Mr. Woods on the Brilish Species of Rosa. 171 



adopted on the same principle of exciting the attention of other 

 observers. 



The drawing out into a table the specific characters of a ge- 

 nus possesses a double advantage; it brings them to a test, by 

 which the writer will inevitably discover if unfortunately some 

 of his specific characters should be drawn up without inclu- 

 ding any peculiarities to separate it from others, a fault of which 

 even good botanical works afford too many examples ; and it is 

 of great assistance to the future investigator, as it leads him 

 step by step to the species which is the object of his examina- 

 tion. But in order to accomplish this end, it is necessary that 

 the characters which are most important and most permanent 

 should occupy the first places: it is desirable that the arrange- 

 ment of the table should of itself divide the genus into its most 

 natural families. To combine these advantages is no easy task. 

 To discover characters which shall be permanent, always ca- 

 pable of clear description and determination, and which at the 

 same time shall uniformly bring together the most similar plants, 

 and separate those comparatively dissimilar, is perhaps beyond 

 the power of the human mind. Mr. Brown's arrangement of 

 Proteacece, in the tenth volume of the Society's Transactions, is 

 an excellent specimen of what may be done in this way. La- 

 marck and De CandoUe in their analysis of the genus Rosa in the 

 Flore Fraiifaise, have proceeded on a similar notion, though they 

 have adopted a much inferior form, and seem to have had no 

 higher ambition than to assist in some degree the investigation of 

 the species. Even in this they have effected very little ; because 

 in taking first the colour of the flower, then the shape of the fruit, 

 and then the prickliness of the peduncle, they have adopted for 

 their leading divisions characters which are very variable. The 



z 2 yellow- 



