The sources of our knowledge are the various systematic treatises upon music which have come down to us from Greek antiquity, together with incidental references in other authors, chiefly poets and philosophers. Of the systematic or 'technical' writers the earliest and most important is Aristoxenus, a pupil of Aristotle. His treatise on Harmonics (harmonikÊ) has reached us in a fragmentary condition, but may be supplemented to some extent from later works of the same school. Among the incidental notices of music the most considerable are the passages in the Republic and the Politics already referred to. To these we have to add a few other references in Plato and Aristotle; a long fragment from the Platonic philosopher Heraclides Ponticus, containing some interesting quotations from earlier poets; a number of detached observations collected in the nineteenth section of the Aristotelian Problems; and one or two notices preserved in lexicographical works, such as the Onomasticon of Pollux.

In these groups of authorities the scholars above mentioned find the double use which they believe to have been made of the names Dorian, Phrygian, Lydian and the rest. In Aristoxenus they recognise that these names are applied to a series of keys (tonoi), which differed in pitch only. In Plato and Aristotle they find the same names applied to scales called harmoniai, and these scales, they maintain, differed primarily in the order of their intervals. I shall endeavour to show that there was no such double use: that in the earlier periods of Greek music the scales in use, whether called tonoi or harmoniai, differed primarily in pitch: that the statements of ancient authors about them, down to and including Aristoxenus, agree as closely as there is reason to expect: and that the passages on which the opposite view is based—all of them drawn from comparatively late writers—either do not relate to these ancient scales at all, or point to the emergence in post-classical times of some new forms or tendencies of musical art. I propose in any case to adhere as closely as possible to a chronological treatment of the evidence which is at our command, and I hope to make it probable that the difficulties of the question may be best dealt with on this method.