It is first shown that the observational light curve data of R Scuti,
a star of the RV Tau type, is not multiperiodic and that it cannot hav
e been generated by a linear stochastic (AR) process. By default, the
signal must be a manifestation of deterministic chaos. We use a novel
nonlinear time series analysis, the global flow reconstruction techniq
ue, to probe the properties of the irregular pulsation cycles. We show
in particular that the chaotic dynamics of this star's complicated li
ght curve is captured by a simple four-dimensional polynomial map or f
low (four first-order ordinary differential equations). Another import
ant feature is that the method allows us to quantify an irregular sign
al that has the potential benefit for extracting novel stellar constra
ints from an irregular light curve. Finally, from the low dimensionali
ty (four) of the flow we can infer a simple physical picture of the pu
lsations, and arguments are presented that the pulsations of R Sct are
the result of the nonlinear interaction of two vibrational normal mod
es of the star.