No. 500] BIOMETRY AS A METHOD IN TAXONOMY 539 



africana Theel as synonyms. The old characters have 

 been redefined, new ones added, and those differentiating 

 H. atra and H. floridana clearly stated. The young and 

 old have been segregated and differentiated and the re- 

 sults of growth determined. The nature and extent of 

 the variation of each character has been recorded. An 

 entirely new character for holothurians has been dis- 

 covered in what I have called "pits," in the body-wall 

 of H. atra. 



It is not possible in taxonomic work to give several 

 years to the study of each species and it often happens 

 that the material is not sufficient. If an author will make 

 a thorough biometric analysis of the characters of at least 

 one species in his group, he will gain a rare insight into 

 the relative values of the characters. The method is 

 searching and leads to especial carefulness in investiga- 

 tion and statement. Even if one has not an ideal number 

 of specimens his determinations are checked by their 

 probable errors. In the case of some characters, as for 

 instance, the spicules of holothurians, an abundance of 

 material is present in each specimen. More and more 

 of the anatomy becomes involved as the work progresses. 

 Now that the individuality and continuity of the chromo- 

 somes has been demonstrated, McOlung has suggested 

 that their number and grouping constitute family, gen- 

 eric, and specific characters of just as definite worth as 

 those that heretofore have been employed. The more 

 characters studied the better, and the ideal taxonomy will 

 be based upon the whole life-history. Then the error 

 of describing the young and old of a species as inde- 

 pendent species will not be repeated. 



It naturally follows from biometric analysis that a 

 group of individuals, giving as much as possible of the 

 range in variation, should be established and dt'positfd, 

 preferably in a national museum, as the specific type 

 group rather than some arbitrarily selected type speci- 

 men. We have instances of specific descriptions based 



