Post-Keynesian growth theory is normally seen as originating from Harrod's 1939 'Essay in Dynamic Theory'. Harrod, however, was trying to lay the foundations of a new approach to economic dynamics, and often complained of misinterpretation. In this paper, the grounds of Harrod's argument are examined and compared with the 'textbook' interpretation. The latter is shown to be extremely reductive, as it ignores both Harrod's interest in the trade cycle and his methodological criticism of the 'time-lag theories of the cycle', and it also underrates the interesting implications of his non-linear approach and the epistemic implications of the instability principle.