Newton's method applied to the elementary symmetric functions of polyn
omials generates a class of dynamical systems. These systems have inva
riant lines on which the Lyapunov exponents can be found analytically,
thus predicting the exact parameter values for which these structures
are chaotic attractors (in the sense of Milnor) and precisely when bi
furcations, such as blowout, destroy their stability. Often, blowout b
ifurcations lead to a behavior called on-off intermittency. We also pr
esent evidence that the bursting mechanism of on-off intermittency fre
quently occurs in neighborhoods focal points, a common feature of maps
that have singularities. Finally, because of an embedding property, t
his new class of examples can be extended to construct systems having
this bifurcation in higher dimensions. (C) 1998 Elsevier Science B.V.