Elisabeth M. deBoer of Leiden University has uploaded a 616-page book to academia.edu called The historical development of Japanese tone, with the subtitle "Part 1: From proto-Japanese to the modern dialects; Part 2: The introduction and adaptation of the Middle Chinese tones in Japan." (She seems to have posted it before, too – no-one tells me anything.) Here's the abstract:

The reconstruction of the historical development of the modern Japanese tone systems is one of the major issues in Japanese historical linguistics. The prevalent theory (Kindaichi 1951), which regards the Kyoto type tone system of central Japan as most archaic fails to explain the modern dialect data. In 1979 R.S. Ramsey proposed an alternative theory, which regards the peripheral Tokyo type dialects as archaic. Even though this theory offered a convincing explanation for many problems that remained unsolved in the prevalent theory, it failed to find acceptance.

Part I of Elisabeth M. de Boer’s study shows how data from a host of Japanese dialects, from the north-eastern tip of Japan to the Ryukyu archipelago in the south-west, offer additional proof for Ramsey’s theory. The final chapter deals with evidence from Japanese loanwords in Ainu.

Part II shows how – contrary to what has often been thought – Ramsey’s theory is not in contradiction with the philological data. The final chapter deals with the interpretation of Buddhist chant as a source of historical information on the Japanese tones.

This is a monster slab of learning aimed right at one of my greatest weaknesses when it comes to Japanese linguistics, so I don't have much to say about it. I will note that the Ramsey paper referenced is "The Old Kyoto dialect and the historical development of Japanese accent" (Harvard Journal of Asiatic Studies; 39, 1979, 157-175), available on JSTOR (see link above).

Ramsey's argument is that the Heian accent data (representing the "Old Kyoto dialect") has been misread as showing an early form of the Kyoto-style accent, when really it is closer to the Tokyo-style accents, and it is Kyoto that was the site of innovation (most importantly, the accent nucleus moving one syllable towards the start of the word). This also makes sense under a "center and periphery" model, since Tokyo-style accents are found both east and west of Kyoto — i.e. the map looks exactly like you would expect it to if you had an accent system shared Honshu-wide and then some folks in the middle started messing with it.