196 BOTANICAL GAZETTE ; [SEPTEMBER 
of its bed. Here, at a distance of too to 200 meters from the shore, 
the pioneer tree association becomes established, and persists upon 
dunes of variable size that are usually more or less actively moving. 
This association is characterized by a paucity of species, all having 
strongly xerophytic structures. Populus deltoides, Salix glau- 
cophylla, S. syrticola, Prunus pumila, and the two grasses Cala- 
movilfa longifolia and Ammophila arenaria are at this point the 
only conspicuous members of this rather open cottonwood dune 
association. In it, upon dunes that have become almost com- 
pletely fixed, two stations were established on May 6, and a third 
on July 9, and at the three stations at least four instruments were 
maintained in constant operation until the last day of October. 
These stations were about 200 meters from the lake shore, some 
too meters apart, and about 12 meters above the level of the waters 
of Lake Michigan. At all stations the atmometers received a small 
amount of shade for a few hours of the day, and on account of the 
_ open nature of the association were little sheltered from the wind, 
the cups receiving a rather sharp sand blast during high winds. 
Station no. 1 had some sheltering groups of cottonwoods on a slight 
elevation of sand a few meters southeast of the instruments, and 
no. 3 possessed a similar but smaller shelter at the southwest. 
These differences of exposure to winds probably caused some of 
the variations in the records of the different stations, but affected 
_ very slightly the average rate for the season. 
e graphs for three cottonwood dune stations have been 
plotted upon the same chart (fig. 1), and exhibit a great similarity 
in their general course and in their simultaneous maxima and 
minima. The rainfall at Chicago (20 miles distant) for the same 
period, expressed in centimeters, is shown for periods corresponding 
with those of the intervals between the evaporation readings, but 
as there seems to be no very exact correspondence between the 
amount of precipitation and the amount of evaporation, these data 
are omitted from the other charts. There is certainly a corre- 
spondence between the number of hours of cloudy or rainy weather 
and the amount of evaporation, but this has not been exactly 
determined, nor does it seem important in our present studies. The 
evaporation graphs indicate that the most critical period occurs © 
