Plants are essential for life on Earth, serving as a major source of food supply and contributing to the planet's carbon balance. Mathematical modelling is an important mechanism for predicting and optimising plant growth in agriculture, and Thornley's mathematical model (Thornley, 1997) is one of the most widely used models describing carbon and nitrogen allocation in plants. However, a formal mathematical analysis of the model's behaviour has not been performed. Our analysis of the model provides new insights into how and why the model can be inappropriate. By varying the values of some model parameters, we identify non-physical and even quite chaotic behaviour. In response, we modify Thornley's model by including additional litter terms, resulting in the elimination of these non-physical behaviours.