CVI. DISCUSSION. 
low level in 1872, and follows on, from that by simple 
addition and subtraction, of the excess or defect, above or 
below the mean annual rainfall. But suppose the years 
before 1872 at Cordeaux to have been similar to those at 
Sydney, the curve would have started at a far higher level 
and would have been proportionately raised all along. The 
fallacy it seems to me is, that the effect on the curve of 
the rainfall of any given year goes on for ever, whereas in 
reality such effect is transient. A year of heavy rainfall 
increases its own stream discharge, and perhaps helps the 
next year a little, but has no appreciable effect on the third 
or subsequent years. The year 1876 at Sydney had a rain- 
fall very nearly up to the mean of 76 years, while the 
residual mass curve was decidedly high, higher in fact 
than it had been for a quarter of a century, and yet 
a little peculiarity in the distribution of the rainfall during 
the months of that year caused the absolute stoppage of 
one system of water supply, the Lachlan, and the serious 
falling off of the other. The serious shortage of water in 
1902, which came under my personal observation during a 
visit to Sydney, occurs under a high part of the curve, and 
shortly following the abundant rainfalls of 1899 and 1900. 
These are my objections to the method. All my experience 
of flood discharge and run off, which [ have to some extent 
made a special study, leads me to reject this curve as a 
guide to estimate the probable discharge of springs and 
streams. The paragraph on the Melbourne Water Supply 
Ido not agree with, especially do I object to its closing 
sentence, but the question is too large to discuss here. 
Melbourne has had for the past half century, and has now 
an abundant supply of wholesome water. The suburbs are 
however rapidly growing, and a supplementary scheme is 
being rapidly pushed on which will give an enormous 
increase of supply all through the year without any storaee 
reservoir being needed. 
