PARALLELISM. 209 



It is nevertheless true that the records brought to 

 hght by embryologists are very imperfect, and have to 

 be carefully interpreted in order to furnish reliable evi- 

 dence as to the phylogeny of the species examined. 

 An illustration of this is the fact that the species char- 

 acters appear in many embryos before those which de- 

 fine the order or the family, although it is certain that 

 the latter appeared first in the order of time. Most of 

 the important conclusions as to the phylogeny of Ver- 

 tebrata demonstrated by paleontology have never been 

 observed by embryologists in the records of the spe- 

 cies studied by them. Thus I have shown that it is 

 certain that in the amniote vertebrates the intercen- 

 trum of the vertebral column has been replaced by the 

 centrum ; yet no evidence of this fact has been ob- 

 served by an embryologist. If we could study the em- 

 bryonic development of the vertebral column of the 

 Permian and Triassic Reptilia, the transition would be 

 observed, but in recent forms caenogeny has progressed 

 so far that no trace of the stage where the intercentrum 

 existed can be found. 



Again I have demonstrated by paleontological evi- 

 dence that the lines of the ungulate Mammalia origi- 

 nated from a bunodont pentadactyle plantigrade an- 

 cestor ; but embryonic research has failed to discover 

 the preservation of a record of this fact in the ungu- 

 lates at present existing. The embryo of the horse is 

 not pentadactyle, nor even tridactyle, although tri- 

 dactyle horses persisted late in geologic time. Nor 

 has embryonic research demonstrated a four- toed 

 stage in the Bovidae (oxen, etc.), although there is no 

 doubt that they descended directly from an ancestor 

 so characterized. Any number of similar cases might 

 be cited to show the prevalence of inexact parallelism 



