ROBINSON. — GENERA OF THE EUPATORIEAE. 431 



The presence or absence of a true (though often much reduced) 

 pappus becomes especially important in distinguishing Ageratum § 

 Coelestina and Alomia and leads, as will be seen elsewhere in this 

 paper, to some readjustment of generic lines. The character is be- 

 lie\'ed to be a good one and to lead to a real distinction between groups 

 which ha\'e hitherto been very poorly delimited. 



To see clearly what is here meant it is onl^^ necessary to compare 

 Ageratum micropappum, possessing a true but excessively reduced 

 pappus, with Ageratum heterolepis, A. eehioides, A. microcrphalum, or 

 A. mirrocarpum, all of which so far as ol)ser\ed by the writer have no 

 real pappus but merely a sub-corollar annulus and in consequence 

 are to be referred to Alomia. 



The involucre in the FAipaforium tribe takes on many forms, all 

 more or less intergrading. In FAipaforium. itself, though the involucre 

 goes through the whole gamut of variation, the number of species is 

 so great that grouping by involucral differences has never appeared 

 to be sharp enough to permit any satisfactory generic segregation of 

 the elements concerned. It is very easy to refer certain iiuirked 

 species to Osmia Sch. Bip. and others to Kyrsteuia Necker, but tran- 

 sitions are innumerable. On the other hand in many of the smaller 

 genera involucral characters furnish distinctions of sectional or even 

 generic value. Three types of involucre may be recognized as follows: 

 1) in which the chief scales are definite or subdefinite in number (four 

 in Mikauia and Kanim'ia, five or six in Stcvia) often surrounded at the 

 base by 1-3 considerably reduced scales i. e. calyculate. 2) the chief 

 scales of indefinite number (though rarely very numerous) and sub- 

 equal, appearing to be in 1-3 series and often accompanied by a 

 very few much reduced outer ones. This is the common form of invo- 

 lucre in Ageratum, Kuhnia, etc. 3) scales of indefinite number (usually 

 numerous) conspicuously unequal and graduall^^ diminishing outward, 

 forming apparently several to many series, though of coiu'se in reality 

 spirally arranged. These three types of involucre, while sometimes 

 confluent, are in general pretty readily distinguished. 



There are also considerable differences in the texture of the involu- 

 cral scales and to some extent these may be used, at least as supple- 

 mentary aids, in distinguishing genera. Thus in Ageratum, most 

 Alomias, and several other genera the scales are prevailingly of rather 

 firm texture and pretty definitely 1-3-costate, while in Brickellia, 

 Hofmeisteria, Podophania, Oaxacania, etc. the scales are prevailingly 

 thin, fiat, and finely striate. 



One other character of special classificatory significance in the 



