A new method is proposed for detecting unstable periodic orbits and th
eir linear stability properties from chaotic experimental time series.
Illustrative examples are presented for both numerically and experime
ntally generated time series. The statistical significance of the resu
lts is assessed using surrogate data.