DISCUSSION : FORESTS, RESERVOIRS, AND STREAM FLOW 477 
Table 18 makes a very good prima facie case. The periods con- mr. Chitten- 
sidered on the various streams range from 18 to 34 years, divided into den 
two equal portions, and the mean annual precipitation, number of 
floods, and average duration of floods, for each, are given. In every 
case the precipitation in the second period is less than in the first, 
the deficiency ranging from about 1 to about 16 per cent. In every 
case the number of floods is greater in the second period, the excess 
ranging from 3% on the Tennessee to 140% on the Muskingum. In 
every case but one the average duration of the floods is greater in the 
second period, the excess ranging from about 15% on the Cumberland 
to 184% on the Muskingum. Presumably deforestation had been in 
progress in every case, and the inference of cause and effect was 
irresistible. Post hoc ergo propter hoc. It was indeed an alluring 
bait, and Mr. Leighton and his associates jumped clear out of the 
water to grab it; but the writer is much mistaken if they found any- 
thing more than an ugly hook from which they would now be glad 
enough to shake themselves free. 
While admitting the correctness of Dr. MceGee’s and Mr. Leighton’s 
tables, a prudent investigator would still be very wary about admitting 
the conclusions which they have drawn from them. There are several 
things quite near the surface that do not look well. One of these is 
the brevity of the periods considered and the fact that they do not com- 
mence until deforestation is already far advanced. Ten-year periods 
may satisfy Mr. Leighton, but it will generally be considered that a 
period reaching back to the time when forests were much more 
prevalent would be better. Another disturbing feature is the fact that 
almost no change in forest areas has taken place on some of these 
water-sheds during the period considered. On the Muskingum, where 
the changes in flood conditions have been most marked, there has been 
practically no change in the quantity of woodland. On the Connecticut 
there has been something of an increase. On the Allegheny and 
Monongahela the changes amounted to practically nothing on the 
first stream and 11% of the total area on the second, or 5.5% from 
any point in the first period to the corresponding point in the second. 
Manifestly, no possible conclusions, either one way or the other, can 
be based on such insignificant changes as these. The writer has not 
the data for the changes in the forest areas on the other water-sheds, 
but assumes that they are greater than on those just considered. 
Another suspicious circumstance is the fact that on the Mus- 
kingum, where there were no changes to speak of in woodland areas, 
and where the mean rainfall for the two periods was practically the 
same, the increase in floods is by far the greatest; while on the 
Tennessee, where the difference in rainfall and the change in forest 
areas are much greater, the difference in flood conditions is least. 
Still another, and very important, consideration, which has involved 
