54 PLANT SOCIOLOGY 



At best, not all plant communities are suited to studies of con- 

 stancy, since not all homogeneous areas are large enough, e.g., aquatic 

 societies, rock crevice, gravel, mosaic moor, terrace-turf communities. 

 In such cases it is better to omit the determination of constancy and 

 not sacrifice the unity and similarity of the stands. 



Determination of Minimal Area; Species: Area Curves. — The 

 increase in number of species with increasing area, first scientifically 

 treated by Jaccard (1912, 1928), was later investigated in detail by 

 Palmgren (1917). He showed that in the grazing regions of the Aland 

 islands the number of species in one portion of the area stands in almost 

 direct relation to the size of the sample plots. Palmgren's empirically 

 derived values were shown by Romell (1925, 1930) to correspond with 



8 nor 



Q 



4) 50 

 5> 



c5^ 



size of Sample p/oh- 



FiQ. 32. — Species: area curve; ° = calculated values; • = empirical values. {After 



Romell.) 



the curves theoretically calculated by the law of chance (Fig. 32). 

 From this it is clearly seen that with increasing size of area the curve of 

 the number of species at Ctst rises very rapidly, then swings nearly 

 to the horizontal, and continues with an almost imperceptible rise. 



Similar empirical curves were later constructed by Brenner (1921), 

 Ilvessalo (1922), and Braun-Blanquet (1926). Their form corresponds 

 with Romell's curves, in spite of the fact that they were based on 

 definitely bounded uniform plant communities, 



Kylin (1926), starting with the law of probability, has established 

 from theoretical considerations the curves discovered by purely empir- 

 ical methods. He demonstrated that the communities studied by 

 Brenner (1921), Du Rietz (1921), and Ilvessalo (1922) obey the law of 

 probability with regard to number of species and area. The similarity 

 of the curves theoretically deduced and the empirical species: area 

 curves is really very striking. 



Romell's distribution curve (Fig. 32) is a species-percentage: area 

 curve. The ordinates represent the number of species expressed as 

 percentage of the whole number. The course of this curve is condi- 

 tioned by the size of the sample plots and by the density (distance 

 between individuals) of each species. 



