470 
and when they are of short duration and merely flush the chan- 
nel, as is predominantly the case during the low-water period. 
Lower levels depress production when they introduce stagna- 
tion conditions—as in the winter under the ice, and when they 
cut off contributory backwaters or otherwise reduce the run- 
off of impounded waters of long standing. They increase pro- 
duction when they lend stability to hydrographic conditions, 
increase the relative fertilization (sewage) of the stream, and 
by slackened current afford time for breeding. 
In general terms, production in the backwaters exhibits 
relations to levels similar to those we have described for chan- 
nel waters so long as the backwaters retain an intimate con- 
nection with the channel, that is, generally during high water, 
and for longest periods when, as in Thompson’s Lake, the con- 
nection with the channel is most intimate. The diversification, 
as levels fall, of the several regions examined by us, renders 
generalizations impossible with respect to all of the backwaters, 
since one or another local factor sooner or later comes in to 
modify conditions. Moreover, some of the backwaters, as 
Phelps Lake, are cut off from the channel early and are not 
affected by changes in river levels, and, in general, the effect 
of the changes in channel levels, especially the minor ones, is 
reduced, equalized, or even obliterated, before it reaches the 
backwaters. 
The season at which the initial stages of the major flood of 
the year occurs, affects the subsequent production. Thus in 
1896 and 1897 floods begin early in the year. The result is 
the carrying away in the run-off of great quantities of organic 
matter in suspension (Table X.) before they have had time to 
decay and yield up in solution their nitrogenous and other con- 
stituents for the support of the plankton. Temperatures are 
low in these months, decay is not rapid, and the plankton is 
not produced in large quantities. The result is that the stream 
is locally impoverished by this early run-off of matters in sus- 
pension and to some extent in solution. In 1898, on the other 
hand, the flood does not reach overflow stages till late in Feb- 
