( 402 ) 
number of lobes of the first leaf was counted in the plants that 
had survived in germinating dishes. Besides I determined in adult 
plants of Scandi Pecten-Veneris the number of umbel-rays and 
with Anethum graveolens also the number of umbel-rays and at the 
same time the number of flowers of the umbellet. For the determi- 
nation of this latter character only the umbellets of the oldest umbel 
of each plant were taken. Of Panunculus arvensis the number of 
fruits per flower was determined and of Cardamine hirsuta the length 
of the silique, of each plant the siliques of the principal stem being 
measured. 
For each of the characters mentioned | took of each of the cultures 
on fertile soil and on sandy soil 300 measurements or countings, a 
number which, according to the calculations of Prof. KarrryN, gives 
in investigations of this kind a sufficient guarantee of accuracy. For 
certain characters I had to be contented with a smailer number 
since the material in these cases was deficient. For those cases in 
which the variability concerns the number, the numbers were noted 
increasing by unity; for those characters that vary in leneth, the 
length was determined in fractions of a millimetre, in millimetres 
or in centimetres, depending on the absolute size of the parts. By 
means of the numbers obtained, curves were plotted in order to have 
a general survey of the observations and to facilitate a comparison 
of the observations of the culture on fertile soil with that on sandy 
soil. In most cases the observations were combined into groups, so 
that from seven to seventeen intervals were obtained. In this way 
curves are obtained that admit of easy inspection and in which the 
smaller irregularities have disappeared. Only for the number of 
branches of the third order of /beris amara, tig. V, the observations 
of the plants on the fertile soil had to be combined to 28 groups, 
since only then a comparison with the plants from the sandy soil 
was possible. 
The curves for the various characters are reproduced on the 
accompanying plate. Since for all cases the frequencies have been 
calculated, all the curves have the same area and can be mutually 
compared. For each character the curve of the well-fed plants has 
been drawn as a continuous line and that of the badly-fed plants as 
a dotted one, both having the same absciss. Of both the observations 
have been combined to groups with the same interval. In all the 
figures the size or the number of the part in question increases from 
left to right. 
These curves now show us the way in which the studied characters 
vary and the limits of this variation. 
