Growth and addition in a herding model

A model of herding is introduced which is exceptionally simple, incorporating only two phenomena, growth and addition. At each time step either (i) with probability p the system grows through the introduction of a new agent or (ii) with probability q = 1 - p a free agent already in the system is added at random to a group of size k with rate A_k. Two versions of the model, A _k = k and A _k = 1, are solved and in both versions we find two different types of behaviour. When p > 1/2 all the moments of the distribution of group sizes are linear in time for large time and the group distribution is power-law. When p < 1/2 the system runs out of free agents in a finite time. Received 12 February 2002 Published online 9 July 2002