ARKIV FÖR BOTANIK. BAND 12. N:0 14. 



Number of samples having 



We find from this table that the number of leaves in 

 the rosule is considerably increasing with the number of 

 flower-stalks. The mean number being 5,5 4, 6,3 8 and 7,8 

 respectively. 



Regarding the number of leaves in the rosule there is 

 some irregularity in the counts owing to the fact that the 

 undermost (and smallest) leaf in the rosule sometimes is 

 attached to the stem at some distance beneath the rosule. 

 Generally I counted it as belonging to the rosule as soon as 

 this distance was smaller than 5 mm. 



The numbers of leaves in the rosule for samples with 1 

 flower-stalk form what I have called a frequency- curve of 

 type B, whereas the other characters here considered (ex- 

 cepting the number of flower-stalks § 1 ) give rise to, approxi- 

 mately, frequency curves of type A. There is no reason to 

 charaterize more exactly the nature of these frequency-curves, 

 taking into regard the small number of classes. 1 have not 

 even considered it appropriate to calculate the mean values 

 of each character, which are only exceptionally given above 

 for the number of leaves in the rosule (for different numbers 

 of flower-stalks). 



B) Heterograde statistics. 



The characters examined were the length and thickness 

 of the rootstock, the stem and the flower-stalks, length and 

 breadth of the leaves in the rosule, of the petals and, the 

 sepals, length of the stamens and of their separate parts as 

 well as of the pistils. 



6. Rootstock. The gathering of the radices being rather 

 time-absorbing I have only examined 1 1 samples regarding 

 the dimensions of the radices. The result obtained, in milli- 

 meters as for all other characters, was 



