stem, to see on what years it grew best.

_The Flower-Cluster Scars_. These are the round, somewhat concave, scars,
found terminating the stem where forking occurs, or seemingly in the
axils of branches, on account of one of the forking branches growing more
rapidly and stoutly than the other and thus taking the place of the main
stem, so that this is apparently continued without interruption. If the
pupils have not understood the cause of the flower-cluster scars, show
them their position in shoots where they are plainly on the summit of the
stem, and tell them to compare this with the arrangement of a large
bud. The flower-cluster terminates the axis in the bud, and this scar
terminates a branch. When the terminal bud is thus prevented from
continuing its growth, the nearest axillary buds are developed.[1] One
shoot usually gets the start, and becomes so much stronger that it throws
the other to one side. The tendency of the Horsechestnut to have its
growth carried on by the terminal buds is so strong that I almost feel
inclined to say that vigorous branches are never formed from axillary
buds, in old trees, except where the terminal bud has been prevented from
continuing the branch. This tendency gives to the tree its characteristic
size of trunk and branches, and lack of delicate spray. On looking closely
at the branches also, they will be seen to be quite irregular, wherever
there has been a flower-cluster swerving to one side or the other.

[Footnote 1: The first winter that I examined Horsechestnut buds I found,
in many cases, that the axillary shoots had from a quarter of an inch to
an inch of wood before the first set of rings. I could not imagine what
had formed this wood, and it remained a complete puzzle to me until the
following spring, when I found in the expanding shoots, that, wherever
a flower-cluster was present, there were one or two pairs of leaflets
already well developed in the axils, and that the next season's buds were