In the second part of this article, we moved on to consider mathematical models in renal disease. Although the literature on
this subject is scarce, I found one such disease model proposed by Chaturvedi & Insana (1997) [CI97]. Their work was based on a series of
results from a subtotal nephrectomy rat experiment. There, they modelled the time evolution of a population of nephrons in terms of
the sclerosis index [Raij *et al* (1984)] and glomerular hypertrophy. However, I have argued that this model is unsatisfactory
in various ways and developed my own equivalent model using dynamical systems theory^{4}. One of the unsatisfactory features
of CI97 was that the rate of change of nephrons was independent of the disease process. After an initial insult, my model will *suck*
nephrons into the disease process in order to maintain the glomerular filtration rate. The rate of change of diseased nephrons is
then linked to the actual disease process as it should be and is not prescribed in some other arbitrary way. The development of any
mathematical disease model is an evolutionary process, however, and should be constantly scrutinized, tested and improved.

Mathematical modelling in medicine is at an embryonic stage. As I have tried to illustrate, it is my belief that the subject
has a long and rich life ahead of it as did fluid mechanics over a century ago. It will always remain extremely important to maintain
a proper dialogue between mathematians and medical people, alike, and for minds to be open. Mathematicians should take heart from the
fact that problems arising in medicine *are* very challenging and the clinicians should try to reap the rewards that mathematical
modelling can offer. It must be a happy partnership.