The Kohonen_Order function propagates neurons in a topological order. There are 2 propagation steps. The first step all input units are propagated, which means that the output of all neurons is calculated. The second step consists of the propagation of all hidden units. This propagation step calculates all hidden neuron's activation and output. Please note that the activation and output are normally not required for the Kohonen algorithm. The activation and output values are used for display and evaluation reasons internally. The Act_Euclid activation function for example, copies the Euclidian distance of the unit from the training pattern to the units activation.