THE OLDEST KNOWN MUTATION 97 



petals also cut up. This sharply denned new form suddenly 

 appeared among the plants of Chelidonium majus which the 

 apothecary had cultivated for many years. It was recognised 

 by botanists as something quite new, and eventually it got the 

 name Chelidonium laciniatum ; it was not to be found wild, 

 or anywhere except in the Heidelberg garden. But from the 

 first this new cut-leaved celandine proved constant from seed. 

 It has been naturalised in England and other countries, and is 

 sometimes now found as an " escape." Its origin by mutation 

 seems as certain as its constancy. It is further of interest to 

 note that in crosses with C. majus it follows the law of Mendel. 



Summary. De Vries has done great service in analysing the 

 complex concept of variation ; in sharply contrasting individual 

 fluctuations and mutations; in defining "elementary species," 

 "retrograde varieties," and "ever-sporting varieties"; in ob- 

 serving the actual origin by mutation of stable new varieties or 

 subspecies of (Enothera lamarckiana and some other plants ; in 

 showing by historical research combined with experiment that 

 many stable stocks of cultivated plants have arisen by mutation ; 

 and by corroborating throughout the fundamental idea that " the 

 characters of organisms are composed of units sharply distin- 

 guished from one another." 



The contrast between fluctuations and mutations is so impor- 

 tant that we may state it once more, (i) Fluctuations are 

 continually occurring generation after generation : mutations are 

 rare and occur intermittently. (2) Fluctuations give rise to a 

 series of minute differences which may be arranged on a frequency 

 curve, according to the laws of chance : mutations may be large 

 or small, and their occurrences do not illustrate any ascertained 

 law of frequency. (3) Fluctuations do not lead to a permanent 

 change in the mean of the species unless there be very rigorous 

 selection, and even then, if the selection be slackened, there is 

 regression to the old mean : mutations lead per saltum to a new 

 specific position, and there is no regression to the old mean, 



7 



