\T,%-<x,r^~:'^vrihe Natural Botanical Families. ' I'Z Hi© 



selvesinto natural groups, after cumbering many othef series. 

 If a doubt arise as to which of two groups a family belongs, we 

 must collect the evidence of its relations, and if we find it re- 

 lated (according to the testimony of different botanists as to dif- 

 ferent families) to a great number of contiguous families, we 

 must not hesitate to place it there, notwithstanding some strong 

 opinions of its relation to a single family, or only a few in an- 

 other part of the system. Hence we separate Legiiminosce with- 

 out hesitation from Rosaccce, We must not weigh the strong 

 expressions of writers as to particular relations, in which they 

 may be insensibly influenced by system, especially in enhancing 

 relations required for the support of Jussieu^s method. Every 

 statement of relation should, however, be considered as evidence 

 of relation ; and we may consider in the same light those cases 

 in which botanists have thought it necessary to take distinctions 

 between two orders. 



The authorities are, in this mode of investigation, to be num- 

 bered rather than weighed, are to be taken as a matter of ave- 

 rage, and to derive value from mutual support, evidencing the 

 relation of many families to many others, and not merely the 

 relation of a single family to another. If the result should un- 

 expectedly bring together families differing in the adhesion of 

 the calyx to the stamens, ovary, &c., we must not forget that 

 botanists have, in such cases, given the evidence against their 

 own impressions. My object in the appended sketch has not 

 been to follow my own judgment or that of any writer whatso- 

 ever, but to bring together the greatest possible number of ad- 

 mitted affinities, and, if possible, in continuous succession. Iso- 

 lated and unconnected resemblances, often accidental or unim- 

 portant, and among orders scattered remotely from each other, 

 must be expected, where they want the weight of joint, com- 

 bined, accumulated evidence. The process adopted may no 

 doubt misplace some families, but there is reason to believe that 

 the great majority will be in their true neighbourhood. 



I have not at all satisfied myself in fixing the arrangement of 

 the families within each group, which is a matter not important 

 at present ; and in many cases the limits of the group are am- 

 b'gUDUs, and must so continue until a physiological key shall 

 be found from a careful examination of the whole. The true 



