A note on inverse autocorrelations