afforded most satisfactory and greatly desired con- 

 firmation of the general validity of these relations. 



Pure Types are Necessary 



Without entering on a discussion of the principle itself, therefore, its 

 applicability will be immediately obvious. If the addition of one mineral 

 to another in any quantity, however small, lowers its melting tempera- 

 ture, and the amount of the lowering is considerable, this may be another 

 thread which leads directly to the situation described earlier in this paper, 

 namely, that mineral melting points gathered from all sources exhibit 

 differences sufficiently large to make them wholly untrustworthy. Any 

 change in the composition, even if small, places the melting point on a 

 sliding scale of temperature so steep (the curve ac, figure 1 ; also AB, 

 figure 7) that considerable differences in the measured melting points 

 must inevitably result. Such observations reflect only too faithfully the 

 fact otherwise established in many different ways, that typical minerals 

 are not found without small and variable admixtures of other minerals 

 which affect their melting temperatures differently. This is in no sense 

 a limitation on the study of the characteristics of minerals during forma- 

 tion, or the competence of such studies, undertaken in the vicinity of the 

 melting temperature, to reveal their true relation, but it imposes on us the 

 absolute necessity of first establishing some quantitative relation between 

 the hind and amount of impurity and the Tcind and amount of effect pro- 

 duced hy it. To a certain extent also we may determine relations be- 

 tween some natural types (like eutectics) which have important factors 

 in common, but we can not recognize and quantitatively establish the in- 

 terdependence of those qualities which determine types if they can not 

 first be separately observed. To the laboratory student this is merely a 

 particular case of the general working rule that the number of unknown 

 factors in his system must not be greater than the number of measure- 

 ments available for their determination. He is undertaking to determine 

 not only the character of the operative forces in rock formation, but their 



*3 It is perhaps well to warn the student against the mistake of trying to apply to 

 rock formation the quantUniive laws of solution which have been deduced for very dilute 

 aqiieoKS solutions, and which can not yet be applied to concentrated aqueous solutions, 

 much less to viscous concentrated silicate solutions, at least not until the various par- 

 ticipating factors have been definitely evaluated, which has never been done. 



