218 



TAXONOMIC STUDIES IN THE GENUS CERASTIUM : I 



It should be noted that in considering this character comparable bracts or bracteoles 

 must be used. In the present investigation the uppermost bract, i.e., the uppermost 

 bract subtending a developed bud or flower, has been used. 



The Character Petal Length/Petal Cleft 



This character is seldom, if ever, mentioned in the usual floras. It is implied in 

 Coste (1901) since, although not mentioned in the text, it is clearly shown in illustrations 

 that the petals of C. pumilum are more deeply cleft than those of C. semidecandrum. 

 Ascherson & Graebner (1919) give more details than the other floras including some 

 idea of the range of size of the depth of the cleft in terms of petal length. 



The measurement of this character is complicated by the occurrence of petals divided 

 into more than two segments. This tendency for the petals to be tri- or quadrifid appears 

 to be much more common than previously noted, and is most marked in C. semidecandrum. 



The occurrence of quadrified petals in other Cerastium species is mentioned in the 

 literature. 



In the herbarium specimens examined tri- or quadrifidy occurred in approximately 

 5 per cent of the C. semidecandrum samples, in approximately 1 per cent of C. atrovirens 

 samples and only in about 0-1 per cent of the C. pumilum samples. If the character of 

 petal/cleft ratio is to be employed it is important to know how representative a sample 

 was contained in the herbarium material. This is especially important when the idio- 

 syncrasies of the collector are known. 



Large amounts of material were collected from Newborough Warren, Anglesey, of 

 C. semidecandrum and C. atrovirens, which were then examined for the occurrence of 

 abnormal petals. Seven hundred and fifty-five flowers of C. semidecandrum were examined 

 and quadrifid petals were observed in thirty-four examples, giving an occurrence of 

 4-5 per cent. Two hundred and twelve flowers of C. atrovirens were examined and two 

 had trifid petals, that is 0-94 per cent of the total examined. 



Although only one population has been examined it would appear probable that 

 the distribution of this character in nature differs little from that of the herbarium material 

 used. 



The differences between the means of the groups were found to be significantly 

 different as may be seen from the data below. The constancy of this character was tested 

 by comparing its range and distribution in the experimental plants, no significant difference 

 being found. 



Standard error of Actual difference 



Groups difference between means between means 



A and B 1-93 12-60 



BandC 2-66 24-70 



A and C 2-51 12-10 



The Character Petal Length/Petal Breadth Ratio 



This character is not normally mentioned as being diagnostic. Although it has not 

 been used in the analysis it shows significant though small differences for the primary 

 groups A, B and C. However, this character proved to be of importance in differentiating 

 between C. atrovirens and C. subtetrandrum and will be discussed more fully in a later 

 paper relating to the analysis of these two taxa. 



The Character " Plant Physiognomy " 



To express a concept such as the physiognomy of a plant in mathematical terms 

 obviously presents difficulties. Although this character has not been used in this analysis 



