8 Lioyp: LEAF WATER IN GossyPIUM 
in other tables, a net loss of water would not be discoverable 
(graph abd, FIG. 1). It is evident, however, from the directions of 
the graphs that a negative fluctuation was in progress from the 
Io to the 12 hour, followed by increase afterwards. It is also 
evident that a positive fluctuation begins after the 14 hour, if we. 
suppose the datum for the 16 hour, September 16, to apply ap- 
proximately to the same hour on September 17. 
Although the above evidence is quantitatively somewhat vague, 
this result is probably due to the application of assumed dry 
weights based upon data obtained in the following year. Had 
such data been obtained at the time, their employment would lead 
to a surer inference. This appears from similar manipulations of 
the determinations contained in TABLE 1 of leaf water and 
dry weight for unknown leaf areas. 
In the first three columns are given the known data; in the 
fifth the average dry weights from TABLE v, which were obtained 
two days later, these being relative to 100 square centimeters of 
leaf and assumed to be applicable to August 24. Since the two 
days (August 24 and 26 at Tucson, Ariz.) were practically identical 
meteorologically, this assumption is probably entirely justified: 
By dividing the assumed dry weight into the observed dry weight 
for each hour the approximate area of the leaf sample taken at 
that hour is obtained. This affords one term of a ratio (column 4), 
the other of which is the observed leaf water (column 3) by which 
the leaf water per 100 square centimeters is derived. This, for 
each of the three hours of observation, is found in column 7 and is 
plotted in graph 1, FIG. 2. | ; 
Just above this is graph ta, the coordinate points for which 
were determined as follows. (See TABLE Iv.) The same ratios 
of increase of dry weight were assumed as above. One of the 
observed dry weight readings in TABLE 111 was then taken as a 
standard, and the dry weights for the other two hours were deter- 
mined by means of these ratios. By assuming the leaf water to 
be constant the inverse ratios, expressing leaf water as percentage 
of dry weight, were obtained (column 2, TABLEIv). By comparing 
these with the ratios (column 4) of the data recorded for each 
hour (columns 2 and 4) differences are obtained by subtraction 
which determine the position of the graph of observed leaf water 
