8 JOURNAL OF THE ARNOLD ARBORETUM (vol. vi 



the leaves were adopted for most of the groups, in the order named. 

 This arrangement may not prove entirely satisfactory for a fuller treat- 

 ment, but these characters were selected after much experimenting for 

 two reasons: First, because they have been clearly stated and emphasized 

 in most of the descriptions of recently proposed species, and second, for 

 the sake of definiteness. It is extremely difficult to find in dealing with 

 a genus like Crataegus, in which the number of species is so great and 

 the differences between them in many cases slight, any common character 

 that will hold good throughout a large series. The taxonomic value of 

 characters varies in different groups, but generally there appears to be 

 considerable variability even within the species in such particulars as 

 the size of the flowers, the number in the corymbs and the compactness 

 or laxity of the latter. The shape of the leaves is even less stable and 

 dependable, many types often being found on a single branch; certain 

 sorts of serration characterize some species, but in most there is much 

 variability and differences are only relative; the texture and veining of 

 the ieaves is sometimes characteristic, but this is often modified by 

 ecological conditions; the color of foliage, though not without significance, 

 is subject to similar modification. The fruit is often one of the best 

 guides to group distinctions, but there is quite too much variation 

 in such particulars as shape, size and color for them to be depended upon 

 rigidly as specific criteria. The number and shape of the nutlets is 

 rather constant in some groups and quite variable in others; but these 

 differences likewise, though furnishing in a few exceptional cases u dis- 

 tinction between closely allied species, usually serve merely as a guide 

 to the group. Even the presence or absence of pubescence on the 

 corymbs, often one of the best specific distinctions, cannot always be 

 relied upon. In the color of the anthers, if not too narrowly defined, is 

 to be found at least a distinction that is definite and not one of degree. 

 Certain species seem to be characterized by having anthers of the yellow 

 group of shades and others by those of shades that may generally be 

 called red. These differences are often more or less distinctly cor- 

 related with the color of the fruit and shade of foliage, especially in its 

 autumnal aspect. 



In the number of stamens there is greater variability and they must 

 be taken with less confidence and greater latitude as specific distinctions. 

 As the flowers of this genus are 5-merous these organs normally appear 

 in multiples of five. Thus in certain species, especially in the Tenuifoliae 

 group, five appears to be the normal number, and these vary in individual 

 flowers from four to seven or rarely eight. There is a double series in 

 another large array of species in which the number may be expected 

 to vary through suppression or redundancy between eight and twelve 

 or fifteen, although in the great majority of flowers it is actually ten. 

 In many species the series is again doubled and we find twenty stamens 

 or rarely a few more or less. Considerable field observation and study 



