To sum up the result of this quantitative investigation, we could point out two problems. The first problem is that as we can see in chapter 2, lexical items change not to a single reflex but often to a group of diverse reflexes. Language change often include stages of variants of different quantity appearing gradually and unifying into one final form. Such would be what we would like to describe in lexical diffusion. The second problem is that as we can see in examples in chapter 2, the orthography in many cases are hypercorrections imitating earlier forms, which therefore is often removed from actual pronunciation. This is especially common in chapter 3 where we can see that the orthography in the variants that have been classified as being intermediate and counter to the rule is so detached from the actual pronunciation that it is difficult to capture the main current of the change. But such phenomena should not be regarded as being in conifict with the theory of change by lexical diffusion. For, the theory of lexical diffusion was proposed to mend the contradiction in the Neogrammarian thesis that change occurs abruptly without exception. Data with such counterexamples are therefore more appropriate for testing the theory.