problem.
One thing more and I have done. It may be pointed out that is determining the inter-relationship of languages, there are two ways of approaching the problem. One from the top beginning with those languages which present fuller forms of words and elements of grammatical expression, and then to trace the changes down through all the stages of corruption. This seems to me the natural procedure to be adopted. Secondly, the problem may be approached from the bottom, starting from the most advanced, i. e., the most corrupt dialect and then to work our way up to the original. This method is fraught with many difficulties and and sometimes leads us into by-lanes and blind alleys, as it has done in the case of Caldwell. It had led him nowhere and ended in the formulation of a solecism that the origin of the Dravidian languages should be
