260 THE GEOLOGICAL HISTORY OF PLANTS. 



of preservation, it is still more likely to occur. Every 

 revision of any group of fossils detects numerous syn- 

 onyms, and of these many are incapable of detection 

 without the comparison of large suites of specimens. 



3. We may select from the flora of any geological pe- 

 riod certain forms, which I shall call specific types, which 

 may for such period be regarded as unchanging. Having 

 settled such types, we may compare them with similar 

 forms in other periods, and such comparisons will not be 

 vitiated by the uncertainty which arises from the com- 

 parison of so-called species which may, in many cases, be 

 mere varietal forms, as distinguished from specific types. 

 Our types may be founded on mere fragments, provided 

 that these are of such a nature as to prove that they be- 

 long to distinct forms which cannot pass into each other, 

 at least within the limits of one geological period. 



4. When we compare the specific types of one period 

 with those of another immediately precedent or subse- 

 quent, we shall find that some continue unchanged 

 through long intervals of geological time, that others are 

 represented by allied forms regarded either as varietal or 

 specific, and as derived or otherwise, according to the 

 view which we may entertain as to the permanence of 

 species. On the other hand, we also find new types not 

 rationally deducible on any theory of derivation from 

 those known in other periods. Further, in comparing 

 the types of a poor period with those of one rich in spe- 

 cies, we may account for the appearance of new types in 

 the latter by the deficiency of information as to the for- 

 mer ; where many new types appear in the poorer period 

 this conclusion seems less probable. For example, new 

 types appearing in poor formations, like the Lower Erian 

 and Lower Carboniferous, have greater significance than if 

 they appeared in the Middle Erian or in the Coal Measures. 



5. When specific types disappear without any known 

 successors, under circumstances in which it seems un- 



