302 
floods which followed normal spring rises. The correspond- 
ence of the August pulse of plankton with a heat wave of 10° 
amplitude is well shown in Plate [X. Similar correspondences 
may be detected in some instances elsewhere in the plankton 
and temperature curves, but neither the completeness of our 
temperature data nor the corroborative evidence is sufficient to 
lend much support to a causal nexus between the phenomena. 
The September pulse has a duration of 25 days,—trom Sept. 
7 to Oct. 2—and a maximum amplitude of 3.25 cm.’ on the 14th. 
Its mean falls on the 17th, 26 days after that of the August 
pulse. This isa month of considerable hydrographic disturb- 
ance, the rises of the 6th, 17th, and 27th causing almost twice 
as much movement (8.75 ft.) in river levels (see Table I.) as is 
found in other years of our operations. These accessions of 
flood water in each instance attend a fall in temperature of 5° 
to 8°, though that on the 27th is combined with normal autum- 
nal decline. None of the three is sufficient to cause overflow; 
and each is of but few days’ duration. Their effect upon plank- 
ton production is, however, considerable. In the first place, 
the immediate result of the invasion of flood water is an in- 
stant decline in the plankton, as shown by the change from 
2.07 on the 4th to .69 on the 7th, the flood in this case acceler- 
ating and perhaps continuing the normal decline of the August 
pulse. So also the little rise of the 17th checks the rising curve 
of production, the fall being from 3.25 cm.* on the 14th to .89 cm.’ 
on the 18th. The rise of the 27th evidently occurs towards the 
minimum of a declining plankton pulse, and the fall from 
1.03 em. on the 25th to .87 cm.’ on Oct. 2 is of less extent. The 
location of these floods in the pulse is such that if my conjec- 
tures as to their reducing effect be true they cause a shifting 
of the apex of the curve and of the location of the mean to the 
left of their probable position had not the floods occurred. In 
technical phraseology the mode of the curve of this pulse ex- 
hibits left-handed skewness. In the second place, the general 
effect of these recurrent rises is a reduction in total production 
the extent of which can only be conjectured. It seems proba- 
