There was evidence of nonlinearity in Spring
1990, when the percentage of cracked fruit
was fairly stable from the first to third clusters,
then doubled for the fourth through sixth clusters
(Fig. 2A). However, regression of cracking
percentage vs. cluster count did not support
a quadratic relationship in that the second-
order term was not significantly different
from zero (P that “c” = 0 > 0.2). The increase
also appeared to be linear in 1992, when the
percentage of cracked fruit increased from 2%
in the first cluster to 38%, 41%, and 45% in
clusters 5 , 6, and 7, respectively (Fig. 2B).