In continuation to the experimental work of Part 1, a fracture mechanics model was developed. With this model, it was intended that the adhesive failure at the particle-latex interfaces, the cohesive failures within the bodies of latex and the strength of the pigment-latex composite could be predicted. Both monodisperse and polydisperse geometries were modelled as continuum micro-structures and latex bridges were geometrically simplified to ensure easier calculations. The parameters input into the model were based on the experimental tests and by doing so, a closer and more accurate association was made between the experimental and theoretical phases of research. Network connectivity is a determinant of the global strength properties. In this research the binder bridge dimensions (the amount of binder between neighbouring particles) and the coordination numbers of the particles are found to be a strong indicator of network connectivity. In comparing the percentages of connections that fail at the interfaces against a product of these two variables, it is found that as the product increases, the percentage of connections failing at the interface decreases to a minimum point of inflexion after which it increases. This is essentially an inverse of the CPVC maximum inflexion point, which is also shown by the modelling, and which ties in well to the experimental work. Systems with very low interfacial strengths and thus, weak adhesion, tended to show little variation of strength as other parameters were altered. These results reflect very well those presented in the literature and the experimental phase of work. The product of the contact length and the coordination number is also shown to be linearly proportional to strength and therefore, these two physical variables are deemed highly important variables to optimise in respect of strength.