STUDIES ON THE VEGETATION OF ICELAND 31 
where the freshwater vegetation is not represented at all, while the 
hot springs have another maximum here. 
In a subsequent chapter I shall return to the vegetation spectra 
given in table 8. In this connection it will suffice to point out that 
it is probably the same forces, viz. differences of tempe- 
rature, which have been active in the formation of the 
Icelandic types of vegetation which have determined the 
floristic differences of the altitudinal zones. 
The factors especially causing differences of temperature in a 
given area of Iceland are partly differences with respect to the 
amount of snow and partly differences with respect to the amount 
of water. Hence the first task of an analyst of plant formations, 
after an examination of the vegetation at various heights above sea 
level, will be to investigate the influence on distribution of these 
two factors, and by this means attempt a grouping of the Icelandic 
types of vegetation. 
In the two succeeding chapters I shall therefore give a more 
detailed account of the results I arrived at on analysing the forma- 
tions on a journey in Iceland in the summer of 1925. In yet 
another chapter the influence on the vegetation of differences in 
snow-covering and the moisture of the soil will be more thoroughly 
discussed, and finally the results thus gained will be utilised in 
setting up the types of Icelandic vegetation which have, up to the 
present, been more thoroughly investigated. 
An analytical study of the formations has hitherto been carried 
out in 4 different places in Iceland, viz. on Lyngdalsheidi in Arness- 
sysla in the south country, partly at c. 100 m above sea level, and 
further at c. 250—300 m and c. 400 m above sea level; on Arnar- 
vatnsheidi near Ulfsvatn in the highlands northwest of Langjökull 
at c. 500 m above sea level; in the valley bottom at Lækjamôt in 
Vididalur in the north country, and in the valley bottom at Nordtunga 
in Borgarfjöröur in the south-west country. 
The investigation was carried out by means of Raunkizr’s 
circling method. With a few exceptions, 25 random samples from 
each locality, each of 1/10 sq. m., were analysed. I have not thought 
it appropriate to take into account other methods of analysis, partly 
because those which could here be considered are of a later date 
than Raunkizr’s circling method and to a certain extent resemble 
it, partly because, from a scientific point of view, they must be re- 
garded as retrograde. Up to the present, Raunkiær’s circling 
