The preceding account has shown that most of the men- 

 tioned minerals may, either completely, or at all events very 

 nearly, be interpreted as compounds of metatitanates etc. with 

 metaniobates etc. Consequently the ratio between the oxygen 

 bound to the acids , and that bound to the bases , will be in 

 close connection with the ratio between the two kinds of acids; 

 in a pure metatitanate RTiO.^ this ratio is as 2 : 1 ; in a pure 

 metaniobate RNb^O^ it is as 5 : 1 ; in all mixtures the ratio 

 must accordingly be varying between these two ones. 



For those ot the minerals not containing H^O ov F it 

 has been shown in the preceding that they consist almost com- 

 pletely of metatitanates etc. with mataniobates etc. With regard 

 to these minerals, then, a system founded on the ratio between 

 the two acids will correspond very exactly to a system founded 

 on the ratio between acetous and basic oxygen atoms. Now it 

 is very probable too, that the two ratios will correspond to 

 each other also with regard to all the other minerals. The 

 ratio between the two kinds of acids has the advantage of 

 being easily calculated for all the minerals as being independent 

 of H^O or F. It seems also to be tolerably well connected 

 with the crystallographic qualities of the minerals, as will be 

 shown in the following. 



The ratio between the groups of TiO., etc. and Nh^ Og etc. 

 is exceedingly varying in the mentioned compounds. It has 

 the greatest value in wöhlerite being as 12:1, and the smal- 

 lest value in porychlore from Brevik where it is as 2 : 5. By 

 a closer examination of the individual minerals they appear to 

 separate quite naturally in two groups, one with the ratio 

 smaller than 1 , the other with the ratio larger than 2 ; no 

 mineral is found between these two numbers. 



It is then seen that the minerals of the first group, viz. 

 pyrochlore, lewizite, mauzeliite, chalcolamprite, and endeiolite, 

 all crystallize regularly and moreover in octahedra, and thus 



