514 JOURNAL OF SCIENCE. 
phenomena presented by the river terraces at the Lower Gorge of the 
Waimakariri, which I brought under your notice two years ago.* 
You may remember that just above the gorge a series of terraces, 
widely apart on each side of the river, suddenly contract to the gorge 
and then below the gorge they expand again, making the gorge look 
like the narrow part of an hour glass. At first sight it appears 
remarkable that the river having got so far on one side of the gorge 
did not go completely round it and leave the gorge altogether. The 
explanation is this. The more the river cut away to the right or left 
of the gorge the more its course was lengthened, because the hard 
rocks of the gorge prevented that part from moving. As the length 
of the curve increased the velocity of the river diminished, and at last 
a time came when it could cut no longer in that direction, and began 
to sway back again, going as far as it could in the opposite direction; 
but, so long.as it was gripped at the gorge, increasing length put a 
limit to its lateral excursions. 
We are now, at last, in a position to get some idea as to how 
terraces have been formed. It is clear that so long as the Jand 
remains stationary a river can make but small changes in its bed. 
The sides of the valley may be flattened down by denudation, and the 
source may be slowly moved backward causing a very slight erosion. 
Portions of the channel may be cut away here and filled up there to: 
be perhaps again removed during another lateral swing of the river; 
but no terraces of any importance can be formed nor can any vallies 
be filled up. We must remember that even if the land had previously 
been rising the riverbed would always be nearly at the point of non- 
erosion, so that the changes that would take place after the land had 
become stationary would be small. 
Let us now see what changes would take place while the land 
was being elevated. Suppose that a river has reduced its bed to the 
non-erosion curve. Now if the land is slowly elevated the river will 
commence to deepen its channel throughout its whole length; it will 
cut down to the underlying rock. But as it sways to one side it cuts 
out the rock on that side and leaves, after every flood, a layer of 
gravel on the other, that is on the old channel it is leaving, so that the 
river has a gradually sloping shingle bank on one side and a high 
rocky one on the other. When it sways back in the opposite direction, 
deepening its channel all the time, it cuts out of the sloping shingle 
bank a rock terrace capped with shingle. And if this alternate lateral 
movement takes place several times, we have as the result, a series of 
rock terraces each with a shingle cap. A river valley, therefore, 
which has only undergone elevation may exhibit rock terraces, but 
not alluvial terraces. : 
But if, starting again with the curve of non-erosion, subsidence 
of the land took place instexd of elevation. Then the level of the 
river's mouth will be gradu: lly raised, and in consequence the whole 
curve of the riverbed. The river will commence by depositing 
material instead of excavating; and as it sways right and left will fill 
up the valley to a broad shingle plain and no terrace of any kind will 
be formed. If now the land should be re-elevated, the river would 
begin to cut terraces, not out of the rock, but out of the shingle it had 
* “Trans. N,Z. Inst.” vol. XVI., p. 449. 
