442 M. A.DeCandolle on the Characters which distinguish 



guished in this manner from fifteen to twenty of wliat may 

 be called physiognomic groups of plants. This is a means of 

 facilitating the descriptions of travellers. There is certainly a 

 great number of forms which are not siifficiently marked to 

 enter into any one of these categories in jiarticular ; or rather, 

 there are forms which are so abundant and common, that they 

 include the great majority of the species of every Flora. 

 Hence perhaps the little practical use that can be made of these 

 divisions. 



2. Characters relating to the Families. 



Proportions of the Species of different Families to the Phanero- 

 gamia. — The calculation usually made to express the pi-oportions 

 of the families in a country, supposes implicitly that the species 

 of different families ai'e equally abundant in individuals in the 

 same country. This, however, is not the case, and we should 

 probably arrive at a more correct idea by ascertaining what are 

 the commonest species, and calculating the proportions of the 

 families from these species. Unfortunately, the collection of 

 data as to degi-ees of frequency is a difficult matter; where 

 existing they are rather vague, and, for most countries, they are 

 entirely wanting. 



Besides, the average area of the species varies according to 

 the families and regions. Thus, under similar conditions, the 

 more extended the space under consideration, the greater is 

 the addition of different species belonging to certain families 

 in which the specific areas are limited in comparison with 

 other families in which the areas are larger. In a central 

 region of Europe, for exam])le, we meet with a small proportion 

 of the Leguminoste, Labiatse, or Compositre, which exist in the 

 whole of Europe, but with a large proportion of the Cyperacese, 

 Juncacese, or Graminese ; consequently, the proportions of these 

 families will be very different, according as we regard the sup- 

 posed central region or the entire continent. The former of 

 these families will have a greater number in the whole of Europe ; 

 but, nevertheless, in the particular district, they will be of no 

 greater importance than is shown by the local Floras. Let us 

 see how serious this cause of error may be. Wo. can only appre- 

 ciate it in Europe, as elsewhere the enumerations of species of 

 regions included within others are either wanting or defective. 



I shall confine myself to the comparison of the Leguminosfe, 

 Compositse and Graminese, as the Cyperaceje are often incom- 

 plete, even in European Floras, and the other families are not 

 sufficiently rich in species to render the proportions independent 

 of errors and local circumstances. I shall take my examples of 

 countries from the continent and under the middle latitudes. 



I shall first compare the department of j\Iainc-ct-Loire ac- 



