GENERAL LAWS OF ORIGIN AND MIGRATION. 267 



fines of the groups of Acrogens and Gymnosperms may 

 be supposed favourable to such exchanges. (3) The con- 

 traction and breaking up of floras, as occurred in the 

 Middle Erian and Lower Carboniferous, may have been 

 eminently favourable to the production of such varietal 

 forms as would result from what has been called the 

 " struggle for existence." (4) The elevation of a great 

 expanse of new land at the close of the Middle Erian and 

 the beginning of the coal period would, by permitting 

 the extension of species over wide areas and fertile soils, 

 and by removing the pressure previously existing, be 

 eminently favourable to the production of new, and es- 

 pecially of improved, varieties. 



2. Whatever importance we may attach to the above 

 supposed causes of change, we still require to account 

 for the origin of our specific types. This may forever 

 elude our observation, but we may at least hope to ascer- 

 tain the external conditions favourable to their produc- 

 tion. In order to attain even to this it will be necessary 

 to inquire critically, with reference to every acknowl- 

 edged species, what its claims to distinctness are, so that 

 we may be enabled to distinguish specific types from 

 mere varieties. Having attained to some certainty in 

 this, we may be prepared to inquire whether the condi- 

 tions favourable to the appearance of new varieties were 

 also those favourable to the creation of new types, or the 

 reverse — whether these conditions were those of compres- 

 sion or expansion, or to what extent the appearance of 

 new types may be independent of any external condi- 

 tions, other than those absolutely necessary for their 

 existence. I am not without hope that the further study 

 of fossil plants may enable us thus to approach to a com- 

 prehension of the laws of the creation, as distinguished 

 from those of the continued existence of species. 



3. In the present state of our knowledge we have no 

 good ground either to limit the number of specific types 



