I study the gradual origins of tonality using a large corpus of scores and harmonic analyses, comprising more than 1000 movements (120,000 chords) drawn from the period 1450–1870. My talk focuses on two processes — for simplicity, considering only ‘major’ modes (in modern terminology ‘Ionian’, ‘Lydian’, and ‘Mixolydian’, where tonal-harmonic routines are first evident). The earliest is the development of (‘zeroth order’) preferences for particular chords. During the early 16th century, the frequency of root-position I, V, and IV increases, while that of secondary triads (including iii and iii⁶) decreases, with the change in IV being particularly dramatic. (Interestingly, the prominence of IV is transitory, being supplanted by ii in the seventeenth century.) The decline of iii makes I⁶ the primary chord over bass scale-degree 3, remarkable insofar as root position triads generally predominate this repertoire. (This preponderance of I⁶ may indicate an intuitive and untheorized precursor to root functionality, though other explanations are also possible.) The second process is the development of ‘first-order’ preferences for particular progressions. These propagate backwards along the circle of fifths: one first finds V moving to I, then (in the early 17th century) ii becoming the primary nontonic approach to V, then (in the 18th century) vi–V being eliminated in favor of vi–IV and vi–ii. Thus the functionally tonal ‘harmonic cycle’ is assembled backwards from tonic to dominant to predominant to pre-predominant (vi).