Rapid

(Received July 16,1993; accepted in revised form September 25,1993) In this short communication, the prediction of the permeability of carbon dioxide through different polymers using a neural network is studied. A neural network is a numeric-mathematical construction that can model complex non-linear relationships. Here it is used to correlate the IR spectrum of a polymer to ita permeability. The underlying assumption is that the chemical information hidden in the IR spec-trum is sufficient for the prediction. The best neural network investigated so far does indeed show pre-dictive capabilities. Key words: gas and vapor permeation; theory 1