n, OEDERS AND GENEBA. 23 



with respect to the Covipositce, the genera Carduus, Hie- 

 racium, and Senecio rather suggest sudivisions within the 

 order, than relations simply to a single centre or type. 



In the third line we have three orders which offer no 

 very decided characters, sufficing to connect their in- 

 cluded plants well together, and equally serving to sepa- 

 rate them from aU other the excluded plants. They have 

 usually been regarded as peculiar groups by technical 

 botanists; but the limits of these groups have greatly 

 varied. Rosacece have either included with themselves 

 or excluded the Prunacece, Pyracece, and Sanguisorbacece. 

 In like manner, the EricacecB have included or excluded 

 the Vacciniacea, Pyrolacece, and Monotropacece. And the 

 inclusive or exclusive limits of the Liliacete have been 

 considerably varied according to the fancies of individual 

 systematists. And though certain genera are selected to 

 give names to the orders, the selection seems to be made 

 rather because those genera and their names are them- 

 selves familiar, than because the rest of the genera bear 

 any special resemblance to them more than to each other. 

 The first of the three orders, indeed, might be made into 

 a better designated group by treating Potentilla or Coma- 

 rum as the typical genus, and holding Rosa an aberrant 

 genus, — one almost equally adapted to constitute another 

 ordinal group, as in the case of Pyrus or Prunus with 

 their congeners. 



The principle of a typical genus becomes more clear 

 and complete in the groups named in the fourth and fifth 

 lines ; although, for the first and last of those six orders, 

 the genera Lithospermum and Scabiosa would seem more 

 appropriate tj'pes, than are those of Borago and Dipsacus. 

 These six orders may be held " natural," inasmuch as 

 their included plants have much general resemblance 

 among themselves. But they are separable from some 



