OM A NATURAL SYSTEM. 



racier and compofition may be made equally c- 

 vident. I llential properties depend on the qua- 

 lity of the parts th.it enter into compofition, 

 and their mutual proportion ; and, unlefs we 

 are well aquanucd with thele parts, we fliall la- 

 bour to little purpjie, in our attempts to mould 

 them to our own dciires : Nay, we often meet 

 with diiappointments, becaufe we have not con- 

 fidered that our views are inconfiilcnt with the 

 very nature of the materials fubjecled to experi- 

 ments. 



} xxn. In ivbat manntr the Compofition of FoJJils 

 may be after taincd. 



HAVING fettled thefc points, it remains yet to 

 be determined in what manner we are to judge 

 of the compofition of fo(Fils : Whether the con- 

 nexion between fupcrficial marks,and the intrin- 

 lic character, is fo intimate and confcqucnt, that 

 the former cannot be known, without the other 

 being revealed ? whether it may be nectflary 

 to proceed . by a chemical analyfis in the dry 

 \\;iy ? c.r, llic.uld this not be fufficient, aie we to 

 have lecoiu.e to the -nn^ift way? We will ecu 

 iidci thcie queilions fcparutely. 



x x 1 1 1 . E\tenml Criteria . 



IF, tluoiigli the means of criteria colleeled 



from 



