236 Professor Necker on Mlneralogij considered 



is only by these Jast properties that the identity of matter be- 

 tween the portion of a mineral submitted to analysis, and the 

 remaining parts of the specimen, as well as all other minerals of 

 the same description, can be established. Were it not so, as the 

 part analysed has been entirely dissipated and destroyed, we 

 could know the chemical compositions only of bodies which have 

 ceased to exist, without being able to draw from such a know- 

 ledge any inference as to the composition of still existing bodies. 



Moreover, chemistry cannot proceed in its own especial pur- 

 pose, without employing physical considerations, such as colour, 

 hardness, crystalline forms, &c. to characterize the productions 

 of its decompositions, and recompositions, the precipitates, the 

 solutions; and it is, by-the-by, a singular contrast to see chemists 

 holding so lightly the physical properties which are distinguishing 

 characters of natural bodies, while they give them a great weight 

 when they are to characterize their own artificial productions. 



§ 43. From the whole of the above mentioned considerations 

 we conclude, that in those cases where the physical and chemical 

 properties are found at variance in the same group of mineral in- 

 dividuals, the preference shall be given to the physical characters, 

 until chemistry shall have clearly pointed out the cause of this 

 difference ; but, at the same time, it must be clearly under- 

 stood, that such groups only in which all the members agree to- 

 gether, and at the same time are distinguished from all other 

 groups by the same physical as well as chemical characters, are 

 to be considered as strictly natural divisions. 



§ 44. Let us now trace the plan of the divisions of the classi- 

 fication from the lowest to the highest, and, in so doing, let us 

 not forget that we are to submit to our method only individuals 

 in their perfect state, isolated and not aggregated one with ano- 

 ther, that isj simple crystals supposed complete, perfect, pure, 

 and possessing all their qualities or properties. By this mode of 

 procedure we get rid of the greatest of the difficulties which 

 has always beset those who, classing abstract substances, what- 

 ever external shape they might assume, were obliged to find a 

 character common not only to all the perfect crystals composed 

 of the same species of substance, but also to the mutilated, 

 avorted, deformed, and rolled ; to those acicular, radiated, fi- 

 brous, and granular masses, which. are nothing else, in fact, but 



