Studying the unidimensional waves, at least 3 books, and a lot of address on the web used the following idea (at least i think that it happened):
In order to describe the rate of change of potential energy per width on the string, is assumed the tension and density to be constant as the vibration goes on, also that the amplitude are small, the medium is continuous, and possibility another things, until here everthing is ok, the problem is that to describe this they use a small amount of the wire saying it to have a mass $dm$,..., but when use the mechanical relations, like work, they use the tension on only one side, why not use the tension on both sides? like in the way "to get" the wave equation?
