CRITERIA. 165 



structures undoubtedly furnish characters of much taxonomic value, it is unfortunate that 

 their descriptions have been based upon superficial examinations, usually with the aid 

 only of a hand-lens and not upon exact measurements. This applies particularly to 

 statements of relative length of appendage and stigmatic portion. The tabulation of a 

 large series of measurements now indicates that the ratio between appendage and stig- 

 matic portion is far from constant for any one species or perhaps even for any one sub- 

 species. It also indicates, however, that the ratio is much higher for some species than 

 for others. Pairs of species can be selected in which no overlapping of ratios occurs, 

 so that the character can there be used to advantage. But any attempt to apply the 

 ratio throughout the genus as a never-failing specific character would result in a cleavage 

 running counter to all other features and it would therefore lead to an unnatural arrange- 

 ment of the forms. 



The extent of variation within a single large species is indicated in table 23 (p. 230). The 

 ratio is of further significance when used to indicate tendencies in the various sections. 

 In the Punctati the stigmatic portion and the appendage are approximately equal in 

 length, sometimes one being the longer and sometimes the other. Among the Typici 

 the otherwise anomalous C. albidus has unusually long appendages, while in the other 

 three the appendage is uniformly shorter than the stigmatic portion or only slightly 

 exceeds it in occasional specimens of vaseyi. A similar condition holds in the Pulchelli, 

 indicating as do other features that these two sections are very close phylogenetically. 

 In the Nauseosi, the appendage is always well developed and usually but not always 

 longer than the stigmatic portion. In two subspecies of nauseosus it is regularly shorter 

 and in a third it is sometimes so. The shape is often given in descriptions, but, aside 

 from the difficulty of expressing this accurately, it possesses little value. Long append- 

 ages are naturally more slender than short ones and they are usually more attenuate. 

 The so-called "broad appendages" of certain species are perhaps not broader by measure- 

 ment than others, but are only seemingly so because of their shortness. 



Stamens. — The stamens have thus far furnished characters of diagnostic value in only 

 one instance. The tips are acute in all species except C. albidus, but in this they are 

 obtuse and also decidedly shorter than in the others (except the small-flowered pyra- 

 Tnidatus). It is possible that detailed studies might reveal further stamen differences 

 of value, but a preliminary survey indicates that the prospects are not promising. 



Corolla. — The shape of the corolla is always tubular-funnelform, although there is 

 some variation between the species. In some the throat enlarges more abruptly than 

 in others, but the line of demarkation between tube and throat is so indefinite that the 

 differences can not be expressed in quantitative terms. Since also the shape depends 

 upon the age and condition of the specimen, whether fresh, dry, or boiled, such expres- 

 sions as "abruptly expanding" or "slenderly tapering" are only relative and of slight 

 value. The total length of the corolla is a character of greater significance. WTiile 

 this fluctuates to disappointing extremes in some of the polymorphous species, in certain 

 other cases it is sufficiently constant to be of use and should always be included in de- 

 scriptions of new forms. Thus, in the Punctati the corolla always measures between 6 

 and 6.5 mm. long, whereas in C. nauseosus it measures 6.2 to 12 mm. and in C. parryi 

 8 to 12 mm. The character is especially useful in distinguishing the two species of the 

 Pulchelli, the corolla in one measuring 7 to 9.0 mm., in the other 10 to 14 mm. How- 

 ever, when the whole series of species in the genus is considered, it is found that the 

 overlapping between them is so frequent and the range of variation is so great in certain 

 cases (more than 50 per cent of the length in C. viscidiflorus) that the utilization of this 

 criterion in keys is very limited. Even in applying it as a measure of the extent of 

 differentiation where only two species are concerned, it must be used only after a con- 

 siderable number of flowers from different plants have been measured. The total 



