XXXll PEOCEEDIN&S OP THE 



In point of form, there has been a considerable change, conse- 

 quent upon the gradual but final abandonment of the Linnsean 

 system. That system was originally accompanied by one of Lin- 

 naeus' s admirable methodical contrivances the separation of the 

 diagnosis and the description, which have in the course of time 

 been so signally confounded and perverted. His plan was to 

 contrast the most important or striking characters in short dia- 

 gnoses (never to exceed twelve words) with a view to giving a ready 

 clue to the name of a plant, and to give a detailed description to 

 enable the student to confirm the identity of his plant, and to 

 teach him what is known concerning it. Lamarck and De Can- 

 doUe's plan was in principle the same ; but the diagnosis is given 

 in a tabular form. Until lately, in all general works as well as 

 in local Floras, the diagnosis has been nominally retained, but in 

 a form which deprives it of all its advantages. It is no longer a 

 Linnaean diagnosis, but a description in an ablative form, or a de- 

 scription of certain classes of characters only, the remainder being 

 reserved for what is specially called the description. Many of the 

 more recent Floras have reunited the whole into one description, 

 replacing the diagnosis by italicizing a few contrasted words, as 

 was very well done by Koch in his Synopsis of the Grerman 

 Flora. His example was followed by Grrenier and Grodron in 

 their Flore Frangaise, and by Babington in his British Manual, 

 but with an arrangement of the descriptions according to an a 

 priori estimate of the importance of characters, which is by no 

 means of practical convenience. Of late years also the tabular 

 dichotomous or nearly dichotomous diagnoses, or analytical keys, 

 are getting more and more into vogue, and when well done from 

 actual comparison of specimens, and not extracted only from de- 

 scriptions, are, I believe, the most practically useful. But nothing 

 is more difiicult than to contrast specific characters in a large 

 genus so as not to lead the student astray; and the careless 

 manner in which many recent analytical keys have been drawn up 

 has tended very much to throw discredit on the system. There 

 is also some difference in the form adopted for these keys. The 

 tabular form, indenting the subdivisions, as used occasionally by 

 Koch and other Glermans, is, I believe, the best where the species 

 are not numerous or can be distributed into well-marked sec- 

 tions. The bracketed analytical keys of De Candolle and most 

 French botanists are very convenient where the species are 

 numerous or complicated in their relations, requiring cross 

 references. The least convenient of all is the running the divi- 



