Comments on Raunkiaer's and Similar Methods of Vegetation Analysis and the "Law of Frequency"

Two recent articles in this journal (Kenoyer, '27, Gleason, '29) deal with Raunkier's method of analysis and his of frequency. The following comments, supplementing the articles mentioned, nay be of interest to ecologists concerned with statistical methods of vegetation analysis. The comments are chiefly condensed abstracts of the discussion on such matters which has been going on in recent years in Scandinavian literature, largely inaccessible to foreign readers. The methods here referred to all have this in common, that merely presence or absence of the different species is recorded on each of a number of sample areas scattered over the vegetation to be analyzed. For every species found, the number of hits is expressed in per cent of the total number of throws, and is taken to represent the frequency (sometimes constancy 1) of the species. In the following, the designation F%70-numbers is used for these indexes to avoid confusion with the true frequencies (average numbers of statistical units per unit area). WN e will use the expression F%-curve to designate the curve showing the per cent distribution of the species (found within the sample areas) into F%/'--classes (usually, 5 or IO have been used). The puzzling U-shape of this curve (Raunkixr's law of distribution of frequencies or law of frequency ) over a wide range of sizes of sample areas is well known. As already noted by Raunkioer, not only homogenous plant associations yield this type of curve, but also large tracts with a varied vegetation, when analyzed with correspondingly larger sample areas, for instance the Aland or the Barbsund archipelago when entire islands (Palmgren, 'i6, Brenner, '21 ), or the province of Oland when parishes (Sterner, '2T), are used as sample areas. This strongly suggests that the features of the normal F%curve are conditioned rather by the method employed than by any particular constitutional characteristics of natural plant associations. The sharp drop in the FC-curve from the peak in the highest F%-class to the next lower classes, which has puzzled several authors, is apparently