Bubble formation on a sieve plate with multiple holes is a very complicated process. At low to intermediate gas flow rates, not all holes are actively bubbling all the time. Furthermore, the interval between bubbling events may be highly variable. These phenomena are types of bubble formation transitions, and they involve both active hole location and event interruption. In this investigation, nonlinear analysis was applied to the experimental lime series data obtained from light scattering, which was sensitive to active hole location. The correlation dimensions and the largest Lyapunov exponent were estimated from the delay coordinates. These findings, along with the result of the surrogate data testing suggest that active hole location transitions can be regarded as low dimensional, deterministic chaos. Effects of gas chamber volume below the sieve plate and elevated viscosity of the liquid phase upon both the transition frequency and the chaotic invariants were studied. it was found that either increased gas chamber volume or elevated liquid viscosity may lead to less frequent active hole location transitions.